PRAISEFORTHESINGULARITYISNEAR

OneofCBSNews'sBestFallBooksof2005AmongSt.LouisPost-Dispatch'sBestNonfictionBooksof2005OneofAmazon.com'sBestScienceBooksof2005"AnyonecangraspMr.Kurzweil'smainidea:thatmankind'stechnologicalknowledgehasbeensnowballing,withdizzyingprospectsforthefuture.Thebasicsareclearlyexpressed.Butforthosemoreknowledgeableandinquisitive,theauthorargueshiscaseinfascinatingdetail....TheSingularityIsNearisstartlinginscopeandbravado."

—JANETMASLIN,TheNewYorkTimes"Filledwithimaginative,scientificallygroundedspeculation....TheSingularityIsNearisworthreadingjustforitswealthofinformation,alllucidlypresented....[It's]animportantbook.NoteverythingthatKurzweilpredictsmaycometopass,butalotofitwill,andevenifyoudon'tagreewitheverythinghesays,it'sallworthpayingattentionto."

—ThePhiladelphiaInquirer"[An]exhilaratingandterrifyinglydeeplookatwhereweareheadedasaspecies....Mr.Kurzweilisabrilliantscientistandfuturist,andhemakesacompellingand,indeed,averymovingcaseforhisviewofthefuture."

—TheNewYorkSun"Compelling."

—SanJoseMercuryNews"Kurzweillinksaprojectedascendanceofartificialintelligencetothefutureoftheevolutionaryprocessitself.Theresultisbothfrighteningandenlightening....TheSingularityIsNearisakindofencyclopedicmapofwhatBillGatesoncecalled'theroadahead.'"

—TheOregonian"Aclear-eyed,sharply-focusedvisionofthenot-so-distantfuture."

—TheBaltimoreSun"Thisbookoffersthreethingsthatwillmakeitaseminaldocument.1)Itbrokersanewidea,notwidelyknown,2)Theideaisaboutasbigasyoucanget:theSingularity-allthechangeinthelastmillionyearswillbesupercededbythechangeinthenextfiveminutes,and3)Itisanideathatdemandsinformedresponse.Thebook'sclaimsaresofootnoted,documented,graphed,argued,andplausibleinsmalldetail,thatitrequirestheequalinresponse.Yetitsclaimsaresooutrageousthatiftrue,itwouldmean...well...theendoftheworldasweknowit,andthebeginningofutopia.RayKurzweilhastakenallthestrandsoftheSingularitymemecirculatinginthelastdecadesandhasunitedthemintoasingletomewhichhehasnailedonourfrontdoor.Isuspectthiswillbeoneofthemostcitedbooksofthedecade.LikePaulEhrlich'supsetting1972bookPopulationBomb,fanorfoe,it'sthewaveattheepicenteryou

havetostartwith."—KEVINKELLY,founderofWired

"Really,reallyoutthere.Delightfullyso."

—Businessweek.com"Stunning,utopianvisionofthenearfuturewhenmachineintelligenceoutpacesthebiologicalbrainandwhatthingsmaylooklikewhenthathappens....Approachableandengaging."

—TheUnofficialMicrosoftBlog"Oneofthemostimportantthinkersofourtime,Kurzweilhasfolloweduphisearlierworks...withaworkofstartlingbreadthandaudaciousscope."

—newmediamusings.com"Anattractivepictureofaplausiblefuture."

—KirkusReviews"Kurzweilisatruescientist-alarge-mindedoneatthat....What'sarrestingisn'tthedegreetowhichKurzweil'sheadyandbracingvisionfailstoconvince—giventhescopeofhisprojections,that'sinevitable—butthedegreetowhichitseemsdownrightplausible."

—PublishersWeekly(starredreview)"[T]hroughoutthistourdeforceofboundlesstechnologicaloptimism,oneisimpressedbytheauthor'sadamantineintellectualintegrity....Ifyouareatallinterestedintheevolutionoftechnologyinthiscenturyanditsconsequencesforthehumanswhoarecreatingit,thisiscertainlyabookyoushouldread.”

—JOHNWALKER,inventorofAutodesk,inFourmilabChangeLog"RayKurzweilisthebestpersonIknowatpredictingthefutureofartificialintelligence.Hisintriguingnewbookenvisionsafutureinwhichinformationtechnologieshaveadvancedsofarandfastthattheyenablehumanitytotranscenditsbiologicallimitations-transformingourlivesinwayswecan'tyetimagine."

—BILLGATES"Ifyouhaveeverwonderedaboutthenatureandimpactofthenextprofounddiscontinuitiesthatwillfundamentallychangethewaywelive,work,andperceiveourworld,readthisbook.Kurzweil'sSingularityisatourdeforce,imaginingtheunimaginableandeloquentlyexploringthecomingdisruptiveeventsthatwillalterourfundamentalperspectivesassignificantlyasdidelectricityandthecomputer."—DEANKAMEN,physicistandinventorofthefirstwearableinsulinpump,theHomeChoiceportabledialysismachine,theIBOTMobilitySystem,andtheSegwayHumanTransporter;

recipientoftheNationalMedalofTechnology"OneofourleadingAIpractitioners,RayKurzweil,hasonceagaincreateda'mustread'bookforanyoneinterestedinthefutureofscience,thesocialimpactoftechnology,andindeedthefutureofourspecies.Histhought-provokingbookenvisagesafutureinwhichwetranscendourbiologicallimitations,whilemakingacompellingcasethatahumancivilizationwithsuperhumancapabilitiesiscloserathandthanmostpeoplerealize."

—RAJREDDY,foundingdirector,RoboticsInstitute,CarnegieMellonUniversity;recipientoftheTuringAwardfromtheAssociationforComputingMachinery

"Ray'soptimisticbookwellmeritsbothreadingandthoughtfulresponse.ForthoselikemyselfwhoseviewsdifferfromRay'sonthebalanceofpromiseandperil,TheSingularityIsNearisaclearcallforacontinuingdialoguetoaddressthegreaterconcernsarisingfromtheseacceleratingpossibilities."

—BILLJOY,cofounderandformerchiefscientist,SunMicrosystems

SUMMARY(FRONTFLAP)Attheonsetofthetwenty-firstcentury,humanitystandsonthevergeofthemosttransformingandthrillingperiodinitshistory.Itwillbeanerainwhichtheverynatureofwhatitmeanstobe human will be both enriched and challenged, as our species breaks the shackles of itgenetic legacy and achieves inconceivable heights of intelligence, material progress, andlongevity.Foroverthreedecades,thegreatinventorandfuturistRayKurzweilhasbeenoneofthemost respectedandprovocativeadvocatesof theroleof technology inour future. InhisclassThe Age of Spiritual Machine, he presented the daring argument that with the ever-accelerating rate of technological change, computers would rival the full range of humanintelligence at its best. Now, inThe Singularity Is Near, he examines the next step in thisinexorable evolutionary process: the unionof humanandmachine, inwhich the knowledgeandskillsembedded inourbrainswillbecombinedwith thevastlygreatercapacity,speed,andknowledge-sharingabilityofourowncreations. Themerging is theessenceof theSingularity,anera inwhichour intelligencewillbecome increasingly nonbiological and trillions of timesmorepowerful than it is today—thedawning of a new civilization that will enable us to transcend out biological limitations andamplifyourcreativity.Inthisnewworld,therewillbenocleardistinctionbetweenhumanandmachine,realrealityandvirtualreality.Wewillbeabletoassumedifferentbodiesandtakeona range of personae at will. In practical terms, human aging and illness will be reversed;pollutionwillbestopped;worldhungerandpovertywillbesolved.Nanotechnologywillmakeitpossibletocreatevirtuallyanyphysicalproductusinginexpensiveinformationprocessesandwillultimatelyturnevendeathintoasolubleproblem.Whilethesocialandphilosophicalramificationsofthesechangeswillbeprofound,andthe threats they pose considerable,The Singularity Is Near maintains a radically optimisticviewofthefuturecourseofhumandevelopment.Assuch,itoffersaviewofthecomingagethatisbothadramaticculminationofthecenturiesoftechnologicalingenuityandagenuinelyinspiringvisionofourultimatedestiny.

ABOUTTHEAUTHOR(BACKFLAP)RayKurzweilisoneoftheworld'sleadinginventors,thinkers,andfuturists,withatwenty-yeartrackrecordofaccuratepredictions.Called"therestlessgenius"byTheWallStreetJournaland"theultimatethinkingmachine"byForbesmagazine,Kurzweilwasselectedasoneofthetop entrepreneurs by Inc. magazine, which described him as the "rightful heir to ThomasEdison."PBSselectedhimasoneof"sixteenrevolutionarieswhomadeAmerica,"alongwithotherinventorsofthepasttwocenturies.AninducteeintotheNationalInventorsHallofFameandrecipientoftheNationalMedalofTechnology,theLemelson-MITPrize(theworld'slargestawardforinnovation),thirteenhonorarydoctorates,andawardsfromthreeU.S.presidents,heis the author of four previous books:Fantastic Voyage: Live Long Enough to Live Forever(coauthoredwithTerryGrossman,M.D.),TheAgeofSpiritualMachines,The10%SolutionforaHealthyLife,andTheAgeofIntelligentMachines.

TheSingularityIsNear

ALSOBYRAYKURZWEIL

TheAgeofIntelligentMachinesThe10%SolutionforaHealthyLifeTheAgeofSpiritualMachines:WhenComputersExceedsHumanIntelligenceFantasticVoyage:LiveLongEnoughToLiveForever(withTerryGrossman,M.D.)

RAYKURZWEILTheSingularityIsNear

WHENHUMANSTRANSCENDBIOLOGY

VIKING

V IK I NGPublishedbythePenguinGroupPenguinGroup(USA)Inc.,375HudsonStreet,NewYork,NewYork10014,U.S.A.lPenguinGroup(Canada),10AlcornAvenue,Toronto,Ontario,CanadaM4V3B2(adivisionofPearsonPenguinCanadaInc.)lPenguinBooksLtd,80Strand,LondonWC2R0RL,England l Penguin Ireland, 25 St. Stephen'sGreen,Dublin 2, Ireland (a division of PenguinBooksLtd) l PenguinBooksAustralia Ltd, 250 Camberwell Road, Camberwell, Victoria 3124, Australia (a division of Pearson Australia Group Pty Ltd) lPenguinBooks IndiaPvtLtd, 11CommunityCentre,PanchsheelPark,NewDelhi—110017, India l PenguinGroup (NZ),CnrAirborneandRosedaleRoads,Albany,Auckland1310,NewZealand(adivisionofPearsonNewZealandLtd) lPenguinBooks(SouthAfrica)(Pty)Ltd,24StrudeeAvenue,Rosebank,Johannesburg2196,SouthAfricaPenguinBooksLtd,RegisteredOffices:80Strand,LondonWC2R0RL,EnglandFirstPublishedin2005byVikingPenguin,amemberofPenguinGroup(USA)Inc.10987654321Copyright©RayKurzweil,2005AllrightsreservedPhotographonp.368byHeleneDeLillo,2005Gratefulacknowledgmentismadeforpermissiontoreprintexcerptsfromthefollowingcopyrightedworks:"PlasticFantasticLover"byMartyBalin,performedbyJeffersonAirplane.IceBagPublishingCorp."WhatIAm"byEdieArlisaBrickell,KennethNeilWithrow,JohnBradleyHouser,JohnWalterBush,BrandonAly.©1988byGeffenMusic,EdieBriskellSongs,WithrowPublishing,EnlightenedKittyMusic, StrangeMindProductions.All rights reserved. "Season of theWitch" byDonovanLeitch.©1996byDonovan(Music)Limited.Copyrightrenewed.Internationalcopyrightsecured.Usedbypermission.Allrightsreserved.WorldrightadministeredbyPeermusic(UK)Ltd."SailingtoByzantium"fromTheCollectedWorksofW.B.Yeats,Volume I:ThePoems, revised edited byRichard J. Finneran.Copyright© 1928 byTheMacmillanCompany; copyrightrenewed©1956byGeorgieYeats.ReprintedwithpermissionofScribner,animprintofSimon&SchusterAdultPublishingGroupandA.P.WattLtdonbehalfofMichaelB.Yeats.LIBRARYOFCONGRESSCATALOGING-IN-PUBLICATIONDATAKurzweil,Ray.Thesingularityisnear:whenhumanstrascendbiology/RayKurzweil.p.cm.Includesbibliographicalreferences(p.).ISBN0-670-03384-71.Brain—Evolution.2.Humanevolution.3.Genetics.4.Nanotechnology.5.Robotics.I.Title.QP376.K852005153.9—dc222004061231PrintedintheUnitedStatesofAmericalSetinMinionlDesignedbyAmyHillWithoutlimitingtherightsundercopyrightreservedabove,nopartofthispublicationmayberesproduced,storedinorintroducedintoaretrievalsystem,ortransmitted,inanyformorbyanymeans(electronic,mechanical,photocopying,recordingorotherwise),withoutthepriorwrittenpermissionofboththecopyrightownerandtheabovepublisherofthisbook.The scanning, uploading, and distribution of this book via the Internet or via any other means without the permission of thepublisherisillegalandpunishablebylaw.Pleasepurchaseonlyauthorizedelectroniceditionsanddonotparticipateinorencourageelectronicpiracyofcopyrightablematerials.Yoursupportoftheauthor'srightsisappreciated.

Tomymother,Hannah,whoprovidedmewiththecouragetoseektheideas

toconfrontanychallenge

ContentsAcknowledgments18Prologue21

ThePowerofIdeasCHAPTERONE

TheSixEpochs24TheIntuitiveLinearViewVersustheHistoricalExponentialView25TheSixEpochs28

Epoch One: Physics and Chemistry. Epoch Two: Biology and DNA. Epoch Three:Brains.EpochFour:Technology.EpochFive:TheMergerofHumanTechnologywithHumanIntelligence.EpochSix:TheUniverseWakesUp.

TheSingularityIsNear35CHAPTERTWO

ATheoryofTechnologyEvolution:TheLawofAccelerating

Returns44

TheNatureofOrder.TheLifeCycleofaParadigm.FractalDesigns.FarsightedEvolution.TheS-CurveofaTechnologyasExpressedinItsLifeCycle56

TheLifeCycleofaTechnology.FromGoatSkinstoDownloads.Moore'sLawandBeyond59

Moore'sLaw:Self-FulfillingProphecy?TheFifthParadigm.FractalDimensionsandtheBrain.DNASequencing,Memory,Communications,theInternet,andMiniaturization75

Information,Order,andEvolution:TheInsightsfromWolframandFredkin'sCellularAutomata.CanWeEvolveArtificialIntelligencefromSimpleRules?

TheSingularityasEconomicImperative94

GetEightyTrillionDollars—LimitedTimeOnly.Deflation...aBadThing?CHAPTERTHREE

AchievingtheComputationalCapacityoftheHumanBrain107TheSixthParadigmofComputingTechnology:Three-DimensionalMolecularComputingandEmergingComputationalTechnologies107

TheBridgeto3-DMolecularComputing.NanotubesAreStilltheBestBet.ComputingwithMolecules.Self-Assembly.EmulatingBiology.ComputingwithDNA.ComputingwithSpin.ComputingwithLight.QuantumComputing.

TheComputationalCapacityoftheHumanBrain113

AcceleratingtheAvailabilityofHuman-LevelPersonalComputing.HumanMemoryCapacity.

TheLimitsofComputation116

ReversibleComputing.HowSmartIsaRock?TheLimitsofNanocomputing.SettingaDatefortheSingularity.MemoryandComputationalEfficiency:ARockVersusa

HumanBrain.GoingBeyondtheUltimate:Pico-andFemtotechnologyandBendingtheSpeedofLight.GoingBackinTime.

CHAPTERFOUR

AchievingtheSoftwareofHumanIntelligence:HowtoReverseEngineertheHumanBrain127ReverseEngineeringtheBrain:AnOverviewoftheTask127

NewBrain-ImagingandModelingTools.TheSoftwareoftheBrain.AnalyticVersusNeuromorphicModelingoftheBrain.HowComplexIstheBrain?ModelingtheBrain.PeelingtheOnion.

IstheHumanBrainDifferentfromaComputer?130

TheBrain'sCircuitsAreVerySlow.ButIt'sMassivelyParallel.TheBrainCombinesAnalogandDigitalPhenomena.TheBrainRewiresItself.MostoftheDetailsintheBrainAreRandom.TheBrainUsesEmergentProperties.TheBrainIsImperfect.WeContradictOurselves.TheBrainUsesEvolution.ThePatternsAreImportant.TheBrainIsHolographic.TheBrainIsDeeplyConnected.TheBrainDoesHaveanArchitectureofRegions.TheDesignofaBrainRegionIsSimplerthantheDesignofaNeuron.TryingtoUnderstandOurOwnThinking:TheAcceleratingPaceofResearch.

PeeringintotheBrain135

NewToolsforScanningtheBrain.ImprovingResolution.ScanningUsingNanobots.BuildingModelsoftheBrain142

SubneuralModels:SynapsesandSpines.NeuronModels.ElectronicNeurons.BrainPlasticity.ModelingRegionsoftheBrain.AneuromorphicModel:TheCerebellum.AnotherExample:Watts'sModeloftheAuditoryRegions.TheVisualSystem.OtherWorksinProgress:AnArtificialHippocampusandanArtificialOlivocerebellarRegion.UnderstandingHigher-LevelFunctions:Imitation,Prediction,andEmotion.

InterfacingtheBrainandMachines163TheAcceleratingPaceofReverseEngineeringtheBrain164

TheScalabilityofHumanIntelligence.UploadingtheHuman

Brain166CHAPTERFIVE

GNR:ThreeOverlappingRevolutions170Genetics:TheIntersectionofInformationandBiology171

Life'sComputer.DesignerBabyBoomers.CanWeReallyLiveForever?RNAi(RNAInterference).CellTherapies.GeneChips.SomaticGeneTherapy.ReversingDegenerativeDisease.CombatingHeartDisease.OvercomingCancer.ReversingAging.DNAMutations.ToxicCells.MitochondrialMutations.IntracellularAggregates.ExtracellularAggregates.CellLossandAtrophy.HumanCloning:TheLeastInterestingApplicationofCloningTechnology.WhyIsCloningImportant?PreservingEndangeredSpeciesandRestoringExtinctOnes.TherapeuticCloning.HumanSomatic-CellEngineering.SolvingWorldHunger.HumanCloningRevisited.

Nanotechnology:TheIntersectionofInformationandthePhysicalWorld183

TheBiologicalAssembler.UpgradingtheCellNucleuswithaNanocomputerandNanobot.FatandStickyFingers.TheDebateHeatsUp.EarlyAdopters.PoweringtheSingularity.ApplicationsofNanotechnologytotheEnvironment.NanobotsintheBloodstream.

Robotics:StrongAI203

RunawayAI.TheAIWinter.AI'sToolkit.ExpertSystems.BayesianNets.MarkovModels.NeuralNets.GeneticAlgorithms(GAs).RecursiveSearch.DeepFritzDraws:AreHumansGettingSmarter,orAreComputersGettingStupider?TheSpecialized-HardwareAdvantage.DeepBlueVersusDeepFritz.SignificantSoftwareGains.AreHumanChessPlayersDoomed?CombiningMethods.ANarrowAISampler.MilitaryandIntelligence.SpaceExploration.Medicine.ScienceandMath.Business,Finance,andManufacturing.ManufacturingandRobotics.SpeechandLanguage.EntertainmentandSports.StrongAI.

CHAPTERSIX

TheImpact...227

APanoplyofImpacts.

...ontheHumanBody228

ANewWayofEating.RedesigningtheDigestiveSystem.Programmable.Blood.HaveaHeart,orNot.SoWhat'sLeft?RedesigningtheHumanBrain.WeAreBecomingCyborgs.HumanBodyVersion3.0.

...ontheHumanBrain234

The2010Scenario.The2030Scenario.BecomeSomeoneElse.ExperienceBeamers.ExpandYourMind.

...onHumanLongevity240

TheTransformationtoNonbiologicalExperience.TheLongevityofInformation....onWarfare:TheRemote,Robotic,Robust,Size-Reduced,Virtual-RealityParadigm246

SmartDust.Nanoweapons.SmartWeapons.VR....onLearning250...onWork251

IntellectualProperty.Decentralization....onPlay253...ontheIntelligentDestinyoftheCosmos:WhyWeAreProbablyAloneintheUniverse254

TheDrakeEquation.TheLimitsofComputationRevisited.BiggerorSmaller.ExpandingBeyondtheSolarSystem.TheSpeedofLightRevisited.Wormholes.ChangingtheSpeedofLight.TheFermiParadoxRevisited.TheAnthropicPrincipleRevisited.TheMultiverse.EvolvingUniverses.IntelligenceastheDestinyoftheUniverse.TheUltimateUtilityFunction.HawkingRadiation.WhyIntelligenceIsMorePowerfulthanPhysics.AUniverse-ScaleComputer.TheHolographicUniverse.

CHAPTERSEVEN

Ichbinein

Singularitarian272

StillHuman?TheVexingQuestionofConsciousness276WhoAmI?WhatAmI?280TheSingularityasTranscendence283CHAPTEREIGHT

TheDeeplyIntertwinedPromiseandPerilofGNR286IntertwinedBenefits...289...andDangers290APanoplyofExistentialRisks292

ThePrecautionaryPrinciple.TheSmallertheInteraction,theLargertheExplosivePotential.OurSimulationIsTurnedOff.CrashingtheParty.GNR:TheProperFocusofPromiseVersusPeril.TheInevitabilityofaTransformedFuture.TotalitarianRelinquishment.

PreparingtheDefenses296

StrongAI.ReturningtothePast?TheIdeaofRelinquishment298

BroadRelinquishment.Fine-GrainedRelinquishment.DealingwithAbuse.TheThreatfromFundamentalism.FundamentalistHumanism.

DevelopmentofDefensiveTechnologiesandtheImpactofRegulation301

Protectionfrom"Unfriendly"StrongAI.Decentralization.DistributedEnergy.CivilLibertiesinanAgeofAsymmetricWarfare.

AProgramforGNRDefense305CHAPTERNINE

ResponsetoCritics309APanoplyofCriticisms309TheCriticismfromIncredulity312TheCriticismfromMalthus312

ExponentialTrendsDon'tLastForever.AVirtuallyUnlimitedLimit.TheCriticismfromSoftware313

SoftwareStability.SoftwareResponsiveness.SoftwarePrice-Performance.SoftwareDevelopmentProductivity.SoftwareComplexity.AcceleratingAlgorithms.TheUltimateSourceofIntelligentAlgorithms.

TheCriticismfromAnalogProcessing318TheCriticismfromtheComplexityofNeuralProcessing318

BrainComplexity.AComputer'sInherentDualism.LevelsandLoops.TheCriticismfromMicrotubulesandQuantumComputing323TheCriticismfromtheChurch-TuringThesis325TheCriticismfromFailureRates327TheCriticismfrom"Lock-

In"327TheCriticismfromOntology:CanaComputerBeConscious?328

Kurzweil'sChineseRoom.TheCriticismfromtheRich-PoorDivide335TheCriticismfromtheLikelihoodofGovernmentRegulation336

TheUnbearableSlownessofSocialInstitutions.TheCriticismfromTheism338TheCriticismfromHolism341Epilogue344

HowSingular?HumanCentrality.ResourcesandContactInformation346Appendix:TheLawofAcceleratingReturnsRevisited348Notes352Index[omitted]434

I

Acknowledgements

'dliketoexpressmydeepappreciationtomymother,Hannah,andmyfather,Fredric,forsupportingallofmyearlyideasandinventionswithoutquestion,whichgavemethefreedomtoexperiment;tomysisterEnidforherinspiration;andtomywife,Sonya,andmykids,EthanandAmy,whogivemylifemeaning,love,andmotivation.I'dliketothankthemanytalentedanddevotedpeoplewhoassistedmewiththiscomplexproject:AtViking:myeditor,RickKot,whoprovidedleadership,enthusiasm,andinsightfulediting;Clare

Ferraro,whoprovidedstrongsupportaspublisher;TimothyMennel,whoprovidedexpertcopyediting;BruceGiffordsandJohnJusino,forcoordinatingthemanydetailsofbookproduction;AmyHill,fortheinteriortextdesign;HollyWatson,forhereffectivepublicitywork;Alessandra.Lusardi,whoablyassistedRickKot;PaulBuckley,forhisclearandelegantartdesign;andHerbThomby,whodesignedtheengagingcover.

LorettaBarrett,myliteraryagent,whoseenthusiasticandastuteguidancehelpedguidethisproject.TerryGrossman,M.D.,myhealthcollaboratorandcoauthorofFantasticVoyage:LiveLong

EnoughtoLiveForever,forhelpingmetodevelopmyideasonhealthandbiotechnologythrough10,000e-mailsbackandforth,andamultifacetedcollaboration.

MartineRothblatt,forherdedicationtoallofthetechnologiesdiscussedinthisbookandforourcollaborationindevelopingdiversetechnologiesintheseareas.

AaronKleiner,mylong-termbusinesspartner(since1973),forhisdevotionandcollaborationthroughmanyprojects,includingthisone.

AmaraAngelica,whosedevotedandinsightfuleffortsledourresearchteam.Amaraalsousedheroutstandingeditingskillstoassistmeinarticulatingthecomplexissuesinthisbook.KathrynMyronuk,whosededicatedresearcheffortsmadeamajorcontributiontotheresearchandthenotes.SarahBlackcontributeddiscerningresearchandeditorialskills.Myresearchteamprovidedverycapableassistance:AmaraAngelica,KathrynMyronuk,SarahBlack,DanielPentlarge,EmilyBrown,CeliaBlack-Brooks,NandaBarker-Hook,SarahBrangan,RobertBradbury,JohnTillinghast,ElizabethCollins,BruceDarner,JimRintoul,SueRintoul,LarryKlaes,andChrisWright.AdditionalassistancewasprovidedbyLizBerry,SarahBrangan,RosemaryDrinka,LindaKatz,LisaKirschner,InnaNirenberg,ChristopherSetzer,JoanWalsh,andBeverlyZibrak.

LaksmanFrank,whocreatedmanyoftheattractivediagramsandimagesfrommydescriptions,andformattedthegraphs.

CeliaBlack-Brooks,forprovidingherleadershipinprojectdevelopmentandcommunications.PhilCohenandTedCoyle,forimplementingmyideasfortheillustrationonpage322,andHelene

DeLillo,forthe"SingularityIsNear"photoatthebeginningofchapter7.NandaBarker-Hook,EmilyBrown,andSarahBrangan,whohelpedmanagetheextensivelogistics

oftheresearchandeditorialprocesses.KenLindeandMattBridges,whoprovidedcomputersystemssupporttokeepourintricatework

flowprogressingsmoothly.DeniseScutellaro,JoanWalsh,MariaEllis,andBobBeal,fordoingtheaccountingonthis

complicatedproject.TheKurzweilAI.netteam,whoprovidedsubstantialresearchsupportfortheproject:AaronKleiner,

AmaraAngelica,BobBeal,CeliaBlack-Brooks,DanielPentlarge,DeniseScutellaro,EmilyBrown,JoanWalsh,KenLinde,LaksmanFrank,MariaEllis,MattBridges,NandaBarker-Hook,SarahBlack,andSarahBrangan.

MarkBizzell,DeborahLieberman,KirstenClausen,andDeaEldorado,fortheirassistanceincommunicationofthisbook'smessage.

RobertA.FreitasJr.,forhisthoroughreviewofthenanotechnology-relatedmaterial.PaulLinsay,forhisthoroughreviewofthemathematicsinthisbook.Mypeerexpertreaderswhoprovidedtheinvaluableserviceofcarefullyreviewingthescientific

content:RobertA.FreitasJr.(nanotechnology,cosmology),RalphMerkle(nanotechnology),MartineRothblatt(biotechnology,technologyacceleration),TerryGrossman(health,medicine,biotechnology),TomasoPoggio(brainscienceandbrainreverse-engineering),JohnParmentola(physics,militarytechnology),DeanKamen(technologydevelopment),NeilGershenfeld(computationaltechnology,physics,quantummechanics),JoelGershenfeld(systemsengineering),HansMoravec(artificialintelligence,robotics),MaxMore(technologyacceleration,philosophy),Jean-JacquesE.Slotine(brainandcognitivescience),SherryTurkle(socialimpactoftechnology),SethShostak(SETI,cosmology,astronomy),DamienBroderick(technologyacceleration,theSingularity),andHarryGeorge(technologyentrepreneurship).

Mycapablein-housereaders:AmaraAngelica,SarahBlack,KathrynMyronuk,NandaBarker-Hook,EmilyBrown,CeliaBlack-Brooks,AaronKleiner,KenLinde,JohnChalupa,andPaulAlbrecht.

Mylayreaders,whoprovidedkeeninsights:myson,EthanKurzweil,andDavidDalrymple.BillGates,EricDrexler,andMarvinMinsky,whogavepermissiontoincludetheirdialoguesinthe

book,andfortheirideas,whichwereincorporatedintothedialogues.Themanyscientistsandthinkerswhoseideasandeffortsarecontributingtoourexponentially

expandinghumanknowledgebase.Theabove-namedindividualsprovidedmanyideasandcorrectionsthatIwasabletomakethanks

totheirefforts.Foranymistakesthatremain,Itakesoleresponsibility.

TheSingularityIsNear

A

PROLOGUE

ThePowerofIdeas

I do not think there is any thrill that can go through the human heart like that felt by theinventorasheseessomecreationofthebrainunfoldingtosuccess.

—NIKOLATESLA,1896,INVENTOROFALTERNATINGCURRENTt the age of five, I had the idea that I would become an inventor. I had the notion thatinventionscouldchangetheworld.Whenotherkidswerewonderingaloudwhattheywantedtobe,IalreadyhadtheconceitthatIknewwhatIwasgoingtobe.TherocketshiptothemoonthatIwasthenbuilding(almostadecadebeforePresidentKennedy'schallengetothenation)

didnotworkout.Butataround the timeI turnedeight,my inventionsbecamea littlemore realistic,suchasarobotictheaterwithmechanicallinkagesthatcouldmovesceneryandcharactersinandoutofview,andvirtualbaseballgames. Having fled theHolocaust,my parents, both artists, wanted amoreworldly, less provincial,religiousupbringingforme.1Myspiritualeducation,asaresult,tookplaceinaUnitarianchurch.Wewould spend six months studying one religion—going to its services, reading its books, havingdialogueswithitsleaders—andthenmoveontothenext.Thethemewas"manypathstothetruth."Inoticed,ofcourse,manyparallelsamongtheworld'sreligious traditions,buteventhe inconsistencieswere illuminating. It became clear to me that the basic truths were profound enough to transcendapparentcontradictions.Attheageofeight,IdiscoveredtheTomSwiftJr.seriesofbooks.Theplotsofallofthethirty-threebooks(onlynineofwhichhadbeenpublishedwhenIstartedtoreadthemin1956)werealwaysthesame:Tomwouldgethimselfintoaterriblepredicament,inwhichhisfateandthatofhisfriends,andoftentherestofthehumanrace,hunginthebalance.Tomwouldretreattohisbasementlabandthinkabouthowtosolvetheproblem.This,then,wasthedramatictensionineachbookintheseries:whatingeniousideawouldTomandhisfriendscomeupwithtosavetheday?2Themoralofthesetaleswassimple:therightideahadthepowertoovercomeaseeminglyoverwhelmingchallenge. Tothisday,Iremainconvincedofthisbasicphilosophy:nomatterwhatquandariesweface—business problems, health issues, relationship difficulties, as well as the great scientific, social, andculturalchallengesofourtime—thereisanideathatcanenableustoprevail.Furthermore,wecanfindthatidea.Andwhenwefindit,weneedtoimplementit.Mylifehasbeenshapedbythisimperative.Thepowerofanidea—thisisitselfanidea.AroundthesametimethatIwasreadingtheTomSwiftJr.series,Irecallmygrandfather,whohadalso fled Europe with my mother, coming back from his first return visit to Europe with two keymemories. One was the gracious treatment he received from the Austrians and Germans, the samepeoplewhohadforcedhimtofleein1938.TheotherwasarareopportunityhehadbeengiventotouchwithhisownhandssomeoriginalmanuscriptsofLeonardodaVinci.Bothrecollectionsinfluencedme,butthelatter isoneI'vereturnedtomanytimes.Hedescribedtheexperiencewithreverence,asifhehadtouchedtheworkofGodhimself.This,then,wasthereligionthatIwasraisedwith:venerationforhumancreativityandthepowerofideas.

In1960,attheageoftwelve,Idiscoveredthecomputerandbecamefascinatedwithitsabilitytomodel and re-create the world. I hung around the surplus electronics stores on Canal Street inManhattan(they'restillthere!)andgatheredpartstobuildmyowncomputationaldevices.Duringthe1960s,Iwasasabsorbedinthecontemporarymusical,cultural,andpoliticalmovementsasmypeers,but I became equally engaged in a much more obscure trend: namely, the remarkable sequence ofmachinesthatIBMprofferedduringthatdecade,fromtheirbig"7000"series(7070,7074,7090,7094)to their small 1620, effectively the first "minicomputer." The machines were introduced at yearlyintervals, and each onewas less expensive andmore powerful than the last, a phenomenon familiartoday.IgotaccesstoanIBM1620andbegantowriteprogramsforstatisticalanalysisandsubsequentlyformusiccomposition.Istillrecallthetimein1968whenIwasallowedintothesecure,cavernouschamberhousingwhatwas then themostpowerful computer inNewEngland, a top-of-the-line IBM360Model91,with aremarkable million bytes (one megabyte) of "core" memory, an impressive speed of one millioninstructions per second (oneMIPS), and a rental cost of only one thousand dollars per hour. I haddeveloped a computer program that matched high-school students to colleges, and I watched infascinationasthefront-panellightsdancedthroughadistinctivepatternasthemachineprocessedeachstudent'sapplication.3EventhoughIwasquitefamiliarwitheverylineofcode,itnonethelessseemedasifthecomputerweredeepinthoughtwhenthelightsdimmedforseveralsecondsatthedenouementofeachsuchcycle.Indeed,itcoulddoflawlesslyintensecondswhattookustenhourstodomanuallywithfarlessaccuracy.Asaninventorinthe1970s,Icametorealizethatmyinventionsneededtomakesenseintermsoftheenablingtechnologiesandmarketforcesthatwouldexistwhentheinventionswereintroduced,asthatworldwouldbeaverydifferentonefromtheoneinwhichtheywereconceived.Ibegantodevelopmodels of how distinct technologies—electronics, communications, computer processors, memory,magnetic storage, and others—developed and how these changes rippled through markets andultimatelyoursocialinstitutions.IrealizedthatmostinventionsfailnotbecausetheR&Ddepartmentcan't get them towork but because the timing iswrong. Inventing is a lot like surfing: you have toanticipateandcatchthewaveatjusttherightmoment.Myinterestintechnologytrendsandtheirimplicationstookonalifeofitsowninthe1980s,andIbegantousemymodelstoprojectandanticipatefuturetechnologies,innovationsthatwouldappearin2000, 2010, 2020, and beyond. This enabled me to invent with the capabilities of the future byconceivinganddesigninginventionsusingthesefuturecapabilities.Inthemid-to-late1980s,Iwrotemyfirstbook,TheAgeofIntelligentMachines.4Itincludedextensive(andreasonablyaccurate)predictionsforthe1990sand2000s,andendedwiththespecterofmachineintelligencebecomingindistinguishablefrom that of its human progenitorswithin the first half of the twenty-first century. It seemed like apoignantconclusion,andinanyeventIpersonallyfounditdifficulttolookbeyondsotransforminganoutcome.Overthelasttwentyyears,Ihavecometoappreciateanimportantmeta-idea:thatthepowerofideastotransformtheworldisitselfaccelerating.Althoughpeoplereadilyagreewiththisobservationwhenitissimplystated,relativelyfewobserverstrulyappreciateitsprofoundimplications.Withinthenext severaldecades,wewillhave theopportunity toapply ideas toconquerage-oldproblems—andintroduceafewnewproblemsalongtheway.Duringthe1990s,Igatheredempiricaldataontheapparentaccelerationofallinformation-relatedtechnologiesandsoughttorefinethemathematicalmodelsunderlyingtheseobservations.IdevelopedatheoryIcallthelawofacceleratingreturns,whichexplainswhytechnologyandevolutionaryprocessesingeneralprogressinanexponentialfashion.5InTheAgeofSpiritualMachines(ASM),whichIwrotein1998,Isoughttoarticulatethenatureofhumanlifeasitwouldexistpastthepointwhenmachineandhumancognitionblurred.Indeed,I'veseenthisepochasanincreasinglyintimatecollaborationbetweenourbiologicalheritageandafuturethattranscendsbiology. Since thepublicationofASM, Ihavebegun to reflecton the futureofourcivilizationand itsrelationshiptoourplaceintheuniverse.Althoughitmayseemdifficulttoenvisionthecapabilitiesofa

futurecivilizationwhoseintelligencevastlyoutstripsourown,ourabilitytocreatemodelsofrealityinourmindenablesustoarticulatemeaningfulinsightsintotheimplicationsofthisimpendingmergerofourbiologicalthinkingwiththenonbiologicalintelligencewearecreating.This,then,isthestoryIwishtotellinthisbook.Thestoryispredicatedontheideathatwehavetheabilitytounderstandourownintelligence—toaccessourownsourcecode,ifyouwill—andthenreviseandexpandit.Someobserversquestionwhetherwearecapableofapplyingourownthinkingtounderstandourownthinking.AIresearcherDouglasHofstadtermusesthat"itcouldbesimplyanaccidentoffatethatourbrainsaretooweaktounderstandthemselves.Thinkofthelowlygiraffe,forinstance,whosebrainis obviously far below the level required for self-understanding—yet it is remarkably similar to ourbrain.6However,wehavealreadysucceededinmodelingportionsofourbrain-neuronsandsubstantialneural regions and the complexity of such models is growing rapidly. Our progress in reverseengineeringthehumanbrain,akeyissuethatIwilldescribeindetailinthisbook,demonstratesthatwedo indeed have the ability to understand, tomodel, and to extend our own intelligence. This is oneaspectoftheuniquenessofourspecies:ourintelligenceisjustsufficientlyabovethecriticalthresholdnecessary for us to scale our own ability to unrestricted heights of creative power andwe have theopposableappendage(ourthumbs)necessarytomanipulatetheuniversetoourwill. Awordonmagic:whenIwasreadingtheTomSwiftJr.books,Iwasalsoanavidmagician.Ienjoyedthedelightofmyaudiencesinexperiencingapparentlyimpossibletransformationsofreality.Inmy teen years, I replaced my parlor magic with technology projects. I discovered that unlike meretricks, technology does not lose its transcendent power when its secrets are revealed. I am oftenreminded of Arthur C. Clarke's third law, that "any sufficiently advanced technology isindistinguishablefrommagic."

Consider J. K. Rowling's Harry Potter stories from this perspective. These tales may beimaginary,buttheyarenotunreasonablevisionsofourworldasitwillexistonlyafewdecadesfromnow.EssentiallyallofthePotter"magic"willberealizedthroughthetechnologiesIwillexploreinthisbook.Playingquidditchandtransformingpeopleandobjects intootherformswillbefeasible infull-immersion virtual-reality environments, as well as in real reality, using nanoscale devices. Moredubious is the time reversal (as described inHarry Potter and the Prisoner of Azkaban), althoughseriousproposalshaveevenbeenputforwardforaccomplishingsomethingalongtheselines(withoutgiving rise to causality paradoxes), at least for bits of information, which essentially is what wecomprise.(Seethediscussioninchapter3ontheultimatelimitsofcomputation.)

Consider that Harry unleashes his magic by uttering the right incantation. Of course,discoveringandapplyingtheseincantationsarenosimplematters.Harryandhiscolleaguesneedtogetthesequence,procedures,andemphasisexactlycorrect.Thatprocessispreciselyourexperiencewithtechnology.Ourincantationsaretheformulasandalgorithmsunderlyingourmodern-daymagic.Withjust the right sequence, we can get a computer to read a book out loud, understand human speech,anticipate (and prevent) a heart attack, or predict the movement of a stock-market holding. If anincantationisjustslightlyoffmark,themagicisgreatlyweakenedordoesnotworkatall. Onemightobject to thismetaphorbypointingout thatHogwartian incantations arebrief andthereforedonotcontainmuchinformationcomparedto,say,thecodeforamodernsoftwareprogram.But the essentialmethods ofmodern technology generally share the same brevity. The principles ofoperation of software advances such as speech recognition can be written in just a few pages offormulas.Oftenakeyadvanceisamatterofapplyingasmallchangetoasingleformula.Thesameobservationholdsforthe"inventions"ofbiologicalevolution:considerthatthegeneticdifference between chimpanzees and humans, for example, is only a few hundred thousand bytes ofinformation.Althoughchimpsarecapableofsomeintellectual feats, that tinydifference inourgeneswassufficientforourspeciestocreatethemagicoftechnology.MurielRukeysersaysthat"theuniverseismadeofstories,notofatoms."Inchapter7,Idescribemyself as a "patternist," someonewho views patterns of information as the fundamental reality. Forexample,theparticlescomposingmybrainandbodychangewithinweeks,butthereisacontinuitytothepatternsthattheseparticlesmake.Astorycanberegardedasameaningfulpatternofinformation,

sowecaninterpretMurielRukeyser'saphorismfromthisperspective.Thisbook,then,isthestoryofthedestinyofthehuman-machinecivilization,adestinywehavecometorefertoastheSingularity.

I

CHAPTERONE

TheSixEpochs

Everyonetakesthelimitsofhisownvisionforthelimitsoftheworld.

—ARTHURSCHOPENHAUER

am not sure when I first became aware of the Singularity. I'd have to say it was a progressiveawakening. In the almost half century that I've immersed myself in computer and relatedtechnologies, I've sought tounderstand themeaningandpurposeof thecontinualupheaval that Ihavewitnessedatmanylevels.Gradually, I'vebecomeawareofa transformingevent looming in

thefirsthalfofthetwenty-firstcentury.Justasablackholeinspacedramaticallyaltersthepatternsofmatter and energy accelerating toward its event horizon, this impending Singularity in our future isincreasinglytransformingeveryinstitutionandaspectofhumanlife,fromsexualitytospirituality.What,then,istheSingularity?It'safutureperiodduringwhichthepaceoftechnologicalchangewillbesorapid,itsimpactsodeep,thathumanlifewillbeirreversiblytransformed.Althoughneitherutopiannordystopian, thisepochwill transform theconcepts thatwe relyon togivemeaning toourlives, fromourbusinessmodels to the cycleofhuman life, includingdeath itself.Understanding theSingularitywillalterourperspectiveonthesignificanceofourpastandtheramificationsforourfuture.Totrulyunderstandit inherentlychangesone'sviewof life ingeneralandone'sownparticular life. IregardsomeonewhounderstandstheSingularityandwhohasreflectedonitsimplicationsforhisorherownlifeasa"singularitarian."1IcanunderstandwhymanyobserversdonotreadilyembracetheobviousimplicationsofwhatIhave called the law of accelerating returns (the inherent acceleration of the rate of evolution, withtechnologicalevolutionasacontinuationofbiologicalevolution),Afterall,ittookmefortyyearstobeabletoseewhatwasrightinfrontofme,andIstillcannotsaythatIamentirelycomfortablewithallofitsconsequences.ThekeyideaunderlyingtheimpendingSingularityisthatthepaceofchangeofourhuman-createdtechnologyisacceleratinganditspowersareexpandingatanexponentialpace.Exponentialgrowthisdeceptive.Itstartsoutalmostimperceptiblyandthenexplodeswithunexpectedfury—unexpected,thatis,ifonedoesnottakecaretofollowitstrajectory.(Seethe"Linearvs.ExponentialGrowth"graphonp.10.) Consider thisparable: a lakeownerwants to stayathome to tend to the lake's fishandmakecertainthatthelakeitselfwillnotbecomecoveredwithlilypads,whicharesaidtodoubletheirnumberevery few days. Month after month, he patiently waits, yet only tiny patches of lily pads can bediscerned,andtheydon'tseemtobeexpandinginanynoticeableway.Withthelilypadscoveringlessthan1percentofthelake,theownerfiguresthatit'ssafetotakeavacationandleaveswithhisfamily.Whenhereturnsa fewweeks later,he'sshocked todiscover that theentire lakehasbecomecoveredwith the pads, and his fish have perished. By doubling their number every few days, the last sevendoublingswere sufficient to extend the pads' coverage to the entire lake. (Sevendoublings extendedtheirreach128-fold.)Thisisthenatureofexponentialgrowth. ConsiderGaryKasparov,who scorned the pathetic state of computer chess in 1992.Yet the

relentless doubling of computer power every year enabled a computer to defeat him only five yearslater.2Thelistofwayscomputerscannowexceedhumancapabilitiesisrapidlygrowing.Moreover,theoncenarrowapplicationsofcomputerintelligencearegraduallybroadeninginonetypeofactivityafteranother. For example, computers are diagnosing electrocardiograms andmedical images, flying andlandingairplanes,controllingthetacticaldecisionsofautomatedweapons,makingcreditandfinancialdecisions,andbeinggivenresponsibilityformanyothertasksthatusedtorequirehumanintelligence.The performance of these systems is increasingly based on integrating multiple types of artificialintelligence(AI).ButaslongasthereisanAIshortcominginanysuchareaofendeavor,skepticswillpointtothatareaasaninherentbastionofpermanenthumansuperiorityoverthecapabilitiesofourowncreations. Thisbookwillargue,however,thatwithinseveraldecadesinformation-basedtechnologieswillencompassallhumanknowledgeandproficiency,ultimatelyincludingthepattern-recognitionpowers,problem-solvingskills,andemotionalandmoralintelligenceofthehumanbrainitself. Although impressive inmany respects, the brain suffers from severe limitations.We use itsmassive parallelism (one hundred trillion interneuronal connections operating simultaneously) toquicklyrecognizesubtlepatterns.Butourthinkingisextremelyslow:thebasicneuraltransactionsareseveral million times slower than contemporary electronic circuits. That makes our physiologicalbandwidthforprocessingnewinformationextremelylimitedcomparedtotheexponentialgrowthoftheoverallhumanknowledgebase.Ourversion1.0biologicalbodiesarelikewisefrailandsubjecttoamyriadoffailuremodes,nottomention the cumbersome maintenance rituals they require. While human intelligence is sometimescapableof soaring in its creativity and expressiveness,muchhuman thought is derivative, petty, andcircumscribed.TheSingularitywillallowustotranscendtheselimitationsofourbiologicalbodiesandbrains.Wewillgainpoweroverourfates.Ourmortalitywillbeinourownhands.Wewillbeabletoliveaslongaswewant (a subtlydifferent statement fromsayingwewill live forever).Wewill fullyunderstandhuman thinking and will vastly extend and expand its reach. By the end of this century, thenonbiological portion of our intelligence will be trillions of trillions of times more powerful thanunaidedhumanintelligence. Wearenowintheearlystagesofthistransition.Theaccelerationofparadigmshift(therateatwhichwechangefundamentaltechnicalapproaches)aswellastheexponentialgrowthofthecapacityof information technologyarebothbeginning to reach the "kneeof the curve,"which is the stage atwhich an exponential trend becomes noticeable. Shortly after this stage, the trend quickly becomesexplosive. Before the middle of this century, the growth rates of our technology—which will beindistinguishable from ourselves—will be so steep as to appear essentially vertical. From a strictlymathematicalperspective,thegrowthrateswillstillbefinitebutsoextremethatthechangestheybringabout will appear to rupture the fabric of human history. That, at least, will be the perspective ofunenhancedbiologicalhumanity. The Singularity will represent the culmination of themerger of our biological thinking andexistencewithourtechnology,resultinginaworldthatisstillhumanbutthattranscendsourbiologicalroots.Therewillbenodistinction,post-Singularity,betweenhumanandmachineorbetweenphysicalandvirtual reality. Ifyouwonderwhatwill remainunequivocallyhuman insuchaworld, it's simplythis quality: ours is the species that inherently seeks to extend its physical andmental reach beyondcurrentlimitations.Manycommentatorsonthesechangesfocusonwhattheyperceiveasalossofsomevitalaspectofour humanity that will result from this transition. This perspective stems, however, from amisunderstandingofwhatourtechnologywillbecome.Allthemachineswehavemettodatelacktheessential subtlety of human biological qualities. Although the Singularity has many faces, its mostimportant implication is this: our technology will match and then vastly exceed the refinement andsupplenessofwhatweregardasthebestofhumantraits.

TheIntuitiveLinearViewVersustheHistoricalExponentialView

When the first transhuman intelligence is created and launches itself into recursive self-improvement, a fundamental discontinuity is likely to occur, the likes ofwhich I can't evenbegintopredict.

—MICHAELANISSIMOVIn the 1950s John vonNeumann, the legendary information theorist,was quoted as saying that "theever-accelerating progress of technology ... gives the appearance of approaching some essentialsingularity in the history of the race beyond which human affairs, as we know them, could notcontinue."3VonNeumannmakestwoimportantobservationshere:accelerationandsingularity.Thefirstideaisthathumanprogressisexponential(thatis,itexpandsbyrepeatedlymultiplyingbyaconstant)ratherthanlinear(thatis,expandingbyrepeatedlyaddingaconstant).

Thesecondisthatexponentialgrowthisseductive,startingoutslowlyandvirtuallyunnoticeably,butbeyondthekneeofthecurveitturnsexplosiveandprofoundlytransformative.Thefutureiswidelymisunderstood.Our forebearsexpected it tobeprettymuch like theirpresent,whichhadbeenprettymuch like theirpast.Exponential trendsdidexistone thousandyearsago,but theywereat thatveryearly stage inwhich theywere so flat and so slow that they looked like no trend at all.As a result,observers' expectation of an unchanged future was fulfilled. Today, we anticipate continuoustechnological progress and the social repercussions that follow. But the future will be far moresurprising thanmostpeoplerealize,becausefewobservershave truly internalized the implicationsofthefactthattherateofchangeitselfisaccelerating. Most long-range forecasts ofwhat is technically feasible in future time periods dramaticallyunderestimate thepowerof futuredevelopments because they arebasedonwhat I call the "intuitivelinear" view of history rather than the "historical exponential" view. My models show that we aredoublingtheparadigm-shiftrateeverydecade,asIwilldiscussinthenextchapter.Thusthetwentiethcentury was gradually speeding up to today's rate of progress; its achievements, therefore, wereequivalent toabout twentyyearsofprogressat the rate in2000.We'llmakeanother twentyyearsof

progressinjustfourteenyears(by2014),andthendothesameagaininonlysevenyears.Toexpressthisanotherway,wewon'texperienceonehundredyearsof technologicaladvancein thetwenty-firstcentury;wewillwitnessontheorderoftwentythousandyearsofprogress(again,whenmeasuredbytoday'srateofprogress),oraboutonethousandtimesgreaterthanwhatwasachievedinthetwentiethcentury.4Misperceptionsabouttheshapeofthefuturecomeupfrequentlyandinavarietyofcontexts.Asoneexampleofmany, ina recentdebate inwhich I tookpartconcerning the feasibilityofmolecularmanufacturing, a Nobel Prizewinning panelist dismissed safety concerns regarding nanotechnology,proclaiming that "we'renot going to see self-replicatingnanoengineered entities [devices constructedmolecular fragment by fragment] for a hundred years." I pointed out that one hundred years was areasonable estimate and actually matched my own appraisal of the amount of technical progressrequiredtoachievethisparticularmilestonewhenmeasuredattoday'srateofprogress(fivetimestheaveragerateofchangewesawinthetwentiethcentury).Butbecausewe'redoublingtherateofprogresseverydecade,we'll see theequivalentof a centuryofprogress—at today's rate—inonly twenty-fivecalendaryears. Similarly atTimemagazine's Future ofLife conference, held in 2003 to celebrate the fiftiethanniversaryofthediscoveryofthestructureofDNA,alloftheinvitedspeakerswereaskedwhattheythoughtthenextfiftyyearswouldbelike.5Virtuallyeverypresenterlookedattheprogressofthelastfiftyyearsanduseditasamodelforthenextfiftyyears.Forexample,JamesWatson,thecodiscovererofDNA,saidthatinfiftyyearswewillhavedrugsthatwillallowustoeatasmuchaswewantwithoutgainingweight. Ireplied,"Fiftyyears?"Wehaveaccomplishedthisalreadyinmicebyblockingthefat insulinreceptor gene that controls the storage of fat in the fat cells. Drugs for human use (using RNAinterferenceandothertechniqueswewilldiscussinchapter5)areindevelopmentnowandwillbeinFDAtestsinseveralyears.Thesewillbeavailableinfivetotenyears,notfifty.Otherprojectionswereequallyshortsighted,reflectingcontemporaryresearchprioritiesrather thantheprofoundchangesthatthenexthalfcenturywillbring.Ofall the thinkersat thisconference, itwasprimarilyBillJoyandIwhotookaccountoftheexponentialnatureofthefuture,althoughJoyandIdisagreeontheimportofthesechanges,asIwilldiscussinchapter8.Peopleintuitivelyassumethatthecurrentrateofprogresswillcontinueforfutureperiods.Evenforthosewhohavebeenaround longenough toexperiencehowthepaceofchange increasesover time,unexaminedintuitionleavesonewiththeimpressionthatchangeoccursatthesameratethatwehaveexperienced most recently. From the mathematician's perspective, the reason for this is that anexponentialcurvelookslikeastraightlinewhenexaminedforonlyabriefduration.Asaresult,evensophisticated commentators, when considering the future, typically extrapolate the current pace ofchange over the next ten years or one hundred years to determine their expectations. This is why Idescribethiswayoflookingatthefutureasthe"intuitivelinear"view. But a serious assessment of the history of technology reveals that technological change isexponential. Exponential growth is a feature of any evolutionary process, of which technology is aprimaryexample.Youcanexaminethedataindifferentways,ondifferenttimescales,andforawidevarietyoftechnologies,rangingfromelectronictobiological,aswellasfortheirimplications,rangingfrom the amount of humanknowledge to the size of the economy.The acceleration of progress andgrowthappliestoeachofthem.Indeed,weoftenfindnotjustsimpleexponentialgrowth,but"double"exponentialgrowth,meaningthattherateofexponentialgrowth(thatis,theexponent)isitselfgrowingexponentially (for example, see the discussion on the price-performance of computing in the nextchapter).ManyscientistsandengineershavewhatIcall"scientist'spessimism."Often,theyaresoimmersedin the difficulties and intricate details of a contemporary challenge that they fail to appreciate theultimatelong-termimplicationsoftheirownwork,andthelargerfieldofworkinwhichtheyoperate.They likewise fail toaccount for the farmorepowerful tools theywillhaveavailablewitheachnewgenerationoftechnology.

Scientistsaretrainedtobeskeptical,tospeakcautiouslyofcurrentresearchgoals,andtorarelyspeculate beyond the current generation of scientific pursuit. This may have been a satisfactoryapproachwhenagenerationof science and technology lasted longer thanahumangeneration,but itdoes not serve society's interests now that a generation of scientific and technological progresscomprisesonlyafewyears.Considerthebiochemistswho,in1990,wereskepticalofthegoaloftranscribingtheentirehumangenomeinamerefifteenyears.Thesescientistshadjustspentanentireyear transcribingamereoneten-thousandthofthegenome.So,evenwithreasonableanticipatedadvances,itseemednaturaltothemthatitwouldtakeacentury,ifnotlonger,beforetheentiregenomecouldbesequenced.Orconsidertheskepticismexpressedinthemid-1980sthattheInternetwouldeverbeasignificantphenomenon,given that it then includedonly tensof thousandsofnodes (alsoknownas servers). Infact, thenumberofnodeswasdoublingeveryyear,sothat therewerelikelytobetensofmillionsofnodes tenyears later.But this trendwasnot appreciatedby thosewhostruggledwith state-of-the-arttechnologyin1985,whichpermittedaddingonlyafewthousandnodesthroughouttheworldinasingleyear."6Theconverseconceptualerroroccurswhencertainexponentialphenomenaarefirstrecognizedandare applied in an overly aggressivemannerwithoutmodeling the appropriate pace of growth.Whileexponentialgrowthgainsspeedovertime,itisnotinstantaneous.Therun-upincapitalvalues(thatis,stockmarketprices)duringthe"Internetbubble"andrelatedtelecommunicationsbubble(1997–2000)wasgreatlyinexcessofanyreasonableexpectationofevenexponentialgrowth.AsIdemonstrateinthenextchapter,theactualadoptionoftheInternetande-commercedidshowsmoothexponentialgrowththrough both boom and bust; the overzealous expectation of growth affected only capital (stock)valuations.Wehaveseencomparablemistakesduringearlierparadigmshifts—forexample,duringtheearlyrailroadera(1830s),whentheequivalentoftheInternetboomandbustledtoafrenzyofrailroadexpansion.Anothererrorthatprognosticatorsmakeistoconsiderthetransformationsthatwillresultfromasingletrendintoday'sworldasifnothingelsewillchange.Agoodexampleistheconcernthatradicallifeextensionwillresult inoverpopulationandtheexhaustionof limitedmaterialresources tosustainhumanlife,whichignorescomparablyradicalwealthcreationfromnanotechnologyandstrongAI.Forexample,nanotechnology-basedmanufacturingdevicesinthe2020swillbecapableofcreatingalmostanyphysicalproductfrominexpensiverawmaterialsandinformation.Iemphasizetheexponential-versus-linearperspectivebecauseit'sthemostimportantfailurethatprognosticatorsmake in considering future trends.Most technology forecasts and forecasters ignorealtogetherthishistoricalexponentialviewoftechnologicalprogress.Indeed,almosteveryoneImeethasa linearviewof thefuture.That'swhypeople tendtooverestimatewhatcanbeachievedin theshortterm (becausewe tend to leaveoutnecessarydetails)butunderestimatewhat canbeachieved in thelongterm(becauseexponentialgrowthisignored).TheSixEpochs

Firstwebuildthetools,thentheybuildus.

—MARSHALLMCLUHANThefutureain'twhatitusedtobe.

—YOGIBERRA

Evolution isaprocessofcreatingpatternsof increasingorder. I'lldiscuss theconceptoforder in thenextchapter;theemphasisinthissectionisontheconceptofpatterns.Ibelievethatit'stheevolutionofpatternsthatconstitutestheultimatestoryofourworld.Evolutionworksthroughindirection:eachstage

or epoch uses the information-processing methods of the previous epoch to create the next. Iconceptualizethehistoryofevolution—bothbiologicalandtechnological—asoccurringinsixepochs.Aswewilldiscuss,theSingularitywillbeginwithEpochFiveandwillspreadfromEarthtotherestoftheuniverseinEpochSix.EpochOne:PhysicsandChemistry.Wecantraceouroriginstoastatethatrepresentsinformationinitsbasicstructures:patternsofmatterandenergy.Recenttheoriesofquantumgravityholdthattimeandspacearebrokendownintodiscretequanta,essentiallyfragmentsofinformation.Thereiscontroversyas to whether matter and energy are ultimately digital or analog in nature, but regardless of theresolutionofthisissue,wedoknowthatatomicstructuresstoreandrepresentdiscreteinformation. A fewhundred thousandyears after theBigBang, atomsbegan to form, as electronsbecametrapped in orbits around nuclei consisting of protons and neutrons. The electrical structure of atomsmade them "sticky."Chemistrywas born a fewmillion years later as atoms came together to createrelativelystablestructurescalledmolecules.Ofalltheelements,carbonprovedtobethemostversatile;it'sable to formbonds in fourdirections(versusone to threeformostotherelements),givingrise tocomplicated,information-rich,three-dimensionalstructures.Therulesofouruniverseandthebalanceofthephysicalconstantsthatgoverntheinteractionofbasicforcesaresoexquisitely,delicately,andexactlyappropriateforthecodificationandevolutionofinformation(resultinginincreasingcomplexity)thatonewondershowsuchanextraordinarilyunlikelysituationcameabout.Wheresomeseeadivinehand,othersseeourownhands—namely,theanthropicprinciple,whichholdsthatonlyinauniversethatallowedourownevolutionwouldwebeheretoasksuchquestions.7Recenttheoriesofphysicsconcerningmultipleuniversesspeculatethatnewuniversesare created on a regular basis, eachwith its own unique rules, but thatmost of these either die outquicklyorelsecontinuewithouttheevolutionofanyinterestingpatterns(suchasEarth-basedbiologyhascreated)becausetheirrulesdonotsupporttheevolutionofincreasinglycomplexforms.8It'shardtoimaginehowwecouldtestthesetheoriesofevolutionappliedtoearlycosmology,butit'sclearthatthephysicallawsofouruniversearepreciselywhattheyneedtobetoallowfortheevolutionofincreasinglevelsoforderandcomplexity.9

EpochTwo:BiologyandDNA.Inthesecondepoch,startingseveralbillionyearsago,carbon-basedcompounds became more and more intricate until complex aggregations of molecules formed self-replicatingmechanisms, and life originated. Ultimately, biological systems evolved a precise digitalmechanism(DNA)tostoreinformationdescribingalargersocietyofmolecules.Thismoleculeanditssupporting machinery of codons and ribosomes enabled a record to be kept of the evolutionaryexperimentsofthissecondepoch.EpochThree:Brains.Eachepochcontinuestheevolutionofinformationthroughaparadigmshifttoafurtherlevelof"indirection."(Thatis,evolutionusestheresultsofoneepochtocreatethenext.)Forexample,inthethirdepoch,DNA-guidedevolutionproducedorganismsthatcoulddetectinformationwiththeirownsensoryorgansandprocessandstorethatinformationintheirownbrainsandnervoussystems.Theseweremadepossiblebysecond-epochmechanisms(DNAandepigeneticinformationofproteinsandRNAfragmentsthatcontrolgeneexpression),which(indirectly)enabledanddefinedthird-epoch information-processingmechanisms (the brains and nervous systems of organisms). The thirdepochstartedwith theabilityofearlyanimals to recognizepatterns,whichstill accounts for thevastmajorityoftheactivityinourbrains.10Ultimately,ourownspeciesevolvedtheabilitytocreateabstractmentalmodelsoftheworldweexperienceandtocontemplatetherationalimplicationsofthesemodels.Wehavetheabilitytoredesigntheworldinourownmindsandtoputtheseideasintoaction.

Epoch Four: Technology. Combining the endowment of rational and abstract thought with ouropposablethumb,ourspeciesusheredinthefourthepochandthenextlevelofindirection:theevolutionofhuman-created technology.This startedoutwith simplemechanismsanddeveloped into elaborateautomata (automated mechanical machines). Ultimately, with sophisticated computational andcommunication devices, technology was itself capable of sensing, storing, and evaluating elaboratepatternsof information.Tocompare therateofprogressof thebiologicalevolutionof intelligence tothatoftechnologicalevolution,considerthatthemostadvancedmammalshaveaddedaboutonecubicinchofbrainmattereveryhundredthousandyears,whereasweareroughlydoublingthecomputationalcapacity of computers every year (see the next chapter). Of course, neither brain size nor computercapacityisthesoledeterminantofintelligence,buttheydorepresentenablingfactors.Ifweplacekeymilestonesofbothbiologicalevolutionandhumantechnologicaldevelopmentonasinglegraphplottingboththex-axis(numberofyearsago)andthey-axis(theparadigm-shifttime)onlogarithmicscales,wefindareasonablystraightline(continualacceleration),withbiologicalevolutionleadingdirectlytohuman-directeddevelopment.11

Theabovefiguresreflectmyviewofkeydevelopmentsinbiologicalandtechnologicalhistory.Note, however, that the straight line, demonstrating the continual acceleration of evolution, does notdependonmyparticularselectionofevents.Manyobserversandreferencebookshavecompiledlistsofimportantevents inbiologicaland technologicalevolution,eachofwhichhas itsown idiosyncrasies.Despite the diversity of approaches, however, if we combine lists from a variety of sources (forexample, the Encyclopaedia Britannica, the American Museum of Natural History, Carl Sagan's"cosmiccalendar,"andothers),weobservethesameobvioussmoothacceleration.Thefollowingplotcombines fifteen different lists of key events.12 Since different thinkers assign different dates to thesame event, and different lists include similar or overlapping events selected according to differentcriteria,weseeanexpected"thickening"ofthetrendlineduetothe"noisiness"(statisticalvariance)ofthisdata.Theoveralltrend,however,isveryclear.

Physicistandcomplexity theoristTheodoreModisanalyzed these listsanddetermined twenty-eightclustersofevents(whichhecalledcanonicalmilestones)bycombiningidentical,similar,and/orrelatedeventsfromthedifferentlists.13Thisprocessessentiallyremovesthe"noise"(forexample,thevariabilityofdatesbetweenlists)fromthelists,revealingagainthesameprogression:

Theattributesthataregrowingexponentiallyinthesechartsareorderandcomplexity,conceptswewill explore in thenext chapter.This accelerationmatchesour commonsenseobservations.Abillionyearsago,notmuchhappenedover thecourseofevenonemillionyears.Butaquarter-millionyearsagoepochal events suchas theevolutionofour speciesoccurred in time framesof justonehundredthousand years. In technology, if we go back fifty thousand years, notmuch happened over a one-thousand-year period.But in the recent past,we see new paradigms, such as theWorldWideWeb,progressfrominceptiontomassadoption(meaningthattheyareusedbyaquarterofthepopulationinadvancedcountries)withinonlyadecade.EpochFive:TheMergerofHumanTechnologywithHumanIntelligence.Lookingaheadseveraldecades, the Singularity will begin with the fifth epoch. It will result from the merger of the vastknowledgeembeddedinourownbrainswiththevastlygreatercapacity,speed,andknowledge-sharingabilityofourtechnology.Thefifthepochwillenableourhuman-machinecivilizationtotranscendthehumanbrain'slimitationsofamerehundredtrillionextremelyslowconnections.14 TheSingularitywillallowus toovercomeage-oldhumanproblemsandvastlyamplifyhumancreativity. We will preserve and enhance the intelligence that evolution has bestowed on us whileovercomingtheprofoundlimitationsofbiologicalevolution.ButtheSingularitywillalsoamplifytheabilitytoactonourdestructiveinclinations,soitsfullstoryhasnotyetbeenwritten.

EpochSix:TheUniverseWakesUp.Iwilldiscussthistopicinchapter6,undertheheading"...ontheIntelligentDestinyof theCosmos." In the aftermathof theSingularity, intelligence, derived from itsbiological origins in human brains and its technological origins in human ingenuity, will begin tosaturate thematter and energy in itsmidst. Itwill achieve this by reorganizingmatter and energy toprovideanoptimal levelof computation (basedon limitswewill discuss in chapter3) to spreadoutfromitsoriginonEarth. Wecurrentlyunderstandthespeedoflightasaboundingfactoronthetransferofinformation.Circumventingthislimithastoberegardedashighlyspeculative,buttherearehintsthatthisconstraintmay be able to be superseded.15 If there are even subtle deviations, wewill ultimately harness thissuperluminal ability.Whether our civilization infuses the rest of the universewith its creativity andintelligence quickly or slowly depends on its immutability. In any event the "dumb" matter andmechanismsof theuniversewillbe transformedintoexquisitelysublimeformsof intelligence,whichwillconstitutethesixthepochintheevolutionofpatternsofinformation.ThisistheultimatedestinyoftheSingularityandoftheuniverse.TheSingularityIsNear

Youknow,thingsaregoingtobereallydifferent!...No,no,Imeanreallydifferent!

—MARK MILLER (COMPUTER SCIENTIST) TO ERIC DREXLER,AROUND1986

What are the consequences of this event? When greater-than-human intelligence drivesprogress, thatprogresswillbemuchmorerapid.Infact, thereseemsnoreasonwhyprogressitselfwould not involve the creation of stillmore intelligent entities—on a still-shorter timescale.ThebestanalogythatIseeiswiththeevolutionarypast:Animalscanadapttoproblemsandmakeinventions,butoftennofasterthannaturalselectioncandoitswork—theworldactsasitsownsimulatorinthecaseofnaturalselection.Wehumanshavetheabilitytointernalizetheworldandconduct"whatif's"inourheads;wecansolvemanyproblemsthousandsoftimesfasterthannaturalselection.Now,bycreatingthemeanstoexecutethosesimulationsatmuchhigher speeds, we are entering a regime as radically different from our human past as wehumans are from the lower animals. From the human point of view, this change will be athrowingawayofallthepreviousrules,perhapsintheblinkofaneye,anexponentialrunawaybeyondanyhopeofcontrol.

—VERNORVINGE,"THETECHNOLOGICALSINGULARITY,"1993Letanultraintelligentmachinebedefinedasamachinethatcanfarsurpassalltheintellectualactivitiesofanymanhoweverclever.Sincethedesignofmachinesisoneoftheseintellectualactivities, an ultraintelligent machine could design even better machines; there would thenunquestionably be an "intelligence explosion," and the intelligence ofmanwould be left farbehind.Thusthefirstultraintelligentmachineisthelastinventionthatmanneedevermake.

—IRVING JOHNGOOD, "SPECULATIONSCONCERNINGTHEFIRSTULTRAINTELLIGENTMACHINE,"1965

Toput theconceptofSingularity into furtherperspective, let's explore thehistoryof theword itself."Singularity" isanEnglishwordmeaningauniqueeventwith,well, singular implications.Thewordwas adopted by mathematicians to denote a value that transcends any finite limitation, such as theexplosionofmagnitudethatresultswhendividingaconstantbyanumberthatgetscloserandclosertozero.Consider,forexample,thesimplefunctiony=l/x.Asthevalueofxapproacheszero,thevalueof

thefunction(y)explodestolargerandlargervalues.

Suchamathematicalfunctionneveractuallyachievesaninfinitevalue,sincedividingbyzeroismathematically "undefined" (impossible to calculate). But the value of y exceeds any possible finitelimit(approachesinfinity)asthedivisorxapproacheszero. The next field to adopt thewordwas astrophysics. If amassive star undergoes a supernovaexplosion,itsremnanteventuallycollapsestothepointofapparentlyzerovolumeandinfinitedensity,anda"singularity"iscreatedatitscenter.Becauselightwasthoughttobeunabletoescapethestarafterit reached this infinitedensity,16 itwas called ablackhole.17 It constitutes a rupture in the fabric ofspaceandtime. One theory speculates that the universe itself beganwith such a Singularity.18 Interestingly,however, theeventhorizon(surface)ofablackholeisofJfinitesize,andgravitationalforceisonlytheoretically infinite at the zero-size center of the black hole.At any location that could actually bemeasured,theforcesarefinite,althoughextremelylarge.ThefirstreferencetotheSingularityasaneventcapableofrupturingthefabricofhumanhistoryisJohn von Neumann's statement quoted above. In the 1960s, I. J. Good wrote of an "intelligenceexplosion" resulting from intelligent machines' designing their next generation without humanintervention. Vernor Vinge, a mathematician and computer scientist at San Diego State University,wroteaboutarapidlyapproaching"technologicalsingularity"inanarticleforOmnimagazinein1983andinascience-fictionnovel,MaroonedinRealtime,in1986.19 My1989book,TheAgeof IntelligentMachines, presented a future headed inevitably towardmachines greatly exceeding human intelligence in the first half of the twenty-first century.20 HansMoravec's 1988 bookMindChildren came to a similar conclusion by analyzing the progression ofrobotics.21 In 1993 Vinge presented a paper to a NASA-organized symposium that described theSingularity as an impending event resulting primarily from the advent of "entitieswith greater thanhumanintelligence,"whichVingesawas theharbingerofa runawayphenomenon.22My1999book,The Age of Spiritual Machines: When Computers Exceed Human Intelligence, described the

increasingly intimate connectionbetweenourbiological intelligenceand theartificial intelligencewearecreating.23HansMoravec's bookRobot:MereMachine to TranscendentMind, also published in1999,describedtherobotsofthe2040sasour"evolutionaryheirs,"machinesthatwill"growfromus,learnourskills,andshareourgoalsandvalues,...childrenofourminds."24AustralianscholarDamienBroderick's1997and2001books,bothtitledTheSpike,analyzedthepervasiveimpactoftheextremephaseoftechnologyaccelerationanticipatedwithinseveraldecades.25Inanextensiveseriesofwritings,John Smart has described the Singularity as the inevitable result of what he calls "MEST" (matter,energy,space,andtime)compression.26 Frommyperspective, theSingularityhasmanyfaces.Itrepresents thenearlyverticalphaseofexponentialgrowththatoccurswhentherateissoextremethattechnologyappearstobeexpandingatinfinitespeed.Ofcourse,fromamathematicalperspective,thereisnodiscontinuity,norupture,andthegrowth rates remain finite, althoughextraordinarily large.But fromour currently limited framework,thisimminenteventappearstobeanacuteandabruptbreakinthecontinuityofprogress.Iemphasizetheword"currently"becauseoneofthesalientimplicationsoftheSingularitywillbeachangeinthenatureofourabilitytounderstand.Wewillbecomevastlysmarteraswemergewithourtechnology. Can thepaceof technologicalprogresscontinue tospeedup indefinitely?Isn't thereapointatwhichhumansareunabletothinkfastenoughtokeepup?Forunenhancedhumans,clearlyso.Butwhatwould 1,000 scientists, each 1,000 times more intelligent than human scientists today, and eachoperating 1,000 times faster than contemporary humans (because the information processing in theirprimarily nonbiological brains is faster) accomplish? One chronological year would be like amillenniumforthem.27Whatwouldtheycomeupwith? Well, for one thing, they would come up with technology to become evenmore intelligent(because their intelligence is no longer of fixed capacity). They would change their own thoughtprocessestoenablethemtothinkevenfaster.Whenscientistsbecomeamilliontimesmoreintelligentandoperateamilliontimesfaster,anhourwouldresultinacenturyofprogress(intoday'sterms).TheSingularityinvolvesthefollowingprinciples,whichIwilldocument,develop,analyze,andcontemplatethroughouttherestofthisbook:

·Therateofparadigmshift(technicalinnovation)isaccelerating,rightnowdoublingeverydecade.28·The power (price-performance, speed, capacity, and bandwidth) of information technologies isgrowing exponentially at an even faster pace, nowdoubling about every year.29 This principleappliestoawiderangeofmeasures,includingtheamountofhumanknowledge.

·For information technologies, there is a second level of exponential growth: that is, exponentialgrowth in the rate of exponential growth (the exponent).The reason: as a technology becomesmore cost effective, more resources are deployed toward its advancement, so the rate ofexponential growth increases over time. For example, the computer industry in the 1940sconsistedofahandfulofnowhistoricallyimportantprojects.Todaytotalrevenueinthecomputerindustry ismore thanone trilliondollars, so researchanddevelopmentbudgets are comparablyhigher.

·Human brain scanning is one of these exponentially improving technologies. As I will show inchapter4,thetemporalandspatialresolutionandbandwidthofbrainscanningaredoublingeachyear. We are just now obtaining the tools sufficient to begin serious reverse engineering(decoding)ofthehumanbrain'sprinciplesofoperation.Wealreadyhaveimpressivemodelsandsimulationsofacoupledozenofthebrain'sseveralhundredregions.Withintwodecades,wewillhaveadetailedunderstandingofhowalltheregionsofthehumanbrainwork.

·Wewillhavetherequisitehardwaretoemulatehumanintelligencewithsupercomputersbytheendof thisdecadeandwithpersonal-computer-sizedevicesby theendof thefollowingdecade.Wewillhaveeffectivesoftwaremodelsofhumanintelligencebythemid-2020s.

·Withboth thehardwareandsoftwareneeded to fullyemulatehuman intelligence,wecanexpectcomputerstopasstheTuringtest,indicatingintelligenceindistinguishablefromthatofbiological

humans,bytheendofthe2020s.30·When they achieve this level of development, computerswill be able to combine the traditionalstrengthsofhumanintelligencewiththestrengthsofmachineintelligence.

·The traditionalstrengthsofhuman intelligence includea formidableability to recognizepatterns.Themassivelyparallelandself-organizingnatureofthehumanbrainisanidealarchitectureforrecognizing patterns that are based on subtle, invariant properties.Humans are also capable oflearningnewknowledgebyapplyinginsightsandinferringprinciplesfromexperience,includinginformationgathered through language.Akeycapabilityofhuman intelligence is the ability tocreatementalmodelsofrealityandtoconductmental"what-if"experimentsbyvaryingaspectsofthesemodels.

·The traditional strengthsofmachine intelligence include theability to rememberbillionsof factspreciselyandrecalltheminstantly.

·Anotheradvantageofnonbiologicalintelligenceisthatonceaskillismasteredbyamachine,itcanbeperformedrepeatedlyathighspeed,atoptimalaccuracy,andwithouttiring.

·Perhapsmostimportant,machinescansharetheirknowledgeatextremelyhighspeed,comparedtotheveryslowspeedofhumanknowledge-sharingthroughlanguage.

·Nonbiological intelligencewill be able to download skills and knowledge from othermachines,eventuallyalsofromhumans.

·Machineswillprocessandswitchsignalsatclosetothespeedoflight(aboutthreehundredmillionmeters per second), compared to about one hundredmeters per second for the electrochemicalsignalsusedinbiologicalmammalianbrains.31Thisspeedratioisatleastthreemilliontoone.

·MachineswillhaveaccessviatheInternettoalltheknowledgeofourhuman-machinecivilizationandwillbeabletomasterallofthisknowledge.

·Machines can pool their resources, intelligence, and memories. Two machines—or one millionmachines—canjointogethertobecomeoneandthenbecomeseparateagain.Multiplemachinescandobothatthesametime:becomeoneandseparatesimultaneously.Humanscallthisfallinginlove,butourbiologicalabilitytodothisisfleetingandunreliable.

·Thecombinationofthesetraditionalstrengths(thepattern-recognitionabilityofbiologicalhumanintelligence and the speed, memory capacity and accuracy, and knowledge and skill-sharingabilitiesofnonbiologicalintelligence)willbeformidable.

·Machineintelligencewillhavecompletefreedomofdesignandarchitecture(thatis,theywon'tbeconstrained by biological limitations, such as the slow switching speed of our interneuronalconnectionsorafixedskullsize)aswellasconsistentperformanceatalltimes.

·Oncenonbiological intelligencecombines the traditionalstrengthsofbothhumansandmachines,thenonbiologicalportionofourcivilization's intelligencewill thencontinue tobenefit fromthedoubleexponentialgrowthofmachineprice-performance,speed,andcapacity.

·Oncemachines achieve the ability to design and engineer technology as humans do, only at farhigher speeds and capacities, theywill have access to their own designs (source code) and theability to manipulate them. Humans are now accomplishing something similar throughbiotechnology(changingthegeneticandotherinformationprocessesunderlyingourbiology),butin a much slower and far more limited way than what machines will be able to achieve bymodifyingtheirownprograms.

·Biologyhas inherent limitations.Forexample,every livingorganismmustbebuilt fromproteinsthat are folded from one-dimensional strings of amino acids. Protein-based mechanisms arelackinginstrengthandspeed.Wewillbeabletoreengineeralloftheorgansandsystemsinourbiologicalbodiesandbrainstobevastlymorecapable.

·Aswewilldiscussinchapter4,humanintelligencedoeshaveacertainamountofplasticity(abilitytochangeitsstructure),moresothanhadpreviouslybeenunderstood.Butthearchitectureofthehuman brain is nonetheless profoundly limited. For example, there is room for only about onehundredtrillioninterneuronalconnectionsineachofourskulls.Akeygeneticchangethatallowedfor the greater cognitive ability of humans compared to that of our primate ancestors was the

developmentofalargercerebralcortexaswellasthedevelopmentofincreasedvolumeofgray-matter tissueincertainregionsof thebrain.32Thischangeoccurred,however,on theveryslowtimescale of biological evolution and still involves an inherent limit to the brain's capacity.Machineswillbeabletoreformulatetheirowndesignsandaugmenttheirowncapacitieswithoutlimit.Byusingnanotechnology-baseddesigns,theircapabilitieswillbefargreaterthanbiologicalbrainswithoutincreasedsizeorenergyconsumption.

·Machines will also benefit from using very fast three-dimensional molecular circuits. Today'selectroniccircuitsaremorethanonemilliontimesfasterthantheelectrochemicalswitchingusedinmammalianbrains.Tomorrow'smolecularcircuitswillbebasedondevicessuchasnanotubes,whicharetinycylindersofcarbonatomsthatmeasureabouttenatomsacrossandarefivehundredtimessmallerthantoday'ssilicon-basedtransistors.Sincethesignalshavelessdistancetotravel,theywillalsobeabletooperateatterahertz(trillionsofoperationspersecond)speedscomparedtothefewgigahertz(billionsofoperationspersecond)speedsofcurrentchips.

·Therateoftechnologicalchangewillnotbelimitedtohumanmentalspeeds.Machineintelligencewillimproveitsownabilitiesinafeedbackcyclethatunaidedhumanintelligencewillnotbeabletofollow.

·This cycle ofmachine intelligence's iteratively improving its owndesignwill become faster andfaster.Thisisinfactexactlywhatispredictedbytheformulaforcontinuedaccelerationoftherateofparadigmshift.Oneoftheobjectionsthathasbeenraisedtothecontinuationoftheaccelerationof paradigm shift is that it ultimately becomes much too fast for humans to follow, and sotherefore, it's argued, it cannot happen. However, the shift from biological to nonbiologicalintelligencewillenablethetrendtocontinue.

·Alongwiththeacceleratingimprovementcycleofnonbiologicalintelligence,nanotechnologywillenablethemanipulationofphysicalrealityatthemolecularlevel.

·Nanotechnology will enable the design of nanobots: robots designed at the molecular level,measured in microns (millionths of a meter), such as "respirocytes" (mechanical red-bloodcells).33 Nanobots will have myriad roles within the human body, including reversing humanaging (to the extent that this taskwill not alreadyhavebeencompleted throughbiotechnology,suchasgeneticengineering).

·Nanobots will interact with biological neurons to vastly extend human experience by creatingvirtualrealityfromwithinthenervoussystem.

·Billionsofnanobotsinthecapillariesofthebrainwillalsovastlyextendhumanintelligence.·Oncenonbiological intelligencegets a foothold in thehumanbrain (thishas already startedwithcomputerizedneuralimplants),themachineintelligenceinourbrainswillgrowexponentially(asit has been doing all along), at least doubling in power each year. In contrast, biologicalintelligenceiseffectivelyoffixedcapacity.Thus,thenonbiologicalportionofourintelligencewillultimatelypredominate.

·Nanobotswillalsoenhancetheenvironmentbyreversingpollutionfromearlierindustrialization.·Nanobots called foglets that can manipulate image and sound waves will bring the morphingqualitiesofvirtualrealitytotherealworld.34

·The human ability to understand and respond appropriately to emotion (so-called emotionalintelligence) isoneof the formsofhuman intelligence thatwillbeunderstoodandmasteredbyfuture machine intelligence. Some of our emotional responses are tuned to optimize ourintelligenceinthecontextofourlimitedandfrailbiologicalbodies.Futuremachineintelligencewillalsohave"bodies"(forexample,virtualbodiesinvirtualreality,orprojectionsinrealrealityusingfoglets)inordertointeractwiththeworld,butthesenanoengineeredbodieswillbefarmorecapableanddurable thanbiologicalhumanbodies.Thus, someof the "emotional" responsesoffuture machine intelligence will be redesigned to reflect their vastly enhanced physicalcapabilities.35

·Asvirtualrealityfromwithinthenervoussystembecomescompetitivewithrealrealityintermsofresolutionandbelievability,ourexperienceswillincreasinglytakeplaceinvirtualenvironments.

·Invirtualreality,wecanbeadifferentpersonbothphysicallyandemotionally.Infact,otherpeople(such as your romantic partner)will be able to select a different body for you than youmightselectforyourself(andviceversa).

·The law of accelerating returns will continue until nonbiological intelligence comes dose to"saturating" the matter and energy in our vicinity of the universe with our human-machineintelligence.Bysaturating,Imeanutilizingthematterandenergypatternsforcomputationtoanoptimaldegree,basedonourunderstandingofthephysicsofcomputation.Asweapproachthislimit, the intelligence of our civilizationwill continue its expansion in capability by spreadingoutward toward the rest of the universe. The speed of this expansionwill quickly achieve themaximumspeedatwhichinformationcantravel.

·Ultimately,theentireuniversewillbecomesaturatedwithourintelligence.Thisisthedestinyoftheuniverse.(Seechapter6.)Wewilldetermineourownfateratherthanhaveitdeterminedbythecurrent"dumb,"simple,machinelikeforcesthatrulecelestialmechanics.

·Thelengthoftimeitwilltaketheuniversetobecomeintelligenttothisextentdependsonwhetherornotthespeedoflightisanimmutablelimit.Thereareindicationsofpossiblesubtleexceptions(orcircumventions)tothislimit,which,iftheyexist,thevastintelligenceofourcivilizationatthisfuturetimewillbeabletoexploit.

This, then, is theSingularity.Somewouldsay thatwecannotcomprehend it,at leastwithourcurrent level of understanding. For that reason, we cannot look past its event horizon and makecompletesenseofwhatliesbeyond.ThisisonereasonwecallthistransformationtheSingularity. Ihavepersonallyfounditdifficult,althoughnotimpossible,tolookbeyondthiseventhorizon,evenafterhavingthoughtaboutitsimplicationsforseveraldecades.Still,myviewisthat,despiteourprofound limitations of thought, we do have sufficient powers of abstraction to make meaningfulstatements about the nature of life after the Singularity. Most important, the intelligence that willemerge will continue to represent the human civilization, which is already a human-machinecivilization.Inotherwords,futuremachineswillbehuman,eveniftheyarenotbiological.Thiswillbethenextstepinevolution,thenexthigh-levelparadigmshift,thenextlevelofindirection.Mostoftheintelligenceofourcivilizationwillultimatelybenonbiological.By theendof thiscentury, itwillbetrillions of trillions of times more powerful than human intelligence.36 However, to address often-expressedconcerns,thisdoesnotimplytheendofbiologicalintelligence,evenifitisthrownfromitsperchofevolutionarysuperiority.Eventhenonbiologicalformswillbederivedfrombiologicaldesign.Our civilization will remain human—indeed, in many ways it will be more exemplary of what weregardashumanthanitistoday,althoughourunderstandingofthetermwillmovebeyonditsbiologicalorigins. Manyobservershaveexpressedalarmat theemergenceof formsofnonbiological intelligencesuperiortohumanintelligence(anissuewewillexplorefurtherinchapter9).Thepotentialtoaugmentour own intelligence through intimate connectionwith other thinking substrates does not necessarilyalleviate the concern, as some people have expressed thewish to remain "unenhanced"while at thesame time keeping their place at the top of the intellectual food chain. From the perspective ofbiologicalhumanity,thesesuperhumanintelligenceswillappeartobeourdevotedservants,satisfyingourneedsanddesires.ButfulfillingthewishesofareveredbiologicallegacywilloccupyonlyatrivialportionoftheintellectualpowerthattheSingularitywillbring.MOLLYCIRCA2004:HowwillIknowwhentheSingularityisuponus?Imean,I'llwantsometimeto

prepare.RAY:Why,whatareyouplanningtodo?MOLLY2004:Let'ssee,forstarters,I'llwanttofine-tunemyrésumé.I'llwanttomakeagoodimpressiononthepowersthatbe.GEORGECIRCA2048:Oh,Icantakecareofthatforyou.MOLLY2004:That'sreallynotnecessary.I'mperfectlycapableofdoingitmyself.Imightalsowantto

eraseafewdocuments—youknow,whereI'malittleinsultingtoafewmachinesIknow.GEORGE2048:Oh,themachineswillfindthemanyway—butdon'tworry,we'reveryunderstanding.MOLLY2004: For some reason, that's not entirely reassuring. But I'd still like to know what the

harbingerswillbe.RAY:Okay,youwillknowtheSingularityiscomingwhenyouhaveamillione-mailsinyourin-box.MOLLY2004:Hmm, in that case, it sounds like we're just about there. But seriously, I'm having

troublekeepingupwithallofthisstuffflyingatmeasitis.HowamIgoingtokeepupwiththepaceoftheSingularity?

GEORGE2048:You'llhavevirtualassistants—actually,you'llneedjustone.MOLLY2004:WhichIsupposewillbeyou?GEORGE2048:Atyourservice.MOLLY 2004: That's just great. You'll take care of everything, you won't even have to keep me

informed."Oh,don'tbothertellingMollywhat'shappening,shewon'tunderstandanyway, let'sjustkeepherhappyandinthedark."

GEORGE2048:Oh,thatwon'tdo,notatall.MOLLY2004:Thehappypart,youmean?GEORGE2048:Iwasreferring tokeepingyou in thedark.You'llbeable tograspwhat I'mup to if

that'swhatyoureallywant.MOLLY2004:What,bybecoming...RAY:Enhanced?MOLLY2004:Yes,that'swhatIwastryingtosay.GEORGE2048:Well,ifourrelationshipistobeallthatitcanbe,thenit'snotabadidea.MOLLY2004:AndshouldIwishtoremainasIam?GEORGE2048: I'll be devoted to you in any event. But I can bemore than just your transcendent

servant.MOLLY2004:Actually,you'rebeing"just"mytranscendentservantdoesn'tsoundsobad.CHARLESDARWIN: If Imay interrupt, itoccurred tome thatoncemachine intelligence isgreater

thanhumanintelligence,itshouldbeinapositiontodesignitsownnextgeneration.MOLLY2004:Thatdoesn'tsoundsounusual.Machinesareusedtodesignmachinestoday.CHARLES:Yes,butin2004they'restillguidedbyhumandesigners.Oncemachinesareoperatingat

humanlevels,well,thenitkindofclosestheloop.NEDLUDD37:Andhumanswouldbeoutoftheloop.MOLLY2004:Itwouldstillbeaprettyslowprocess.RAY:Oh,notatall. Ifanon-biological intelligencewasconstructedsimilarly toahumanbrainbut

usedevencirca2004circuitry,it—MOLLYCIRCA2104:Youmean"she."RAY:Yes,ofcourse...she...wouldbeabletothinkatleastamilliontimesfaster.TIMOTHYLEARY:Sosubjectivetimewouldbeexpanded.RAY:Exactly.MOLLY2004:Soundslikealotofsubjectivetime.Whatareyoumachinesgoingtodowithsomuchof

it?GEORGE2048:Oh,there'splentytodo.Afterall,IhaveaccesstoallhumanknowledgeontheInternet.

MOLLY2004:Justthehumanknowledge?Whataboutallthemachineknowledge?GEORGE2048:Weliketothinkofitasonecivilization.CHARLES:So,itdoesappearthatmachineswillbeabletoimprovetheirowndesign.MOLLY2004:Oh,wehumansarestartingtodothatnow.RAY:Butwe'rejusttinkeringwithafewdetails.Inherently,DNA-basedintelligenceisjustsoveryslow

andlimited.CHARLES:Sothemachineswilldesigntheirownnextgenerationratherquickly.GEORGE2048:Indeed,in2048,thatiscertainlythecase.CHARLES:JustwhatIwasgettingat,anewlineofevolutionthen.

NED:Soundsmorelikeaprecariousrunawayphenomenon.CHARLES:Basically,that'swhatevolutionis.NED:Butwhatoftheinteractionofthemachineswiththeirprogenitors?Imean,Idon'tthinkI'dwant

toget intheirway.IwasabletohidefromtheEnglishauthorities forafewyearsintheearly1800s,butIsuspectthatwillbemoredifficultwiththese...

GEORGE2048:Guys.MOLLY2004:Hidingfromthoselittlerobots—RAY:Nanobots,youmean.MOLLY2004:Yes,hidingfromthenanobotswillbedifficult,forsure.RAY:Iwould expect the intelligence that arises from the Singularity to have great respect for their

biologicalheritage.GEORGE2048:Absolutely,it'smorethanrespect,it's...reverence.MOLLY2004:That'sgreat,George,I'llbeyourreveredpet.NotwhatIhadinmind.NED:That'sjusthowTedKaczynskiputsit:we'regoingtobecomepets.That'sourdestiny,tobecome

contentedpetsbutcertainlynotfreemen.MOLLY2004:Andwhatabout thisEpochSix? If I staybiological, I'llbeusingupall thisprecious

matterandenergy inamost inefficientway.You'llwant to turnme into, like,abillionvirtualMollysandGeorges,eachofthemthinkingalotfasterthanIdonow.Seemsliketherewillbealotofpressuretogoovertotheotherside.

RAY: Still, you represent only a tiny fraction of the available matter and energy. Keeping youbiologicalwon'tappreciablychangetheorderofmagnitudeofmatterandenergyavailabletotheSingularity.Itwillbewellworthittomaintainthebiologicalheritage.

GEORGE2048:Absolutely.RAY:Justliketodayweseektopreservetherainforestandthediversityofspecies.MOLLY2004:That'sjustwhatIwasafraidofImean,we'redoingsuchawonderfuljobwiththerain

forest.Ithinkwestillhavealittlebitofitleft.We'llenduplikethoseendangeredspecies.NED:Orextinctones.MOLLY2004:Andthere'snotjustme.HowaboutallthestuffIuse?Igothroughalotofstuff.GEORGE2048:That'snotaproblem,we'lljustrecycleallyourstuff.We'llcreatetheenvironmentsyou

needasyouneedthem.MOLLY2004:Oh,I'llbeinvirtualreality?RAY:No,actually,fogletreality.MOLLY2004:I'llbeinafog?RAY:No,no,foglets.MOLLY2004:Excuseme?RAY:I'llexplainlaterinthebook.MOLLY2004:Well,givemeahint.RAY:Fogletsarenanobots—robots the sizeofbloodcells—thatcanconnect themselves to replicate

anyphysicalstructure.Moreover,theycandirectvisualandauditoryinformationinsuchawayastobringthemorphingqualitiesofvirtualrealityintorealreality.38

MOLLY2004:I'msorryIasked.But,asIthinkaboutit,Iwantmorethanjustmystuff.Iwantalltheanimalsandplants,too.EvenifIdon'tgettoseeandtouchthemall,Iliketoknowthey'rethere.

GEORGE2048:Butnothingwillbelost.MOLLY2004: I know you keep saying that. But Imean actually there—you know, as in biological

reality.RAY:Actually, the entire biosphere is less than onemillionth of thematter and energy in the solar

system.CHARLES:Itincludesalotofthecarbon.RAY:It'sstillworthkeepingallofittomakesurewehaven'tlostanything.GEORGE2048:Thathasbeentheconsensusforatleastseveralyearsnow.MOLLY2004:So,basically,I'llhaveeverythingIneedatmyfingertips?

GEORGE2048:Indeed.MOLLY2004:SoundslikeKingMidas.Youknow,everythinghetouchedturnedtogold.NED:Yes,andasyouwillrecallhediedofstarvationasaresult.MOLLY2004:Well, if I do end up going over to the other side, with all of that vast expanse of

subjectivetime,IthinkI'lldieofboredom.GEORGE2048:Oh,thatcouldneverhappen.Iwillmakesureofit.

T

CHAPTERTWO

ATheoryofTechnologicalEvolution

TheLawofAcceleratingReturns

Thefurtherbackwardyoulook,thefurtherforwardyoucansee.

—WINSTONCHURCHILLTwobillionyearsago,ourancestorsweremicrobes;ahalf-billionyearsago, fish;ahundredmillion years ago, something likemice; tenmillion years ago, arboreal apes; and amillionyearsago,proto-humanspuzzlingoutthetamingoffire.Ourevolutionarylineageismarkedbymasteryofchange.Inourtime,thepaceisquickening.

—CARLSAGANOursoleresponsibilityistoproducesomethingsmarterthanweare;anyproblemsbeyondthatarenotourstosolve....[T]herearenohardproblems,onlyproblemsthatarehardtoacertainlevel of intelligence. Move the smallest bit upwards [in level of intelligence], and someproblems will suddenly move from "impossible" to "obvious." Move a substantial degreeupwards,andallofthemwillbecomeobvious.

—ELIEZER S. YUDNOWSKY, STARING INTO THE SINGULARITY,1996

"Thefuturecan'tbepredicted"isacommonrefrain....But...when[thisperspective]iswrong,itisprofoundlywrong.

—JOHNSMART1heongoingaccelerationoftechnologyistheimplicationandinevitableresultofwhatIcallthelawofacceleratingreturns,whichdescribestheaccelerationofthepaceofandtheexponentialgrowthoftheproducts of an evolutionary process. These products include, in particular, information-bearingtechnologiessuchascomputation,andtheiraccelerationextendssubstantiallybeyondthepredictionsmadebywhathasbecomeknownasMoore'sLaw.TheSingularityistheinexorableresultofthelaw

ofacceleratingreturns,soitisimportantthatweexaminethenatureofthisevolutionaryprocess.TheNatureofOrder.Thepreviouschapterfeaturedseveralgraphsdemonstratingtheaccelerationofparadigmshift.(Paradigmshiftsaremajorchangesinmethodsandintellectualprocessestoaccomplishtasks;examples includewritten languageand thecomputer.)Thegraphsplottedwhat fifteen thinkersandreferenceworksregardedasthekeyeventsinbiologicalandtechnologicalevolutionfromtheBigBang to the Internet. We see some expected variation, but an unmistakable exponential trend: keyeventshavebeenoccurringatanever-hasteningpace. Thecriteriaforwhatconstituted"keyevents"variedfromone thinker's list toanother.But it's

worthconsideringtheprinciplestheyusedinmakingtheirselections.Someobservershavejudgedthatthe truly epochal advances in the history of biology and technology have involved increases incomplexity.2 Although increased complexity does appear to follow advances in both biological andtechnologicalevolution, Ibelieve that thisobservation isnotpreciselycorrect.But let's firstexaminewhatcomplexitymeans. Not surprisingly, the concept of complexity is complex. One concept of complexity is theminimum amount of information required to represent a process. Let's say you have a design for asystem(forexample,acomputerprogramoracomputer-assisteddesignfileforacomputer),whichcanbedescribedbyadatafilecontainingonemillionbits.Wecouldsayyourdesignhasacomplexityofonemillionbits.But supposewenotice that theonemillionbits actually consist of a patternof onethousandbits that isrepeatedonethousandtimes.Wecouldnotetherepetitions,removetherepeatedpatterns,andexpresstheentiredesigninjustoveronethousandbits, therebyreducingthesizeofthefilebyafactorofaboutonethousand.Themostpopulardata-compressiontechniquesusesimilarmethodsoffindingredundancywithininformation.3But after you've compressed a data file in thisway, can yoube absolutely certain thattherearenootherrulesormethodsthatmightbediscoveredthatwouldenableyoutoexpressthefileinevenmorecompactterms?Forexample,supposemyfilewassimply"pi"(3.1415...)expressedtoonemillionbitsofprecision.Mostdata-compressionprogramswould fail to recognize this sequenceandwould not compress themillion bits at all, since the bits in a binary expression of pi are effectivelyrandomandthushavenorepeatedpatternaccordingtoalltestsofrandomness.Butifwecandeterminethatthefile(oraportionofthefile)infactrepresentspi,wecaneasilyexpress it (or thatportionof it)verycompactlyas"pi toonemillionbitsofaccuracy."Sincewecanneverbesurethatwehavenotoverlookedsomeevenmorecompactrepresentationofaninformationsequence,anyamountofcompressionsetsonlyanupperboundforthecomplexityoftheinformation.MurrayGell-Mannprovidesonedefinitionofcomplexityalongtheselines.Hedefinesthe"algorithmicinformationcontent"(Ale)ofasetofinformationas"thelengthoftheshortestprogramthatwillcauseastandarduniversalcomputertoprintoutthestringofbitsandthenhalt."4However,Gell-Mann'sconceptisnotfullyadequate.Ifwehaveafilewithrandominformation,itcannot be compressed. That observation is, in fact, a key criterion for determining if a sequence ofnumbers is truly random.However, if any randomsequencewilldo foraparticulardesign, then thisinformation can be characterized by a simple instruction, such as "put random sequence of numbershere."Sotherandomsequence,whetherit'stenbitsoronebillionbits,doesnotrepresentasignificantamountofcomplexity,becauseitischaracterizedbyasimpleinstruction.Thisisthedifferencebetweenarandomsequenceandanunpredictablesequenceofinformationthathaspurpose.Togainsomefurtherinsightintothenatureofcomplexity,considerthecomplexityofarock.Ifwewere tocharacterizeallof theproperties (precise location,angularmomentum,spin,velocity,andsoon) of every atom in the rock, wewould have a vast amount of information. A one-kilogram (2.2-pound) rockhas1025 atomswhich, as Iwill discuss in thenext chapter, canholdup to1027bitsofinformation. That's one hundred million billion times more information than the genetic code of ahuman(evenwithoutcompressingthegeneticcode).5Butformostcommonpurposes,thebulkofthisinformation is largely random and of little consequence. So we can characterize the rock for mostpurposeswithfar less information justbyspecifying itsshapeand the typeofmaterialofwhich it ismade.Thus,itisreasonabletoconsiderthecomplexityofanordinaryrocktobefarlessthanthatofahumaneventhoughtherocktheoreticallycontainsvastamountsofinformation.6Oneconceptofcomplexityistheminimumamountofmeaningful,non-random,butunpredictableinformationneededtocharacterizeasystemorprocess.InGell-Mann'sconcept,theAlCofamillion-bitrandomstringwouldbeaboutamillionbitslong.So I am adding toGell-Mann'sAlC concept the idea of replacing each random stringwith a simpleinstructionto"putrandombits"here.However,eventhisisnotsufficient.Anotherissueisraisedbystringsofarbitrarydata,suchas

names and phone numbers in a phone book, or periodic measurements of radiation levels ortemperature.Suchdataisnotrandom,anddata-compressionmethodswillonlysucceedinreducingittoasmalldegree.Yetitdoesnotrepresentcomplexityasthattermisgenerallyunderstood.Itisjustdata.Soweneedanothersimpleinstructionto"putarbitrarydatasequence"here.Tosummarizemyproposedmeasureofthecomplexityofasetofinformation,wefirstconsideritsAlCasGell-Mannhasdefinedit.Wethenreplaceeachrandomstringwithasimpleinstructiontoinsertarandomstring.Wethendothesameforarbitrarydatastrings.Nowwehaveameasureofcomplexitythatreasonablymatchesourintuition. It isafairobservationthatparadigmshiftsinanevolutionaryprocesssuchasbiology—anditscontinuationthroughtechnology—eachrepresentanincreaseincomplexity,asIhavedefineditabove.Forexample,theevolutionofDNAallowedformorecomplexorganisms,whosebiologicalinformationprocesses couldbe controlledby theDNAmolecule's flexible data storage.TheCambrian explosionprovidedastablesetofanimalbodyplans(inDNA),sothattheevolutionaryprocesscouldconcentrateonmorecomplexcerebraldevelopment.Intechnology,theinventionofthecomputerprovidedameansforhumancivilization tostoreandmanipulateevermorecomplexsetsof information.TheextensiveinterconnectednessoftheInternetprovidesforevengreatercomplexity. "Increasingcomplexity"on itsown isnot,however, theultimategoalor end-productof theseevolutionary processes. Evolution results in better answers, not necessarily more complicated ones.Sometimesasuperiorsolutionisasimplerone.Solet'sconsideranotherconcept:order.Orderisnotthesameas theoppositeofdisorder. Ifdisorder representsa randomsequenceofevents, theoppositeofdisordershouldbe"notrandomness."Informationisasequenceofdatathatismeaningfulinaprocess,suchastheDNAcodeofanorganismorthebitsinacomputerprogram."Noise,"ontheotherhand,isarandomsequence.Noiseisinherentlyunpredictablebutcarriesnoinformation.Information,however,isalsounpredictable.Ifwecanpredictfuturedatafrompastdata,thatfuturedatastopsbeinginformation.Thus,neitherinformationnornoisecanbecompressed(andrestoredtoexactlythesamesequence).Wemight consider a predictably alternating pattern (such as 0101010...) to be orderly, but it carries noinformationbeyondthefirstcoupleofbits. Thus, orderliness does not constitute order, because order requires information. Order isinformationthatfitsapurpose.Themeasureoforderisthemeasureofhowwelltheinformationfitsthepurpose. In the evolution of lifeforms, the purpose is to survive. In an evolutionary algorithm (acomputerprogramthatsimulatesevolutiontosolveaproblem)appliedto,say,designingajetengine,thepurposeistooptimizeengineperformance,efficiency,andpossiblyothercriteria.7Measuringorderismoredifficultthanmeasuringcomplexity.Thereareproposedmeasuresofcomplexity,asIdiscussedabove.Fororder,weneedameasureof"success" thatwouldbe tailored toeachsituation.Whenwecreate evolutionary algorithms, the programmer needs to provide such a successmeasure (called the"utilityfunction").Intheevolutionaryprocessoftechnologydevelopment,wecouldassignameasureofeconomicsuccess. Simplyhavingmoreinformationdoesnotnecessarilyresultinabetterfit.Sometimes,adeeperorder—a better fit to a purpose—is achieved through simplification rather than further increases incomplexity.Forexample,anew theory that ties togetherapparentlydisparate ideas intoonebroader,morecoherent theory reducescomplexitybutnonethelessmay increase the"order forapurpose." (Inthiscase,thepurposeistoaccuratelymodelobservedphenomena.)Indeed,achievingsimplertheoriesis a driving force in science. (As Einstein said, "Make everything as simple as possible, but nosimpler.") An important example of this concept is one that represented a key step in the evolution ofhominids: the shift in the thumb's pivot point, which allowed more precise manipulation of theenvironment."8Primatessuchaschimpanzeescangraspbuttheycannotmanipulateobjectswitheithera "power grip," or sufficient fine-motor coordination to write or to shape objects. A change in thethumb's pivot point did not significantly increase the complexity of the animal but nonetheless didrepresentanincreaseinorder,enabling,amongotherthings,thedevelopmentoftechnology.Evolutionhas shown, however, that the general trend toward greater order does typically result in greater

complexity.9 Thus improving a solution to a problem—which usually increases but sometimes decreasescomplexity—increasesorder.Nowweareleftwiththeissueofdefiningtheproblem.Indeed,thekeytoan evolutionary algorithm (and to biological and technological evolution in general) is exactly this:definingtheproblem(whichincludestheutilityfunction).Inbiologicalevolutiontheoverallproblemhasalwaysbeentosurvive.Inparticularecologicalnichesthisoverridingchallengetranslatesintomorespecific objectives, such as the ability of certain species to survive in extreme environments or tocamouflagethemselvesfrompredators.Asbiologicalevolutionmovedtowardhumanoids,theobjectiveitselfevolvedtotheabilitytooutthinkadversariesandtomanipulatetheenvironmentaccordingly. Itmayappear that this aspectof the lawof accelerating returnscontradicts the second lawofthermodynamics, which implies that entropy (randomness in a closed system) cannot decrease and,therefore,generallyincreases.10However,thelawofacceleratingreturnspertainstoevolution,whichisnotaclosedsystem. It takesplaceamidgreatchaosand indeeddependson thedisorder in itsmidst,from which it draws its options for diversity. And from these options, an evolutionary processcontinually prunes its choices to create ever greater order. Even a crisis, such as the periodic largeasteroidsthathavecrashedintotheEarth,althoughincreasingchaostemporarily,endupincreasing—deepening—theordercreatedbybiologicalevolution.Tosummarize,evolutionincreasesorder,whichmayormaynotincreasecomplexity(butusuallydoes).Aprimaryreasonthatevolution—oflife-formsoroftechnology—speedsupisthatitbuildsonits own increasing order, with ever more sophisticated means of recording and manipulatinginformation.Innovationscreatedbyevolutionencourageandenablefasterevolution.Inthecaseoftheevolution of life-forms, the most notable early example is DNA, which provides a recorded andprotected transcription of life's design fromwhich to launch further experiments. In the case of theevolution of technology, ever-improving humanmethods of recording information have fostered yetfurtheradvances in technology.The first computersweredesignedonpaperandassembledbyhand.Today,theyaredesignedoncomputerworkstations,withthecomputersthemselvesworkingoutmanydetails of the next generation's design, and are then produced in fully automated factorieswith onlylimitedhumanintervention.Theevolutionaryprocessoftechnologyimprovescapacitiesinanexponentialfashion.Innovatorsseek to improvecapabilitiesbymultiples. Innovation ismultiplicative,notadditive.Technology, likeanyevolutionaryprocess,buildsonitself.ThisaspectwillcontinuetoacceleratewhenthetechnologyitselftakesfullcontrolofitsownprogressioninEpochFive.11Wecansummarizetheprinciplesofthelawofacceleratingreturnsasfollows:

·Evolution applies positive feedback: the more capable methods resulting from one stage ofevolutionaryprogressareusedtocreatethenextstage.Asdescribedinthepreviouschapter,eachepochofevolutionhasprogressedmorerapidlybybuildingontheproductsofthepreviousstage.Evolution works through indirection: evolution created humans, humans created technology,humans are nowworkingwith increasingly advanced technology to create new generations oftechnology. By the time of the Singularity, there won't be a distinction between humans andtechnology.Thisisnotbecausehumanswillhavebecomewhatwethinkofasmachinestoday,butrather machines will have progressed to be like humans and beyond. Technology will be themetaphoricalopposablethumbthatenablesournextstepinevolution.Progress(furtherincreasesinorder)will thenbebasedonthinkingprocesses thatoccurat thespeedof lightrather than inveryslowelectrochemicalreactions.Eachstageofevolutionbuildsonthefruitsofthelaststage,sotherateofprogressofanevolutionaryprocessincreasesatleastexponentiallyovertime.Overtime, the"order"of the informationembedded in theevolutionaryprocess (themeasureofhowwelltheinformationfitsapurpose,whichinevolutionissurvival)increases.

·Anevolutionaryprocessisnotaclosedsystem;evolutiondrawsuponthechaosinthelargersystemin which it takes place for its options for diversity. Because evolution also builds on its ownincreasingorder,inanevolutionaryprocessorderincreasesexponentially.

·Acorrelateof theaboveobservation is that the "returns"ofanevolutionaryprocess (suchas thespeed, efficiency, cost-effectiveness, or overall "power" of a process) also increase at leastexponentiallyovertime.WeseethisinMoore'sLaw,inwhicheachnewgenerationofcomputerchip(whichnowappearsapproximatelyeverytwoyears)providestwiceasmanycomponentsperunitcost,eachofwhichoperatessubstantiallyfaster(becauseofthesmallerdistancesrequiredforthe electrons to travel within and between them and other factors). As I illustrate below, thisexponentialgrowthinthepowerandprice-performanceofinformation-basedtechnologiesisnotlimited tocomputersbut is true foressentiallyall information technologiesand includeshumanknowledge,measuredmanydifferentways.Itisalsoimportanttonotethattheterm"informationtechnology" is encompassing an increasingly broad class of phenomena and will ultimatelyincludethefullrangeofeconomicactivityandculturalendeavor.

·Inanotherpositive-feedbackloop,themoreeffectiveaparticularevolutionaryprocessbecomes—forexample,thehigherthecapacityandcost-effectivenessthatcomputationattains—thegreatertheamountofresourcesthataredeployedtowardthefurtherprogressofthatprocess.Thisresultsinasecondlevelofexponentialgrowth;thatis,therateofexponentialgrowth—theexponent—itselfgrowsexponentially.Forexample,asseeninthefigureonp.67,"Moore'sLaw:TheFifthParadigm,"ittookthreeyearstodoubletheprice-performanceofcomputationatthebeginningofthetwentiethcenturyandtwoyearsinthemiddleofthecentury.Itisnowdoublingaboutonceperyear.Notonlyiseachchipdoublinginpowereachyearforthesameunitcost,butthenumberofchips beingmanufactured is also growing exponentially; thus, computer research budgets havegrowndramaticallyoverthedecades.

·Biologicalevolutionisonesuchevolutionaryprocess.Indeed,itisthequintessentialevolutionaryprocess.Becauseittookplaceinacompletelyopensystem(asopposedtotheartificialconstraintsinanevolutionaryalgorithm),manylevelsofthesystemevolvedatthesametime.Notonlydoestheinformationcontainedinaspecies'genesprogresstowardgreaterorder,buttheoverallsystemimplementing the evolutionary process itself evolves in this way. For example, the number ofchromosomesand the sequenceofgeneson thechromosomeshavealsoevolvedover time.Asanother example, evolution has developed ways to protect genetic information from excessivedefects(althoughasmallamountofmutationisallowed,sincethisisabeneficialmechanismforongoing evolutionary improvement). One primary means of achieving this is the repetition ofgenetic information on paired chromosomes. This guarantees that, even if a gene on onechromosome is damaged, its corresponding gene is likely to be correct and effective.Even theunpairedmaleYchromosomehasdevisedmeansofbackingupitsinformationbyrepeatingitontheYchromosomeitself.12Onlyabout2percentofthegenomecodesforproteins.13Therestofthegeneticinformationhasevolvedelaboratemeanstocontrolwhenandhowtheprotein-codinggenes express themselves (produceproteins) in a processwe are onlybeginning to understand.Thus,theprocessofevolution,suchastheallowedrateofmutation,hasitselfevolvedovertime.

·Technological evolution is another such evolutionary process. Indeed, the emergence of the firsttechnology-creatingspeciesresultedinthenewevolutionaryprocessoftechnology,whichmakestechnological evolution an outgrowth of—and a continuation of—biological evolution. Homosapiensevolvedoverthecourseofafewhundredthousandyears,andearlystagesofhumanoid-created technology (such as thewheel, fire, and stone tools) progressedbarely faster, requiringtensofthousandsofyearstoevolveandbewidelydeployed.Ahalfmillenniumago,theproductofaparadigmshiftsuchastheprintingpresstookaboutacenturytobewidelydeployed.Today,theproductsofmajorparadigmshifts,suchascellphonesandtheWorldWideWeb,arewidelyadoptedinonlyafewyears'time.

·Aspecificparadigm(amethodorapproachtosolvingaproblem;forexample,shrinkingtransistorsonanintegratedcircuitasawaytomakemorepowerfulcomputers)generatesexponentialgrowthuntil its potential is exhausted. When this happens, a paradigm shift occurs, which enablesexponentialgrowthtocontinue.

TheLifeCycleofaParadigm.Eachparadigmdevelopsinthreestages:

1.Slowgrowth(theearlyphaseofexponentialgrowth)2. Rapidgrowth (the late, explosivephaseof exponentialgrowth), as seen in theS-curve

figurebelow3.Alevelingoffastheparticularparadigmmatures

TheprogressionofthesethreestageslooksliketheletterS,stretchedtotheright.TheS-curveillustrationshowshowanongoingexponentialtrendcanbecomposedofacascadeofS-curves.EachsuccessiveS-curveisfaster(takeslesstimeonthetime,orx,axis)andhigher(takesupmoreroomontheperformance,ory,axis).

S-curvesaretypicalofbiologicalgrowth:replicationofasystemofrelativelyfixedcomplexity(suchasanorganismofaparticularspecies),operatinginacompetitivenicheandstrugglingforfinitelocal resources. This often occurs, for example, when a species happens upon a new hospitableenvironment. Its numbers will grow exponentially for a while before leveling off. The overallexponential growth of an evolutionary process (whether molecular, biological, cultural, ortechnological)supersedesthelimitstogrowthseeninanyparticularparadigm(aspecificS-curve)asaresultoftheincreasingpowerandefficiencydevelopedineachsuccessiveparadigm.Theexponentialgrowth of an evolutionary process, therefore, spans multiple S-curves. The most importantcontemporaryexampleofthisphenomenonisthefiveparadigmsofcomputationdiscussedbelow.Theentireprogressionofevolutionseeninthechartsontheaccelerationofparadigmshiftintheprevious

chapter represents successiveS-curves.Eachkeyevent, suchaswritingorprinting, representsanewparadigmandanewS-curve.Theevolutionarytheoryofpunctuatedequilibrium(PE)describesevolutionasprogressingthroughperiodsofrapidchangefollowedbyperiodsofrelativestasis.14Indeed,thekeyeventsontheepochal-event graphs do correspond to renewed periods of exponential increase in order (and, generally, ofcomplexity), followed by slower growth as each paradigm approaches its asymptote (limit ofcapability). So PE does provide a better evolutionarymodel than amodel that predicts only smoothprogressionthroughparadigmshifts. But the key events in punctuated equilibrium,while giving rise tomore rapid change, don'trepresentinstantaneousjumps.Forexample,theadventofDNAallowedasurge(butnotanimmediatejump)ofevolutionaryimprovementinorganismdesignandresultingincreasesincomplexity.Inrecenttechnological history, the invention of the computer initiated another surge, still ongoing, in thecomplexityofinformationthatthehuman-machinecivilizationiscapableofhandling.Thislattersurgewillnotreachanasymptoteuntilwesaturatethematterandenergyinourregionoftheuniversewithcomputation,basedonphysicallimitswe'lldiscussinthesection"...ontheIntelligentDestinyoftheCosmos"inchapter6.15Duringthisthirdormaturingphaseinthelifecycleofaparadigm,pressurebeginstobuildforthenextparadigmshift.Inthecaseoftechnology,researchdollarsareinvestedtocreatethenextparadigm.Wecansee this in theextensive researchbeingconducted today toward three-dimensionalmolecularcomputing, despite the fact that we still have at least a decade left for the paradigm of shrinkingtransistorsonaflatintegratedcircuitusingphotolithography. Generally, by the time a paradigm approaches its asymptote in price-performance, the nexttechnicalparadigmisalreadyworkinginnicheapplications.Forexample,inthe1950sengineerswereshrinkingvacuumtubestoprovidegreaterprice-performanceforcomputers,until theprocessbecamenolongerfeasible.Atthispoint,around1960,transistorshadalreadyachievedastrongnichemarketinportableradiosandweresubsequentlyusedtoreplacevacuumtubesincomputers. The resources underlying the exponential growth of an evolutionary process are relativelyunbounded.Onesuchresourceisthe(ever-growing)orderoftheevolutionaryprocessitself(since,asIpointedout,theproductsofanevolutionaryprocesscontinuetogrowinorder).Eachstageofevolutionprovidesmore powerful tools for the next. For example, in biological evolution, the advent ofDNAenabledmorepowerfuland fasterevolutionary"experiments."Or to takeamore recentexample, theadventofcomputer-assisteddesigntoolsallowsrapiddevelopmentofthenextgenerationofcomputers. The other required resource for continued exponential growth of order is the "chaos" of theenvironmentinwhichtheevolutionaryprocesstakesplaceandwhichprovidestheoptionsforfurtherdiversity. The chaos provides the variability to permit an evolutionary process to discover morepowerful and efficient solutions. In biological evolution, one source of diversity is the mixing andmatching of gene combinations through sexual reproduction. Sexual reproduction itself was anevolutionary innovation that accelerated the entire process of biological adaptation and provided forgreaterdiversityofgenetic combinations thannonsexual reproduction.Other sourcesofdiversity aremutations and ever-changing environmental conditions. In technological evolution, human ingenuitycombinedwithvariablemarketconditionskeepstheprocessofinnovationgoing.FractalDesigns.Akeyquestionconcerningtheinformationcontentofbiologicalsystemsishowitispossibleforthegenome,whichcontainscomparativelylittleinformation,toproduceasystemsuchasahuman, which is vastly more complex than the genetic information that describes it. One way ofunderstandingthisistoviewthedesignsofbiologyas"probabilisticfractals."Adeterministicfractalisa design inwhich a single design element (called the "initiator") is replacedwithmultiple elements(together called the "generator"). In a second iteration of fractal expansion, each element in thegenerator itselfbecomesan initiatorand is replacedwith theelementsof thegenerator (scaled to thesmallersizeofthesecond-generationinitiators).Thisprocessisrepeatedmanytimes,witheachnewlycreatedelementofageneratorbecomingan initiatorandbeingreplacedwithanewscaledgenerator.

Eachnewgenerationof fractalexpansionaddsapparentcomplexitybut requiresnoadditionaldesigninformation.Aprobabilisticfractaladdstheelementofuncertainty.Whereasadeterministicfractalwilllook thesameevery time it is rendered,aprobabilistic fractalwill lookdifferenteach time,althoughwith similar characteristics. Inaprobabilistic fractal, theprobabilityof eachgenerator elementbeingappliedislessthan1.Inthisway,theresultingdesignshaveamoreorganicappearance.Probabilisticfractals are used in graphics programs to generate realistic-looking images of mountains, clouds,seashores,foliage,andotherorganicscenes.Akeyaspectofaprobabilisticfractalisthatitenablesthegenerationofagreatdealofapparentcomplexity,includingextensivevaryingdetail,fromarelativelysmall amount of design information. Biology uses this same principle. Genes supply the designinformation,butthedetailinanorganismisvastlygreaterthanthegeneticdesigninformation. Some observersmisconstrue the amount of detail in biological systems such as the brain byarguing,forexample,thattheexactconfigurationofeverymicrostructure(suchaseachtubule)ineachneuron ispreciselydesignedandmustbeexactly thewayit is for thesystemtofunction. Inorder tounderstandhowabiologicalsystemsuchasthebrainworks,however,weneedtounderstanditsdesignprinciples, which are far simpler (that is, contain far less information) than the extremely detailedstructuresthatthegeneticinformationgeneratesthroughtheseiterative,fractal-likeprocesses.Thereareonlyeighthundredmillionbytesofinformationintheentirehumangenome,andonlyaboutthirtytoonehundredmillionbytesafterdatacompression isapplied.This isaboutonehundredmillion timesless information than is represented by all of the interneuronal connections and neurotransmitterconcentrationpatternsinafullyformedhumanbrain.Considerhowtheprinciplesofthelawofacceleratingreturnsapplytotheepochswediscussedinthefirstchapter.ThecombinationofaminoacidsintoproteinsandofnucleicacidsintostringsofRNAestablishedthebasicparadigmofbiology.StringsofRNA(andlaterDNA)thatself-replicated(EpochTwo) provided a digital method to record the results of evolutionary experiments. Later on, theevolutionofaspeciesthatcombinedrationalthought(EpochThree)withanopposableappendage(thethumb)causedafundamentalparadigmshiftfrombiologytotechnology(EpochFour).Theupcomingprimary paradigm shift will be from biological thinking to a hybrid combining biological andnonbiological thinking (Epoch Five), which will include "biologically inspired" processes resultingfromthereverseengineeringofbiologicalbrains. Ifwe examine the timingof these epochs,we see that theyhavebeenpart of a continuouslyacceleratingprocess.Theevolutionoflife-formsrequiredbillionsofyearsforitsfirststeps(primitivecells,DNA),andthenprogressaccelerated.DuringtheCambrianexplosion,majorparadigmshiftstookonlytensofmillionsofyears.Later,humanoidsdevelopedoveraperiodofmillionsofyears,andHomosapiensoveraperiodofonlyhundredsofthousandsofyears.Withtheadventofatechnology-creatingspeciestheexponentialpacebecametoofastforevolutionthroughDNA-guidedproteinsynthesis,andevolution moved on to human-created technology. This does not imply that biological (genetic)evolutionisnotcontinuing,justthatitisnolongerleadingthepaceintermsofimprovingorder(oroftheeffectivenessandefficiencyofcomputation).16FarsightedEvolution.Therearemanyramificationsoftheincreasingorderandcomplexitythathaveresultedfrombiologicalevolutionanditscontinuationthroughtechnology.Considertheboundariesofobservation.Earlybiological lifecouldobservelocaleventsseveralmillimetersaway,usingchemicalgradients.Whensightedanimalsevolved,theywereabletoobserveeventsthatweremilesaway.Withthe invention of the telescope, humans could see other galaxies millions of light-years away.Conversely,usingmicroscopes,theycouldalsoseecellular-sizestructures.Todayhumansarmedwithcontemporary technology can see to the edge of the observable universe, a distance of more thanthirteenbillionlight-years,anddowntoquantum-scalesubatomicparticles. Consider thedurationofobservation.Single-cell animals could remember events for seconds,based on chemical reactions. Animals with brains could remember events for days. Primates withculturecouldpassdowninformation throughseveralgenerations.Earlyhumancivilizationswithoralhistorieswereable topreservestories forhundredsofyears.With theadventofwritten language the

permanenceextendedtothousandsofyears.Asoneofmanyexamplesoftheaccelerationofthetechnologyparadigm-shiftrate,ittookaboutahalf century for the late-nineteenth-century invention of the telephone to reach significant levels ofusage(seethefigurebelow).17

Incomparison,thelate-twentieth-centuryadoptionofthecellphonetookonlyadecade.18

Overallweseeasmoothaccelerationintheadoptionratesofcommunicationtechnologiesoverthepastcentury.19

Asdiscussedinthepreviouschapter,theoverallrateofadoptingnewparadigms,whichparallelsthe rateof technologicalprogress, iscurrentlydoublingeverydecade.That is, the time toadoptnewparadigms isgoingdownbyhalfeachdecade.At this rate, technologicalprogress in the twenty-firstcentury will be equivalent (in the linear view) to two hundred centuries of progress (at the rate ofprogressin2000).20,21

TheS-CurveofaTechnologyasExpressedinitsLifeCycle

AmachineisasdistinctivelyandbrilliantlyandexpressivelyhumanasaviolinsonotaoratheoreminEuclid.—GREGORYVLASTOS

Itisafarcryfromthemonkishcalligrapher,workinginhiscellinsilence,tothebrisk“click,click”ofthemodernwritingmachine,whichinaquarterofacenturyhasrevolutionizedandreformedbusiness.

—SCIENTIFICAMERICAN,1905Nocommunicationtechnologyhaseverdisappearedbutinsteadbecomesincreasinglylessimportantasthetechnologicalhorizonwidens.

—ArthurC.ClarkeIalwayskeepastackofbooksonmydeskthatIleafthroughwhenIrunoutofideas,feelrestless,orotherwiseneedashotofinspiration.PickingupafatvolumethatIrecentlyacquired,Iconsiderthebookmaker'scraft:470finelyprintedpagesorganizedinto16-pagesignatures,allofwhicharesewntogetherwithwhitethreadandgluedontoagraycanvascord.Thehardlinen-boundcovers,stampedwithgoldletters,areconnectedtothesignatureblockbydelicatelyembossedendsheets.Thisisatechnologythatwasperfectedmanydecadesago.Booksconstitutesuchanintegralelementofoursociety—bothreflectingandshapingitsculture—thatitishardtoimaginelifewithoutthem.Buttheprintedbook,likeanyothertechnology,willnotliveforever.

TheLifeCycleofaTechnologyWecanidentifysevendistinctstagesinthelifecycleofatechnology.

1.Duringtheprecursorstage,theprerequisitesofatechnologyexist,anddreamersmaycontemplatetheseelementscomingtogether.Wedonot,however,regarddreamingtobe thesameas inventing,even if thedreamsarewrittendown.LeonardodaVincidrewconvincingpicturesofairplanesandautomobiles,butheisnotconsideredtohaveinventedeither.

2.Thenextstage,onehighlycelebratedinourculture,isinvention,averybriefstage,similarinsomerespectstotheprocessofbirthafteranextendedperiodoflabor.Heretheinventorblendscuriosity,scientificskills,determination,andusuallyofshowmanshiptocombinemethodsinanewwayandbringsanewtechnologytolife.

3. Thenextstage isdevelopment,duringwhich the invention isprotectedandsupportedbydotingguardians (whomay include theoriginal inventor). Often this stage is more crucial than invention and may involve additional creation that can have greatersignificancethantheinventionitself.Manytinkerershadconstructedfinelyhand-tunedhorselesscarriages,butitwasHenryFord'sinnovationofmassproductionthatenabledtheautomobiletotakerootandflourish.

4.Thefourthstageismaturity.Althoughcontinuingtoevolve,thetechnologynowhasalifeofitsownandhasbecomeanestablishedpartofthecommunity.Itmaybecometointertwinedinthefabricoflifethatitappearstomanyobserversthatitwilllastforever.Thiscreatesaninterestingdramawhenthenextstagearrives,whichIcallthestageofthefalsepretenders.

5.Hereanupstartthreatenstoeclipsetheoldertechnology.Itsenthusiastsprematurelypredictvictory.Whileprovidingsomedistinctbenefits,thenewertechnologyisfoundonreflectiontobelackingsomekeyelementoffunctionalityorquality.Whenitindeedfailstodislodgetheestablishedorder,thetechnologyconservativestakethisasevidencethattheoriginalapproachwillindeedliveforever.

6. Thisisusuallyashort-livedvictory for theagingtechnology.Shortly thereafter,anothernew technology typicallydoessucceed inrenderingtheoriginaltechnologytothestageofobsolescence.Inthispartofthelifecycle,thetechnologylivesoutitssenioryearsingradualdecline,itsoriginalpurposeandfunctionalitynowsubsumedbyamoresprycompetitor.

7.Inthisstage,whichmaycomprise5to10percentofatechnology'slifecycle,itfinallyyieldstoantiquity(asdidthehorseandbuggy)theharpsichord,thevinylrecord,andthemanualtypewriter).

Inthemid-nineteenthcenturytherewereseveralprecursorstothephonograph,includingLéonScottdeMartinville'sphonautograph,adevice that recordedsoundvibrationsasaprintedpattern. ItwasThomasEdison,however,whobroughtall of theelements togetherandinvented the first device that couldboth record and reproduce sound in 1877. Further refinementswere necessary for the phonograph tobecomecommerciallyviable.Itbecameafullymaturetechnologyin1949whenColumbiaintroducedthe33-rpmlong-playingrecording(LP)andRCAintroducedthe45-rpmdisc.Thefalsepretenderwasthecassettetape,introducedinthe1960sandpopularizedduringthe1970s.Earlyenthusiastspredictedthatitssmallsizeandabilitytobererecordedwouldmaketherelativelybulkyandscratchablerecordobsolete.Despitethesebenefits,cassetteslackrandomaccessandarepronetotheirownformsofdistortionandlackfidelity.Thecompactdisc(CD)delivered themortalblow.With theCDprovidingboth randomaccessanda levelofqualityclose to the limitsof thehumanauditorysystem,thephonographquicklyenteredthestageofobsolescence.Althoughstillproduced,thetechnologythatEdisongavebirthtoalmost130yearsagohasnowreachedantiquity. Consider the piano, an area of technology that I have been personally involved with replicating. In the early eighteenth centuryBartolommeoCristiforiwasseekingawaytoprovideatouchresponsetothethen-popularharpsichordsothatthevolumeofthenoteswouldvarywiththeintensityofthetouchoftheperformer.Calledgravicembalocalpianoeforte("harpsichordwithsoftandloud"),hisinventionwasnotanimmediatesuccess.Furtherrefinements,includingStein'sVienneseactionandZumpe'sEnglishaction,helpedtoestablishthe"piano"as the preeminent keyboard instrument. It reached maturity with the development of the complete cast-iron frame, patented in 1825 byAlpheusBabcock,andhasseenonlysubtlerefinementssincethen.Thefalsepretenderwastheelectricpianooftheearly1980s.Itofferedsubstantially greater functionality. Compared to the single (piano) sound of the acoustic piano, the electronic variant offered dozens ofinstrumentsounds,sequencersthatallowedtheusertoplayanentireorchestraatonce,automatedaccompaniment,educationprogramtoteachkeyboardskills,andmanyotherfeatures.Theonlyfeatureitwasmissingwasagood-qualitypianosound.Thiscrucialflawandtheresultingfailureofthefirstgenerationofelectronicpianosledtothewidespreadconclusionthatthepianowouldneverbereplacedbyelectronics.But the"victory"of theacousticpianowillnotbepermanent.With their fargreater rangeof featuresandprice-performance,digitalpianosalreadyexceedsalesofacousticpianosinhomes.Manyobserversfeelthatthequalityofthe"piano"soundondigitalpianosnowequalsorexceedsthatoftheuprightacousticpiano.Withtheexceptionofconcertandluxurygrandpianos(asmallpartofthemarket),thesaleofacousticpianosisindecline.

FromGoatSkinstoDownloadsSowhereinthetechnologylifecycleisthebook?AmongitsprecursorswereMesopotamianclaytabletsandEgyptianpapyrusscrolls.InthesecondcenturyB.C.,thePtolemiesofEgyptcreatedagreatlibraryofscrollsatAlexandriaandoutlawedtheexportofpapyrustodiscouragecompetition.WhatwereperhapsthefirstbookswerecreatedbyEumenesII,rulerofancientGreekPerganum,usingpagesofvellummadefromthe

skinsofgoatsandsheep,whichweresewntogetherbetweenwoodencovers.ThistechniqueenabledEumenestocompilealibraryequaltothatofAlexandria.Aroundthesametime,theChinesehasalsodevelopedacrudeformofbookmadefrombamboostrips. Thedevelopmentandmaturationofbookshasinvolvedthreegreatadvances.Printing,firstexperimentedwithbytheChineseintheeightcenturyA.D.using raisedwoodblocks,allowedbooks tobe reproduced inmuch largerquantities, expanding their audiencebeyondgovernment and religious leaders. Of even greater significance was the advent of movable type, which the Chinese and Koreansexperimentedwithbytheeleventhcentury,butthecomplexityofAsiancharacterspreventedtheseearlyattemptsfrombeingfullysuccessful.JohannesGutenbergworkinginthefifteenthcentury,benefitedfromtherelativesimplicityoftheRomancharacterset.HeproducedhisBible,thefirstlarge-scaleworkprintedentirelywithmovabletype,in1455.Whiletherehasbeenacontinualstreamofevolutionaryimprovementsinthemechanicalandelectromechanicalprocessofprinting,thetechnologyofbookmakingdidnotseeanotherqualitativeleapuntiltheavailabilityofcomputertypesetting,whichdidawaywithmovabletypeabouttwodecadesago.Typographyisnowregardedasapartofdigitalimageprocessing.Withbooksafullymaturetechnology,thefalsepretendersarrivedabouttwentyyearsagowiththefirstwaveof"electronicbooks."Asisusually thecase, these falsepretendersoffereddramaticqualitativeandquantitativebenefits.CD-ROM-or flashmemory-basedelectronicbookscanprovidetheequivalentofthousandsofbookswithpowerfulcomputer-basedsearchandknowledgenavigationfeatures.WithWeb-orCD-ROM-andDVD-basedencyclopedias,Icanperformrapidwordsearchesusingextensivelogicrules,somethingisjustnotpossiblewiththethirty-three-volume"book"versionIpossess.Electronicbookscanprovidepicturesthatareanimatedandthatrespondtoourinput.Pagesarenotnecessarilyorderedsequentiallybutcanbeexploredalongmoreintuitiveconnections.Aswiththephonographrecordandthepiano,thisfirstgenerationoffalsepretenderswas(andstillis)missinganessentialqualityoftheoriginal,whichinthiscaseisthesuperbvisualcharacteristicsofpaperandink.Paperdoesnotflicker,whereasthetypicalcomputerscreenisdisplayingsixtyormorefieldspersecond.Thisisaproblembecauseofanevolutionaryadaptationoftheprimatevisualsystem.Weareabletoseeonlyaverysmallportionofthevisualfieldwithhighresolution.Thisportion,imagedbythefoveaintheretina,isfocusedonanareaabout the sizeof a singlewordat twenty-two inches away.Outside of the fovea,we have very little resolution but exquisite sensitivity tochangesinbrightness,anabilitythatallowedourprimateforebearstoquicklydetectapredatorthatmightbeattacking.Theconstantflickerofa video graphics array (VGA) computer screen is detected by our eyes as motion and causes constant movement of the fovea. Thissubstantiallyslowsdownreadingspeeds.which isone reason that readingonascreen is lesspleasant than readingaprintedbook.Thisparticularissuehasbeensolvedwithflat-paneldisplays,whichdonotflicker.Othercrucialissuesincludecontrast—agood-qualitybookhasanink-to-papercontrastofabout120:1;typicalscreensareperhapshalfofthat—andresolution.Printandillustrationsinabookrepresentaresolutionofabout600to1000dotsperinch(dpi),whilecomputerscreensareaboutonetenthofthat.Thesizeandweightofcomputerizeddevicesareapproachingthoseofbooks,butthedevicesarestillheavierthanapaperbackbook.Paperbooksalsodonotrunoutofbatterypower. Most important, there is thematter of the available software, bywhich Imean the enormous installed base of printed books. Fiftythousandnewprint booksarepublishedeach year in theUnitedStates, andmillionsof booksarealready in circulation.Therearemajoreffortsunderwaytoscananddigitizeprintmaterials,butitwillbealongtimebeforetheelectronicdatabaseshaveacomparablewealthofmaterial.Thebiggestobstaclehere is theunderstandablehesitationofpublishers tomake theelectronicversionsof theirbooksavailable,giventhedevastatingeffectthatillegalfilesharinghashadonthemusic-recordingindustry.Solutionsareemergingtoeachoftheselimitations.New,inexpensivedisplaytechnologieshavecontrast,resolution,lackofflicker,andviewing and comparable to high-quality paper documents. Fuel-cell power for portable electronics is being introduced, which will keepelectronicdevicespoweredforhundredsofhoursbetweenfuel-cartridgechanges.Portableelectronicdevicesarealreadycomparabletothesize and weight of a book. The primary issue is going to be finding secure means of making electronic information available. This is afundamental concern for every level of our economy.Everything—including physical products, once nanotechnology-basedmanufacturingbecomesarealityinabouttwentyyears—isbecominginformation.

Moore'sLawandBeyond

WhereacalculatorontheENIACisequippedwith18,000vacuumtubesandweighs30tons,computersinthefuturemayhaveonly1,000vacuumtubesandperhapsweigh1.5tons.

—POPULARMECHANICS,1949ComputerScienceisnomoreaboutcomputersthanastronomyisabouttelescopes.

—E.W.DIJKSTRABefore considering further the implications of the Singularity, let's examine the wide range oftechnologiesthataresubjecttothelawofacceleratingreturns.Theexponentialtrendthathasgainedthegreatest public recognition has becomeknown asMoore'sLaw. In themid-1970s,GordonMoore, aleadinginventorofintegratedcircuitsandlaterchairmanofIntel,observedthatwecouldsqueezetwiceas many transistors onto an integrated circuit every twenty-four months (in the mid-1960s, he hadestimated twelvemonths). Given that the electrons would consequently have less distance to travel,circuitswouldalsorunfaster,providinganadditionalboosttooverallcomputationalpower.Theresultis exponential growth in the price-performance of computation. This doubling rate—about twelvemonths—ismuch faster than thedoubling rate forparadigmshift that I spokeaboutearlier,which isabout tenyears.Typically,wefind that thedoubling timefordifferentmeasures—price-performance,bandwidth,capacity—ofthecapabilityofinformationtechnologyisaboutoneyear. TheprimarydrivingforceofMoore'sLawisareductionofsemiconductorfeaturesizes,whichshrinkbyhalfevery5.4yearsineachdimension.(Seethefigurebelow.)Sincechipsarefunctionallytwo-dimensional,thismeansdoublingthenumberofelementspersquaremillimeterevery2.7years.22

The following charts combine historical data with the semiconductor-industry road map(International Technology Roadmap for Semiconductors [ITRS] from Sematech), which projectsthrough2018.

The cost of DRAM (dynamic random access memory) per square millimeter has also been comingdown.ThedoublingtimeforbitsofDRAMperdollarhasbeenonly1.5years.23

Asimilartrendcanbeseenwithtransistors.Youcouldbuyonetransistorforadollarin1968;in2002adollarpurchasedabouttenmilliontransistors.SinceDRAMisaspecializedfieldthathasseenits own innovation, the halving time for average transistor price is slightly slower than for DRAM,about1.6years(seethefigurebelow).24

Thisremarkablysmoothaccelerationinprice-performanceofsemiconductorshasprogressedthroughaseriesofstagesofprocesstechnologies(definedbyfeaturesizes)ateversmallerdimensions.Thekeyfeature size is now dipping below one hundred nanometers, which is considered the threshold of"nanotechnology."25

UnlikeGertrudeStein'srose,itisnotthecasethatatransistorisatransistorisatransistor.Astheyhavebecomesmallerandlessexpensive, transistorshavealsobecomefasterbyafactorofaboutonethousandover thecourseof thepast thirtyyears (see thefigurebelow)—again,because theelectronshavelessdistancetotravel.26

Ifwecombinetheexponentialtrendstowardless-expensivetransistorsandfastercycletimes,wefindahalvingtimeofonly1.1yearsinthecostpertransistorcycle(seethefigurebelowl.27Thecostper transistor cycle is a more accurate overall measure of price-performance because it takes intoaccount both speed and capacity. But the cost per transistor cycle still does not take into accountinnovation at higher levels of design (such as microprocessor design) that improves computationalefficiency.

ThenumberoftransistorsinIntelprocessorshasdoubledeverytwoyears(seethefigurebelow).Several other factors have boosted price-performance, including clock speed, reduction in cost permicroprocessor,andprocessordesigninnovations.28

Processor performance in MIPS has doubled every 1.8 years per processor (see the figure below).Again,notethatthecostperprocessorhasalsodeclinedthroughthisperiod.29

IfIexaminemyownfour-plusdecadesofexperienceinthisindustry,IcancomparetheMITcomputerIusedasastudentinthelate1960stoarecentnotebook.In1967Ihadaccesstoamultimillion-dollarIBM7094with32K(36-bit)wordsofmemoryandaquarterofaMIPSprocessorspeed.In2004Iuseda $2,000 personal computerwith a half-billion bytes ofRAMand a processor speed of about 2,000MIPS.TheMITcomputerwasaboutonethousandtimesmoreexpensive,sotheratioofcostperMIPSisabouteightmilliontoone.

Myrecentcomputerprovides2,000MIPSofprocessingatacostthatisabout224lowerthanthatofthecomputerIusedin1967.That's24doublingsin37years,orabout18.5monthsperdoubling.Ifwefactorintheincreasedvalueoftheapproximately2,000timesgreaterRAM,vastincreasesindiskstorage,andthemorepowerfulinstructionsetofmycirca2004computer,aswellasvastimprovementsincommunication speeds,morepowerful software, andother factors, thedoubling timecomesdownevenfurther.Despitethismassivedeflationinthecostofinformationtechnologies,demandhasmorethankeptup.Thenumberofbitsshippedhasdoubledevery1.1years,fasterthanthehalvingtimeincostperbit,whichis1.5years.30Asaresult,thesemiconductorindustryenjoyed18percentannualgrowthintotalrevenue from 1958 to 2002.31 The entire information-technology (IT) industry has grown from 4.2percent of the gross domestic product in 1977 to 8.2 percent in 1998.32 IThas become increasinglyinfluentialinalleconomicsectors.Theshareofvaluecontributedbyinformationtechnologyformostcategoriesofproductsandservicesisrapidlyincreasing.Evencommonmanufacturedproductssuchastables and chairs have an information content, represented by their computerized designs and theprogramming of the inventory-procurement systems and automated-fabrication systems used in theirassembly.

Moore'sLaw:Self-FulfillingProphecy?

SomeobservershavestatedthatMoore'sLawisnothingmorethanaself-fulfillingprophecy:thatindustryparticipantsanticipatewheretheyneedtobeatparticulartimeinthefuture,andorganizetheirresearchdevelopmentaccordingly.Theindustry'sownwrittenroadmapisagoodexampleofthis.34However,theexponentialtrendsininformationtechnologyarefarbroaderthanthosecoveredbyMoore'sLaw.Weseethesametypeoftrendsinessentiallyeverytechnologyormeasurementthatdealswithinformation.Thisincludesmanytechnologiesinwhichaperceptionofacceleratingprice-performancedoesnotexistorhasnotpreviouslybeenarticulated(seebelow).Evenwithincomputingitself,thegrowthincapabilityperunitcostismuchbroaderthanwhatMoore'sLawalonewouldpredict.

TheFifthParadigm35

Moore'sLawisactuallynotthefirstparadigmincomputationalsystems.Youcanseethisifyouplottheprice-performance—measuredby instructionsper secondper thousandconstantdollars—of forty-nine famous computational systems and computers spanning the twentieth century (see the figurebelow).

As the figure demonstrates, therewere actually four different paradigms—electromechanical,relays, vacuum tubes, and discrete transistors—that showed exponential growth in the price-performanceofcomputinglongbeforeintegratedcircuitswereeveninvented.AndMoore'sparadigmwon't be the last.WhenMoore'sLaw reaches the endof itsS-curve, nowexpectedbefore2020, theexponential growthwill continuewith three-dimensionalmolecular computing,whichwill constitutethesixthparadigm.

FractalDimensionsandtheBrain

Notethattheuseofthethirddimensionincomputingsystemsisnotaneither-orchoicebutacontinuumbetweentwoandthreedimensions.Intermsofbiologicalintelligence,thehumancortexisactuallyratherflat,withonlysixthinlayersthatareelaboratelyfolded,anarchitecturethatgreatly increases the surfacearea.The foldingoneway touse the thirddimension. In "fractal" systems (systems inwhich a drawingreplacementorfoldingruleisiterativelyapplied),structuresthatareelaboratelyfoldedareconsideredtoconstituteapartialdimension.Fromthat perspective, the convoluted surface of the human cortex represents a number of dimensions in between two and three.Other brainstructures,suchasthecerebellum,arethree-dimensionalbutcomprisearepeatingstructurethatisessentiallytwo-dimensional.Itislikely thatour future computational systems will also combine systems that are highly folded two-dimensional systems with fully three-dimensionalstructures.

Noticethatthefigureshowsanexponentialcurveonalogarithmicscale,indicatingtwolevelsofexponentialgrowth.36Inotherwords,thereisagentlebutunmistakableexponentialgrowthintherateof exponential growth. (A straight line on a logarithmic scale shows simple exponential growth; anupwardly curving line shows higher-than-simple exponential growth.) As you can see, it took threeyearstodoubletheprice-performanceofcomputingatthebeginningofthetwentiethcenturyandtwoyearsinthemiddle,andittakesaboutoneyearcurrently.37HansMoravecprovidesthefollowingsimilarchart(seethefigurebelow),whichusesadifferentbutoverlappingsetofhistoricalcomputersandplotstrendlines(slopes)atdifferentpointsintime.Aswiththefigureabove,theslopeincreaseswithtime,reflectingthesecondlevelofexponentialgrowth.38

Ifweprojectthesecomputationalperformancetrendsthroughthisnextcentury,wecanseeinthefigure below that supercomputers will match human brain capability by the end of this decade andpersonal computing will achieve it by around 2020—or possibly sooner, depending on howconservative an estimate of human brain capacity we use. (We'll discuss estimates of human braincomputationalspeedinthenextchapter.39

Theexponentialgrowthofcomputingisamarvelousquantitativeexampleoftheexponentiallygrowingreturnsfromanevolutionaryprocess.Wecanexpresstheexponentialgrowthofcomputingintermsofitsacceleratingpace:ittookninetyyearstoachievethefirstMIPSperthousanddollars;nowweaddoneMIPSperthousanddollarseveryfivehours.40

IBM'sBlueGene/Psupercomputerisplannedtohaveonemilliongigaflops(billionsoffloating-pointoperationsper second),or1015 calculationsper secondwhen it launches in 2007.41 That's onetenthofthe1016calculationspersecondneededtoemulatethehumanbrain(seethenextchapter).Andifweextrapolatethisexponentialcurve,weget1016calculationspersecondearlyinthenextdecade. Asdiscussedabove,Moore'sLawnarrowlyreferstothenumberoftransistorsonanintegratedcircuitoffixedsizeandsometimeshasbeenexpressedevenmorenarrowlyintermsoftransistorfeaturesize.Butthemostappropriatemeasuretotrackprice-performanceiscomputationalspeedperunitcost,anindexthattakesintoaccountmanylevelsof"cleverness"(innovation,whichistosay,technologicalevolution).Inadditiontoalloftheinventioninvolvedinintegratedcircuits,therearemultiplelayersofimprovementincomputerdesign(forexample,pipelining,parallelprocessing,instructionlook-ahead,instructionandmemorycaching,andmanyothers).Thehumanbrainusesaveryinefficientelectrochemical,digital-controlledanalogcomputationalprocess.Thebulkofitscalculationsarecarriedoutintheinterneuronalconnectionsataspeedofonlyabout two hundred calculations per second (in each connection),which is at least onemillion timesslower than contemporary electronic circuits. But the brain gains its prodigious powers from itsextremelyparallelorganizationinthreedimensions.Therearemanytechnologiesinthewingsthatwillbuildcircuitryinthreedimensions,whichIdiscussinthenextchapter. Wemightaskwhetherthereareinherentlimitstothecapacityofmatterandenergytosupportcomputationalprocesses.This isan important issue,butaswewill see in thenextchapter,wewon'tapproachthoselimitsuntillateinthiscentury.ItisimportanttodistinguishbetweentheS-curvethatis

characteristic of any specific technological paradigm and the continuing exponential growth that ischaracteristic of the ongoing evolutionary process within a broad area of technology, such ascomputation.Specificparadigms,suchasMoore'sLaw,doultimatelyreachlevelsatwhichexponentialgrowthisnolongerfeasible.Butthegrowthofcomputationsupersedesanyofitsunderlyingparadigmsandisforpresentpurposesanongoingexponential.Inaccordancewiththelawofacceleratingreturns,paradigmshift(alsocalledinnovation)turnstheS-curve of any specific paradigm into a continuing exponential. A new paradigm, such as three-dimensionalcircuits,takesoverwhentheoldparadigmapproachesitsnaturallimit,whichhasalreadyhappened at least four times in the history of computation. In such nonhuman species as apes, themasteryofatool-makingor-usingskillbyeachanimalischaracterizedbyanS-shapedlearningcurvethat ends abruptly; human-created technology, in contrast, has followed an exponential pattern ofgrowthandaccelerationsinceitsinception.DNASequencing,Memory,Communications,theInternet,andMiniaturization

Civilizationadvancesbyextendingthenumberofimportantoperationswhichwecanperformwithoutthinkingaboutthem.

—ALFREDNORTHWHITEHEAD,191142Thingsaremoreliketheyarenowthantheyeverwerebefore.

—DWIGHTD.EISENHOWERThelawofacceleratingreturnsappliestoalloftechnology,indeedtoanyevolutionaryprocess.Itcanbechartedwithremarkableprecisionininformation-basedtechnologiesbecausewehavewell-definedindexes (for example, calculations per second per dollar, or calculations per second per gram) tomeasure them.There are a greatmany examples of the exponential growthA impliedby the lawofaccelerating returns, inareasasvariedaselectronicsofallkinds,DNAsequencing,communications,brain scanning, brain reverse engineering, the size and scope of human knowledge, and the rapidlyshrinkingsizeoftechnology.Thelattertrendisdirectlyrelatedtotheemergenceofnanotechnology. ThefutureGNR(Genetics,Nanotechnology,Robotics)age(seechapter5)willcomeaboutnotfromtheexponentialexplosionofcomputationalonebutratherfromtheinterplayandmyriadsynergiesthatwill result frommultiple intertwined technological advances.Aseverypointon theexponential-growthcurvesunderlyingthispanoplyoftechnologiesrepresentsanintensehumandramaofinnovationandcompetition,wemustconsideritremarkablethatthesechaoticprocessesresultinsuchsmoothandpredictable exponential trends. This is not a coincidence but is an inherent feature of evolutionaryprocesses.Whenthehuman-genomescangotunderwayin1990criticspointedoutthatgiventhespeedwithwhichthegenomecouldthenbescanned,itwouldtakethousandsofyearstofinishtheproject.Yetthefifteen-yearprojectwascompletedslightlyaheadofschedule,withafirstdraftin2003.43ThecostofDNAsequencingcamedownfromabouttendollarsperbasepairin1990toacoupleofpenniesin2004andisrapidlycontinuingtofall(seethefigurebelow).44

TherehasbeensmoothexponentialgrowthintheamountofDNAsequencedatathathasbeencollected (see the figure below).45 A dramatic recent example of this improving capacity was thesequencing of the SARS virus,which took only thirty-one days from the identification of the virus,comparedtomorethanfifteenyearsfortheHIVvirus.46

Ofcourse,weexpecttoseeexponentialgrowthinelectronicmemoriessuchasRAM.Butnotehow the trend on this logarithmic graph (below) proceeds smoothly through different technologyparadigms:vacuumtubetodiscretetransistortointegratedcircuit.47

However,growth in theprice-performanceofmagnetic (disk-drive)memory isnot a resultofMoore'sLaw.This exponential trend reflects the squeezing of data onto amagnetic substrate, ratherthan transistors onto an integrated circuit, a completely different technical challenge pursued bydifferentengineersanddifferentcompanies.48

Exponentialgrowthincommunicationstechnology(measuresforcommunicatinginformation;seethefigurebelow)hasformanyyearsbeenevenmoreexplosivethaninprocessingormemorymeasuresofcomputationandisnolesssignificantinitsimplications.Again,thisprogressioninvolvesfarmorethanjustshrinkingtransistorsonanintegratedcircuitbutincludesacceleratingadvancesinfiberoptics,opticalswitching,electromagnetictechnologies,andotherfactors.49Wearecurrentlymovingawayfromthetangleofwiresinourcitiesandinourdailylivesthroughwireless communication, the power ofwhich is doubling every ten to elevenmonths (see the figurebelow).

ThefiguresbelowshowtheoverallgrowthoftheInternetbasedonthenumberofhosts(Web-servercomputers).Thesetwochartsplotthesamedata,butoneisonalogarithmicaxisandtheotherislinear.Ashasbeendiscussed,whiletechnologyprogressesexponentially,weexperienceitinthelineardomain. From the perspective ofmost observers, nothingwas happening in this area until themid-1990s,whenseeminglyoutofnowheretheWorldWideWebande-mailexplodedintoview.Buttheemergence of the Internet into a worldwide phenomenon was readily predictable by examiningexponentialtrenddataintheearly1980sfromtheARPANET,predecessortotheIntemet.50

Thisfigureshowsthesamedataonalinearscale.51

Inadditiontoservers,theactualdatatrafficontheInternethasalsodoubledeveryyear.52

Toaccommodatethisexponentialgrowth,thedatatransmissionspeedoftheInternetbackbone(asrepresentedbythefastestannouncedbackbonecommunicationchannelsactuallyusedfortheInternet)has itselfgrownexponentially.Notethat in thefigure"InternetBackboneBandwidth"below,wecanactually see theprogressionofS-curves: theacceleration fosteredbyanewparadigm, followedbyaleveling off as the paradigm runs out of steam, followed by renewed acceleration through paradigmshift.53

Anothertrendthatwillhaveprofoundimplicationsforthetwenty-firstcenturyis thepervasivemovement toward miniaturization. The key feature sizes of a broad range of technologies, bothelectronic and mechanical, are decreasing, and at an exponential rate. At present, we are shrinkingtechnologybyafactorofaboutfourperlineardimensionperdecade.Thisminiaturizationisadrivingforce behindMoore's Law, but it's also reflected in the size of all electronic systems—for example,magneticstorage.Wealsoseethisdecreaseinthesizeofmechanicaldevices,asthefigureonthesizeofmechanicaldevicesillustrates.54

Asthesalientfeaturesizeofawiderangeoftechnologiesmovesinexorablyclosertothemulti-nanometerrange(lessthanonehundrednanometers—billionthsofameter),ithasbeenaccompaniedbya rapidlygrowing interest innanotechnology.Nanotechnologysciencecitationshavebeen increasingsignificantlyoverthepastdecade,asnotedinthefigurebelow.55

Weseethesamephenomenoninnanotechnology-relatedpatents(below).56

As we will explore in chapter 5, the genetics (or biotechnology) revolution is bringing theinformationrevolution,withitsexponentiallyincreasingcapacityandprice-performance,tothefieldofbiology. Similarly, the nanotechnology revolution will bring the rapidly increasing mastery ofinformationtomaterialsandmechanicalsystems.Therobotics(or"strongAI")revolutioninvolvesthereverse engineering of the human brain, which means coming to understand human intelligence ininformationtermsandthencombiningtheresultinginsightswithincreasinglypowerfulcomputationalplatforms.Thus,allthreeoftheoverlappingtransformations—genetics,nanotechnology,androbotics—thatwilldominatethefirsthalfofthiscenturyrepresentdifferentfacetsoftheinformationrevolution.

Information,Order,andEvolution:

TheInsightsfromWolframandFredkin'sCellularAutomata:

AsI'vedescribedinthischapter,everyaspectof informationandinformationtechnology isgrowingatanexponentialpace. Inherent inourexpectationofaSingularitytakingplaceinhumanhistoryisthepervasiveimportanceof informationtothefutureofhumanexperience.Wesee informationatevery level of existence.Every formofhumanknowledgeandartistic expression—scientific andengineering ideasanddesigns,literature,music,pictures,movies—canbeexpressedasdigitalinformation. Ourbrainsalsooperatedigitally, throughdiscrete firingsofourneurons.Thewiringofour interneuronalconnectionscanbedigitallydescribed,andthedesignofourbrainsisspecifiedbyasurprisinglysmalldigitalgeneticcode.57Indeed,allofbiologyoperatesthroughlinearsequencesof2-bitDNAbasepairs,which in turncontrol thesequencingofonly twentyamino acids in proteins.Molecules form discrete arrangements of atoms. The carbon atom, with its four positions establishingmolecularconnections, is particularlyadeptat creatinga varietyof three-dimensional shapes,whichaccounts for its central role inbothbiologyandtechnology.Withintheatom,electronstakeondiscreteenergylevels.Othersubatomicparticles,suchasprotons,comprisediscretenumbersofvalencequarks. Althoughthe formulasofquantummechanicsareexpressed in termsofbothcontinuous fieldsanddiscrete levels,wedoknow that

continuous levels can be expressed to any desired degree of accuracy using binary data.58 In fact, quantum mechanics, as the word"quantum"implies,isbaseondiscretevalues.Physicist-mathematicianStephenWolframprovidesextensiveevidencetoshowhowincreasingcomplexitycanoriginatefromauniversethatisatitscoreadeterministic,algorithmicsystem(asystembasedonfixedruleswithpredeterminedoutcomes).InhisbookANewKindofScience,Wolframoffersacomprehensiveanalysisofhowtheprocessesunderlyingamathematicalconstructioncalled"acellularautomaton"havethepotentialtodescribeeverylevelofournaturalworld.59(Acellularautomatonisasimplecomputationalmechanismthat,forexample,changesthecolorofeachcellonagridbasedonthecolorofadjacentnearbycellsaccordingtoatransformationrule.)Inhisview,itisfeasibletoexpressallinformationprocessesintermsofoperationsoncellularautomata,soWolfram'sinsightsbearonseveral key issues related to information and its pervasiveness.Wolfram postulates that the universe itself is a giant cellular-automatoncomputer.Inhishypothesisthereisadigitalbasicforapparentlyanalogphenomena(suchasmotionandtime)andforformulasinphysics,andwecanmodelourunderstandingofphysicsasthesimpletransformationofacellularautomaton.Othershaveproposedthispossibility.RichardFeynmanwonderedaboutitinconsideringtherelationshipofinformationtomatterandenergy.NorbertWienerheraldedafundamentalchangeinfocusfromenergytoinformationinhis1948bookCyberneticandsuggestedthatthe transformationof information,notenergy,was the fundamentalbuildingblockof theuniverse.60Perhaps the first topostulate that theuniverse is being computed on a digital computer was Konrad Zuse in 1967.61 Zuse is best known as the inventor of the first workingprogrammablecomputer,whichhedevelopedfrom1935to1941.Anenthusiasticproponentofaninformation-basedtheoryofphysicswasEdwardFredkin,whointheearly1980sproposeda"newtheoryofphysics"foundedontheideathattheuniverseisultimatelycomposedofsoftware.Weshouldnotthinkofrealityasconsistingofparticlesandforces,accordingtoFredkin,butratherasbitsofdatamodifiedaccordingtocomputationrules.FredkinwasquotesbyRobertWrightinthe1980sassaying,

Therearethreegreatphilosophicalquestions.Whatislife?Whatisconsciousnessandthinkingandmemoryandallofthat?Andhowdoestheuniversework?...[The]"informationviewpoint"encompassesall three....WhatI'msayingisthatatthemostbasiclevelofcomplexityaninformationprocessrunswhatwethinkofasphysics.Atthemuchhigherlevelofcomplexity,life,DNA—youknow,thebiochemicalfunction—arecontrolledbyadigitalinformationprocess.Then,atanotherlevel,outthoughtprocessesarebasicallyinformationprocessing....Ifindthesupportingevidenceformybeliefsintenthousanddifferentplaces....Andtomeit'sjusttotallyoverwhelming.It'slikethere'sananimalIwanttofind.I'vefoundhisfootprints.I'vefoundhisdroppings.I’vefoundthehalf-chewedfood.Ifindpiecesofhisfur,andsoon.Ineverycaseitfitsonekindofanimal,andit'snotlikeanyanimalanyone'severseen.Peoplesay,Where is this animal? I say,Well, hewashere, he's about this big, this that, and the other. And I know athousandthingsabouthim.Idon'thaveinhand,butIknowhe'sthere....WhatIseeissocompellingthatitcan'tbeacreatureofmyimagination.62

IncommentingonFredkin'stheoryofdigitalphysics,Wrightwrites,

Fredkin...istalkingaboutaninterestingcharacteristicofsomecomputerprograms,includingmanycellularautomata:thereisnoshortcuttofindingoutwhattheywillleadto.This,indeed,isabasicdifferencebetweenthe"analytical"approachassociatedwithtraditionalmathematics, includingdifferentialequations,and the "computational"approachassociatedwithalgorithms.Youcanpredictafuturestateofasystemsusceptibletotheanalyticapproachwithoutfiguringoutwhatstatesitwilloccupybetweennowandthen,butinthecaseofmanycellularautomata,youmustgothroughalltheintermediatestatestofindoutwhattheendwillbelike:thereisnowaytoknowthefutureexcepttowatchitunfold....Fredkinexplains:"Thereisnowaytoknowtheanswertosomequestionany faster thanwhat'sgoingon. "...Fredkinbelieves that theuniverse isvery literallyacomputerand that it isbeingusedbysomeone,orsomething,tosolveaproblem.Itsoundslikeagood-news/bad-newsjoke:thegoodnewsisthatourliveshavepurpose;thebasnewsisthattheirpurposeistohelpsomeremotehackerestimatepitoninejilliondecimalplaces.63

Fredkinwenton toshow thatalthoughenergy isneeded for informationstorageand retrieval,wecanarbitrarily reduce theenergyrequiredtoperformanyparticularexampleofinformationprocessing.andthatthisoperationhasnolowerlimit.64Thatimpliesthatinformationrather thanmatterandenergymaybe regardedas themore fundamental reality.65 Iwill return toFredkin's insight regarding theextremelower limitofenergy required forcomputationandcommunication inchapter3,since itpertains to theultimatepowerof intelligence in theuniverse.Wolframbuildshistheoryprimarilyonasingle,unifiedinsight.ThediscoverythathassoexcitedWolframisasimplerulehecallscellularautomatarules110anditsbehavior.(Therearesomeotherinterestingautomatarules,butrule110makesthepointwellenough.)MostofWolfram'sanalysesdealwiththesimplestpossiblecellularautomata,specificallythosethat involvejustaone-dimensional lineofcells,twopossiblecolors (blackandwhite),and rulesbasedonlyon the two immediatelyadjacentcells.Foreach transformation, the color of a celldependsonlyonitsownpreviouscolorandthatofthecellontheleftandthecellontheright.Thus,thereareeightpossibleinputsituations(thatis,threecombinationsoftwocolors).Eachrulesmapsallcombinationsoftheseeightinputsituationstoanoutput(blackorwhite).Sothereare28(256)possiblerulesforsuchaone-dimension,two-color,adjacent-cellautomaton.Halfthe256possiblerulesmapontotheotherhalfbecauseof left-right-symmetry.Wecanmaphalfof themagainbecauseofblack-whiteequivalence,soweare leftwith64 rule types.Wolframillustratestheactionoftheseautomatawithtwo-dimensionalpatternsinwhicheachline(alongthey-axis)representsasubsequentgenerationofapplyingtheruletoeachcellinthatline.Mostoftherulesaredegenerate,meaningtheycreaterepetitivepatternsofnointerest,suchascellsofasinglecolor,oracheckerboardpattern.Wolframcallstheserulesclass1automata.Somerulesproducearbitrarilyspacedstreaksthatremainstable,andWolframclassifiestheseasbelongingtoclass2.Class3rulesareabitmoreinteresting,inthatrecognizablefeatures(suchastriangles)appearintheresultingpatterninaessentiallyrandomorder.However,itwasclass4automatathatgaverisetothe"aha"experiencethatresultedinWolfram'sdevotingadecadetothetopic.Theclass4automata,ofwhichrule110isthequintessentialexample,producesurprisinglycomplexpatternsthatdonotrepeatthemselves.Wesee in themartifactssuchas linesat variousangles,aggregationsof triangles,andother interestingconfigurations.Theresultingpattern,however,isneitherregularnorcompletelyrandom;itappearstohavesomeorderbutisneverpredictable.

Whyisthisimportantorinteresting?Keepinmindthatwebeganwiththesimplestpossiblestartingpoint:asingleblackcell.Theprocessinvolvesrepetitiveapplicationofaverysimplerule.66Fromsucharepetitiveanddeterministicprocess,onewouldexpect repetitiveandpredictablebehavior.Thereare twosurprising resultshere.One is that theresultsproduceapparent randomness.However, the resultsaremore interesting thanpure randomness,which itselfwouldbecomeboringveryquickly.Therearediscernibleandinterestingfeaturesinthedesignsproduced,sothatthepattern has someorder andapparent intelligence.Wolfram include a number of example of these images,many ofwhichareratherlovelytolookat.Wolframmakesthefollowingpointrepeatedly:"Wheneveraphenomenonisencounteredthatseemscomplexitistakenalmostforgrantedthatthephenomenonmustbetheresultofsomeunderlyingmechanismthatisitselfcomplex.Butmydiscoverythatsimpleprogramscanproducegreatcomplexitymakesitclearthatthisisnotinfactcorrect."67Idofindthebehaviorofrule110ratherdelightful.Furthermore,theideathatacompletelydeterministicprocesscanproduceresultsthatarecompletelyunpredictableisofgreatimportance,asitprovidesanexplanationforhowtheworld can be inherently unpredictable while still based on fully deterministic rules.68 However, I am not entirelysurprised by the idea that simplemechanisms can produce results more complicated than their starting conditions.We'veseenthisphenomenoninfractals,chaosandcomplexitytheory,andself-organizingsystems(suchasneuralnetsandMarkov models), which start with simple networks but organize themselves to produce apparently intelligentbehavior.Atadifferentlevel,weseeit inthehumanbrainitself,whichstartswithonlyaboutthirtytoonehundredmillionbytesofspecificationinthecompressedgenomeyetendsupwithacomplexitythatisaboutabilliontimesgreater.69Itisalsonotsurprisingthatadeterministicprocesscanproduceapparentlyrandomresults.Wehavehasrandom-numbergenerators (forexample, the "randomize" function inWolfram's programMathematics) that use deterministicprocessestoproducesequencesthatpassstatistical testsforrandomness.Theseprogramsdatebacktotheearliestdaysofcomputersoftware,suchasthefirstversionofFortran.However,Wolframdoesprovideathoroughtheoreticalfoundationforthisobservation.Wolframgoesontodescribehowsimplecomputationalmechanismscanexistinnatureatdifferentlevels,andheshowsthatthesesimpleanddeterministicmechanismscanproduceallofthecomplexitythatweseeandexperience.Heprovidesmyriadexamples,suchas thepleasingdesignsofpigmentationonanimals, theshapeandmarkingsonshells,andpatternsofturbulence(suchas thebehaviorofsmoke in theair).Hemakes thepoint thatcomputation isessentially simple and ubiquitous. The repetitive application of simple computational transformations, according toWolfram,isthetruesourceofcomplexityintheworld.Myownviewisthatthisisonlypartycorrect.IagreewithWolframthatcomputationisallaroundus,andthatsomeof the patternswe see are created by the equivalent of cellular automata. But a key issue to ask is this:Just howcomplexaretheresultsofclassautomata?

Wolframeffectivelysidesteps the issueofdegreesofcomplexity. Iagree thatadegeneratepatternsuchasachessboard has no complexity. Wolfram also acknowledges that mere randomness does not represent complexityeither,because pure randomness becomes predictable in its pure lack of predictability. It is true that the interestingfeaturesofclass4automataareneitherrepeatingnorpurelyrandom,soIwouldagreethattheyaremorecomplexthantheresultsproducedbyotherclassesofautomata.However,thereisnonethelessadistinctlimittothecomplexityproducedbyclass4automata.Themanyimagesofsuch automata in Wolfram's book all have a similar look to them, and although they are nonrepeating, they areinteresting (and intelligent)only toadegree.Moreover, theydonotcontinue toevolve intoanythingcomplex,nordotheydevelopnew typesof features.One could run these for trillions or even trillions of trillions of iterationsand theimagewouldremainat thesamelimited levelcomplexity.Theydonotevolve into,say, insectsorhumansorChopinpreludesor anything else thatwemight consider of a higher order of complexity than the streaksand interminglingtrianglesdisplayedintheseimages.Complexityisacontinuum.HereIdefine"order"as"informationthatfitsapurpose."70Acompletelypredictableprocesshaszeroorder.Ahighlevelofinformationalonedoesnotnecessarilyimplyahighlevelofordereither.Aphonebookhasa lotof information,buttheleveloforderof that information isquite low.Arandomsequence isessentiallypureinformation(sinceit isnotpredictable)buthasnoorder.Theoutputofclass4automatadoespossessacertainleveloforder,anditdoessurvivelikeotherpersistingpatterns.Butthepatternsrepresentedbyahumanbeinghasafarhigherleveloforder,andofcomplexity.Humanbeingsfulfillahighlydemandingpurpose:theysurviveinachallengingecologicalniche.Humanbeingsrepresentanextremelyintricateandelaboratehierarchyofotherpatterns.Wolframregardsanypatternsthatcombinesomerecognizablefeaturesandunpredictableelementstobeeffectivelyequivalenttoonanother.Buthedoesnotshowhowaclass4automatoncaneverincreaseitcomplexity,letalonebecomeapatternascomplexasahumanbeing.Thereisamissinglinkhere,onethatwouldaccountforhowonegetsfromtheinterestingbutultimatelyroutinepatternsofacellularautomatontothecomplexityofpersistingstructuresthatdemonstratehigherlevelsofintelligence.Forexample,theseclass4patternsarenotcapableofsolvinginterestingproblems,andnoamountofiterationmovesthemclosertodoingso.Wolframwouldcounterthanarule110automatoncouldbeusedasa"universalcomputer."71However, by itself, a universal computer is not capable of solving intelligent programs without what I would call"software."Itisthecomplexityofthesoftwarethatrunsonauniversalcomputerthatispreciselytheissue.Onemightpointoutthatclass4patternsresultfromthesimplestpossibleautomata(one-dimensional,two-color,two-neighbor rules). What happens if we increase the dimensionality—for example, go to multiple colors or evengeneralize thesediscretecellularautomata tocontinuous function?Wolframaddressall of this quite thoroughly.Theresultsproducedfrommorecomplexautomataareessentiallythesameasthoseoftheverysimpleones.Wegetthesamesortsofinterestingbyultimatelyquitelimitedpatterns.Wolframmakestheintriguingpointthatwedonotneedtousemorecomplexrulestogetcomplexityintheendresult.ButIwouldmaketheconversepointthatweareunabletoincreasethecomplexityof theendresultsthrougheithermorecomplexrulesorfurther iteration.Socellularautomatagetusonlysofar.

CanWeEvolveArtificialIntelligencefromSimpleRules?SohowdowegetfromtheseinterestingbutlimitedpatternstothoseofinsectsorChopininterludes?Oneconceptweneedintoconsiderationisconflict—thatis,evolution.Ifweaddanothersimpleconcept—anevolutionaryalgorithm—tothatofWolfram'ssimplecellularautomata,westarttogetfarmoreexcitingandmoreintelligentresults.Wolframsaythattheclass4automataandanevolutionaryalgorithmare"computationallyequivalent."ButthatistrueonlyonwhatIconsiderthe"hardware"level.Onthesoftwarelevel,theotherofthepatternsproducedareclearlydifferentanofadifferentorderofcomplexityandusefulness.Anevolutionaryalgorithmcanstartwithrandomlygeneratedpotentialsolutionstoaproblem,whichareencodedinadigitalgeneticcode.Wethenhavethesolutionscompetewithoneanotherinasimulatedevolutionarybattle.Thebettersolutionssurviveandprocreateinasimulatedsexualreproductioninwhichoffspringsolutionsarecreated,drawingtheirgeneticcode(encodedsolutions)fromtwoparents.Wecanalsointroducearateofgeneticmutation.Varioushigh-levelparametersofthisprocess,suchastherateofmutation,therateofoffspring,andsoon,areappropriatelycalled"Godparameters,"anditisthejoboftheengineerdesigningtheevolutionaryalgorithmtosetthemtoreasonablyoptimalvalues.Theprocessisrunformanythousandsofgenerationsofsimulatedevolution,andattheendoftheprocessoneislikelytofindsolutionsthatareofadistinctlyhigherorderthanthestartingones.Theresultsoftheseevolutionary(sometimescalledgenetic)algorithmscanbeelegant,beautiful,andintelligentsolutionstocomplexproblems.Theyhavebeenused,forexample,tocreateartisticdesignsanddesignsforartificiallife-forms,aswellastoexecuteawiderangeofpracticalassignmentssuchasdesigningjetengines.Geneticalgorithmsareoneapproachto"narrow"artificialintelligence—thatis,creatingsystemsthatcanperformparticularfunctionsthatusedtorequiretheapplicationofhumanintelligence.Butsomethingisstillmissing.Althoughgeneticalgorithmsareausefultoolinsolvingspecificproblems,theyhaveneverachievedanythingresembling"strongAI"—thatis,aptituderesemblingthebroad,deep,andsubtlefeaturesofhumanintelligence,particularlyitspowerofpatternrecognitionandcommandlanguage.Istheproblemthatwearenotrunningtheevolutionaryalgorithmslongenough?Afterall,humansevolvedthroughaprocessthattookbillionsofyears.Perhapswecannotre-createthatprocesswithjustafewdaysorweeksofcomputersimulation.Thiswon'twork,

however,becauseconventionalgeneticalgorithmsreachanasymptoteintheirlevelofperformance,sorunningthemforalongerperiodoftimewon'thelp.Athirdlevel(beyondtheabilityofcellularprocessestoproduceapparentrandomnessandgeneticalgorithmstoproducefocusedintelligentsolutions)istoperformevolutiononmultiplelevels.Conventionalgeneticalgorithmsallowevolutiononlywithintheconfinesofanarrowproblemandasinglemeansofevolution.Thegeneticcodeitselfneedstoevolve;therulesofevolutionneedtoevolve.Naturedidnotstaywithasinglechromosome,forexample.Therehavebeenmanylevelsofindirectionincorporatedinthenaturalevolutionaryprocess.Andwerequireacomplexenvironmentinwhichtheevolutiontakesplace.TobuildstrongAIwewillhavetheopportunitytoshort-circuitthisprocess,however,byreverse-engineeringthehumanbrain,aprojectwellunderway,therebybenefitingfromtheevolutionaryprocessthathasalreadytakenplace.Wewillbeapplyingevolutionaryalgorithmswithinthesesolutionsjustasthehumanbraindoes.Forexample,thefetalwiringisinitiallyrandomwithinconstraintsspecifiedinthegenomeinatleastsomeregions.Recentresearchshowsthatareashavingtodowithlearningundergomorechange,whereasstructureshavingtodowithsensoryprocessingexperiencelesschangeafterbirth.72Wolframmakethevalidpointthatcertain(indeed,most)computationalprocessesarenotpredictable.Inotherwords,wecannotpredictfuturestatewithoutrunningtheentireprocess,Iagreewithhimthatwecanknowtheanswerinadvanceonlyifsomehowwecansimulateaprocessatafasterspeed.Giventhattheuniverserunsatthefastestspeeditcanrun,thereisusuallynowaytoshort-circuittheprocess.However,wehavethebenefitsofthebillionsofyearsofevolutionthathavealreadytakenplace,whichareresponsibleforthegreatlyincreasedorderofcomplexityinthenaturalworld.Wecannowbenefitfromitbyusingoutevolvedtoolstoreverseengineertheproductsofbiologicalevolution(mostimportantly,thehumanbrain).Yes,itistruethatsomephenomenainnaturethatmayappearcomplexatsomelevelaremerelytheresultsofsimpleunderlyingcomputationalmechanismsthatareessentiallycellularautomataatwork.Theinterestingpatternoftrianglesona"tentoliuve"(citedextensivelybyWolfram)ortheintricateandvariedpatternsofasnowflakearegoodexample.Idon'tthinkthisisanewobservation,inthatwe'vealwaysregardedthedesignofsnowflakestoderivefromasimplemolecularcomputation-likebuildingprocess.However,Wolframdoesprovideuswithacompellingtheoreticalfoundationforexpressingtheseprocessesandtheirresultingpatterns.Butthereismoretobiologythanclass4patterns.AnotherimportantthesebyWolframliesinhisthoroughtreatmentofcomputationasasimpleandubiquitousphenomenon.Ofcourse,we'veknownformorethanacenturythatcomputationisinherentlysimple:wecanbuildanypossiblelevelofcomplexityfromafoundationofthesimplestpossiblemanipulationsofinformation.Forexample,CharlesBabbage'slate-nineteenth-centurymechanicalcomputer(whichneverran)providedonlyahandfulofoperationcodes,yetprovided(withinitsmemorycapacityandspeed)thesamekindsoftransformationsthatmoderncomputersdo.ThecomplexityofBabbage'sinventionstemmedonlyfromthedetailsofitsdesign,whichindeedprovedtoodifficultforBabbagetoimplementusingthetechnologyavailabletohim.TheTuringmachine,AlanTuring'stheoreticalconceptionofauniversalcomputerin1950,providesonlysevenverybasiccommands,yetcanbeorganizedtoperformanypossiblecomputation.73Theexistenceofa"universalTuringmachine,"whichcansimulateanypossibleTuringmachinethatisdescribedonitstapememory,isafurtherdemonstrationoftheuniversalityandsimplicityofinformation.74InTheAgeofIntelligentMachines,Ishowedhowanycomputercouldbeconstructedfrom"asuitablenumberof[a]verysimpledevice,"namely,the"nor"gate.75ThisisnotexactlythesamedemonstrationasauniversalTuringmachine,butitdoesdemonstratethatanycomputationcanbeperformedbyacascadeofthisverysimpledevice(whichissimplerthanrule110),giventherightsoftware(whichwouldincludetheconnectiondescriptionofthenorgates).76Althoughweneedadditionalconceptstodescribeanevolutionaryprocessthatcreateintelligentsolutionstoproblems,Wolfram'sdemonstrationofthesimplicityanubiquityofcomputationisanimportantcontributioninourunderstandingofthefundamentalsignificanceofinformationintheworld.MOLLY2004:You'vegotmachinesevolvingatanacceleratingpace.Whatabouthumans?RAY:Youmeanbiologicalhumans?MOLLY2004:Yes.CHARLESDARWIN:Biologicalevolutionispresumablycontinuing,isitnot?RAY:Well,biologyatthislevelisevolvingsoslowlythatithardlycounts.Imentionedthatevolution

worksthroughindirection.Itturnsoutthattheolderparadigmssuchasbiologicalevolutiondocontinuebutattheiroldspeed,sotheyareeclipsedbythenewparadigms.Biologicalevolutionforanimalsascomplexashumanstakestensofthousandsofyearstomakenoticeable,albeitstillsmall, differences. The entire history of human cultural and technological evolution has takenplaceonthattimescale.Yetwearenowpoisedtoascendbeyondthefragileandslowcreationsofbiologicalevolutioninamereseveraldecades.Currentprogressisonascalethatisathousand

toamilliontimesfasterthanbiologicalevolution.NEDLUDD:Whatifnoteveryonewantstogoalongwiththis?RAY:Iwouldn'texpecttheywould.Therearealwaysearlyandlateadopters.There'salwaysaleading

edge and a trailing edge to technology or to any evolutionary change. We still have peoplepushingplows,butthathasn'tsloweddowntheadoptionofcellphones,telecommunications,theInternet,biotechnology,andsoon.However,thelaggingedgedoesultimatelycatchup.WehavesocietiesinAsiathatjumpedfromagrarianeconomiestoinformationeconomies,withoutgoingthroughindustrialization.77

NED:Thatmaybeso,butthedigitaldivideisgettingworse.RAY:Iknowthatpeoplekeepsayingthat,buthowcanthatpossiblybetrue?Thenumberofhumansis

growing only very slowly. The number of digitally connected humans, no matter how youmeasureit,isgrowingrapidly.Alargerandlargerfractionoftheworld'spopulationisgettingelectroniccommunicatorsandleapfroggingourprimitivephone-wiringsystembyhookinguptotheInternetwirelessly,sothedigitaldivideisrapidlydiminishing,notgrowing.

MOLLY2004:Istillfeelthatthehave/havenotissuedoesn'tgetenoughattention.There'smorewecando.

RAY:Indeed,buttheoverriding,impersonalforcesofthelawofacceleratingreturnsarenonethelessmoving in the right direction. Consider that technology in a particular area starts outunaffordable and notworking verywell. Then it becomesmerely expensive andworks a littlebetter. The next step is the product becomes inexpensive and works really well. Finally, thetechnology becomes virtually free and works great. It wasn't long ago that when you sawsomeoneusingaportablephoneinamovie,heorshewasamemberofthepowerelite,becauseonlythewealthycouldaffordportablephones.Orasamorepoignantexample,considerdrugsforAIDS.Theystartedoutnotworkingverywellandcostingmorethantenthousanddollarsperyearperpatient.Nowtheyworkalotbetterandaredowntoseveralhundreddollarsperyearinpoor countries.78 Unfortunately with regard to AIDS, we're not yet at the working great andcostingalmostnothingstage.TheworldisbeginningtotakesomewhatmoreeffectiveactiononAIDS,butithasbeentragicthatmorehasnotbeendone.Millionsoflives,mostinAfrica,havebeenlostasaresult.Buttheeffectofthelawofacceleratingreturnsisnonethelessmovingintherightdirection.And the timegapbetween leadingand laggingedge is itselfcontracting.RightnowIestimatethislagataboutadecade.Inadecade,itwillbedowntoabouthalfadecade.

TheSingularityasEconomicImperative

The reasonable man adapts himself to the world; the unreasonable one persists intrying to adapt the world to himself. Therefore all progress depends on theunreasonableman.

—GEORGE BERNARD SHAW, "MAXIMS FORREVOLUTIONISTS",MANANDSUPERMAN,1903

Allprogressisbaseduponauniversalinnatedesireonthepartofeveryorganismtolivebeyonditsincome.

—SAMUELBUTLER,NOTEBOOKS,1912If Iwere just setting out today tomake that drive to theWestCoast to start a newbusiness,Iwouldbelookingatbiotechnologyandnanotechnology.

—JEFFBEZOS,FOUNDERANDCEOOFAMAZON.COM

GetEightyTrillionDollars—LimitedTimeOnlyYouwill get eighty trillion dollars just by reading this section and understanding what it says. Forcompletedetails,seebelow.(It'struethatanauthorwilldojustaboutanythingtokeepyourattention,butI'mseriousaboutthisstatement.UntilIreturntoafurtherexplanation,however,doreadthefirstsentenceofthisparagraphcarefully.) The lawofacceleratingreturns is fundamentallyaneconomic theory.Contemporaryeconomictheoryandpolicyarebasedonoutdatedmodels that emphasize energycosts, commodityprices, andcapital investment in plant and equipment as key driving factors, while largely overlookingcomputationalcapacity,memory,bandwidth, thesizeof technology, intellectualproperty,knowledge,andotherincreasinglyvital(andincreasinglyincreasing)constituentsthataredrivingtheeconomy. It's the economic imperative of a competitive marketplace that is the primary force drivingtechnologyforwardandfuelingthelawofacceleratingreturns.Inturn,thelawofacceleratingreturnsistransforming economic relationships.Economic imperative is the equivalent of survival in biologicalevolution.Wearemovingtowardmoreintelligentandsmallermachinesastheresultofmyriadsmalladvances,eachwithitsownparticulareconomicjustification.Machinesthatcanmorepreciselycarryout theirmissions have increased value,which explainswhy they are being built. There are tens ofthousandsofprojectsthatareadvancingthevariousaspectsofthelawofacceleratingreturnsindiverseincrementalways.Regardlessofnear-termbusinesscycles,supportfor"hightech"inthebusinesscommunity,andinparticular for software development, has grown enormously. When I started my optical characterrecognition (OCR) and speech-synthesis company (Kurzweil Computer Products) in 1974, high-techventuredealsintheUnitedStatestotaledlessthanthirtymilliondollars(in1974dollars).Evenduringthe recent high-tech recession (2000–2003), the figurewas almost one hundred times greater.79 Wewouldhavetorepealcapitalismandeveryvestigeofeconomiccompetitiontostopthisprogression.Itisimportanttopointoutthatweareprogressingtowardthe"new"knowledge-basedeconomyexponentiallybutnonethelessgradually.80Whentheso-calledneweconomydidnottransformbusinessmodelsovernight,manyobserverswerequicktodismisstheideaasinherentlyflawed.Itwillbeanothercouple of decades before knowledge dominates the economy, but it will represent a profoundtransformationwhenithappens.WesawthesamephenomenonintheInternetandtelecommunicationsboom-and-bustcycles.Thebooms were fueled by the valid insight that the Internet and distributed electronic communicationrepresented fundamental transformations.Butwhen these transformationsdidnotoccur inwhatwereunrealistictimeframes,morethantwotrilliondollarsofmarketcapitalizationvanished.AsIpointoutbelow, the actual adoption of these technologies progressed smoothlywith no indicationof boomorbust.VirtuallyalloftheeconomicmodelstaughtineconomicsclassesandusedbytheFederalReserveBoard to set monetary policy, by government agencies to set economic policy, and by economicforecastersofallkindsarefundamentallyflawedintheirviewoflong-termtrends.That'sbecausetheyarebasedonthe"intuitivelinear"viewofhistory(theassumptionthatthepaceofchangewillcontinueat the current rate) rather than the historically based exponential view. The reason that these linearmodelsappeartoworkforawhileisthesamereasonmostpeopleadopttheintuitivelinearviewinthefirstplace:exponential trendsappear tobe linearwhenviewedandexperienced forabriefperiodoftime,particularlyintheearlystagesofanexponentialtrend,whennotmuchishappening.Butoncethe"kneeofthecurve"isachievedandtheexponentialgrowthexplodes,thelinearmodelsbreakdown.Asthisbookisbeingwritten,thecountryisdebatingchangingtheSocialSecurityprogrambasedonprojectionsthatgooutto2042,approximatelythetimeframeI'veestimatedfortheSingularity(seethenextchapter).Thiseconomicpolicyreviewisunusualintheverylongtimeframesinvolved.Thepredictions are based on linear models of longevity increases and economic growth that are highlyunrealistic. On the one hand, longevity increases will vastly outstrip the government's modest

expectations. On the other hand, people won't be seeking to retire at sixty-five when they have thebodies and brains of thirty-year-olds. Most important, the economic growth from the "GNR"technologies(seechapter5)willgreatlyoutstripthe1.7percentperyearestimatesbeingused(whichunderstatebyhalfevenourexperienceoverthepastfifteenyears).Theexponentialtrendsunderlyingproductivitygrowtharejustbeginningthisexplosivephase.TheU.S. real gross domestic product has grown exponentially, fostered by improving productivity fromtechnology,asseeninthefigurebelow.81

SomecriticscreditpopulationgrowthwiththeexponentialgrowthinGDP,butweseethesametrendonaper-capitabasis(seethefigurebelow).82

Note that theunderlyingexponentialgrowth in theeconomyisafarmorepowerful force thanperiodic recessions. Most important, recessions, including depressions, represent only temporarydeviations from theunderlyingcurve.Even theGreatDepression representsonlyaminorblip in thecontextoftheunderlyingpatternofgrowth.Ineachcase,theeconomyendsupexactlywhereitwouldhavebeenhadtherecession/depressionneveroccurred. Theworldeconomyiscontinuingtoaccelerate.TheWorldBankreleasedareport inlate2004indicating that the past year had been more prosperous than any year in history with worldwideeconomicgrowthof4percent.83Moreover, the highest rateswere in the developing countries:morethan6percent.EvenomittingChinaandIndia,theratewasover5percent.IntheEastAsianandPacificregion,thenumberofpeoplelivinginextremepovertywentfrom470millionin1990to270millionin2001,andisprojectedbytheWorldBanktobeunder20millionby2015.Otherregionsareshowingsimilar,althoughsomewhatlessdramatic,economicgrowth.Productivity(economicoutputperworker)hasalsobeengrowingexponentially.Thesestatisticsareinfactgreatlyunderstatedbecausetheydonotfullyreflectsignificantimprovementsinthequalityand features of products and services. It is not the case that "a car is a car"; there have beenmajorupgradesinsafety,reliability,andfeatures.Certainly,onethousanddollarsofcomputationtodayisfarmorepowerful thanonethousanddollarsofcomputationtenyearsago(byafactorofmorethanonethousand). There are many other such examples. Pharmaceutical drugs are increasingly effectivebecause they are now being designed to precisely carry out modifications to the exact metabolicpathwaysunderlyingdiseaseandagingprocesseswithminimalsideeffects(notethatthevastmajority

of drugs on themarket today still reflect the old paradigm; see chapter 5). Products ordered in fiveminuteson theWebanddelivered toyourdoorareworthmore thanproducts thatyouhave to fetchyourself.Clothescustom-manufacturedforyouruniquebodyareworthmorethanclothesyouhappentofindonastorerack.Thesesortsof improvementsare takingplace inmostproductcategories,andnoneofthemisreflectedintheproductivitystatistics. The statistical methods underlying productivity measurements tend to factor out gains byessentiallyconcludingthatwestillgetonlyonedollarofproductsandservicesforadollar,despitethefactthatwegetmuchmoreforthatdollar.(Computersareanextremeexampleofthisphenomenon,butit is pervasive.)University ofChicago professor PeteKlenow andUniversity ofRochester professorMarkBilsestimatethatthevalueinconstantdollarsofexistinggoodshasbeenincreasingat1.5percentperyearforthepasttwentyyearsbecauseofqualitativeimprovements.84Thisstilldoesnotaccountforthe introduction of entirely new products and product categories (for example, cell phones, pagers,pocket computers, downloaded songs, and software programs). It does not consider the burgeoningvalueoftheWebitself.Howdowevaluetheavailabilityoffreeresourcessuchasonlineencyclopediasandsearchenginesthatincreasinglyprovideeffectivegatewaystohumanknowledge?TheBureauofLaborStatistics,whichisresponsiblefortheinflationstatistics,usesamodelthatincorporatesanestimateofqualitygrowthofonly0.5percentperyear.85IfweuseKlenowandBils'sconservativeestimate, this reflectsasystematicunderestimateofquality improvementanda resultingoverestimateofinflationbyatleast1percentperyear.Andthatstilldoesnotaccountfornewproductcategories. Despite theseweaknesses in theproductivitystatisticalmethods,gains inproductivityarenowactuallyreachingthesteeppartoftheexponentialcurve.Laborproductivitygrewat1.6percentperyearuntil 1994, then rose at 2.4 percent per year, and is nowgrowing evenmore rapidly.Manufacturingproductivityinoutputperhourgrewat4.4percentannuallyfrom1995to1999,durablesmanufacturingat6.5percentperyear.Inthefirstquarterof2004,theseasonallyadjustedannualrateofproductivitychangewas4.6percentinthebusinesssectorand5.9percentindurablegoodsmanufacturing.86 Weseesmoothexponentialgrowthinthevalueproducedbyanhouroflaboroverthelasthalfcentury(seethefigurebelow).Again,thistrenddoesnottakeintoaccountthevastlygreatervalueofadollar'spower inpurchasing information technologies(whichhasbeendoublingaboutonceayear inoverallprice-performance).87

Deflation...aBadThing?In1846webelievetherewasnotasinglegarmentinourcountrysewedbymachinery;inthatyearthefirstAmericanpatentofasewingmachinewasissued.Atthepresentmomentthousandsarewearingclotheswhichhavebeenstitchedbyironfingers,withadelicacyrivalingthatofaCashmeremaiden.

—SCIENTIFICAMERICAN,1853Asthisbookisbeingwritten,aworryofmanymainstreameconomistsonboththepoliticalrightandtheleftisdeflation.Onthefaceofit,havingyourmoneygofurtherwouldappeartobeagoodthing.Theeconomists'concernisthatifconsumerscanbuywhattheyneedandwantwithfewerdollars,theeconomywillshrink(asmeasuredindollars).Thisignores,however,theinherentlyinsatiableneedsanddesires of human consumers. The revenues of the semiconductor industry,which "suffers" 40 to 50percentdeflationperyear,havenonethelessgrownby17percenteachyearoverthepasthalfcentury.88Since the economy is in fact expanding, this theoretical implication of deflation should not causeconcern. The1990sandearly2000shave seen themostpowerfuldeflationary forces inhistory,whichexplains why we are not seeing significant rates of inflation. Yes, it's true that historically low

unemployment,highassetvalues,economicgrowth,andothersuchfactorsare inflationary,but thesefactors are offset by the exponential trends in the price-performance of all information-basedtechnologies: computation, memory, communications, biotechnology, miniaturization, and even theoverall rate of technical progress. These technologies deeply affect all industries. We are alsoundergoingmassivedisintermediation in thechannelsofdistribution through theWebandothernewcommunicationtechnologies,aswellasescalatingefficienciesinoperationsandadministration.Sincetheinformationindustryisbecomingincreasinglyinfluentialinallsectorsoftheeconomy,weareseeingtheincreasingimpactoftheITindustry'sextraordinarydeflationrates.DeflationduringtheGreatDepressioninthe1930swasduetoacollapseofconsumerconfidenceandacollapseofthemoney supply.Today'sdeflation is a completelydifferentphenomenon, causedby rapidly increasingproductivityandtheincreasingpervasivenessofinformationinallitsforms. Allof the technology trendcharts in thischapter representmassivedeflation.Therearemanyexamplesof the impactof theseescalatingefficiencies.BPAmoco'scost for findingoil in2000waslessthanonedollarperbarrel,downfromnearlytendollarsin1991.ProcessinganInternettransactioncostsabankonepenny,comparedtomorethanonedollarusingateller. It is important to point out that a key implication of nanotechnology is that itwill bring theeconomics of software to hardware—that is, to physical products. Software prices are deflating evenmorequicklythanthoseofhardware(seethefigurebelow).

Theimpactofdistributedandintelligentcommunicationshasbeenfeltperhapsmostintenselyintheworldofbusiness.DespitedramaticmoodswingsonWallStreet,theextraordinaryvaluesascribedtoso-callede-companiesduringthe1990sboomerareflectedavalidperception:thebusinessmodelsthat have sustained businesses for decades are in the early phases of a radical transformation. Newmodelsbasedondirectpersonalizedcommunicationwith thecustomerwill transformevery industry,resulting in massive disintermediation of the middle layers that have traditionally separated thecustomerfromtheultimatesourceofproductsandservices.Thereis,however,apacetoallrevolutions,andtheinvestmentsandstockmarketvaluationsinthisareaexpandedwaybeyondtheearlyphasesofthiseconomicS-curve.Theboom-and-bustcycleintheseinformationtechnologieswasstrictlyacapital-markets(stock-value)phenomenon.Neitherboomnorbust isapparent in theactualbusiness-to-consumer(B2C)andbusiness-to-business(B2B)data(seethefigureonthenextpage).ActualB2Crevenuesgrewsmoothlyfrom$1.8billionin1997to$70billionin2002.B2Bhadsimilarlysmoothgrowthfrom$56billionin1999 to $482 billion in 2002.90 In 2004 it is approaching $1 trillion.We certainly do not see anyevidence of business cycles in the actual price-performance of the underlying technologies, as Idiscussedextensivelyabove.

Expanding access to knowledge is also changing power relationships. Patients increasinglyapproachvisits totheirphysicianarmedwithasophisticatedunderstandingoftheirmedicalconditionand their options. Consumers of virtually everything from toasters, cars, and homes to banking andinsurance are now using automated software agents to quickly identify the right choices with theoptimal features and prices.Web services such as eBay are rapidly connecting buyers and sellers inunprecedentedways.Thewishesanddesiresofcustomers,oftenunknowneventothemselves,arerapidlybecomingthedrivingforceinbusinessrelationships.Well-connectedclothesshoppers,forexample,arenotgoingtobesatisfiedformuchlongerwithsettlingforwhateveritemshappentobelefthangingontherackoftheir local store. Instead, they will select just the right materials and styles by viewing how manypossiblecombinationslookonathree-dimensionalimageoftheirownbody(basedonadetailedbodyscan),andthenhavingthechoicescustom-manufactured. The current disadvantages ofWeb-based commerce (for example, limitations in the ability todirectly interactwith products and the frequent frustrations of interactingwith inflexiblemenus andformsinsteadofhumanpersonnel)willgraduallydissolveasthetrendsmoverobustlyinfavoroftheelectronicworld.Bytheendofthisdecade,computerswilldisappearasdistinctphysicalobjects,withdisplaysbuiltinoureyeglasses,andelectronicswoveninourclothing,providingfull-immersionvisualvirtualreality.Thus,"goingtoaWebsite"willmeanenteringavirtual-realityenvironment—atleastforthevisualandauditorysenses—wherewecandirectlyinteractwithproductsandpeople,bothrealand

simulated.Althoughthesimulatedpeoplewillnotbeup tohumanstandards—at leastnotby2009—theywillbequitesatisfactoryassalesagents,reservationclerks,andresearchassistants.Haptic(tactile)interfaceswillenableustotouchproductsandpeople.Itisdifficulttoidentifyanylastingadvantageoftheoldbrick-and-mortarworld thatwillnotultimatelybeovercomeby the rich interactive interfacesthataresoontocome. Thesedevelopmentswillhavesignificant implications for the real-estate industry.Theneed tocongregateworkersinofficeswillgraduallydiminish.Fromtheexperienceofmyowncompanies,weare already able to effectively organize geographically disparate teams, something thatwas farmoredifficultadecadeago.The full-immersionvisual-auditoryvirtual-realityenvironments,whichwillbeubiquitous during the second decade of this century,will hasten the trend toward people living andworkingwherevertheywish.Oncewehavefull-immersionvirtual-realityenvironmentsincorporatingallofthesenses,whichwillbefeasiblebythelate2020s,therewillbenoreasontoutilizerealoffices.Realestatewillbecomevirtual.AsSunTzupointedout,"knowledgeispower,"andanotherramificationofthelawofacceleratingreturnsistheexponentialgrowthofhumanknowledge,includingintellectualproperty.

Hereisacloseupoftheupper-rightsectionoftheabovefigure–ed.:

Noneof thismeans thatcyclesof recessionwilldisappear immediately.Recently, thecountryexperiencedaneconomicslowdownandtechnology-sectorrecessionandthenagradualrecovery.Theeconomyisstillburdenedwithsomeoftheunderlyingdynamicsthathistoricallyhavecausedcyclesofrecession: excessive commitments such as overinvestment in capital-intensive projects and theoverstocking of inventories. However, because the rapid dissemination of information, sophisticatedformsofonlineprocurement,andincreasinglytransparentmarketsinallindustrieshavediminishedtheimpactof thiscycle,"recessions"are likely tohave lessdirect impactonourstandardof living.Thatappears tohavebeen thecase in themini-recession thatweexperienced in1991–1993andwasevenmoreevidentinthemostrecentrecessionintheearly2000s.Theunderlyinglong-termgrowthratewillcontinueatanexponentialrate. Moreover, innovation and the rate of paradigm shift are not noticeably affectedby theminordeviationscausedbyeconomiccycles.Allofthetechnologiesexhibitingexponentialgrowthshowninthe above charts are continuing without losing a beat through recent economic slowdowns. Marketacceptance also shows no evidence of boom and bust. The overall growth of the economy reflectscompletelynewformsandlayersofwealthandvaluethatdidnotpreviouslyexist,oratleastthatdidnotpreviouslyconstituteasignificantportionoftheeconomy,suchasnewformsofnanoparticle-basedmaterials, genetic information, intellectual property, communication portals, Web sites, bandwidth,software,databases,andmanyothernewtechnology-basedcategories. Theoverall information-technologysectorisrapidlyincreasingitsshareoftheeconomyandisincreasinglyinfluentialonallothersectors,asnotedinthefigurebelow.92

Another implicationof the lawof accelerating returns is exponential growth in educationandlearning.Overthepast120years,wehaveincreasedourinvestmentinK-12education(perstudentandinconstantdollars)byafactoroften.Therehasbeenahundredfoldincreaseinthenumberofcollegestudents. Automation started by amplifying the power of our muscles and in recent times has beenamplifyingthepowerofourminds.Soforthepasttwocenturies,automationhasbeeneliminatingjobsatthebottomoftheskillladderwhilecreatingnew(andbetter-paying)jobsatthetopoftheskillladder.The ladder has been moving up, and thus we have been exponentially increasing investments ineducationatalllevels(seethefigurebelow).93

Oh,andaboutthat"offer"atthebeginningofthisprécis,considerthatpresentstockvaluesarebasedonfutureexpectations.Giventhatthe(literally)shortsightedlinearintuitiveviewrepresentstheubiquitousoutlook,thecommonwisdomineconomicexpectationsisdramaticallyunderstated.Sincestockpricesreflect theconsensusofabuyer-sellermarket, thepricesreflect theunderlying linearassumption thatmostpeopleshareregardingfutureeconomicgrowth.Butthelawofacceleratingreturnsclearlyimpliesthatthegrowthratewillcontinuetogrowexponentially,becausetherateofprogresswillcontinuetoaccelerate.MOLLY 2004: But wait a second, you said that I would get eighty trillion dollars if I read and

understoodthissectionofthechapter.RAY:That'sright.Accordingtomymodels,ifwereplacethelinearoutlookwiththemoreappropriate

exponential outlook, current stock prices should triple.94 Since there's (conservatively) fortytrilliondollarsintheequitymarkets,that'seightytrillioninadditionalwealth.

MOLLY2004:ButyousaidIwouldgetthatmoney.RAY:No,Isaid"you"wouldgetthemoney,andthat'swhyIsuggestedreadingthesentencecarefully.

TheEnglishword"you"canbesingularorplural.Imeantitinthesenseof"allofyou."MOLLY2004:Hmm,that'sannoying.Youmeanallofusasinthewholeworld?Butnoteveryonewill

readthisbook.RAY:Well, but everyone could. So if all of you read this book and understand it, then economic

expectationswouldbebasedon the historical exponentialmodel, and thus stock valueswouldincrease.

MOLLY2004:Youmeanifeveryoneunderstandsitandagreeswithit.Imeanthemarketisbasedonexpectations,right?

RAY:Okay,IsupposeIwasassumingthat.MOLLY2004:Soisthatwhatyouexpecttohappen?RAY:Well,actually,no.Puttingonmyfuturisthatagain,mypredictionisthatindeedtheseviewson

exponentialgrowthwillultimatelyprevailbutonlyovertime,asmoreandmoreevidenceoftheexponential nature of technology and its impact on the economy becomes apparent. This willhappengraduallyoverthenextdecade,whichwillrepresentastronglong-termupdraftforthemarket.

GEORGE 2048: I don't know, Ray. You were right that the price-performance of informationtechnologyinallofitsformskeptgrowingatanexponentialrate,andwithcontinuedgrowthalsoin the exponent. And indeed, the economy kept growing exponentially, thereby more thanovercomingaveryhighdeflationrate.Anditalsoturnedoutthatthegeneralpublicdidcatchontoallofthesetrends.Butthisrealizationdidn'thavethepositiveimpactonthestockmarketthatyou'redescribing.Thestockmarketdidincreasealongwiththeeconomy,buttherealizationofahighergrowthratedidlittletoincreasestockprices.

RAY:Whydoyousupposeitturnedoutthatway?GEORGE2048:Becauseyouleftonethingoutofyourequation.Althoughpeoplerealizedthatstock

valueswouldincreaserapidly,thatsamerealizationalsoincreasedthediscountrate(therateatwhichweneedtodiscountvaluesinthefuturewhenconsideringtheirpresentvalue).Thinkaboutit.Ifweknowthatstocksaregoingtoincreasesignificantlyinafutureperiod,thenwe'dliketohave the stocks now so thatwe can realize those future gains. So the perception of increasedfuture equity values also increases the discount rate. And that cancels out the expectation ofhigherfuturevalues.

MOLLY2104:Uh,George,thatwasnotquiterighteither.Whatyousaymakeslogicalsense,butthepsychological reality is that the heightened perception of increased future values did have agreaterpositiveimpactonstockpricesthanincreasesinthediscountratehadanegativeeffect.Sothegeneralacceptanceofexponentialgrowthinboththeprice-performanceoftechnologyandthe rate of economic activity did provide an upward draft for the equitiesmarket, but not thetriplingthatyouspokeabout,Ray,duetotheeffectthatGeorgewasdescribing.

MOLLY2004:Okay,I'msorryIasked.IthinkI'lljustholdontothefewsharesI'vegotandnotworryaboutit.

RAY:Whathaveyouinvestedin?MOLLY2004:Let'ssee,there'sthisnewnaturallanguage-basedsearch-enginecompanythathopesto

takeonGoogle.AndI'vealsoinvestedinafuel-cellcompany.Also,acompanybuildingsensorsthatcantravelinthebloodstream.

RAY:Soundslikeaprettyhigh-risk,high-techportfolio.MOLLY2004: I wouldn't call it a portfolio. I'm just dabbling with the technologies you're talking

about.RAY:Okay, but keep inmind thatwhile the trendspredictedby the lawof accelerating returnsare

remarkablysmooth,thatdoesn'tmeanwecanreadilypredictwhichcompetitorswillprevail.MOLLY2004:Right,that'swhyI'mspreadingmybets.

I

CHAPTERTHREE

AchievingtheComputationalCapacityoftheHumanBrain

AsIdiscussinEnginesofCreation,ifyoucanbuildgenuineAI,therearereasonstobelievethat you can build things like neurons that are a million times faster. That leads to theconclusionthatyoucanmakesystemsthatthinkamilliontimesfasterthanaperson.WithAI,thesesystemscoulddoengineeringdesign.Combiningthiswiththecapabilityofasystemtobuildsomethingthatisbetterthanit,youhavethepossibilityforaveryabrupttransition.Thissituationmay bemore difficult to dealwith even than nanotechnology, but it ismuchmoredifficulttothinkaboutitconstructivelyatthispoint.Thus,ithasn'tbeenthefocusofthingsthatIdiscuss,althoughIperiodicallypointtoitandsay:"That'simportanttoo."

—ERICDREXLER,1989TheSixthParadigmofComputingTechnology:Three-DimensionalMolecularComputingandEmergingComputationalTechnologies

n the April 19, 1965, issue of Electronics, Gordon Moore wrote, "The future of integratedelectronics is the future of electronics itself. The advantages of integration will bring about aproliferationofelectronics,pushingthisscienceintomanynewareas,"1Withthosemodestwords,Mooreusheredinarevolutionthatisstillgainingmomentum.Togivehisreaderssomeideaofhow

profoundthisnewsciencewouldbe,Moorepredictedthat"by1975,economicsmaydictatesqueezingasmanyas65,000componentsonasinglesiliconchip."Imaginethat. Moore's article described the repeated annual doubling of the number of transistors (used forcomputational elements, or gates) that could be fitted onto an integrated circuit. His 1965 "Moore'sLaw"predictionwascriticizedatthetimebecausehislogarithmicchartofthenumberofcomponentsonachiphadonlyfivereferencepoints(from1959through1965),soprojectingthisnascenttrendallthewayout to1975was seen aspremature.Moore's initial estimatewas incorrect, andhe revised itdownward a decade later. But the basic idea—the exponential growth of the price-performance ofelectronics based on shrinking the size of transistors on an integrated circuit—was both valid andprescient.2Today,wetalkaboutbillionsofcomponentsratherthanthousands.Inthemostadvancedchipsof2004, logic gates are only fifty nanometers wide, already well within the realm of nanotechnology(whichdealswithmeasurementsofonehundrednanometersorless).ThedemiseofMoore'sLawhasbeenpredictedonaregularbasis,buttheendofthisremarkableparadigmkeepsgettingpushedoutintime.PaoloGargini,IntelFellow,directorofInteltechnologystrategy,andchairmanoftheinfluentialInternationalTechnologyRoadmapforSemiconductors(ITRS),recentlystated,"Weseethatforatleastthe next 15 to 20 years,we can continue staying onMoore'sLaw. In fact, ... nanotechnology offersmanynewknobswecanturntocontinueimprovingthenumberofcomponentsonadie.3Theaccelerationofcomputationhastransformedeverythingfromsocialandeconomicrelationsto

political institutions, as Iwill demonstrate throughout this book.ButMoore did not point out in hispapers that the strategy of shrinking feature sizes was not, in fact, the first paradigm to bringexponentialgrowth to computationandcommunication. Itwas the fifth, andalready,wecan see theoutlinesof thenext:computingat themolecular levelandin threedimensions.Eventhoughwehavemorethanadecadeleftofthefifthparadigm,therehasalreadybeencompellingprogressinalloftheenablingtechnologiesrequiredforthesixthparadigm.Inthenextsection,Iprovideananalysisoftheamountofcomputationandmemoryrequiredtoachievehumanlevelsofintelligenceandwhywecanbeconfidentthattheselevelswillbeachievedininexpensivecomputerswithintwodecades.Eventheseverypowerfulcomputerswillbefarfromoptimal,andinthelastsectionofthischapterI'llreviewthelimitsofcomputationaccordingtothelawsofphysicsasweunderstandthemtoday.Thiswillbringustocomputerscircathelatetwenty-firstcentury.TheBridgeto3-DMolecularComputing.Intermediatestepsarealreadyunderway:newtechnologiesthatwill lead to thesixthparadigmofmolecular three-dimensionalcomputing includenanotubesandnanotube circuitry, molecular computing, self-assembly in nanotube circuits, biological systemsemulatingcircuitassembly,computingwithDNA,spintronics(computingwith thespinofelectrons),computing with light, and quantum computing. Many of these independent technologies can beintegratedintocomputationalsystemsthatwilleventuallyapproachthetheoreticalmaximumcapacityofmatter and energy to perform computation andwill far outpace the computational capacities of ahumanbrain. One approach is to build three-dimensional circuits using "conventional" silicon lithography.Matrix Semiconductor is already selling memory chips that contain vertically stacked planes oftransistorsratherthanoneflatlayer.4Sinceasingle3-Dchipcanholdmorememory,overallproductsizeisreduced,soMatrixisinitiallytargetingportableelectronics,whereitaimstocompetewithflashmemory(usedincellphonesanddigitalcamerasbecauseitdoesnotloseinformationwhenthepoweristurnedoff).Thestackedcircuitryalsoreduces theoverallcostperbit.AnotherapproachcomesfromoneofMatrix'scompetitors,FujioMasuoka,a formerToshibaengineerwho invented flashmemory.Masuokaclaimsthathisnovelmemorydesign,whichlookslikeacylinder,reducesthesizeandcost-per-bitofmemorybyafactoroftencomparedtoflatchips.5Workingprototypesofthree-dimensionalsilicon chips have also been demonstrated at Rensselaer Polytechnic Institute's Center forGigascaleIntegrationandattheMITMediaLab. Tokyo'sNipponTelegraphandTelephoneCorporation(NTT)hasdemonstratedadramatic3-Dtechnology using electron-beam lithography, which can create arbitrary three-dimensional structureswithfeaturesizes(suchastransistors)assmallastennanometers.6NTTdemonstratedthetechnologybycreatingahigh-resolutionmodeloftheEarthsixtymicronsinsizewithten-nanometerfeatures.NTTsays the technology is applicable tonanofabricationof electronicdevices suchas semiconductors, aswellascreatingnanoscalemechanicalsystems.Nanotubes Are Still the Best Bet. In The Age of Spiritual Machines, I cited nanotubes—usingmoleculesorganized in threedimensions tostorememorybitsand toactas logicgates—as themostlikely technology to usher in the era of three-dimensional molecular computing. Nanotubes, firstsynthesizedin1991,aretubesmadeupofahexagonalnetworkofcarbonatomsthathavebeenrolledup to make a seamless cylinder.7 Nanotubes are very small: single-wall nanotubes are only onenanometerindiameter,sotheycanachievehighdensities.Theyarealsopotentiallyveryfast.PeterBurkeandhiscolleaguesattheUniversityofCaliforniaatIrvine recently demonstrated nanotube circuits operating at 2.5 gigahertz (GHz). However, inNanoLetters, a peer-reviewed journal of theAmericanChemicalSociety,Burke says the theoretical speedlimit for thesenanotube transistors "shouldbe terahertz (1THz=1,000GHz),which is about1,000times faster than modern computer speeds."8 One cubic inch of nanotube circuitry, once fullydeveloped,wouldbeuptoonehundredmilliontimesmorepowerfulthanthehumanbrain.9

Nanotubecircuitrywascontroversialwhen Idiscussed it in1999,but therehasbeendramaticprogressinthetechnologyoverthepastsixyears.Twomajorstridesweremadein2001.Ananotube-based transistor (with dimensions of one by twenty nanometers), operating at room temperature andusingonlyasingleelectrontoswitchbetweenonandoffstates,wasreportedintheJuly6,2001,issueofScience.10Around the same time, IBMalso demonstrated an integrated circuitwith one thousandnanotube-basedtransistors.11Morerecently,wehaveseenthefirstworkingmodelsofnanotube-basedcircuitry.InJanuary2004researchers at theUniversity ofCalifornia atBerkeley andStanfordUniversity created an integratedmemory circuit based on nanotubes.12 One of the challenges in using this technology is that somenanotubesareconductive(thatis,simplytransmitelectricity),whileothersactlikesemiconductors(thatis,arecapableofswitchingandabletoimplementlogicgates).Thedifferenceincapabilityisbasedonsubtlestructuralfeatures.Untilrecently,sortingthemoutrequiredmanualoperations,whichwouldnotbepracticalforbuildinglarge-scalecircuits.TheBerkeleyandStanfordscientistsaddressedthisissuebydevelopingafullyautomatedmethodofsortinganddiscardingthenonsemiconductornanotubes.Liningupnanotubesisanotherchallengewithnanotubecircuits,sincetheytendtogrowineverydirection.In2001IBMscientistsdemonstratedthatnanotubetransistorscouldbegrowninbulk,similarto silicon transistors.Theyusedaprocess called "constructivedestruction,"whichdestroysdefectivenanotubesrightonthewaferinsteadofsortingthemoutmanually.ThomasTheis,directorofphysicalsciencesatIBM'sThomasJ.WatsonResearchCenter,saidatthetime,"WebelievethatIBMhasnowpassed amajormilestoneon the road towardmolecular-scale chips....Ifweareultimately successful,thencarbonnanotubeswillenableustoindefinitelymaintainMoore'sLawintermsofdensity,becausethereisverylittledoubtinmymindthatthesecanbemadesmallerthananyfuturesilicontransistor.13InMay2003Nantero,asmallcompanyinWoburn,Massachusetts,cofoundedbyHarvardUniversityresearcherThomasRueckes, tooktheprocessastepfurtherwhenitdemonstratedasingle-chipwaferwithtenbillionnanotubejunctions,allalignedintheproperdirection.TheNanterotechnologyinvolvesusing standard lithography equipment to remove automatically the nanotubes that are incorrectlyaligned.Nantero's use of standard equipment has excited industry observers because the technologywouldnotrequireexpensivenewfabricationmachines.TheNanterodesignprovidesrandomaccessaswellasnonvolatility(dataisretainedwhenthepowerisoff),meaningthatitcouldpotentiallyreplacealloftheprimaryformsofmemory:RAM,flash,anddisk.ComputingwithMolecules.Inadditiontonanotubes,majorprogresshasbeenmadeinrecentyearsincomputingwithjustoneorafewmolecules.Theideaofcomputingwithmoleculeswasfirstsuggestedintheearly1970sbyIBM'sAviAviramandNorthwesternUniversity'sMarkA.Ratner.14Atthattime,wedidnothavetheenablingtechnologies,whichrequiredconcurrentadvancesinelectronics,physics,chemistry,andeventhereverseengineeringofbiologicalprocessesfortheideatogaintraction. In 2002 scientists at theUniversity ofWisconsin andUniversity ofBasel created an "atomicmemory drive" that uses atoms to emulate a hard drive. A single silicon atom could be added orremoved from a block of twenty others using a scanning tunnelingmicroscope. Using this process,researchers believe, the system could be used to store millions of times more data on a disk ofcomparablesize—adensityofabout250terabitsofdatapersquareinch—althoughthedemonstrationinvolvedonlyasmallnumberofbits.15 The one-terahertz speed predicted by Peter Burke for molecular circuits looks increasinglyaccurate, given the nanoscale transistor created by scientists at the University of Illinois at Urbana-Champaign.Itrunsatafrequencyof604gigahertz(morethanhalfaterahertz).16 Onetypeofmoleculethatresearchershavefoundtohavedesirablepropertiesforcomputingiscalled a "rotaxane," which can switch states by changing the energy level of a ringlike structurecontained within the molecule. Rotaxane memory and electronic switching devices have beendemonstrated,andtheyshowthepotentialofstoringonehundredgigabits(1011bits)persquareinch.Thepotentialwouldbeevengreateriforganizedinthreedimensions.

Self-Assembly.Self-assembling of nanoscale circuits is another key enabling technique for effectivenanoelectronics. Self-assembly allows improperly formed components to be discarded automaticallyandmakesitpossibleforthepotentiallytrillionsofcircuitcomponentstoorganizethemselves,ratherthan be painstakingly assembled in a top-down process. It would enable large-scale circuits to becreatedintesttubesratherthaninmultibillion-dollarfactories,usingchemistryratherthanlithography,according to UCLA scientists.17 Purdue University researchers have already demonstrated self-organizingnanotube structures, using the sameprinciple that causesDNAstrands to link together instablestructures.18 HarvardUniversity scientists took a key step forward in June 2004when they demonstratedanother self-organizing method that can be used on a large scale.19 The technique starts withphotolithography to create an etched array of interconnects (connections between computationalelements). A large number of nanowire field-effect transistors (a common form of transistors) andnanoscale interconnectsare thendepositedonthearray.Thesethenconnect themselvesin thecorrectpattern.In2004researchersattheUniversityofSouthernCaliforniaandNASA'sAmesResearchCenterdemonstrated a method that self-organizes extremely dense circuits in a chemical solution.20 Thetechniquecreatesnanowiresspontaneouslyandthencausesnanoscalememorycells,eachabletoholdthreebitsofdata,toself-assembleontothewires.Thetechnologyhasastoragecapacityof258gigabitsofdatapersquareinch(whichresearchersclaimcouldbeincreasedtenfold),comparedto6.5gigabitsonaflashmemorycard.Alsoin2003IBMdemonstratedaworkingmemorydeviceusingpolymersthatself-assembleintotwenty-nanometer-widehexagonalstructures.21It'salsoimportantthatnanocircuitsbeself-configuring.Thelargenumberofcircuitcomponentsandtheirinherentfragility(duetotheirsmallsize)makeitinevitablethatsomeportionsofacircuitwillnotfunctioncorrectly.Itwillnotbeeconomicallyfeasibletodiscardanentirecircuitsimplybecauseasmallnumberoftransistorsoutofatrillionarenonfunctioning.Toaddressthisconcern,futurecircuitswill continuously monitor their own performance and route information around sections that areunreliableinthesamemannerthatinformationontheInternetisroutedaroundnonfunctioningnodes.IBM has been particularly active in this area of research and has already developedmicroprocessordesignsthatautomaticallydiagnoseproblemsandreconfigurechipresourcesaccordingly.22EmulatingBiology.Theideaofbuildingelectronicormechanicalsystemsthatareself-replicatingandself-organizing is inspired by biology, which relies on these properties. Research published in theProceedings of the National Academy of Sciences described the construction of self-replicatingnanowiresbasedonprions,whichareself-replicatingproteins. (Asdetailed inchapter4,oneformofprionappearstoplayaroleinhumanmemory,whereasanotherformisbelievedtoberesponsibleforvariantCreutzfeldt-Iakob disease, the human form ofmad-cow disease.)23 The team involved in theproject used prions as a model because of their natural strength. Because prions do not normallyconductelectricity,however,thescientistscreatedageneticallymodifiedversioncontainingathinlayerofgold,whichconductselectricitywith lowresistance.MITbiologyprofessorSusanLindquist,whoheaded the study, commented, "Mostof thepeopleworkingonnanocircuits are trying tobuild themusing'top-down'fabricationtechniques.Wethoughtwe'dtrya'bottom-up'approach,andletmolecularself-assemblydothehardworkforus." The ultimate self-replicating molecule from biology is, of course, DNA. Duke Universityresearcherscreatedmolecularbuildingblockscalled"tiles"outof self-assemblingDNAmolecules.24Theywereabletocontrolthestructureoftheresultingassembly,creating"nanogrids,"Thistechniqueautomatically attaches protein molecules to each nanogrid's cell, which could be used to performcomputing operations.They also demonstrated a chemical process that coated theDNAnanoribbonswith silver to create nanowires. Commenting on the article in the September 26, 2003, issue of thejournal Science, lead researcher Hao Yan said, "To use DNA self-assembly to template protein

molecules or other molecules has been sought for years, and this is the first time it has beendemonstratedsoclearly."25Computing with DNA. DNA is nature's own nanoengineered computer, and its ability to storeinformation and conduct logical manipulations at the molecular level has already been exploited inspecialized"DNAcomputers."ADNAcomputerisessentiallyatesttubefilledwithwatercontainingtrillionsofDNAmolecules,witheachmoleculeactingasacomputer.Thegoalofthecomputationistosolveaproblem,withthesolutionexpressedasasequenceofsymbols.(Forexample,thesequenceofsymbolscouldrepresentamathematicalprooforjustthedigitsofanumber.)Here'showaDNAcomputerworks.AsmallstrandofDNAiscreated,usingauniquecode for each symbol. Each such strand is replicated trillions of times using a process called"polymerasechainreaction"(PCR).ThesepoolsofDNAare thenput intoa test tube.BecauseDNAhasanaffinitytolinkstrandstogether,longstrandsformautomatically,withsequencesofthestrandsrepresentingthedifferentsymbols,eachofthemapossiblesolutiontotheproblem.Sincetherewillbemanytrillionsofsuchstrands,therearemultiplestrandsforeachpossibleanswer(thatis,eachpossiblesequenceofsymbols). Thenext stepof theprocess is to testallof thestrandssimultaneously.This isdonebyusingspecially designed enzymes that destroy strands that do not meet certain criteria. The enzymes areapplied to the test tubesequentially,andbydesigningapreciseseriesofenzymes theprocedurewilleventuallyobliteratealltheincorrectstrands,leavingonlytheoneswiththecorrectanswer.(Foramorecompletedescriptionoftheprocess,seethisnote:26)ThekeytothepowerofDNAcomputingisthatitallowsfortestingeachofthetrillionsofstrandssimultaneously. In 2003 Israeli scientists led by Ehud Shapiro at theWeizmann Institute of SciencecombinedDNAwithadenosinetriphosphate(ATP),thenaturalfuelforbiologicalsystemssuchasthehumanbody.27Withthismethod,eachoftheDNAmoleculeswasabletoperformcomputationsaswellas provide its own energy. TheWeizmann scientists demonstrated a configuration consisting of twospoonfuls of this liquid supercomputing system, which contained thirty million billion molecularcomputers andperformed a total of 660 trillion calculationsper second (6.6 Î 1014 cps). The energyconsumptionofthesecomputersisextremelylow,onlyfiftymillionthsofawattforall thirtymillionbillioncomputers.There'salimitation,however,toDNAcomputing:eachofthemanytrillionsofcomputershastoperformthesameoperationatthesametime(althoughondifferentdata),sothatthedeviceisa"singleinstructionmultipledata"(SIMD)architecture.WhilethereareimportantclassesofproblemsthatareamenabletoaSIMDsystem(forexample,processingeverypixelinanimageforimageenhancementor compression, and solving combinatorial-logic problems), it is not possible to program them forgeneral-purposealgorithms,inwhicheachcomputerisabletoexecutewhateveroperationisneededforits particular mission. (Note that the research projects at Purdue University and Duke University,described earlier, that use self-assembling DNA strands to create three-dimensional structures aredifferentfromtheDNAcomputingdescribedhere.ThoseresearchprojectshavethepotentialtocreatearbitraryconfigurationsthatarenotlimitedtoSIMDcomputing.)ComputingwithSpin.Inadditiontotheirnegativeelectricalcharge,electronshaveanotherpropertythatcanbeexploitedformemoryandcomputation:spin.According toquantummechanics,electronsspinonanaxis,similartothewaytheEarthrotatesonitsaxis.Thisconceptistheoretical,becauseanelectronisconsideredtooccupyapointinspace,soitisdifficulttoimagineapointwithnosizethatnonethelessspins.However,whenanelectricalchargemoves,itcausesamagneticfield,whichisrealandmeasurable.Anelectroncanspininoneoftwodirections,describedas"up"and"down,”sothispropertycanbeexploitedforlogicswitchingortoencodeabitofmemory.Theexcitingpropertyofspintronicsisthatnoenergyisrequiredtochangeanelectron'sspinstate.Stanford University physics professor Shoucheng Zhang and University of Tokyo professor NaotoNagaosaputitthisway:"Wehavediscoveredtheequivalentofanew'Ohm'sLaw'[theelectronicslaw

that states that current in awire equals voltage divided by resistance]....[It] says that the spin of theelectroncanbetransportedwithoutanylossofenergy,ordissipation.Furthermore,thiseffectoccursatroom temperature in materials already widely used in the semiconductor industry, such as galliumarsenide.That'simportantbecauseitcouldenableanewgenerationofcomputingdevices."28Thepotential,then,istoachievetheefficienciesofsuperconducting(thatis,movinginformationator close to the speed of light without any loss of information) at room temperature. It also allowsmultiple properties of each electron to be used for computing, thereby increasing the potential formemoryandcomputationaldensity. One formof spintronics isalready familiar tocomputerusers:magnetoresistance (achange inelectrical resistance caused by a magnetic field) is used to store data on magnetic hard drives. AnexcitingnewformofnonvolatilememorybasedonspintronicscalledMRAM(magneticrandom-accessmemory)isexpectedtoenterthemarketwithinafewyears.Likeharddrives,MRAMmemoryretainsitsdatawithoutpowerbutusesnomovingpartsandwillhavespeedsandrewritabilitycomparabletoconventionalRAM.MRAMstoresinformationinferromagneticmetallicalloys,whicharesuitablefordatastoragebutnot for the logicaloperationsof amicroprocessor.Theholygrailof spintronicswouldbe to achievepracticalspintronicseffectsinasemiconductor,whichwouldenableustousethetechnologybothformemory and for logic. Today's chip manufacturing is based on silicon, which does not have therequisitemagneticproperties.InMarch2004aninternationalgroupofscientistsreportedthatbydopingablendof siliconand ironwithcobalt, thenewmaterialwasable todisplay themagneticpropertiesneeded for spintronics while still maintaining the crystalline structure silicon requires as aserniconductor.29 Animportant roleforspintronics in thefutureofcomputermemory isclear,and it is likely tocontributetologicsystemsaswell.Thespinofanelectronisaquantumproperty(subjecttothelawsofquantum mechanics), so perhaps the most important application of spintronics will be in quantumcomputingsystems,usingthespinofquantum-entangledelectronstorepresentqubits,whichIdiscussbelow.Spinhasalsobeenusedtostoreinformationinthenucleusofatoms,usingthecomplexinteractionof their protons' magnetic moments. Scientists at the University of Oklahoma also demonstrated a"molecular photography" technique for storing 1,024 bits of information in a single liquid-crystalmoleculecomprisingnineteenhydrogenatoms.30ComputingwithLight.AnotherapproachtoSIMDcomputingistousemultiplebeamsoflaserlightinwhich information is encoded in each stream of photons. Optical components can then be used toperform logical and arithmetic functionson the encoded information streams.For example, a systemdeveloped by Lenslet, a small Israeli company, uses 256 lasers and can perform eight trillioncalculationspersecondbyperformingthesamecalculationoneachofthe256streamsofdata.31Thesystemcanbeusedforapplicationssuchasperformingdatacompressionon256videochannels.SIMDtechnologiessuchasDNAcomputersandopticalcomputerswillhaveimportantspecializedrolestoplayinthefutureofcomputation.Thereplicationofcertainaspectsofthefunctionalityofthehumanbrain,suchasprocessingsensorydata,canuseSIMDarchitectures.Forotherbrainregions,suchasthosedealingwithlearningandreasoning,general-purposecomputingwithits"multipleinstructionmultipledata"(MIMD)architectureswillberequired.Forhigh-performanceMIMDcomputing,wewillneedtoapplythethree-dimensionalmolecular-computingparadigmsdescribedabove.QuantumComputing.QuantumcomputingisanevenmoreradicalformofSIMDparallelprocessing,butonethatisinamuchearlierstageofdevelopmentcomparedtotheothernewtechnologieswehavediscussed.Aquantumcomputercontainsa seriesofqubits,whichessentiallyarezeroandoneat thesame time. The qubit is based on the fundamental ambiguity inherent in quantum mechanics. In aquantumcomputer,thequbitsarerepresentedbyaquantumpropertyofparticles—forexample,thespinstateofindividualelectrons.Whenthequbitsareinan"entangled"state,eachoneissimultaneouslyin

bothstates.Inaprocesscalled"quantumdecoherence"theambiguityofeachqubitisresolved,leavinganunambiguoussequenceofonesandzeroes.Ifthequantumcomputerissetupintherightway,thatdecohered sequence will represent the solution to a problem. Essentially, only the correct sequencesurvivestheprocessofdecoherence.AswiththeDNAcomputerdescribedabove,akeytosuccessfulquantumcomputingisacarefulstatement of the problem, including a precise way to test possible answers. The quantum computereffectivelytestseverypossiblecombinationofvaluesforthequbits.Soaquantumcomputerwithonethousand qubits would test 21,000 (a number approximately equal to one followed by 301 zeroes)potentialsolutionssimultaneously.Athousand-bitquantumcomputerwouldvastlyoutperformanyconceivableDNAcomputer,orforthatmatteranyconceivablenonquantumcomputer.Therearetwolimitationstotheprocess,however.Thefirstisthat,liketheDNAandopticalcomputersdiscussedabove,onlyaspecialsetofproblemsisamenable to being presented to a quantum computer. In essence, we need to I be able to test eachpossibleanswerinasimpleway.Theclassicexampleofapracticaluseforquantumcomputingisinfactoringverylargenumbers(finding which smaller numbers, when multiplied together, result in the large number). Factoringnumberswithmore than512bits iscurrentlynotachievableonadigitalcomputer,evenamassivelyparallel one.32 Interesting classes of problems amenable to quantum computing include breakingencryptioncodes(whichrelyonfactoringlargenumbers).Theotherproblemisthatthecomputationalpowerofaquantumcomputerdependson thenumberofentangledqubits, and the stateof theart iscurrently limited toaround tenbits.A ten-bitquantumcomputer isnotveryuseful, since210 isonly1,024. In a conventional computer, it is a straightforwardprocess to combinememorybits and logicgates.We cannot, however, create a twenty-qubit quantum computer simply by combining two ten-qubitmachines.All of the qubits have to be quantum-entangled together, and that has proved to bechallenging.Akeyquestionis:howdifficultisittoaddeachadditionalqubit?Thecomputationalpowerofaquantum computer grows exponentially with each added qubit, but if it turns out that adding eachadditional qubitmakes the engineering task exponentiallymore difficult,wewill not be gaining anyleverage.(Thatis,thecomputationalpowerofaquantumcomputerwillbeonlylinearlyproportionaltotheengineeringdifficulty.)Ingeneral,proposedmethodsforaddingqubitsmaketheresultingsystemssignificantlymoredelicateandsusceptibletoprematuredecoherence.Thereareproposalstoincreasesignificantlythenumberofqubits,althoughthesehavenotyetbeenprovedinpractice.Forexample,StephanGuldeandhiscolleaguesattheUniversityofInnsbruckhavebuiltaquantumcomputerusingasingleatomofcalciumthathasthepotentialtosimultaneouslyencodedozensofqubits—possiblyuptoonehundred—usingdifferentquantumpropertieswithintheatom.33The ultimate role of quantum computing remains unresolved. But even if a quantum computerwithhundredsofentangledqubitsprovesfeasible,itwillremainaspecial-purposedevice,althoughonewithremarkablecapabilitiesthatcannotbeemulatedinanyotherway.WhenIsuggestedinTheAgeofSpiritualMachinesthatmolecularcomputingwouldbethesixthmajorcomputingparadigm,theideawasstillcontroversial.Therehasbeensomuchprogressinthepastfive years that there has been a sea change in attitude among experts, and this is now amainstreamview. We already have proofs of concept for all of the major requirements for three-dimensionalmolecular computing: single-molecule transistors, memory cells based on atoms, nanowires, andmethodstoself-assembleandself-diagnosethetrillions(potentiallytrillionsoftrillions)ofcomponents.Contemporaryelectronicsproceedsfromthedesignofdetailedchiplayoutstophotolithographytothemanufacturingofchipsinlarge,centralizedfactories.Nanocircuitsaremorelikelytobecreatedinsmallchemistryflasks,adevelopmentthatwillbeanotherimportantstepinthedecentralizationofourindustrial infrastructure and will maintain the law of accelerating returns through this century andbeyond.

TheComputationalCapacityoftheHumanBrain

Itmay seem rash to expect fully intelligentmachines in a fewdecades,when the computershavebarelymatchedinsectmentalityinahalf-centuryofdevelopment.Indeed,forthatreason,many long-time artificial intelligence researchers scoff at the suggestion, and offer a fewcenturiesasamorebelievableperiod.Butthereareverygoodreasonswhythingswillgomuchfasterinthenextfiftyyearsthantheyhaveinthelastfifty....Since1990,thepoweravailabletoindividualAIandroboticsprogramshasdoubledyearly,to30MIPSby1994and500MIPSby1998.Seedslongagoallegedbarrenaresuddenlysprouting.Machinesreadtext,recognizespeech,even translate languages.Robotsdrivecross-country,crawlacrossMars, and trundledownofficecorridors.In1996atheorem-provingprogramcalledEQPrunningfiveweeksona50 MIPS computer at Argonne National Laboratory found a proof of a Boolean algebraconjecturebyHerbertRobbins thathadeludedmathematicians for sixtyyears.And it is stillonlySpring.WaituntilSummer.

—HANSMORAVEC,“WHENWILLCOMPUTERHARDWAREMATCHTHEHUMANBRAIN?”1997

Whatisthecomputationalcapacityofahumanbrain?Anumberofestimateshavebeenmade,basedonreplicating the functionality of brain regions that have been reverse engineered (that is, themethodsunderstood)athumanlevelsofperformance.Oncewehaveanestimateofthecomputationalcapacityforaparticularregion,wecanextrapolatethatcapacitytotheentirebrainbyconsideringwhatportionofthebrainthatregionrepresents.Theseestimatesarebasedonfunctionalsimulation,whichreplicatestheoverallfunctionalityofaregionratherthansimulatingeachneuronandinterneuronalconnectioninthatregion.Althoughwewouldnotwanttorelyonanysinglecalculation,wefindthatvariousassessmentsofdifferentregionsofthebrainallprovidereasonablycloseestimatesfortheentirebrain.Thefollowingareorder-of-magnitudeestimates,meaningthatweareattemptingtodeterminetheappropriatefigurestotheclosestmultipleoften.Thefactthatdifferentwaysofmakingthesameestimateprovidesimilaranswerscorroboratestheapproachandindicatesthattheestimatesareinanappropriaterange. Thepredictionthat theSingularity—anexpansionofhumanintelligencebyafactorof trillionsthrough merger with its nonbiological form—will occur within the next several decades does notdepend on the precision of these calculations. Even if our estimate of the amount of computationrequired to simulate the human brain was too optimistic (that is, too low) by a factor of even onethousand(whichIbelieveisunlikely),thatwoulddelaytheSingularitybyonlyabouteightyears.34Afactorofonemillionwouldmeanadelayofonlyaboutfifteenyears,andafactorofonebillionwouldbeadelayofabouttwenty-oneyears.35 Hans Moravec, legendary roboticist at Carnegie Mellon University, has analyzed thetransformationsperformedbytheneuralimage-processingcircuitrycontainedintheretina.36Theretinais about two centimeters wide and a half millimeter thick.Most of the retina's depth is devoted tocapturingan image,butone fifthof it is devoted to imageprocessing,which includesdistinguishingdarkandlight,anddetectingmotioninaboutonemillionsmallregionsoftheimage. The retina, according toMoravec's analysis, performs tenmillion of these edge andmotiondetectionseachsecond.Basedonhisseveraldecadesofexperienceincreatingroboticvisionsystems,heestimatesthattheexecutionofaboutonehundredcomputerinstructionsisrequiredtore-createeachsuch detection at human levels of performance, meaning that replicating the image-processingfunctionalityofthisportionoftheretinarequires1,000MIPS.Thehumanbrainisabout75,000timesheavier than the0.02gramsofneurons in thisportionof the retina, resulting inanestimateofabout1014(100trillion)instructionspersecondfortheentirebrain.37AnotherestimatecomesfromtheworkofLloydWattsandhiscolleaguesoncreatingfunctionalsimulationsofregionsofthehumanauditorysystem,whichIdiscussfurtherinchapter4.38Oneofthe

functions of the softwareWatts has developed is a task called "stream separation,"which is used inteleconferencingandotherapplicationstoachievetelepresence(thelocalizationofeachparticipantinaremoteaudioteleconference),Toaccomplishthis,Wattsexplains,means"preciselymeasuringthetimedelaybetweensoundsensorsthatareseparatedinspaceandthatbothreceivethesound."Theprocessinvolvespitchanalysis,spatialposition,andspeechcues,includinglanguage-specificcues."Oneoftheimportant cues used by humans for localizing the position of a sound source is the Interaural TimeDifference(ITD),thatis,thedifferenceintimeofarrivalofsoundsatthetwoears."39Watts'sowngrouphascreatedfunctionallyequivalentre-creationsofthesebrainregionsderivedfromreverseengineering.Heestimatesthat1011cpsarerequiredtoachievehuman-levellocalizationofsounds.Theauditorycortexregionsresponsibleforthisprocessingcompriseatleast0.1percentofthebrain'sneurons.Soweagainarriveataballparkestimateofaround1014cpsÎ103). Yet another estimate comes from a simulation at theUniversity of Texas that represents thefunctionalityofacerebellumregioncontaining104neurons;thisrequiredabout108cps,orabout104

cpsperneuron.Extrapolatingthisoveranestimated1011neuronsresultsinafigureofabout1015cpsfortheentirebrain. Wewill discuss the stateofhuman-brain reverse engineering later, but it is clear thatwecanemulatethefunctionalityofbrainregionswithlesscomputationthanwouldberequiredtosimulatetheprecisenonlinearoperationofeachneuronandalloftheneuralcomponents(thatis,allofthecomplexinteractionsthattakeplaceinsideeachneuron).Wecometothesameconclusionwhenweattempttosimulatethefunctionalityoforgansinthebody.Forexample,implantabledevicesarebeingtestedthatsimulatethefunctionalityofthehumanpancreasinregulatinginsulinlevels.40Thesedevicesworkbymeasuringglucoselevelsinthebloodandreleasinginsulininacontrolledfashiontokeepthelevelsinanappropriaterange.Whiletheyfollowamethodsimilartothatofabiologicalpancreas,theydonot,however,attempttosimulateeachpancreaticisletcell,andtherewouldbenoreasontodoso.Theseestimatesallresultincomparableordersofmagnitude(1014to1015cps).Giventheearlystage of human-brain reverse engineering, Iwill use amore conservative figure of 1016 cps for oursubsequentdiscussions.Functionalsimulationofthebrainissufficienttore-createhumanpowersofpatternrecognition,intellect, and emotional intelligence.On the other hand, ifwewant to "upload" a particular person'spersonality(thatis,captureallofhisorherknowledge,skills,andpersonality,aconceptIwillexploreingreaterdetailattheendofchapter4),thenwemayneedtosimulateneuralprocessesatthelevelofindividual neurons and portions of neurons, such as the soma (cell body), axon (output connection),dendrites(treesofincomingconnections),andsynapses(regionsconnectingaxonsanddendrites).Forthis,weneedtolookatdetailedmodelsofindividualneurons.The"fanout"(numberofinterneuronalconnections) per neuron is estimated at 103. With an estimated 1011 neurons, that's about 1014

connections.Witharesettimeoffivemilliseconds,thatcomestoabout1016synaptictransactionspersecond.Neuron-modelsimulationsindicatetheneedforabout103calculationspersynaptictransactiontocapturethenonlinearities(complexinteractions)inthedendritesandotherneuronregions,resultinginanoverallestimateofabout1019cpsforsimulatingthehumanbrainat this level.41Wecan thereforeconsiderthisanupperbound,but1014to1016cpstoachievefunctionalequivalenceofallbrainregionsislikelytobesufficient. IBM'sBlueGene/Lsupercomputer,nowbeingbuiltandscheduledtobecompletedaroundthetimeof thepublicationof thisbook, isprojectedtoprovide360trillioncalculationspersecond(3.6 Î1014cps).42Thisfigureisalreadygreaterthanthelowerestimatesdescribedabove.BlueGene/Lwillalso have around one hundred terabytes (about 1015 bits) of main storage, more than our memoryestimateforfunctionalemulationof thehumanbrain(seebelow).In linewithmyearlierpredictions,supercomputerswill achievemymore conservative estimate of 1016 cps for functional human-brainemulationbyearlyinthenextdecade(seethe"SupercomputerPower"figureonp.71).

Accelerating the Availability of Human-Level Personal Computing. Personal computers todayprovidemore than109 cps.According to the projections in the "ExponentialGrowth ofComputing"chart (p.70),wewillachieve1016cpsby2025.However, thereareseveralways this timelinecanbeaccelerated.Rather thanusinggeneral-purposeprocessors,onecanuseapplication-specific integratedcircuits (ASICs) to provide greater price-performance for very repetitive calculations. Such circuitsalready provide extremely high computational throughput for the repetitive calculations used ingenerating moving images in video games. ASICs can increase price-performance a thousandfold,cuttingabouteightyearsoffthe2025date.Thevariedprogramsthatasimulationofthehumanbrainwill comprise will also include a great deal of repetition and thus will be amenable to ASICimplementation.Thecerebellum,forexample,repeatsabasicwiringpatternbillionsoftimes. Wewill also be able to amplify the power of personal computers by harvesting the unusedcomputation power of devices on the Internet. New communication paradigms such as "mesh"computingcontemplatetreatingeverydeviceinthenetworkasanoderatherthanjusta"spoke."43Inotherwords,insteadofdevices(suchaspersonalcomputersandPDAs)merelysendinginformationtoandfromnodes,eachdevicewillactasanodeitself,sendinginformationtoandreceivinginformationfromeveryotherdevice.Thatwillcreateveryrobust,self-organizingcommunicationnetworks.Itwillalsomake iteasier forcomputersandotherdevices to tapunusedCPUcyclesof thedevices in theirregionofthemesh.Currentlyatleast99percent,ifnot99.9percent,ofthecomputationalcapacityofallthecomputersontheInternetliesunused.Effectivelyharnessingthiscomputationcanprovideanotherfactorof102or103inincreasedprice-performance.Forthesereasons,itisreasonabletoexpecthumanbraincapacity,atleastintermsofhardwarecomputationalcapacity,foronethousanddollarsbyaround2020. Yet another approach to accelerate the availability of human-level computation in a personalcomputeristousetransistorsintheirnative"analog"mode.Manyoftheprocessesinthehumanbrainareanalog,notdigital.Althoughwecanemulateanalogprocesses toanydesireddegreeofaccuracywith digital computation, we lose several orders of magnitude of efficiency in doing so. A singletransistorcanmultiply twovaluesrepresentedasanalog levels;doingsowithdigitalcircuits requiresthousands of transistors. California Institute of Technology's CarverMead has been pioneering thisconcept.44OnedisadvantageofMead'sapproachisthattheengineeringdesigntimerequiredforsuchnativeanalogcomputingislengthy,somostresearchersdevelopingsoftwaretoemulateregionsofthebrainusuallyprefertherapidturnaroundofsoftwaresimulations.HumanMemoryCapacity.Howdoescomputationalcapacitycomparetohumanmemorycapacity?Itturnsoutthatwearriveatsimilartime-frameestimatesifwelookathumanmemoryrequirements.Thenumberof"chunks"ofknowledgemasteredbyanexpertinadomainisapproximately105foravarietyofdomains.Thesechunksrepresentpatterns(suchasfaces)aswellasspecificknowledge.Forexample,aworld-classchessmaster isestimated tohavemasteredabout100,000boardpositions.Shakespeareused29,000wordsbutclose to100,000meaningsof thosewords.Developmentofexpertsystems inmedicineindicatethathumanscanmasterabout100,000conceptsinadomain.Ifweestimatethatthis"professional"knowledgerepresentsaslittleas1percentoftheoverallpatternandknowledgestoreofahuman,wearriveatanestimateof107chunks.Basedonmyownexperienceindesigningsystemsthatcanstoresimilarchunksofknowledgeineitherrule-basedexpertsystemsorself-organizingpattern-recognitionsystems,areasonableestimateisabout106bitsperchunk(patternoritemofknowledge),foratotalcapacityof1013(10trillion)bitsforahuman'sfunctionalmemory.AccordingtotheprojectionsfromtheITRSroadmap(seeRAMchartonp.57),wewillbeabletopurchase1013bitsofmemoryforonethousanddollarsbyaround2018.Keepinmindthatthismemorywillbemillionsoftimesfasterthantheelectrochemicalmemoryprocessusedinthehumanbrainandthuswillbefarmoreeffective.

Again,ifwemodelhumanmemoryonthelevelofindividualinterneuronalconnections,wegetahigher estimate.We can estimate about 104 bits per connection to store the connection patterns andneurotransmitter concentrations. With an estimated 1014 connections, that comes to 1018 (a billionbillion)bits.Basedontheaboveanalyses,itisreasonabletoexpectthehardwarethatcanemulatehuman-brainfunctionalitytobeavailableforapproximatelyonethousanddollarsbyaround2020.Aswewilldiscussinchapter4,thesoftwarethatwillreplicatethatfunctionalitywilltakeaboutadecadelonger.However,theexponentialgrowthoftheprice-performance,capacity,andspeedofourhardwaretechnologywillcontinueduringthatperiod,soby2030itwilltakeavillageofhumanbrains(aroundonethousand)tomatchathousanddollars'worthofcomputing.By2050,onethousanddollarsofcomputingwillexceedthe processing power of all human brains onEarth.Of course, this figure includes those brains stillusingonlybiologicalneurons.Whilehumanneuronsarewondrouscreations,wewouldn't(anddon't)designcomputingcircuitsusing thesameslowmethods.Despite the ingenuityof thedesignsevolved throughnaturalselection,theyaremanyordersofmagnitudelesscapablethanwhatwewillbeabletoengineer.Aswereverseengineerourbodiesandbrains,wewillbeinapositiontocreatecomparablesystemsthatarefarmoredurableandthatoperatethousandstomillionsoftimesfasterthanournaturallyevolvedsystems.Ourelectronic circuits are already more than one million times faster than a neuron's electrochemicalprocesses,andthisspeediscontinuingtoaccelerate.Mostofthecomplexityofahumanneuronisdevotedtomaintainingitslife-supportfunctions,notits information-processing capabilities.Ultimately,wewill be able to port ourmental processes to amoresuitablecomputationalsubstrate.Thenourmindswon'thavetostaysosmall.TheLimitsofComputation

Ifamostefficientsupercomputerworksalldaytocomputeaweathersimulationproblem,whatistheminimumamountofenergythatmustbedissipatedaccordingtothelawsofphysics?Theanswerisactuallyverysimpletocalculate,sinceitisunrelatedtotheamountofcomputation.Theanswerisalwaysequaltozero.

—EDWARDFREDKIN,PHYSICIST45We've already had five paradigms (electromechanical calculators, relay-based computing, vacuumtubes,discrete transistors,andintegratedcircuits) thathaveprovidedexponentialgrowthto theprice-performanceandcapabilitiesofcomputation.Eachtimeaparadigmreacheditslimits,anotherparadigmtookitsplace.Wecanalreadyseetheoutlinesofthesixthparadigm,whichwillbringcomputingintothemolecular thirddimension.Becausecomputationunderlies the foundationsofeverythingwecareabout, fromtheeconomy tohuman intellectandcreativity,wemightwellwonder:are thereultimatelimitstothecapacityofmatterandenergytoperformcomputation?Ifso,whataretheselimits,andhowlongwillittaketoreachthem?Ourhumanintelligenceisbasedoncomputationalprocessesthatwearelearningtounderstand.Wewillultimatelymultiplyourintellectualpowersbyapplyingandextendingthemethodsofhumanintelligenceusingthevastlygreatercapacityofnonbiologicalcomputation.Sotoconsidertheultimatelimitsofcomputationisreallytoask:whatisthedestinyofourcivilization?Acommonchallengetotheideaspresentedinthisbookisthattheseexponentialtrendsmustreacha limit, as exponential trends commonly do.When a species happens upon a new habitat, as in thefamousexampleofrabbits inAustralia, itsnumbersgrowexponentiallyforawhile.But iteventuallyreachesthelimitsofthatenvironment'sabilitytosupportit.Surelytheprocessingofinformationmusthave similar constraints. It turns out that, yes, there are limits to computation based on the laws ofphysics.Butthesestillallowforacontinuationofexponentialgrowthuntilnonbiologicalintelligenceistrillions of trillions of times more powerful than all of human civilization today, contemporary

computersincluded.Amajorfactorinconsideringcomputationallimitsistheenergyrequirement.TheenergyrequiredperMIPSforcomputingdeviceshasbeenfallingexponentially,asshowninthefollowingfigure.46

However, we also know that the number ofMIPS in computing devices has been growingexponentially.Theextenttowhichimprovementsinpowerusagehavekeptpacewithprocessorspeeddependsontheextenttowhichweuseparallelprocessing.Alargernumberofless-powerfulcomputerscaninherentlyruncoolerbecausethecomputationisspreadoutoveralargerarea.Processorspeedisrelated to voltage, and the power required is proportional to the square of the voltage. So running aprocessor at a slower speed significantly reduces power consumption. If we invest in more parallelprocessingratherthanfastersingleprocessors,itisfeasibleforenergyconsumptionandheatdissipationtokeeppacewiththegrowingMIPSperdollar,asthefigure"ReductioninWattsperMIPS"shows.Thisisessentiallythesamesolutionthatbiologicalevolutiondevelopedinthedesignofanimalbrains.Humanbrainsuseaboutonehundred trillioncomputers (the interneuronalconnections,wheremost of the processing takes place). But these processors are very low in computational power andthereforerunrelativelycool. UntiljustrecentlyIntelemphasizedthedevelopmentoffasterandfastersingle-chipprocessors,which have been running at increasingly high temperatures. Intel is gradually changing its strategytoward parallelization by puttingmultiple processors on a single chip.Wewill see chip technologymoveinthisdirectionasawayofkeepingpowerrequirementsandheatdissipationincheck.47ReversibleComputing. Ultimately, organizing computation with massive parallel processing, as isdone in the human brain,will not by itself be sufficient to keep energy levels and resulting thermaldissipationatreasonablelevels.Thecurrentcomputerparadigmreliesonwhatisknownasirreversiblecomputing,meaningthatweareunableinprincipletorunsoftwareprogramsbackward.Ateachstepintheprogressionofaprogram,theinputdataisdiscarded—erased—andtheresultsofthecomputationpass to the next step.Programsgenerally donot retain all intermediate results, as thatwoulduse uplargeamountsofmemoryunnecessarily.Thisselectiveerasureofinputinformationisparticularlytrueforpattern-recognitionsystems.Visionsystems,forexample,whetherhumanormachine,receiveveryhigh ratesof input (from the eyesorvisual sensors)yet produce relatively compactoutputs (such asidentification of recognized patterns). This act of erasing data generates heat and therefore requires

energy.Whenabitof information iserased, that informationhas togosomewhere.According to thelawsofthermodynamics,theerasedbitisessentiallyreleasedintothesurroundingenvironment,therebyincreasing its entropy, which can be viewed as a measure of information (including apparentlydisordered information) in an environment. This results in a higher temperature for the environment(becausetemperatureisameasureofentropy).If,ontheotherhand,wedon'teraseeachbitofinformationcontainedintheinputtoeachstepofanalgorithmbutinsteadjustmoveittoanotherlocation,thatbitstaysinthecomputer,isnotreleasedintotheenvironment,andthereforegeneratesnoheatandrequiresnoenergyfromoutsidethecomputer.RolfLandauershowedin1961thatreversiblelogicaloperationssuchasNOT(turningabitintoitsopposite)couldbeperformedwithoutputtingenergyinortakingheatout,butthatirreversiblelogicaloperations suchasAND(generatingbitC,which is a1 if andonly ifboth inputsAandBare1)dorequireenergy.48In1973CharlesBennettshowedthatanycomputationcouldbeperformedusingonlyreversible logical operations.49 A decade later, Ed Fredkin and Tommaso Toffoli presented acomprehensive reviewof the ideaof reversible computing.50 The fundamental concept is that if youkeep all the intermediate results and then run the algorithm backward when you've finished yourcalculation,youendupwhereyoustarted,haveusednoenergy,andgeneratednoheat.Alongtheway,however,you'vecalculatedtheresultofthealgorithm.HowSmartIsaRock?Toappreciatethefeasibilityofcomputingwithnoenergyandnoheat,considerthecomputationthattakesplaceinanordinaryrock.Althoughitmayappearthatnothingmuchisgoingoninsidearock,theapproximately1025(tentrilliontrillion)atomsinakilogramofmatterareactuallyextremely active. Despite the apparent solidity of the object, the atoms are all in motion, sharingelectronsbackandforth,changingparticlespins,andgeneratingrapidlymovingelectromagneticfields.Allofthisactivityrepresentscomputation,evenifnotverymeaningfullyorganized.We'vealreadyshownthatatomscanstoreinformationatadensityofgreaterthanonebitperatom,suchasincomputingsystemsbuiltfromnuclearmagnetic-resonancedevices.UniversityofOklahomaresearchersstored1,024bitsinthemagneticinteractionsoftheprotonsofasinglemoleculecontainingnineteenhydrogenatoms.51Thus,thestateoftherockatanyonemomentrepresentsatleast1027bitsofmemory.Intermsofcomputation,andjustconsideringtheelectromagneticinteractions,thereareatleast1015changesinstateperbitpersecondgoingoninsidea2.2-poundrock,whicheffectivelyrepresentsabout1042 (amillion trillion trillion trillion)calculationspersecond.Yet therockrequiresnoenergyinputandgeneratesnoappreciableheat.Ofcourse,despiteallthisactivityattheatomiclevel,therockisnotperforminganyusefulworkasidefromperhapsactingasapaperweightoradecoration.Thereasonforthisisthatthestructureofthe atoms in the rock is for themostpart effectively random. If, on theotherhand,weorganize theparticlesinamorepurposefulmanner,wecouldhaveacool,zero-energy-consumingcomputerwithamemory of about a thousand trillion trillion bits and a processing capacity of 1042 operations persecond,whichisabouttentrilliontimesmorepowerfulthanallhumanbrainsonEarth,evenifweusethemostconservative(highest)estimateof1019cps.52 EdFredkindemonstrated thatwedon'tevenhavetobotherrunningalgorithmsinreverseafterobtaining a result.53 Fredkin presented several designs for reversible logic gates that perform thereversalsastheycomputeandthatareuniversal,meaningthatgeneral-purposecomputationcanbebuiltfromthem.54Fredkingoesontoshowthattheefficiencyofacomputerbuiltfromreversiblelogicgatescanbedesignedtobeveryclose(at least99percent) to theefficiencyofonesbuilt fromirreversiblegates.Hewrites:

itispossibleto...implement...conventionalcomputermodelsthathavethedistinctionthatthebasiccomponentsaremicroscopically.reversible.Thismeansthatthemacroscopicoperation

of the computer is also reversible.This fact allowsus to address the ... question ... "what isrequiredforacomputertobemaximallyefficient?"Theansweristhatifthecomputerisbuiltout of microscopically reversible components then it can be perfectly efficient. How muchenergydoes aperfectly efficient computerhave todissipate inorder to compute something?Theansweristhatthecomputerdoesnotneedtodissipateanyenergy.55

Reversiblelogichasalreadybeendemonstratedandshowstheexpectedreductionsinenergyinputandheatdissipation.56Fredkin'sreversiblelogicgatesanswerakeychallengetotheideaofreversiblecomputing: that it would require a different style of programming. He argues that we can, in fact,construct normal logic andmemory entirely from reversible logic gates,whichwill allow the use ofexistingconventionalsoftware-developmentmethods.Itishardtooverstatethesignificanceofthisinsight.AkeyobservationregardingtheSingularityisthat information processes—computation—will ultimately drive everything that is important. Thisprimaryfoundationforfuturetechnologythusappearstorequirenoenergy.Thepracticalrealityisslightlymorecomplicated.Ifweactuallywanttofindouttheresultsofacomputation—that is, to receive output from a computer—the process of copying the answer andtransmitting it outsideof thecomputer is an irreversibleprocess,one thatgeneratesheat for eachbittransmitted. However, for most applications of interest, the amount of computation that goes intoexecutinganalgorithmvastlyexceedsthecomputationrequiredtocommunicatethefinalanswers,sothelatterdoesnotappreciablychangetheenergyequation. However,becauseofessentiallyrandomthermalandquantumeffects, logicoperationshaveaninherenterrorrate.Wecanovercomeerrorsusingerror-detectionand-correctioncodes,buteachtimewe correct a bit, the operation is not reversible,whichmeans it requires energy and generates heat.Generally,errorratesarelow.Buteveniferrorsoccurattherateof,say,oneper1010operations,wehave only succeeded in reducing energy requirements by a factor of 1010, not in eliminating energydissipationaltogether.Asweconsiderthelimitsofcomputation,theissueoferrorratebecomesasignificantdesignissue.Certainmethodsofincreasingcomputationalrate,suchasincreasingthefrequencyoftheoscillationofparticles,alsoincreaseerrorrates,sothisputsnaturallimitsontheabilitytoperformcomputationusingmatterandenergy.Anotherimportanttrendwithrelevanceherewillbethemovingawayfromconventionalbatteriestoward tiny fuel cells (devices storing energy in chemicals, such as forms of hydrogen, which iscombined with available oxygen). Fuel cells are already being constructed using MEMS(microelectronicmechanicalsystems)technology.57Aswemovetowardthree-dimensional,molecularcomputingwithnanoscale features,energy resources in the formofnano-fuelcellswillbeaswidelydistributedthroughoutthecomputingmediumamongthemassivelyparallelprocessors.Wewilldiscussfuturenanotechnology-basedenergytechnologiesinchapter5.TheLimitsofNanocomputing.Evenwith the restrictionswehavediscussed, theultimate limits ofcomputers are profoundly high. Building onwork byUniversity of California at Berkeley professorHansBremermannandnanotechnologytheoristRobertFreitas,MITprofessorSethLloydhasestimatedthemaximumcomputationalcapacity,accordingtotheknownlawsofphysics,ofacomputerweighingonekilogramandoccupyingoneliterofvolume—aboutthesizeandweightofasmalllaptopcomputer—whathecalls the"ultimate laptop."58Thepotentialamountofcomputationriseswith theavailableenergy.Wecanunderstandthelinkbetweenenergyandcomputationalcapacityasfollows.Theenergyinaquantityofmatter is theenergyassociatedwitheachatom(andsubatomicparticle).Sothemoreatoms,themoreenergy.Asdiscussedabove,eachatomcanpotentiallybeusedforcomputation.Sothemoreatoms,themorecomputation.Theenergyofeachatomorparticlegrowswiththefrequencyofitsmovement: the more movement, the more energy. The same relationship exists for potentialcomputation:thehigherthefrequencyofmovement,themorecomputationeachcomponent(whichcan

beanatom)canperform.(Weseethisincontemporarychips:thehigherthefrequencyofthechip,thegreateritscomputationalspeed.)Sothereisadirectproportionalrelationshipbetweentheenergyofanobjectanditspotentialtoperform computation. The potential energy in a kilogram ofmatter is very large, as we know fromEinstein'sequationE=mc2, The speed of light squared is a very large number: approximately 1017

meter2/second2.Thepotentialofmattertocomputeisalsogovernedbyaverysmallnumber,Planck'sconstant:6.6Î10-34joule-seconds(ajouleisameasureofenergy).Thisisthesmallestscaleatwhichwe can apply energy for computation. We obtain the theoretical limit of an object to performcomputationbydividingthetotalenergy(theaverageenergyofeachatomorparticletimesthenumberofsuchparticles)byPlanck'sconstant.LloydshowshowthepotentialcomputingcapacityofakilogramofmatterequalspitimesenergydividedbyPlanck'sconstant.SincetheenergyissuchalargenumberandPlanck'sconstantissosmall,thisequationgeneratesanextremelylargenumber:about5Î1050operationspersecond.59Ifwerelatethatfiguretothemostconservativeestimateofhumanbraincapacity(1019cpsand1010humans),itrepresentstheequivalentofaboutfivebilliontrillionhumancivilizations.60Ifweusethefigureof1016cpsthatIbelievewillbesufficientforfunctionalemulationofhumanintelligence,theultimate laptop would function at the equivalent brain power of five trillion trillion humancivilizations.61 Such a laptop could perform the equivalent of all human thought over the last tenthousandyears(thatis,tenbillionhumanbrainsoperatingfortenthousandyears)inoneten-thousandthofananosecond.62Again,afewcaveatsareinorder.Convertingallofthemassofour2.2-poundlaptopintoenergyisessentiallywhathappensinathermonuclearexplosion.Ofcourse,wedon'twantthelaptoptoexplodebuttostaywithinitsone-literdimension.Sothiswillrequiresomecarefulpackaging,tosaytheleast.Byanalyzingthemaximumentropy(degreesoffreedomrepresentedbythestateofalltheparticles)insuchadevice,Lloyd shows that sucha computerwouldhavea theoreticalmemorycapacityof1031bits.It'sdifficulttoimaginetechnologiesthatwouldgoallthewayinachievingtheselimits.Butwecanreadilyenvisiontechnologies thatcomereasonablyclose todoingso.AstheUniversityofOklahomaproject shows,wealreadydemonstrated theability to storeat least fiftybitsof informationperatom(althoughonlyonasmallnumberofatoms,sofar).Storing1027bitsofmemoryinthe1025atomsinakilogramofmattershouldthereforebeeventuallyachievable.Butbecausemanypropertiesofeachatomcouldbeexploitedtostoreinformation—suchasthepreciseposition, spin,andquantumstateofallof itsparticles—wecanprobablydosomewhatbetterthan1027bits.NeuroscientistAndersSandbergestimates thepotentialstoragecapacityofahydrogenatomataboutfourmillionbits.Thesedensitieshavenotyetbeendemonstrated,however,sowe'llusethemoreconservativeestimate.63Asdiscussedabove,1042calculationspersecondcouldbeachievedwithoutproducingsignificantheat.Byfullydeployingreversiblecomputingtechniques,usingdesignsthatgeneratelowlevelsoferrors,andallowingforreasonableamountsofenergydissipation,weshouldendupsomewherebetween1042and1050calculationspersecond.Thedesignterrainbetweenthesetwolimitsiscomplex.Examiningthetechnicalissuesthatariseasweadvancefrom1042to1050isbeyondthescopeofthischapter.Weshouldkeepinmind,however,thatthewaythiswillplayoutisnotbystartingwiththeultimatelimitof1050andworkingbackwardbasedonvariouspracticalconsiderations.Rather,technologywillcontinuetorampup,alwaysusingitslatestprowesstoprogresstothenextlevel.Sooncewegettoacivilizationwith1042cps(forevery2.2pounds),thescientistsandengineersofthatdaywillusetheiressentiallyvastnonbiologicalintelligencetofigureouthowtoget1043,then1044,andsoon.Myexpectationisthatwewillgetveryclosetotheultimatelimits. Even at 1042 cps, a 2.2-pound "ultimate portable computer" would be able to perform theequivalentofallhumanthoughtoverthelasttenthousandyears(assumedattenbillionhumanbrains

fortenthousandyears)intenmicroseconds.64Ifweexaminethe"ExponentialGrowthofComputing"chart(p.70),weseethatthisamountofcomputingisestimatedtobeavailableforonethousanddollarsby2080.Amoreconservativebutcompellingdesignforamassivelyparallel,reversiblecomputerisEricDrexler'spatentednanocomputerdesign,whichisentirelymechanical.65Computationsareperformedbymanipulatingnanoscale rods,whichare effectively spring-loaded.After eachcalculation, the rodscontaining intermediate values return to their original positions, thereby implementing the reversecomputation. The device has a trillion (1012) processors and provides an overall rate of 1021 cps,enoughtosimulateonehundredthousandhumanbrainsinacubiccentimeter.SettingaDatefortheSingularity.Amoremodestbutstillprofoundthresholdwillbeachievedmuchearlier.Intheearly2030sonethousanddollars'worthofcomputationwillbuyabout1017cps(probablyaround1020cpsusingASICsandharvestingdistributedcomputationviatheInternet).Todaywespendmorethan$1011 ($100billion)oncomputation inayear,whichwillconservatively rise to$1012($1trillion)by2030.Sowewillbeproducingabout1026to1029cpsofnonbiologicalcomputationperyearintheearly2030s.Thisisroughlyequaltoourestimateforthecapacityofalllivingbiologicalhumanintelligence.Evenifjustequalincapacitytoourownbrains,thisnonbiologicalportionofourintelligencewillbemorepowerfulbecauseitwillcombinethepattern-recognitionpowersofhumanintelligencewiththememory- and skill-sharingability andmemoryaccuracyofmachines.Thenonbiologicalportionwillalwaysoperateatpeakcapacity,whichisfarfromthecaseforbiologicalhumanitytoday;the1026cpsrepresentedbybiologicalhumancivilizationtodayispoorlyutilized.Thisstateofcomputationintheearly2030swillnotrepresenttheSingularity,however,becauseitdoesnotyetcorrespondtoaprofoundexpansionofourintelligence.Bythemid-2040s,however,thatone thousanddollars'worthof computationwill be equal to1026 cps, so the intelligence createdperyear (at a total cost of about $1012) will be about one billion timesmore powerful than all humanintelligencetoday.66

Thatwill indeedrepresentaprofoundchange,andit isfor thatreasonthatIsetthedatefortheSingularity—representingaprofoundanddisruptivetransformationinhumancapability—as2045.Despitetheclearpredominanceofnonbiologicalintelligencebythemid-2040s,ourswillstillbeahumancivilization.Wewilltranscendbiology,butnotourhumanity.I'llreturntothisissueinchapter7.Returningtothelimitsofcomputationaccordingtophysics,theestimatesabovewereexpressedintermsoflaptop-sizecomputersbecausethatisafamiliarformfactortoday.Bytheseconddecadeofthiscentury,however,mostcomputingwillnotbeorganizedinsuchrectangulardevicesbutwillbehighlydistributedthroughouttheenvironment.Computingwillbeeverywhere:inthewalls,inourfurniture,inourclothing,andinourbodiesandbrains. And,ofcourse,humancivilizationwillnotbe limited tocomputingwith justa fewpoundsofmatter.Inchapter6,we'llexaminethecomputationalpotentialofanEarth-sizeplanetandcomputersonthescaleofsolarsystems,ofgalaxies,andoftheentireknownuniverse.Aswewillsee,theamountoftime required for our human civilization to achieve scales of computation and intelligence that gobeyondourplanetandintotheuniversemaybealotshorterthanyoumightthink.IsetthedatefortheSingularity—representingaprofoundanddisruptivetransformationinhumancapability—as2045.Thenonbiologicalintelligencecreatedinthatyearwillbeonebilliontimesmorepowerfulthanallhumanintelligencetoday.MemoryandComputationalEfficiency:ARockVersusaHumanBrain.With the limitsofmatterand energy to perform computation in mind, two useful metrics are the memory efficiency andcomputationalefficiencyofanobject.Thesearedefinedas thefractionsofmemoryandcomputationtakingplaceinanobjectthatareactuallyuseful.Also,weneedtoconsidertheequivalenceprinciple:evenifcomputationisuseful,ifasimplermethodproducesequivalentresults,thenweshouldevaluatethecomputationagainstthesimpleralgorithm.Inotherwords,iftwomethodsachievethesameresultbut one uses more computation than the other, the more computationally intensive method will beconsideredtouseonlytheamountofcomputationofthelessintensivemethod.67Thepurposeofthesecomparisonsistoassessjusthowfarbiologicalevolutionhasbeenabletogofrom systems with essentially no intelligence (that is, an ordinary rock, which performs no usefulcomputation)totheultimateabilityofmattertoperformpurposefulcomputation.Biologicalevolution

took us part of the way, and technological evolution (which, as I pointed out earlier, represents acontinuationofbiologicalevolution)willtakeusveryclosetothoselimits.Recallthata2.2-poundrockhasontheorderof1027bitsofinformationencodedinthestateofitsatoms and about 1042 cps represented by the activity of its particles. Sincewe are talking about anordinarystone,assumingthatitssurfacecouldstoreaboutonethousandbitsisaperhapsarbitrarybutgenerousestimate.68Thisrepresents10–24ofitstheoreticalcapacity,oramemoryefficiencyof10–24.69Wecanalsouseastonetodocomputation.Forexample,bydroppingthestonefromaparticularheight,wecancomputetheamountoftimeittakestodropanobjectfromthatheight.Ofcourse,thisrepresentsverylittlecomputation:perhaps1cps,meaningitscomputationalefficiencyis10–42.70Incomparison,whatcanwesayabouttheefficiencyofthehumanbrain?Earlierinthischapterwediscussedhoweachoftheapproximately1014interneuronalconnectionscanstoreanestimated104bitsin the connection's neurotransmitter concentrations and synaptic anddendritic nonlinearities (specificshapes),foratotalof1018bits.Thehumanbrainweighsaboutthesameasourstone(actuallycloserto3 pounds than 2.2, but since we're dealing with orders of magnitude, the measurements are closeenough).Itrunswarmerthanacoldstone,butwecanstillusethesameestimateofabout1027bitsoftheoretical memory capacity (estimating that we can store one bit in each atom). This results in amemory efficiency of 10–9. However, by the equivalence principle, we should not use the brain'sinefficientcodingmethodstorateitsmemoryefficiency.Usingourfunctionalmemoryestimateaboveof 1013 bits, we get amemory efficiency of 10–14. That's about halfway between the stone and theultimatecoldlaptoponalogarithmicscale.However,eventhoughtechnologyprogressesexponentially,ourexperiencesareinalinearworld,andonalinearscalethehumanbrainisfarclosertothestonethantotheultimatecoldcomputer. Sowhat is the brain's computational efficiency?Again,we need to consider the equivalenceprincipleandusetheestimateof1016cpsrequiredtoemulatethebrain'sfunctionality,ratherthanthehigher estimate (1019 cps) required to emulate all of the nonlinearities in every neuron. With thetheoreticalcapacityofthebrain'satomsestimatedat1042cps,thisgivesusacomputationalefficiencyof10–26.Again,that'sclosertoarockthantothelaptop,evenonalogarithmicscale. Ourbrainshaveevolvedsignificantly in theirmemoryandcomputationalefficiencyfrompre-biologyobjectssuchasstones.Butweclearlyhavemanyordersofmagnitudeofimprovementtotakeadvantageofduringthefirsthalfofthiscentury.Going Beyond the Ultimate: Pico- and Femtotechnology and Bending the Speed of Light. Thelimitsofaround1042cpsforaone-kilogram,one-litercoldcomputerandaround1050fora(very)hotone are based on computing with atoms. But limits are not always what they seem. New scientificunderstandinghasawayofpushingapparentlimitsaside.Asoneofmanysuchexamples,earlyinthehistoryofaviation,aconsensusanalysisofthelimitsofjetpropulsionapparentlydemonstratedthatjetaircraftwereinfeasible.71 The limits I discussed above represent the limits of nanotechnology based on our currentunderstanding. But what about picotechnology, measured in trillionths (10–12) of a meter, andfemtotechnology, scales of 10–15 of a meter? At these scales, we would require computing withsubatomicparticles.Withsuchsmallersizecomesthepotentialforevengreaterspeedanddensity. Wedohaveat least severalveryearly-adopterpicoscale technologies.Germanscientistshavecreatedanatomic-forcemicroscope(AFM)thatcanresolvefeaturesofanatomthatareonlyseventy-sevenpicometersacross72Anevenhigher-resolution technologyhasbeencreatedbyscientistsat theUniversity of California at Santa Barbara, who have developed an extremely sensitivemeasurementdetectorwithaphysicalbeammadeofgallium-arsenidecrystalandasensingsystemthatcanmeasureaflexingofthebeamofaslittleasonepicometer.ThedeviceisintendedtoprovideatestofHeisenberg'suncertaintyprinciple.73

In the timedimensionCornellUniversity scientistshavedemonstratedan imaging technologybased onX-ray scattering that can recordmovies of themovement of a single electron. Each framerepresentsonlyfourattoseconds(10–18seconds,eachoneabillionthofabillionthofasecond.74Thedevicecanachievespatialresolutionofoneangstrom(10–10meter,whichis100picometers).However,ourunderstandingofmatteratthesescales,particularlyinthefemtometerrange,isnotsufficientlywelldeveloped toproposecomputingparadigms.AnEnginesofCreation (EricDrexler'sseminal1986bookthatprovidedthefoundationsfornanotechnology)forpico-orfemtotechnologyhasnotyetbeenwritten.However,eachofthecompetingtheoriesforthebehaviorofmatterandenergyatthesescales isbasedonmathematicalmodels thatarebasedoncomputable transformations.Manyofthetransformationsinphysicsdoprovidethebasisforuniversalcomputation(that is, transformationsfromwhichwe can build general-purpose computers), and itmay be that behavior in the pico- andfemtometerrangewilldosoaswell. Of course, even if the basic mechanisms of matter in these ranges provide for universalcomputation in theory, we would still have to devise the requisite engineering to create massivenumbersofcomputingelementsandlearnhowtocontrolthem.Thesearesimilartothechallengesonwhichwe are now rapidlymakingprogress in the field of nanotechnology.At this time,wehave toregard the feasibility of pico- and femtocomputing as speculative. But nanocomputing will providemassive levels of intelligence, so if it's at all possible to do, our future intelligencewill be likely tofigureoutthenecessaryprocesses.Thementalexperimentweshouldbemakingisnotwhetherhumansaswe know them todaywill be capable of engineering pico- and femtocomputing technologies, butwhether the vast intelligence of future nanotechnology-based intelligence (which will be trillions oftrillions of timesmore capable than contemporary biological human intelligence)will be capable ofrenderingthesedesigns.AlthoughIbelieveitislikelythatourfuturenanotechnology-basedintelligencewillbeable toengineercomputationatscalesfiner thannanotechnology, theprojections in thisbookconcerningtheSingularitydonotrelyonthisspeculation.Inadditiontomakingcomputingsmaller,wecanmakeitbigger—thatis,wecanreplicatethesevery small devices on amassive scale.With full-scale nanotechnology, computing resources can bemadeself-replicatingandthuscanrapidlyconvertmassandenergyintoanintelligentform.However,werunupagainstthespeedoflight,becausethematterintheuniverseisspreadoutovervastdistances.Aswewilldiscusslater,thereareatleastsuggestionsthatthespeedoflightmaynotbeimmutable.Physicists Steve Lamoreaux and Justin Torgerson of the Los Alamos National Laboratory haveanalyzeddatafromanoldnaturalnuclearreactorthattwobillionyearsagoproducedafissionreactionlastingseveralhundredthousandyears inwhat isnowWestAfrica.75Examiningradioactive isotopesleft over from the reactor andcomparing them to isotopes fromsimilarnuclear reactions today, theydetermined that thephysics constant alpha (also called the fine-structure constant),whichdeterminesthestrengthoftheelectromagneticforce,apparentlyhaschangedovertwobillionyears.Thisisofgreatsignificancetotheworldofphysics,becausethespeedoflightisinverselyproportionaltoalpha,andbothhavebeenconsideredunchangeableconstants.Alphaappearstohavedecreasedby4.5partsoutof108.Ifconfirmed,thiswouldimplythatthespeedoflighthasincreased.Ofcourse,theseexploratoryresultswillneedtobecarefullyverified.Iftrue,theymayholdgreatimportanceforthefutureofourcivilization.Ifthespeedoflighthasincreased,ithaspresumablydonesonotjustasaresultofthepassageoftimebutbecausecertainconditionshavechanged.Ifthespeedoflight has changeddue to changing circumstances, that cracks open the door just enough for the vastpowers of our future intelligence and technology to swing the doorwidely open.This is the type ofscientific insight that technologists can exploit. Human engineering often takes a natural, frequentlysubtle,effect,andcontrolsitwithaviewtowardgreatlyleveragingandmagnifyingit.Evenifwefinditdifficulttosignificantlyincreasethespeedoflightoverthelongdistancesofspace, doing so within the small confines of a computing device would also have importantconsequences forextending thepotential forcomputation.Thespeedof light isoneof the limits thatconstrain computing devices even today, so the ability to boost itwould extend further the limits ofcomputation. We will explore several other intriguing approaches to possibly increasing, or

circumventing, thespeedof light inchapter6.Expandingthespeedof light is,ofcourse,speculativetoday,andnoneoftheanalysesunderlyingourexpectationoftheSingularityrelyonthispossibility.Going Back in Time. Another intriguing—and highly speculative—possibility is to send acomputational process back in time through a "wormhole" in space-time.Theoretical physicist ToddBrunoftheInstituteforAdvancedStudiesatPrincetonhasanalyzedthepossibilityofcomputingusingwhathecallsa"closedtimelikecurve"(CTC).AccordingtoBrun,CTCscould"sendinformation(suchastheresultofcalculations)intotheirownpastlightcones."76Brundoesnotprovideadesignforsuchadevicebutestablishesthatsuchasystemisconsistentwith the lawsofphysics.His time-travelingcomputeralsodoesnotcreate the"grandfatherparadox,"oftencitedindiscussionsoftimetravel.Thiswell-knownparadoxpointsoutthatifpersonAgoesbackintime,hecouldkillhisgrandfather,causingAnottoexist,resultinginhisgrandfathernotbeingkilledbyhim,soAwouldexistandthuscouldgobackandkillhisgrandfather,andsoon,adinfinitum.Brun'stime-stretchingcomputationalprocessdoesnotappeartointroducethisproblembecauseitdoesnotaffect thepast. Itproducesadeterminateandunambiguousanswerin thepresent toaposedquestion.Thequestionmusthaveadearanswer,andtheanswerisnotpresenteduntilafterthequestionisasked,althoughtheprocesstodeterminetheanswercantakeplacebeforethequestionisaskedusingtheCTC.Conversely, theprocesscould takeplaceafter thequestion isaskedand thenuseaCTCtobring the answer back into the present (but not before the question was asked, because that wouldintroduce the grandfather paradox). Theremay verywell be fundamental barriers (or limitations) tosuchaprocessthatwedon'tyetunderstand,but thosebarriershaveyettobeidentified.Iffeasible, itwouldgreatlyexpandthepotentialof localcomputation.Again,allofmyestimatesofcomputationalcapacitiesandofthecapabilitiesoftheSingularitydonotrelyonBrun'stentativeconjecture.ERICDREXLER: I don't know, Ray. I'm pessimistic on the prospects for picotechnology.With the

stable particles we know of, I don't see how there can be picoscale structure without theenormouspressures found inacollapsedstar—awhitedwarforaneutronstar—and thenyouwouldgetasolidchunkofstufflikeametal,butamilliontimesdenser.Thisdoesn'tseemveryuseful,evenifitwerepossibletomakeitinoursolarsystem.Ifphysicsincludedastableparticlelikean electronbut ahundred timesmoremassive, itwouldbeadifferent story, butwedon'tknowofone.

RAY:Wemanipulate subatomic particles todaywith accelerators that fall significantly short of theconditions in a neutron star. Moreover, we manipulate subatomic particles such as electronstodaywithtabletopdevices.Scientistsrecentlycapturedandstoppedaphotondeadinitstracks.

ERIC: Yes, but what kind of manipulation? If we count manipulating small particles, then alltechnology is already picotechnology, because all matter is made of subatomic particles.Smashingparticlestogetherinacceleratorsproducesdebris,notmachinesorcircuits.

RAY:Ididn’tsaywe'vesolvedtheconceptualproblemsofpicotechnology.I'vegotyoupenciledintodothatin2072.

ERIC:Oh,good,thenIseeyouhavemelivingalongtime.RAY:Yes,well,ifyoustayonthesharpleadingedgeofhealthandmedicalinsightsandtechnology,as

I'mtryingtodo,Iseeyoubeinginrathergoodshapearoundthen.MOLLY2104:Yes,quiteafewofyoubabyboomersdidmakeitthrough.Butmostwereunmindfulof

the opportunities in 2004 to extend human mortality long enough to take advantage of thebiotechnology revolution, which hit its stride a decade later, followed by nanotechnology adecadeafterthat.

MOLLY2004:So,Molly2104,youmustbequitesomething,consideringthatonethousanddollarsofcomputation in 2080 can perform the equivalent of ten billion human brains thinking for tenthousand years in a matter of ten microseconds. That presumably will have progressed evenfurther by 2104, and I assume you have access to more than one thousand dollars' worth ofcomputation.

MOLLY2104:Actually,millionsofdollarsonaverage—billionswhenIneedit.MOLLY2004:That'sprettyhardtoimagine.MOLLY2104:Yeah,well,IguessI'mkindofsmartwhenIneedtobe.MOLLY2004:Youdon'tsoundthatbright,actually.MOLLY2104:I'mtryingtorelateonyourlevel.MOLLY2004:Now,waitasecond,MissMollyofthefuture....GEORGE2048:Ladies,please,you'rebothveryengaging.MOLLY2004:Yes,well,tellthattomycounterparthere—shefeelsshe'sajilliontimesmorecapable

thanIam.GEORGE2048:She is your future, you know.Anyway, I've always felt therewas something special

aboutabiologicalwoman.MOLLY2104: Yeah, what would you know about biological women anyway?GEORGE2048: I've

readagreatdealaboutitandengagedinsomeveryprecisesimulations.MOLLY2004: It occurs to me that maybe you're bothmissing something that you're not aware of

GEORGE2048:Idon'tseehowthat'spossible.MOLLY2104:Definitelynot.MOLLY2004:Ididn'tthinkyouwould.ButthereisonethingIunderstandyoucandothatIdofind

cool.MOLLY 2104: Just one?MOLLY 2004: One that I'm thinking of, anyway. You can merge your

thinkingwithsomeoneelseandstillkeepyourseparateidentityatthesametime.MOLLY2104:Ifthesituation—andtheperson—isright,then,yes,it'saverysublimethingtodo.MOLLY2004:Likefallinginlove?MOLLY2104:Likebeinginlove.It'stheultimatewaytoshare.GEORGE2048:Ithinkyou'llgoforit,Molly2004.MOLLY2104:Yououghttoknow,George,sinceyouwerethefirstpersonIdiditwith.

CHAPTERFOUR

AchievingtheSoftwareofHumanIntelligence:HowtoReverseEngineertheHumanBrain

There aregood reasons tobelieve thatweare at a turningpoint, and that itwill bepossiblewithin thenext twodecades to formulateameaningfulunderstandingofbrain function.Thisoptimisticviewisbasedonseveralmeasurabletrends,andasimpleobservationwhichhasbeenproven repeatedly in thehistoryof science:Scientificadvancesare enabledbya technologyadvancethatallowsustoseewhatwehavenotbeenabletoseebefore.Atabouttheturnofthetwenty-firstcentury,wepassedadetectableturningpointinbothneuroscienceknowledgeandcomputingpower.For the first time inhistory,we collectivelyknowenough about ourownbrains,andhavedevelopedsuchadvancedcomputing technology, thatwecannowseriouslyundertaketheconstructionofaverifiable,real-time,high-resolutionmodelofsignificantpartsofourintelligence.

—LLOYDWATTS,NEUROSCIENTIST1Now,forthefirsttime,weareobservingthebrainatworkinaglobalmannerwithsuchclaritythatweshouldbeabletodiscovertheoverallprogramsbehinditsmagnificentpowers.

—J.G.TAYLOR,B.HORWITZ,K.J.FRISTON,NEUROSCIENTISTS2

The brain is good: it is an existence proof that a certain arrangement ofmatter can producemind,performintelligentreasoning,patternrecognition, learningandalotofother importanttasks of engineering interest. Hencewe can learn to build new systems by borrowing ideasfrom thebrain....Thebrain isbad: it is anevolved,messy systemwherea lotof interactionshappen because of evolutionary contingencies. ... On the other hand, itmust also be robust(sincewecansurvivewith it)andbeable tostandfairlymajorvariationsandenvironmentalinsults, so the trulyvaluable insight fromthebrainmightbehow tocreate resilientcomplexsystemsthatself-organizewell....Theinteractionswithinaneuronarecomplex,butonthenextlevel neurons seem to be somewhat simple objects that can be put together flexibly intonetworks. The cortical networks are a real mess locally, but again on the next level theconnectivity isn't that complex. Itwould be likely that evolution has produced a number ofmodulesor repeating themes thatarebeing re-used,andwhenweunderstand themand theirinteractionswecandosomethingsimilar.

—ANDERS SANDBERG, COMPUTATIONAL NEUROSCIENTIST,ROYALINSTITUTEOFTECHNOLOGY,SWEDEN

T

ReverseEngineeringtheBrain:AnOverviewoftheTaskhecombinationofhuman-levelintelligencewithacomputer'sinherentsuperiorityinspeed,accuracy,andmemory-sharingabilitywillbeformidable.Todate,however,mostAIresearchanddevelopmenthasutilizedengineeringmethodsthatarenotnecessarilybasedonhowthehumanbrainfunctions,forthesimplereasonthatwehavenothadtheprecisetoolsneededtodevelopdetailedmodelsofhumancognition.

Ourability to reverseengineer thebrain—tosee inside,model it, andsimulate its regions—isgrowing exponentially.Wewill ultimately understand the principles of operation underlying the fullrangeofourownthinking,knowledgethatwillprovideuswithpowerfulproceduresfordevelopingthesoftwareofintelligentmachines.Wewillmodify,refine,andextendthesetechniquesasweapplythemtocomputationaltechnologiesthatarefarmorepowerfulthantheelectrochemicalprocessingthattakesplaceinbiologicalneurons.Akeybenefitofthisgrandprojectwillbethepreciseinsightsitoffersintoourselves.Wewill alsogainpowerful newways to treat neurological problems such asAlzheimer's,stroke, Parkinson's disease, and sensory disabilities, and ultimatelywill be able to vastly extend ourintelligence.NewBrain-ImagingandModelingTools.Thefirststepinreverseengineeringthebrainistopeerintothebraintodeterminehowitworks.Sofar,ourtoolsfordoingthishavebeencrude,but that isnowchanging, as a significantnumberofnewscanning technologies featuregreatly improved spatial andtemporal resolution, price-performance, and bandwidth. Simultaneouslywe are rapidly accumulatingdata on the precise characteristics and dynamics of the constituent parts and systems of the brain,rangingfromindividualsynapsestolargeregionssuchasthecerebellum,whichcomprisesmorethanhalfofthebrain'sneurons.Extensivedatabasesaremethodicallycatalogingourexponentiallygrowingknowledgeofthebrain.3Researchershavealsoshowntheycanrapidlyunderstandandapplythisinformationbybuildingmodels andworking simulations. These simulations of brain regions are based on themathematicalprinciples of complexity theory and chaotic computing and are alreadyproviding results that closelymatchexperimentsperformedonactualhumanandanimalbrains.Asnotedinchapter2,thepowerofthescanningandcomputationaltoolsneededforthetaskofreverse engineering the brain is accelerating, similar to the acceleration in technology thatmade thegenomeprojectfeasible.Whenwegettothenanobotera(see"ScanningUsingNanobots"onp.163),wewill be able to scan from inside the brainwith exquisitely high spatial and temporal resolution.4Therearenoinherentbarrierstoourbeingabletoreverseengineertheoperatingprinciplesofhumanintelligence and replicate these capabilities in the more powerful computational substrates that willbecomeavailableinthedecadesahead.Thehumanbrainisacomplexhierarchyofcomplexsystems,butitdoesnotrepresentalevelofcomplexitybeyondwhatwearealreadycapableofhandling.The Software of theBrain. The price-performance of computation and communication is doublingeveryyear.Aswesawearlier,thecomputationalcapacityneededtoemulatehumanintelligencewillbeavailable in less than two decades.5 A principal assumption underlying the expectation of theSingularity is thatnonbiologicalmediumswillbeable toemulate the richness,subtlety,anddepthofhumanthinking.Butachievingthehardwarecomputationalcapacityofasinglehumanbrain—orevenof thecollective intelligenceofvillagesandnations—willnotautomaticallyproducehuman levelsofcapability. (By “human levels" I include all the diverse and subtle ways humans are intelligent,includingmusicalandartisticaptitude,creativity,physicalmotionthroughtheworld,andunderstandingandrespondingappropriately toemotions.)Thehardwarecomputationalcapacity isnecessarybutnotsufficient.Understandingtheorganizationandcontentoftheseresources—thesoftwareofintelligence—isevenmorecriticalandistheobjectiveofthebrainreverse-engineeringundertaking. Onceacomputerachievesahumanlevelof intelligence, itwillnecessarilysoarpast it.Akey

advantageofnonbiologicalintelligenceisthatmachinescaneasilysharetheirknowledge.IfyoulearnFrenchorreadWarandPeace,youcan'treadilydownloadthatlearningtome,asIhavetoacquirethatscholarship the same painstaking way that you did. I can't (yet) quickly access or transmit yourknowledge,whichisembeddedinavastpatternofneurotransmitterconcentrations(levelsofchemicalsinthesynapsesthatallowoneneurontoinfluenceanother)andinterneuronalconnections(portionsoftheneuronscalledaxonsanddendritesthatconnectneurons).Butconsiderthecaseofamachine'sintelligence.Atoneofmycompanies,wespentyearsteachingoneresearchcomputerhowtorecognizecontinuoushumanspeech,usingpattern-recognitionsoftware.6Weexposedittothousandsofhoursofrecordedspeech,correcteditserrors,andpatientlyimproveditsperformancebytrainingits"chaotic"self-organizingalgorithms(methodsthatmodifytheirownrules,based on processes that use semirandom initial information, and with results that are not fullypredictable).Finally, thecomputerbecamequiteadeptat recognizingspeech.Now, ifyouwantyourown personal computer to recognize speech, you don't have to put it through the same painstakinglearningprocess(aswedowitheachhumanchild);youcansimplydownloadthealreadyestablishedpatternsinseconds.AnalyticVersusNeuromorphicModelingoftheBrain.Agoodexampleofthedivergencebetweenhuman intelligence and contemporary AI is how each undertakes the solution of a chess problem.Humansdosobyrecognizingpatterns,whilemachinesbuildhugelogical"trees"ofpossiblemovesandcountermoves.Mosttechnology(ofallkinds)todatehasusedthislattertypeof"top-down,"analytic,engineeringapproach.Ourflyingmachines,forexample,donotattempttore-createthephysiologyandmechanicsofbirds.Butasourtoolsforreverseengineeringthewaysofnaturearegrowingrapidlyinsophistication,technologyismovingtowardemulatingnaturewhileimplementingthesetechniquesinfarmorecapablesubstrates.Themostcompellingscenarioformasteringthesoftwareofintelligenceistotapdirectlyintotheblueprint of the best example we can get our hands on of an intelligent process: the human brain.Althoughittookitsoriginal"designer"(evolution)severalbillionyearstodevelopthebrain,it'sreadilyavailabletous,protectedbyaskullbutwiththerighttoolsnothiddenfromourview.Itscontentsarenot yet copyrighted or patented. (We can, however, expect that to change; patent applications havealready been filed based on brain reverse engineering.)7Wewill apply the thousands of trillions ofbytes of information derived from brain scans and neural models at many levels to design moreintelligentparallelalgorithmsforourmachines,particularlythosebasedonself-organizingparadigms. With this self-organizing approach,we don't have to attempt to replicate every single neuralconnection.Thereisagreatdealofrepetitionandredundancywithinanyparticularbrainregion.Wearediscoveringthathigher-levelmodelsofbrainregionsareoftensimplerthanthedetailedmodelsoftheirneuronalcomponents.HowComplexIstheBrain?Althoughtheinformationcontainedinahumanbrainwouldrequireontheorderofonebillionbillionbits(seechapter3),theinitialdesignofthebrainisbasedontherathercompacthumangenome.Theentiregenomeconsistsofeighthundredmillionbytes,butmostof it isredundant, leaving only about thirty to one hundred million bytes (less than 109 bits) of uniqueinformation(aftercompression),whichissmallerthantheprogramforMicrosoftWord.8Tobefair,weshouldalsotakeintoaccount"epigenetic"data,whichisinformationstoredinproteinsthatcontrolgeneexpression(thatis,thatdeterminewhichgenesareallowedtocreateproteinsineachcell),aswellastheentire protein-replication machinery, such as the ribosomes and a host of enzymes. However, suchadditionalinformationdoesnotsignificantlychangetheorderofmagnitudeofthiscalculation.9Slightlymore than half of the genetic and epigenetic information characterizes the initial state of the humanbrain.Ofcourse,thecomplexityofourbrainsgreatlyincreasesasweinteractwiththeworld(byafactorofaboutonebillionoverthegenome).10Buthighlyrepetitivepatternsarefoundineachspecificbrain

region, so it is not necessary to capture each particular detail to successfully reverse engineer therelevantalgorithms,whichcombinedigitalandanalogmethods(forexample,thefiringofaneuroncanbeconsideredadigitaleventwhereasneurotransmitterlevelsinthesynapsecanbeconsideredanalogvalues).Thebasicwiringpatternofthecerebellum,forexample,isdescribedinthegenomeonlyoncebutrepeatedbillionsoftimes.Withtheinformationfrombrainscanningandmodelingstudies,wecandesignsimulated"neuromorphic"equivalentsoftware(thatis,algorithmsfunctionallyequivalenttotheoverallperformanceofabrainregion). Thepaceofbuildingworkingmodelsandsimulationsisonlyslightlybehindtheavailabilityofbrain-scanningandneuron-structureinformation.Therearemorethanfiftythousandneuroscientistsintheworld,writingarticlesformorethanthreehundredjournals.11Thefieldisbroadanddiverse,withscientists and engineers creating new scanning and sensing technologies and developingmodels andtheories at many levels. So even people in the field are often not completely aware of the fulldimensionsofcontemporaryresearch.ModelingtheBrain.Incontemporaryneuroscience,modelsandsimulationsarebeingdevelopedfromdiverse sources, including brain scans, interneuronal connection models, neuronal models, andpsychophysicaltesting.Asmentionedearlier,auditory-systemresearcherLloydWattshasdevelopedacomprehensive model of a significant portion of the human auditory-processing system fromneurobiologystudiesofspecificneurontypesandinterneuronal-connectioninformation.Watts'smodelincludesfiveparallelpathsandtheactualrepresentationsofauditoryinformationateachstageofneuralprocessing.Wattshasimplementedhismodel inacomputerasreal-timesoftwarethatcanlocateandidentifysoundsandfunctions,similartothewayhumanhearingoperates.Althoughaworkinprogress,themodelillustratesthefeasibilityofconvertingneurobiologicalmodelsandbrain-connectiondataintoworkingsimulations.AsHansMoravecandothershavespeculated,theseefficientfunctionalsimulationsrequireaboutonethousandtimeslesscomputationthanwouldberequiredifwesimulatedthenonlinearitiesineachdendrite,synapse,andothersubneuralstructureintheregionbeingsimulated.(AsIdiscussedinchapter3,wecanestimatethecomputationrequiredforfunctionalsimulationofthebrainat1016calculationspersecond[cps],versus1019cpstosimulatethesubneuralnonlinearities.)12 The actual speed ratiobetween contemporary electronics and the electrochemical signaling inbiologicalinterneuronalconnectionsisatleastonemilliontoone.Wefindthissameinefficiencyinallaspects of our biology, because biological evolution built all of itsmechanisms and systemswith aseverelyconstrainedsetofmaterials:namely,cells,whichare themselvesmadefroma limitedsetofproteins.Althoughbiologicalproteinsarethree-dimensional, theyarerestrictedtocomplexmoleculesthatcanbefoldedfromalinear(one-dimensional)sequenceofaminoacids.PeelingtheOnion.Thebrain isnotasingle information-processingorganbut ratheran intricateandintertwined collection of hundreds of specialized regions. The process of "peeling the onion" tounderstand the functions of these interleaved regions is well under way. As the requisite neurondescriptionsandbrain-interconnectiondatabecomeavailable,detailedandimplementablereplicassuchasthesimulationoftheauditoryregionsdescribedbelow(see"AnotherExample:Watts'sModeloftheAuditoryRegions”onp.183)willbedevelopedforallbrainregions.Mostbrain-modelingalgorithmsarenotthesequential,logicalmethodsthatarecommonlyusedindigitalcomputingtoday.Thebraintendstouseself-organizing,chaotic,holographicprocesses(thatis,informationnot located inoneplacebutdistributed throughoutaregion). It isalsomassivelyparalleland utilizes hybrid digital-controlled analog techniques. However, a wide range of projects hasdemonstratedourabilitytounderstandthesetechniquesandtoextractthemfromourrapidlyescalatingknowledgeofthebrainanditsorganization.Afterthealgorithmsofaparticularregionareunderstood,theycanberefinedandextendedbeforebeingimplementedinsyntheticneuralequivalents.Theycanberunonacomputationalsubstratethatisalready far faster than neural circuitry. (Current computers perform computations in billionths of a

second,comparedtothousandthsofasecondforinterneuronaltransactions.)Andwecanalsomakeuseofthemethodsforbuildingintelligentmachinesthatwealreadyunderstand.IstheHumanBrainDifferentfromaComputer?Theanswertothisquestiondependsonwhatwemeanbytheword"computer."Mostcomputerstodayarealldigitalandperformone(orperhapsafew)computationsata timeatextremelyhighspeed.Incontrast,thehumanbraincombinesdigitalandanalogmethodsbutperformsmostcomputationsintheanalog(continuous)domain,usingneurotransmittersandrelatedmechanisms.Althoughtheseneuronsexecutecalculationsatextremelyslowspeeds(typicallytwohundredtransactionspersecond),thebrainas awhole ismassively parallel:most of its neuronswork at the same time, resulting in up to onehundredtrillioncomputationsbeingcarriedoutsimultaneously.Themassiveparallelismofthehumanbrainisthekeytoitspattern-recognitionability,whichisoneofthepillarsofourspecies'thinking.Mammalianneuronsengageinachaoticdance(thatis,withmanyapparentlyrandominteractions),andif theneuralnetworkhas learnedits lessonswell,astablepatternwillemerge,reflectingthenetwork'sdecision.Atthepresent,paralleldesignsforcomputersaresomewhat limited. But there is no reason why functionally equivalent nonbiological re-creations ofbiologicalneuralnetworkscannotbebuiltusingtheseprinciples.Indeed,dozensofeffortsaroundtheworld have already succeeded in doing so. My own technical field is pattern recognition, and theprojects that I have been involved in for about forty years use this form of trainable andnondeterministiccomputing.Manyofthebrain'scharacteristicmethodsoforganizationcanalsobeeffectivelysimulatedusingconventionalcomputingofsufficientpower.Duplicatingthedesignparadigmsofnaturewill,Ibelieve,be akey trend in future computing.We shouldkeep inmind, aswell, that digital computing canbefunctionallyequivalent toanalogcomputing—that is,wecanperformallof the functionsofahybriddigital-analognetworkwithanall-digitalcomputer.Thereverseisnottrue:wecan'tsimulateallofthefunctionsofadigitalcomputerwithananalogone. However,analogcomputingdoeshaveanengineeringadvantage: it ispotentially thousandsoftimesmoreefficient.Ananalogcomputationcanbeperformedbyafewtransistorsor, in thecaseofmammalian neurons, specific electrochemical processes. A digital computation, in contrast, requiresthousandsortensofthousandsoftransistors.Ontheotherhand,thisadvantagecanbeoffsetbytheeaseofprogramming(andmodifying)digitalcomputer-basedsimulations.Thereareanumberofotherkeywaysinwhichthebraindiffersfromaconventionalcomputer:

·Thebrain'scircuitsareveryslow.Synaptic-resetandneuron-stabilizationtimes(theamountoftimerequiredforaneuronanditssynapsestoresetthemselvesaftertheneuronfires)aresoslowthatthereareveryfewneuron-firingcyclesavailabletomakepattern-recognitiondecisions.Functionalmagnetic-resonance imaging (fMRI) and magnetoencephalography (MEG) scans show thatjudgments thatdonot requireresolvingambiguitiesappear tobemade inasingleneuron-firingcycle (less than twenty milliseconds), involving essentially no iterative (repeated) processes.Recognition of objects occurs in about 150milliseconds, so that even if we "think somethingover," the number of cycles of operation is measured in hundreds or thousands at most, notbillions,aswithatypicalcomputer.

·But it's massively parallel. The brain has on the order of one hundred trillion interneuronalconnections, each potentially processing information simultaneously. These two factors (slowcycle time andmassive parallelism) result in a certain level of computational capacity for thebrain,aswediscussedearlier.

Todayour largest supercomputers are approaching this range.The leading supercomputers(includingthoseusedbythemostpopularsearchengines)measureover1014cps,whichmatchesthe lower range of the estimates I discussed in chapter 3 for functional simulation. It is not

necessary,however,tousethesamegranularityofparallelprocessingasthebrainitselfsolongaswematch theoverallcomputationalspeedandmemorycapacityneededandotherwisesimulatethebrain'smassivelyparallelarchitecture.

·The brain combines analog and digital phenomena. The topology of connections in the brain isessentially digital—a connection exists, or it doesn't. An axon firing is not entirely digital butcloselyapproximatesadigitalprocess.Mosteveryfunctioninthebrainisanalogandisfilledwithnonlinearities(suddenshiftsinoutput,ratherthanlevelschangingsmoothly)thataresubstantiallymore complex than the classical model that we have been using for neurons. However, thedetailed,nonlineardynamicsofaneuronandallof itsconstituents (dendrites, spines,channels,andaxons) canbemodeled through themathematicsofnonlinear systems.Thesemathematicalmodels can then be simulated on a digital computer to any desired degree of accuracy. As Imentioned,ifwesimulatetheneuralregionsusingtransistorsintheirnativeanalogmoderatherthan throughdigital computation, thisapproachcanprovide improvedcapacityby threeor fourordersofmagnitude,asCarverMeadhasdemonstrated.13

·Thebrainrewiresitself.Dendritesarecontinuallyexploringnewspinesandsynapses.Thetopologyandconductanceofdendritesandsynapsesarealsocontinuallyadapting.Thenervoussystemisself-organizing at all levels of its organization. While the mathematical techniques used incomputerized pattern-recognition systems such as neural nets and Markov models are muchsimpler than those used in the brain,we do have substantial engineering experiencewith self-organizing models.14 Contemporary computers don't literally rewire themselves (althoughemerging "self-healing systems" are starting to do this), but we can effectively simulate thisprocessinsoftware.15 In the future,wecan implement this inhardware,aswell,although theremay be advantages to implementing most self-organization in software, which provides moreflexibilityforprogrammers.

·Mostofthedetailsinthebrainarerandom.Whilethereisagreatdealofstochastic(randomwithincarefullycontrolledconstraints)processineveryaspectofthebrain,itisnotnecessarytomodelevery"dimple"onthesurfaceofeverydendrite,anymorethanitisnecessarytomodeleverytinyvariation in the surface of every transistor in understanding the principles of operation of acomputer.Butcertaindetailsarecriticalindecodingtheprinciplesofoperationofthebrain,whichcompelsustodistinguishbetweenthemandthosethatcomprisestochastic"noise"orchaos.Thechaotic (random and unpredictable) aspects of neural function can be modeled using themathematicaltechniquesofcomplexitytheoryandchaostheory.16

·The brain uses emergent properties. Intelligent behavior is an emergent property of the brain'schaoticandcomplexactivity.Considertheanalogytotheapparentlyintelligentdesignoftermiteandantcolonies,withtheirdelicatelyconstructedinterconnectingtunnelsandventilationsystems.Despite their clever and intricate design, ant and termite hills have no master architects; thearchitecture emerges from the unpredictable interactions of all the colony members, eachfollowingrelativelysimplerules.

·Thebrainisimperfect.Itisthenatureofcomplexadaptivesystemsthattheemergentintelligenceofits decisions is suboptimal. (That is, it reflects a lower level of intelligence than would berepresentedbyanoptimalarrangementofitselements.}Itneedsonlytobegoodenough,whichinthe case of our species meant a level of intelligence sufficient to enable us to outwit thecompetitorsinourecologicalniche(forexample,primateswhoalsocombineacognitivefunctionwithanopposableappendagebutwhosebrainsarenotasdevelopedashumansandwhosehandsdonotworkaswell).

·We contradict ourselves.A variety of ideas and approaches, including conflicting ones, leads tosuperioroutcomes.Ourbrainsarequitecapableofholdingcontradictoryviews.Infact,wethriveonthisinternaldiversity.Considertheanalogytoahumansociety,particularlyademocraticone,withitsconstructivewaysofresolvingmultipleviewpoints.

·Thebrainusesevolution.Thebasiclearningparadigmusedbythebrainisanevolutionaryone:thepatternsofconnectionsthataremostsuccessfulinmakingsenseoftheworldandcontributingto

recognitions and decisions survive. A newborn's brain contains mostly randomly linkedinterneuronalconnections,andonlyaportionofthosesurviveinthetwo-year-oldbrain.17

·Thepatternsareimportant.Certaindetailsofthesechaoticself-organizingmethods,expressedasmodelconstraints (rulesdefining the initial conditionsand themeans for self-organization), arecrucial,whereasmanydetailswithin theconstraints are initially set randomly.The system thenself-organizes and gradually represents the invariant features of the information that has beenpresentedto thesystem.Theresultinginformationisnotfoundinspecificnodesorconnectionsbutratherisadistributedpattern.

·Thebrainisholographic.Thereisananalogybetweendistributedinformationinahologramandthe method of information representation in brain networks. We find this also in the self-organizing methods used in computerized pattern recognition, such as neural nets, Markovmodels,andgeneticalgorithms.18

·Thebrainisdeeplyconnected.Thebraingetsitsresiliencefrombeingadeeplyconnectednetworkin which information has many ways of navigating from one point to another. Consider theanalogy to the Internet,which has become increasingly stable as the number of its constituentnodeshasincreased.Nodes,evenentirehubsoftheInternet,canbecomeinoperativewithouteverbringing down the entire network. Similarly,we continually lose neuronswithout affecting theintegrityoftheentirebrain.

·Thebraindoeshaveanarchitectureofregions.Althoughthedetailsofconnectionswithinaregionare initially random within constraints and self-organizing, there is an architecture of severalhundred regions that perform specific functions, with specific patterns of connections betweenregions.

·Thedesignofabrainregionissimplerthanthedesignofaneuron.Modelsoftengetsimpleratahigherlevel,notmorecomplex.Considerananalogywithacomputer.Wedoneedtounderstandthedetailedphysicsofsemiconductorstomodelatransistor,andtheequationsunderlyingasinglereal transistor are complex. However, a digital circuit that multiplies two numbers, althoughinvolvinghundredsoftransistors,canbemodeledfarmoresimply,withonlyafewformulas.Anentirecomputerwithbillionsoftransistorscanbemodeledthroughitsinstructionsetandregisterdescription, which can be described on a handful of written pages of text and mathematicaltransformations.

The software programs for an operating system, language compilers, and assemblers arereasonably complex, butmodeling a particular program—for example, a speech-recognition programbasedonMarkovmodeling—maybedescribed inonlya fewpagesofequations.Nowhere insuchadescriptionwouldbefoundthedetailsofsemiconductorphysics.Asimilarobservationalsoholdstrueforthebrain.Aparticularneuralarrangementthatdetectsaparticularinvariantvisualfeature(suchasaface)orthatperformsaband-passfiltering(restrictinginputtoaspecificfrequencyrange)operationonauditoryinformationorthatevaluatesthetemporalproximityoftwoeventscanbedescribedwithfargreatersimplicity thantheactualphysicsandchemicalrelationscontrolling theneurotransmittersandothersynapticanddendriticvariablesinvolvedintherespectiveprocesses.Althoughallofthisneuralcomplexitywillhavetobecarefullyconsideredbeforeadvancingtothenexthigherlevel(modelingthebrain),muchofitcanbesimplifiedoncetheoperatingprinciplesofthebrainareunderstood.

TryingtoUnderstandOurOwnThinking

TheAcceleratingPaceofResearchWearenowapproachingthekneeofthecurve(theperiodofrapidexponentialgrowth)intheacceleratingpaceofunderstandingthehumanbrain,butourattemptsinthisareahavealonghistory.Ourabilitytoreflectonandbuildmodelsofourthinkingisauniqueattributeofourspecies.Earlymentalmodelswereofnecessitybasedonsimplyobservingourexternalbehavior(forexample,Aristotle'sanalysisofthehumanabilitytoassociateideas,written2,350yearsago).19Atthebeginningofthetwentiethcenturywedevelopedthetoolstoexaminethephysicalprocessesinsidethebrain.Anearlybreakthroughwasthemeasurementoftheelectricaloutputofnervecells,developedin1928byneurosciencepioneerE.D.Adrian,which

demonstratedthattherewereelectricalprocessestakingplaceinsidethebrain.20AsAdrianwrite,"Ihadarrangedsomeelectrodesontheopticnerveofatoadinconnectionwithsomeexperimentontheretina.TheroomwasnearlydarkandIwaspuzzledtohearrepeatednoisesintheloudspeakerattachedtotheamplifier,noisesindicatingthatagreatdealofimpulseactivitywasgoingon.ItwasnotuntilIcomparedthenoiseswithmyownmovementsaroundtheroomthatIrealizedIwasinthefieldofvisionofthetoad'seyeandthatitwassignalingwhatIwasdoing."Adrian'skeyinsightfromthisexperimentremainsacornerstoneofneurosciencetoday:thefrequencyoftheimpulsesfromthesensorynerveisproportionaltotheintensityofthesensoryphenomenabeingmeasured.Frexample,thehighertheintensityofthelight,thehigherthefrequency(pulsespersecond)oftheneuralimpulsesfromtheretinatothebrain.ItwasastudentofAdrian,HoraceBarlow,whocontributedanotherlastinginsight,"triggerfeatures"inneurons,withthediscoverythattheretinasoffrogsandrabbitshassingleneuronsthatwouldtriggeron"seeing"specificshapes,directions,orvelocities.Inotherwords,perceptioninvolvesaseriesofstages,witheachlayerofneuronsrecognizingmoresophisticatedfeaturesoftheimage.In1939webegantodevelopanideaofhowneuronsperform:byaccumulating(adding)theirinputsandthenproducingaspikeofmembraneconductance(asuddenincreaseintheabilityoftheneuron'smembranetoconductasignal)anvoltagealongtheneuron'saxon(whichconnectstootherneuron'sviaasynapse).A.L.HodgkinandA.F.Huxleydescribedtheirtheoryoftheaxon's"actionpotential"(voltage).21Theyalsomadeanactualmeasurementofanactionpotentialonananimalneuralaxonin1952.Theychosesquidneuronsbecauseoftheirsizeandaccessibleanatomy.BuildingonHodgkinandHuxley'sinsightW.S.McCullochandW.Pittsdevelopedin1943asimplifiedmodelofneuralnetsthatmotivatedahalfcenturyofworkonartificial(simulated)neuralnets(usingacomputerprogramtosimulatethewayneuronsworkinthebrainasanetwork).ThismodelwasfurtherrefinedbyHodgkinandHuxleyin1952.Althoughwenowrealizethatactualneuronsarefarmorecomplexthattheseearlymodels,theoriginalconcepthasheldupwell.Thisbasicneural-netmodelhasaneural"weight"(representingthe"strength"oftheconnection)foreachsynapseandanonlinearity(firingthreshold)intheneuronsoma(cellbody).Asthesumoftheweightedinputstotheneuronsomaincreases,thereisrelativelylittleresponsefromtheneuronuntilacriticalthresholdisreached,aswhichpointtheneuronrapidlyincreasedtheoutputofitsaxonandfires.Differentneuronshavedifferentthresholds.Althoughrecentresearchshowsthattheactualresponseismorecomplexthanthis,theMcCulloch-PittsandHodgkin-Huxleymodelsremainessentiallyvalid.Theseinsightsledtoanenormousamountofearlyworkincreatingartificialneuralnets,inafieldthatbecameknownasconnectionism.Thiswasperhapsthefirstself-organizingparadigmintroducedtothefieldofcomputation.Akeyrequirementforaself-organizingsystemisanonlinearity:somemeansofcreatingoutputsthatarenotsimpleweightssumsoftheinputs.Theearlyneural-netmodelsprovidedthisnonlinearityintheirreplicaoftheneuronnucleus.23(Thebasicneural-netmethodisstraightforward.)24WorkinitiatedbyAlanTuringontheoreticalmodelsofcomputationaroundthesametimealsoshowedthatcomputationrequiresanonlinearity.Asystemthatsimplecreatesweightedsumsofitsinputscannotperformtheessentialrequirementsofcomputation.Wenowknowthatactualbiologicalneuronshavemanyothernonlinearitiesresultingfromtheelectrochemicalactionofthesynapsesandthemorphology(shape)ofthedendrites.Differentarrangementsofbiologicalneuronscanperformcomputations,includingsubtracting,multiplying,averaging,filtering,normalizing,andthresholdingsignals,amongothertypesoftransformations.Theabilityofneuronstoperformmultiplicationisimportantbecauseitallowedthebehaviorofonenetworkofneuronsinthebraintobemodulated(influenced)bytheresultsofcomputationsofanothernetwork.Experimentsusingelectrophysiologicalmeasurementsonmonkeysprovideevidencethattherateofsignalingbyneuronsinthevisualcortexwhenprocessinganimageisincreasedordecreasedbywhetherornotthemonkeyispayingattentiontoaparticularareaofthatimage.25HumanfMRIstudieshavealsoshownthatpayingattentiontoaparticularareaofanimageincreasestheresponsivenessoftheneuronsprocessingthatimageinacorticalregioncalledV5,whichisresponsibleformotiondetection.26Theconnectionismmovementexperiencedasetbackin1969withthepublicationofthebookPerceptronsbyMIT'sMarvinMinskyandSeymourPapert.27Itincludedakeytheoremdemonstratingthatthemostcommon(andsimplest)typeofneuralnetusedatthetime(calledaPerceptron,pioneeredbyCornell'sFrankRosenblatt),wasunabletosolvethesimpleproblemofdeterminingwhetherornotalinedrawingwasfullyconnected.28Theneural-netmovementhadaresurgenceinthe1980susingamethodcalled"backpropagation,"inwhichthestrengthofeachsimulatedsynapsewasdeterminedusingalearningalgorithmthatadjustedtheweight(thestrengthoftheoutputofeachofartificialneuronaftereachtrainingtrialsothenetworkcould"learn"tomorecorrectlymatchtherightanswer.However,backpropagationisnotafeasiblemodeloftrainingsynapticweightinanactualbiologicalneuralnetwork,becausebackwardconnectionstoactuallyadjustthestrengthofthesynapticconnectionsdonotappeartoexistinmammalianbrains.Incomputers,however,thistypeofself-organizingsystemcansolveawiderangeofpattern-recognitionproblems,andthepowerofthissimplemodelofself-organizinginterconnectedneuronshasbeendemonstrated.LesswellknowisHebb'ssecondformoflearning:ahypothesizedloopinwhichheexcitationoftheneuronwouldfeedbackonitself(possiblythroughotherlayers),causingareverberation(acontinuedreexcitationcouldbethesourceofshort-termlearning).Healsosuggestedthatthisshort-termreverberationcouldleadtolong-termmemories:"Letusassumethenthatthepersistenceorrepetitionofareverberatoryactivity(or'trace')tendstoinducelastingcellularchangesthataddtoitsstability.Theassumptioncanbepreciselystatedasfollows:WhenanaxonofcellAisnearenoughtoexciteacellBandrepeatedlyorpersistentlytakepartinfiringit,somegrowthprocessormetabolicchangetakesplaceinoneorbothcellssuchthatA'sefficiency,asoneofthecell'sfiringB,isincreased."AlthoughHebbianreverberatorymemoryisnotaswellestablishedasHebb'ssynapticlearning,instanceshavebeenrecentlydiscovered.Forexample,setsofexcitatoryneurons(onesthatstimulateasynapse)andinhibitoryneurons(onesthatblockastimulus)beginanoscillationwhencertainvisualpatternsarepresented.29AndresearchersatMITandLucentTechnologies'BellLabshavecreatedanelectronicintegratedcircuit,composedoftransistors,thatsimulatestheactionofsixteenexcitatoryneuronsandoneinhibitoryneurontomimicthebiologicalcircuitryofthecerebralcortex.30Theseearlymodelsofneuronsandneuralinformationprocessing,althoughoverlysimplifiedandinaccurateinsomerespects,wereremarkable,giventhelackofdataandtoolswhenthesetheoriesweredeveloped.

PeeringintotheBrain

We'vebeenabletoreducedriftandnoiseinourinstrumentstosuchanextentthatwecanseethe tiniest motions of these molecules, through distances that are less than their owndiameters....[T]hesekindsofexperimentswerejustpipedreams15yearsago.

—STEVENBLOCK,PROFESSOROFBIOLOGICALSCIENCESANDOF

APPLIEDPHYSICS,STANFORDUNIVERSITYImagine thatwewere trying to reverse engineer a computerwithout knowing anything about it (the"blackbox" approach).Wemight start byplacingarraysofmagnetic sensors around thedevice.Wewouldnotice thatduringoperations thatupdatedadatabase,significantactivitywas takingplace inaparticular circuit board.Wewould be likely to take note that therewas also action in the hard diskduringtheseoperations.(Indeed,listeningtotheharddiskhasalwaysbeenonecrudewindowintowhatacomputerisdoing.)Wemightthentheorizethatthediskhadsomethingtodowiththelong-termmemorythatstoresthe databases and that the circuit board that is active during these operations was involved intransformingthedatatobestored.Thistellsusapproximatelywhereandwhentheoperationsaretakingplacebutrelativelylittleabouthowthesetasksareaccomplished.Ifthecomputer'sregisters(temporarymemorylocations)wereconnectedtofront-panellights(aswasthecasewithearlycomputers),wewouldseecertainpatternsoflightflickeringthatindicatedrapidchanges in the states of these registers during periods when the computer was analyzing data butrelatively slowchangeswhen the computerwas transmittingdata.Wemight then theorize that theselights reflectedchanges in logic stateduring somekindof analyticbehavior.Such insightswouldbeaccuratebutcrudeandwould fail toprovideuswitha theoryofoperationorany insightsas tohowinformationisactuallycodedortransformed.Thehypotheticalsituationdescribedabovemirrorsthesortofeffortsthathavebeenundertakentoscanandmodelthehumanbrainwiththecrudetoolsthathavehistoricallybeenavailable.Mostmodelsbased on contemporary brain-scanning research (utilizing such methods as fMRI,MEG, and othersdiscussed below) are only suggestive of the underlying mechanisms. Although these studies arevaluable,theircrudespatialandtemporalresolutionisnotadequateforreverseengineeringthesalientfeaturesofthebrain.NewToolsforScanningtheBrain.Nowimagine,inourcomputerexampleabove,thatweareabletoactually place precise sensors at specific points in the circuitry and that these sensors are capable oftrackingspecificsignalsatveryhighspeeds.Wewouldnowhavethetoolsneededtofollowtheactualinformationbeingtransformedinrealtime,andwewouldbeabletocreateadetaileddescriptionofhowthecircuitsactuallywork.Thisis,infact,exactlyhowelectricalengineersgoaboutunderstandinganddebuggingcircuitssuchascomputerboards(toreverseengineeracompetitor'sproduct,forexample),usinglogicanalyzersthatvisualizecomputersignals.Neurosciencehasnotyethadaccesstosensortechnologythatwouldachievethistypeofanalysis,but that situation is about to change. Our tools for peering into our brains are improving at anexponentialpace.Theresolutionofnoninvasivebrain-scanningdevicesisdoublingabouteverytwelvemonths(perunitvolume).31

Weseecomparableimprovementsinthespeedofbrainscanningimagereconstruction:

Themostcommonlyusedbrain-scanningtoolisfMRI,whichprovidesrelativelyhighspatialresolutionof one to three millimeters (not high enough to image individual neurons) but low temporal (time)resolutionofafewseconds.RecentgenerationsoffMRI technologyprovide timeresolutionofaboutonesecond,oratenthofasecondforathinbrainslice. Anothercommonlyused technique isMEG,whichmeasuresweakmagnetic fieldsoutside theskull, produced principally by the pyramidal neurons of the cortex. MEG is capable of rapid (onemillisecond)temporalresolutionbutonlyverycrudespatialresolution,aboutonecentimeter.FritzSommer,aprincipalinvestigatoratRedwoodNeuroscienceInstitute,isdevelopingmethodsof combining fMRI andMEG to improve the spatiotemporal precision of the measurements. OtherrecentadvanceshavedemonstratedfMRItechniquescapableofmappingregionscalledcolumnarandlaminarstructures,whichareonlyafractionofamillimeterwide,andofdetectingtasksthattakeplaceintensofmilliseconds.32fMRIandarelatedscanningtechniqueusingpositronscalledpositron-emissiontomography(PET)both gauge neuronal activity through indirect means. PET measures regional cerebral blood flow(rCBF), while tMRImeasures blood-oxygen levels.33 Although the relationship of these blood-flowamounts toneural activity is the subjectof somecontroversy, theconsensus is that they reflect localsynapticactivity,notthespikingofneurons.Therelationshipofneuralactivitytobloodflowwasfirstarticulated in the latenineteenthcentury.34A limitationof tMRI,however, is that the relationshipofbloodflowtosynapticactivityisnotdirect:avarietyofmetabolicmechanismsaffecttherelationship

betweenthetwophenomena.However,bothPETandtMRIarebelievedtobemostreliableformeasuringrelativechangesinbrainstate.Theprimarymethodtheyuseisthe"subtractionparadigm,"whichcanshowregionsthataremostactiveduringparticulartasks.35Thisprocedureinvolvessubtractingdataproducedbyascanwhenthesubjectisnotperforminganactivityfromdataproducedwhilethesubjectisperformingaspecifiedmentalactivity.Thedifferencerepresentsthechangeinbrainstate. An invasive technique thatprovideshigh spatial and temporal resolution is "optical imaging,"which involves removing part of the skull, staining the living brain tissuewith a dye that fluorescesuponneuralactivity,and then imaging theemitted lightwithadigitalcamera.Sinceoptical imagingrequiressurgery,ithasbeenusedmainlyinanimal,particularlymouse,experiments. Anotherapproachtoidentifyingbrainfunctionalityindifferentregionsistranscranialmagneticstimulation (TMS), which involves applying a strong-pulsed magnetic field from outside the skull,usingamagneticcoilpreciselypositionedover thehead.Byeither stimulatingor inducinga"virtuallesion"of(bytemporarilydisabling)smallregionsofthebrain,skillscanbediminishedorenhanced.36TMScanalsobeusedtostudytherelationshipofdifferentareasofthebrainonspecifictasksandcaneveninducesensationsofmysticalexperiences.37BrainscientistAllanSnyderhasreportedthatabout40 percent of his test subjects hooked up toTMSdisplay significant new skills,many ofwhich areremarkable,suchasdrawingabilities.38 Ifwehave theoptionofdestroying thebrain thatwearescanning,dramaticallyhigherspatialresolution becomes possible. Scanning a frozen brain is feasible today, though not yet at sufficientspeed or bandwidth to fully map all interconnections. But again, in accordance with the law ofacceleratingreturns,thispotentialisgrowingexponentially,asareallotherfacetsofbrainscanning.CarnegieMellonUniversity'sAndreasNowatzykisscanningthenervoussystemofthebrainandbody of a mouse with a resolution of less than two hundred nanometers, which is approaching theresolution needed for full reverse engineering. Another destructive scanner called the "Brain TissueScanner" developed at the BrainNetworks Laboratory at TexasA&MUniversity is able to scan anentiremousebrainataresolutionof250nanometersinonemonth,usingslices.39ImprovingResolution.Manynewbrain-scanning technologiesnow indevelopmentaredramaticallyimprovingbothtemporalandspatialresolution.Thisnewgenerationofsensingandscanningsystemsisprovidingthetoolsneededtodevelopmodelswithunprecedentedfinelevelsofdetail.Followingisasmallsampleoftheseemergingimagingandsensingsystems. One particularly exciting new scanning camera is being developed at the University ofPennsylvania Neuroengineering Research Laboratory, led by Leif H. Pinkel.40 The optical system'sprojected spatial resolutionwill be high enough to image individual neurons and at one-millisecondtimeresolution,whichissufficienttorecordthefiringofeachneuron. Initialversionsareabletoscanaboutonehundredcellssimultaneously,atadepthofuptotenmicrons from the camera. A future version will image up to one thousand simultaneous cells, at adistanceofupto150micronsfromthecameraandatsubmillisecondtimeresolution.Thesystemcanscanneuraltissueinvivo(inalivingbrain)whileananimalisengagedinamentaltask,althoughthebrainsurfacemustbeexposed.Theneuraltissueisstainedtogeneratevoltage-dependentfluorescence,which is pickedupby thehigh-resolution camera.The scanning systemwill beused to examine thebrainsofanimalsbeforeandafter they learnspecificperceptualskills.Thissystemcombines thefast(one millisecond) temporary resolution of MEG while being able to image individual neurons andconnections.Methodshavealsobeendevelopedtononinvasivelyactivateneuronsorevenaspecificpartofaneuron in a temporally and spatially precisemanner.One approach, involvingphotons, uses a direct"two-photon" excitation, called "two-photon laser scanning microscopy" (TPLSM).41 This creates asinglepoint of focus in three-dimensional space that allowsveryhigh-resolution scanning. It utilizeslaser pulses lasting only one millionth of one billionth of a second (10–15 second) to detect the

excitation of single synapses in the intact brain bymeasuring the intracellular calcium accumulationassociatedwith the activation of synaptic receptors.42Although themethod destroys an insignificantamount of tissue, it provides extremely high-resolution images of individual dendritic spines andsynapsesinaction.Thistechniquehasbeenusedtoperformultrapreciseintracellularsurgery.PhysicistEricMazurandhiscolleaguesatHarvardUniversityhavedemonstrateditsabilitytoexecuteprecisemodificationsofcells,suchasseveringaninterneuronalconnectionordestroyingasinglemitochondrion(thecell'senergy source) without affecting other cellular components. "It generates the heat of the sun," saysMazur'scolleagueDonaldIngber,"butonlyforquintillionthsofasecond,andinaverysmallspace." Another technique, called "multielectrode recording," uses an array of electrodes to recordsimultaneously the activity of a large number of neurons with very high (submillisecond) temporalresolution.43Also, a noninvasive technique called second-harmonic generation (SHG)microscopy isable"tostudycellsinaction,"explainsleaddeveloperDanielDombeck,agraduatestudentatCornellUniversity. Yet another technique, called optical coherence imaging (OCI), uses coherent light(lightwaves thatareallaligned in thesamephase) tocreateholographic three-dimensional imagesofcellclusters.Scanning Using Nanobots. Although these largely noninvasive means of scanning the brain fromoutside theskullare rapidly improving, themostpowerfulapproach tocapturingeverysalientneuraldetailwillbetoscanitfrominside.Bythe2020snanobottechnologywillbeviable,andbrainscanningwillbeoneofitsprominentapplications.Asdescribedearliernanobotsarerobotsthatwillbethesizeofhumanbloodcells(seventoeightmicrons)orevensmaller.44Billionsofthemcouldtravelthrougheverybraincapillary, scanningeach relevantneural feature fromupclose.Usinghigh-speedwirelesscommunication,thenanobotswouldcommunicatewithoneanotherandwithcomputerscompilingthebrain-scandatabase. (Inotherwords, thenanobotsandcomputerswill allbeonawireless localareanetwork.)45 Akeytechnicalchallenge to interfacingnanobotswithbiologicalbrainstructures is theblood-brainbarrier (BBB). In the latenineteenthcentury, scientistsdiscovered thatwhen they injectedbluedye into an animal's bloodstream, all the organs of the animal turned bluewith the exception of thespinalcordandbrain.Theyaccuratelyhypothesizedabarrierthatprotectsthebrainfromawiderangeofpotentiallyharmfulsubstancesintheblood,includingbacteria,hormones,chemicalsthatmayactasneurotransmitters, and other toxins. Only oxygen, glucose, and a very select set of other smallmoleculesareabletoleavethebloodvesselsandenterthebrain.Autopsiesearlyinthetwentiethcenturyrevealedthattheliningofthecapillariesinthebrainandother nervous-system tissues is indeed packed much more tightly with endothelial cells thancomparable-sizevessels inotherorgans.More recent studieshave shown that theBBB is a complexsystem that featuresgatewayscompletewithkeysandpasswords that allowentry into thebrain.Forexample,twoproteinscalledzonulinandzothavebeendiscoveredthatreactwithreceptorsinthebraintotemporarilyopentheBBBatselectsites.Thesetwoproteinsplayasimilarroleinopeningreceptorsinthesmallintestinetoallowdigestionofglucoseandothernutrients.AnydesignfornanobotstoscanorotherwiseinteractwiththebrainwillhavetoconsidertheBBB.Idescribehereseveralstrategies thatwillbeworkable,givenfuturecapabilities.Undoubtedly,otherswillbedevelopedoverthenextquartercentury.

·Anobvious tactic is tomake thenanobot smallenough toglide through theBBB,but this is theleast practical approach, at least with nanotechnology as we envision it today. To do this, thenanobotwouldhavetobetwentynanometersorlessindiameter,whichisaboutthesizeofonehundred carbon atoms. Limiting a nanobot to these dimensions would severely limit itsfunctionality.

·AnintermediatestrategywouldbetokeepthenanobotinthebloodstreambuttohaveitprojectaroboticarmthroughtheBBBandintotheextracellularfluidthatlinestheneuralcells.Thiswould

allow the nanobot to remain large enough to have sufficient computational and navigationalresources. Since almost all neurons lie within two or three cell-widths of a capillary, the armwouldneedtoreachonlyuptoaboutfiftymicrons.AnalysesconductedbyRobFreitasandothersshow that it is quite feasible to restrict the width of such a manipulator to under twentynanometers.

·Another approach is to keep the nanobots in the capillaries and use noninvasive scanning. Forexample,thescanningsystembeingdesignedbyFinkelandhisassociatescanscanatveryhighresolution (sufficient to see individual interconnections) to a depth of 150 microns, which isseveraltimesgreaterthanweneed.Obviouslythistypeofoptical-imagingsystemwouldhavetobe significantly miniaturized (compared to contemporary designs), but it uses charge-coupleddevicesensors,whichareamenabletosuchsizereduction.

·Anothertypeofnoninvasivescanningwouldinvolveonesetofnanobotsemittingfocusedsignalssimilartothoseofatwo-photonscannerandanothersetofnanobotsreceivingthetransmission.The topology of the intervening tissue could be determined by analyzing the impact on thereceivedsignal.

·Anothertypeofstrategy,suggestedbyRobertFreitas,wouldbeforthenanobotliterallytobargeitswaypasttheBBBbybreakingaholeinit,exitthebloodvessel,andthenrepairthedamage.Sincethe nanobot can be constructed using carbon in a diamondoid configuration, it would be farstronger than biological tissues. Freitas writes, "To pass between cells in cell-rich tissue, it isnecessaryforanadvancingnanorobot todisruptsomeminimumnumberofcell-to-celladhesivecontacts that lie ahead in its path. After that, and with the objective of minimizingbiointrusiveness,thenanorobotmustresealthoseadhesivecontactsinitswake,crudelyanalogoustoaburrowingmole."46

·Yetanotherapproachissuggestedbycontemporarycancerstudies.CancerresearchersarekeenlyinterestedinselectivelydisruptingtheBBBtotransportcancer-destroyingsubstancesto tumors.RecentstudiesoftheBBBshowthatitopensupinresponsetoavarietyoffactors,whichincludecertain proteins, as mentioned above; localized hypertension; high concentrations of certainsubstances;microwavesandotherformsofradiation;infection;andinflammation.Therearealsospecializedprocessesthatferryoutneededsubstancessuchasglucose.Ithasalsobeenfoundthatthesugarmannitolcausesatemporaryshrinkingofthetightlypackedendothelialcellstoprovidea temporary breach of the BBB. By exploiting thesemechanisms, several research groups aredevelopingcompoundsthatopentheBBB.47Althoughthisresearchisaimedatcancertherapies,similarapproachescanbeusedtoopenthegatewaysfornanobotsthatwillscanthebrainaswellasenhanceourmentalfunctioning.

·WecouldbypassthebloodstreamandtheBBBaltogetherbyinjectingthenanobotsintoareasofthebrainthathavedirectaccesstoneuraltissue.AsImentionbelow,newneuronsmigratefromtheventriclestootherpartsofthebrain.Nanobotscouldfollowthesamemigrationpath.

·RobFreitashasdescribedseveraltechniquesfornanobotstomonitorsensorysignals.48Thesewillbe important both for reverse engineering the inputs to the brain, as well as for creating full-immersionvirtualrealityfromwithinthenervoussystem.

►Toscanandmonitorauditorysignals,Freitasproposes"mobilenanodevices...[that]swimintothespiralarteryoftheearanddownthroughitsbifurcationstoreachthecochlearcanal,thenpositionthemselvesasneuralmonitorsinthevicinityofthespiralnervefibersandthenervesenteringtheepitheliumoftheorganofCorti[cochlearorauditorynerves]withinthespiralganglion.Thesemonitorscandetect,record,orrebroadcasttoothernanodevicesinthecommunicationsnetworkallauditoryneuraltrafficperceivedbythehumanear."

► For the body's "sensations of gravity, rotation, and acceleration," he envisions"nanomonitorspositionedattheafferentnerveendingsemanatingfromhaircellslocatedinthe...semicircularcanals."

► For"kinestheticsensorymanagement...motorneuronscanbemonitoredtokeeptrackof

limbmotionsandpositions,orspecificmuscleactivities,andeventoexertcontrol."► "Olfactoryandgustatorysensoryneuraltrafficmaybeeavesdropped[on]bynanosensoryinstruments."

►"Painsignalsmayberecordedormodifiedasrequired,ascanmechanicalandtemperaturenerveimpulsesfrom...receptorslocatedintheskin."

►Freitaspointsoutthattheretinaisrichwithsmallbloodvessels,"permittingreadyaccesstobothphotoreceptor(rod,cone,bipolarandganglion)andintegrator...neurons."Thesignalsfrom the optic nerve representmore than one hundredmillion levels per second, but thislevelofsignalprocessingisalreadymanageable.AsMIT'sTomasoPoggioandothershaveindicated,wedonotyetunderstandthecodingoftheopticnerve'ssignals.Oncewehavetheabilitytomonitorthesignalsforeachdiscretefiberintheopticnerve,ourabilitytointerpretthesesignalswillbegreatlyfacilitated.Thisiscurrentlyanareaofintenseresearch.

AsIdiscussbelow,therawsignalsfromthebodygothroughmultiplelevelsofprocessingbeforebeing aggregated in a compact dynamic representation in two small organs called the right and leftinsula,locateddeepinthecerebralcortex.Forfull-immersionvirtualreality,itmaybemoreeffectivetotapintothealready-interpretedsignalsintheinsularatherthantheunprocessedsignalsthroughoutthebody.Scanningthebrainforthepurposeofreverseengineeringitsprinciplesofoperationisaneasieractionthanscanningitforthepurposeof"uploading"aparticularpersonality,whichIdiscussfurtherbelow(seethe"UploadingtheHumanBrain"section,p.198).Inordertoreverseengineerthebrain,weonlyneedtoscantheconnectionsinaregionsufficientlytounderstandtheirbasicpattern.Wedonotneedtocaptureeverysingleconnection.Onceweunderstandtheneuralwiringpatternswithinaregion,wecancombinethatknowledgewithadetailedunderstandingofhoweachtypeofneuroninthatregionoperates.Althoughaparticularregionofthebrainmayhavebillionsofneurons,itwillcontainonlyalimitednumberofneurontypes.Wehavealreadymadesignificantprogressinderivingthemechanismsunderlyingspecificvarietiesofneuronsandsynapticconnectionsbystudyingthesecellsinvitro(inatestdish),aswellasinvivousingsuchmethodsastwo-photonscanning.Thescenariosaboveinvolvecapabilitiesthatexistatleastinanearlystagetoday.Wealreadyhavetechnology capable of producing very high-resolution scans for viewing the precise shape of everyconnectioninaparticularbrainarea,ifthescannerisphysicallyproximatetotheneuralfeatures.Withregard to nanobots, there are already fourmajor conferences dedicated to developing blood cell-sizedevices for diagnostic and therapeutic purposes.49 As discussed in chapter 2, we can project theexponentiallydecliningcostofcomputationandtherapidlydecliningsizeandincreasingeffectivenessof both electronic and mechanical technologies. Based on these projections, we can conservativelyanticipate the requisite nanobot technology to implement these types of scenarios during the 2020s.Once nanobot-based scanning becomes a reality,wewill finally be in the same position that circuitdesignersareintoday:wewillbeabletoplacehighlysensitiveandveryhigh-resolutionsensors(intheformofnanobots)atmillionsorevenbillionsoflocationsinthebrainandthuswitnessinbreathtakingdetaillivingbrainsinaction.BuildingModelsoftheBrain

Ifweweremagicallyshrunkandput intosomeone'sbrainwhileshewasthinking,wewouldseeall thepumps,pistons,gearsandleversworkingaway,andwewouldbeabletodescribetheir workings completely, in mechanical terms, thereby completely describing the thoughtprocessesofthebrain.Butthatdescriptionwouldnowherecontainanymentionofthought!Itwouldcontainnothingbutdescriptionsofpumps,pistons,levers!

—G.W.LEIBNIZ(1646–1716)Howdo...fieldsexpresstheirprinciples?Physicistsusetermslikephotons,electrons,quarks,quantum wave function, relativity, and energy conservation. Astronomers use terms likeplanets, stars, galaxies, Hubble shift, and black holes. Thermodynamicists use terms likeentropy, first law, second law, and Carnot cycle. Biologists use terms like phylogeny,ontogeny, DNA, and enzymes. Each of these terms is actually the title of a story! Theprinciplesofafieldareactuallyasetofinterwovenstoriesaboutthestructureandbehavioroffieldelements.

—PETERJ.DENNING,PASTPRESIDENTOFTHEASSOCIATIONFORCOMPUTING MACHINERY, IN "GREAT PRINCIPLES OFCOMPUTING"

Itisimportantthatwebuildmodelsofthebrainattherightlevel.Thisis,ofcourse,trueforallofourscientificmodels.Althoughchemistry is theoreticallybasedonphysicsandcouldbederivedentirelyfromphysics, thiswouldbeunwieldyand infeasible inpractice.Sochemistryuses itsownrulesandmodels.Weshouldlikewise,intheory,beabletodeducethelawsofthermodynamicsfromphysics,butthis is a far-from-straightforward process. Once we have a sufficient number of particles to callsomethingagasratherthanabunchofparticles,solvingequationsforeachparticleinteractionbecomesimpractical, whereas the laws of thermodynamics work extremelywell. The interactions of a singlemolecule within the gas are hopelessly complex and unpredictable, but the gas itself, comprisingtrillionsofmolecules,hasmanypredictableproperties.Similarly,biology,whichisrootedinchemistry,usesitsownmodels.Itisoftenunnecessarytoexpresshigher-levelresultsusingthe intricaciesof thedynamicsof the lower-levelsystems,althoughonehastothoroughlyunderstandthelowerlevelbeforemovingtothehigherone.Forexample,wecancontrol certain genetic features of an animal by manipulating its fetal DNA without necessarilyunderstandingallofthebiochemicalmechanismsofDNA,letalonetheinteractionsoftheatomsintheDNAmolecule. Often, the lower level ismore complex. A pancreatic islet cell, for example, is enormouslycomplicated,intermsofallitsbiochemicalfunctions(mostofwhichapplytoallhumancells,sometoall biological cells). Yet modeling what a pancreas does—with its millions of cells—in terms ofregulating levels of insulin anddigestive enzymes, althoughnot simple, is considerably less difficultthanformulatingadetailedmodelofasingleisletcell.Thesameissueappliestothelevelsofmodelingandunderstandinginthebrain,fromthephysicsofsynapticreactionsuptothetransformationsofinformationbyneuralclusters.Inthosebrainregionsforwhichwe have succeeded in developing detailedmodels, we find a phenomenon similar to thatinvolving pancreatic cells. The models are complex but remain simpler than the mathematicaldescriptions of a single cell or even a single synapse.Aswe discussed earlier, these region-specificmodels also require significantly less computation than is theoretically implied by the computationalcapacityofallofthesynapsesandcells. Gilles Laurent of theCalifornia Institute of Technology observes, "Inmost cases, a system'scollectivebehaviorisverydifficulttodeducefromknowledgeofitscomponents....[Neuroscienceis...ascienceofsystemsinwhichfirst-orderandlocalexplanatoryschemataareneededbutnotsufficient."Brainreverse-engineeringwillproceedbyiterativerefinementofbothtop-to-bottomandbottom-to-topmodelsandsimulations,aswerefineeachlevelofdescriptionandmodeling. Untilvery recentlyneurosciencewascharacterizedbyoverly simplisticmodels limitedby thecrudenessofour sensingand scanning tools.This ledmanyobservers todoubtwhetherour thinkingprocesseswere inherentlycapableofunderstanding themselves.PeterD.Kramerwrites, "If themindweresimpleenoughforustounderstand,wewouldbetoosimpletounderstandit."50Earlier,IquotedDouglasHofstadter's comparison of our brain to that of a giraffe, the structure of which is not thatdifferentfromahumanbrainbutwhichclearlydoesnothavethecapabilityofunderstandingitsown

methods.However,recentsuccessindevelopinghighlydetailedmodelsatvariouslevels—fromneuralcomponentssuchassynapsestolargeneuralregionssuchasthecerebellum—demonstratethatbuildingprecise mathematical models of our brains and then simulating these models with computation is achallengingbutviabletaskoncethedatacapabilitiesbecomeavailable.Althoughmodelshavealonghistory in neuroscience, it is only recently that they have become sufficiently comprehensive anddetailedtoallowsimulationsbasedonthemtoperformlikeactualbrainexperiments.SubneuralModels:SynapsesandSpinesInanaddresstotheannualmeetingoftheAmericanPsychologicalAssociationin2002,psychologistandneuroscientistJosephLeDouxofNewYorkUniversitysaid,

Ifwhoweareisshapedbywhatweremember,andifmemoryisafunctionofthebrain,thensynapses—theinterfacesthroughwhichneuronscommunicatewitheachotherandthephysicalstructures inwhichmemories are encoded—are the fundamental units of the self....Synapsesare pretty low on the totem pole of how the brain is organized, but I think they're prettyimportant....Theselfisthesumofthebrain'sindividualsubsystems,eachwithitsownformof"memory," together with the complex interactions among the subsystems.Without synapticplasticity—theabilityofsynapsestoaltertheeasewithwhichtheytransmitsignalsfromoneneuron to another—the changes in those systems that are required for learning would beimpossible.51

Althoughearlymodelingtreatedtheneuronastheprimaryunitoftransforminginformation,thetidehasturnedtowardemphasizingitssubcellularcomponents.ComputationalneuroscientistAnthonyJ.Bell,forexample,argues:

Molecularandbiophysicalprocessescontrolthesensitivityofneuronstoincomingspikes(bothsynaptic efficiency and post-synaptic responsivity), the excitability of the neuron to producespikes, thepatternsof170spikes itcanproduceand the likelihoodofnewsynapsesforming(dynamicrewiring),tolistonlyfourofthemostobviousinterferencesfromthesubneurallevel.Furthermore, transneural volume effects such as local electric fields and the transmembranediffusionofnitricoxidehavebeenseentoinfluence,responsively,coherentneuralfiring,andthedeliveryofenergy(bloodflow)tocells,thelatterofwhichdirectlycorrelateswithneuralactivity.Thelistcouldgoon.Ibelievethatanyonewhoseriouslystudiesneuromodulators,ionchannels, or synaptic mechanism and is honest, would have to reject the neuron level as aseparatecomputinglevel,evenwhilefindingittobeausefuldescriptivelevel.52

Indeed,anactualbrainsynapseisfarmorecomplexthanisdescribedintheclassicMcCulloch-Pittsneural-netmodel.Thesynapticresponseisinfluencedbyarangeoffactors,includingtheactionofmultiplechannelscontrolledbyavarietyof ionicpotentials (voltages)andmultipleneurotransmittersand neuromodulators. Considerable progress has been made in the past twenty years, however, indevelopingthemathematicalformulasunderlyingthebehaviorofneurons,dendrites,synapses,andtherepresentation of information in the spike trains (pulses by neurons that have been activated). PeterDayan and Larry Abbott have recently written a summary of the existing nonlinear differentialequations thatdescribeawiderangeofknowledgederivedfromthousandsofexperimentalstudies.53Well-substantiated models exist for the biophysics of neuron bodies, synapses, and the action offeedforwardnetworksofneurons,suchas thosefound in theretinaandopticnerves,andmanyotherclassesofneurons.AttentiontohowthesynapseworkshasitsrootsinHebb'spioneeringwork.Hebbaddressedthequestion, How does short-term (also calledworking)memory function? The brain region associated

withshort-termmemoryistheprefrontalcortex,althoughwenowrealizethatdifferentformsofshort-term information retention have been identified inmost other neural circuits that have been closelystudied.MostofHebb'sworkfocusedonchangesinthestateofsynapsestostrengthenorinhibitreceivedsignalsandonthemorecontroversialreverberatorycircuitinwhichneuronsfireinacontinuousloop.54AnothertheoryproposedbyHebbisachangeinstateofaneuronitself—thatis,amemoryfunctioninthe cell soma (body). The experimental evidence supports the possibility of all of these models.ClassicalHebbiansynapticmemoryandreverberatorymemoryrequireatimedelaybeforetherecordedinformationcanbeused.Invivoexperimentsshowthatinatleastsomeregionsofthebrainthereisaneural response that is too fast to be accounted for by such standard learningmodels, and thereforecouldonlybeaccomplishedbylearning-inducedchangesinthesoma.55AnotherpossibilitynotdirectlyanticipatedbyHebbisreal-timechangesintheneuronconnectionsthemselves.Recent scanning results show rapid growthof dendrite spikes andnew synapses, so thismust be considered an important mechanism. Experiments have also demonstrated a rich array oflearningbehaviorsonthesynapticlevelthatgobeyondsimpleHebbianmodels.Synapsescanchangetheirstaterapidly,buttheythenbegintodecayslowlywithcontinuedstimulation,orinsomealackofstimulation,ormanyothervariations.56AlthoughcontemporarymodelsarefarmorecomplexthanthesimplesynapsemodelsdevisedbyHebb, his intuitions have largely proved correct. In addition to Hebbian synaptic plasticity, currentmodelsincludeglobalprocessesthatprovidearegulatoryfunction.Forexample,synapticscalingkeepssynapticpotentialsfrombecomingzero(andthusbeingunable tobe increasedthroughmultiplicativeapproaches)orbecomingexcessivelyhighandtherebydominatinganetwork.Invitroexperimentshavefound synaptic scaling in cultured networks of neocortical, hippocampal, and spinal-cord neurons.57Othermechanisms are sensitive to overall spike timing and the distribution of potential acrossmanysynapses. Simulations have demonstrated the ability of these recently discovered mechanisms toimprovelearningandnetworkstability.Themostexcitingnewdevelopmentinourunderstandingofthesynapseisthatthetopologyofthesynapses and the connections they form are continually changing. Our first glimpse into the rapidchanges in synaptic connections was revealed by an innovative scanning system that requires ageneticallymodifiedanimalwhoseneuronshavebeenengineeredtoemitafluorescentgreenlight.Thesystem can image living neural tissue and has a sufficiently high resolution to capture not only thedendrites(interneuronalconnections)butthespines:tinyprojectionsthatsproutfromthedendritesandinitiatepotentialsynapses. NeurobiologistKarelSvoboda andhis colleagues atColdSpringHarborLaboratoryonLongIslandused thescanningsystemonmice to investigatenetworksofneurons thatanalyze informationfromthewhiskers,astudythatprovidedafascinatinglookatneurallearning.Thedendritescontinuallygrewnewspines.Mostoftheselastedonlyadayortwo,butonoccasionaspinewouldremainstable."Webelievethatthehighturnoverthatweseemightplayanimportantroleinneuralplasticity,inthatthe sprouting spines reach out to probe different presynaptic partners on neighboring neurons,” saidSvoboda."Ifagivenconnectionisfavorable,thatis,reflectingadesirablekindofbrainrewiring,thenthesesynapsesarestabilizedandbecomemorepermanent.Butmostofthesesynapsesarenotgoingintherightdirection,andtheyareretracted."58Anotherconsistentphenomenonthathasbeenobservedisthatneuralresponsesdecreaseovertime,if a particular stimulus is repeated.This adaptationgivesgreatest priority to newpatternsof stimuli.Similar work by neurobiologist Wen-Biao Gan at New York University's School of Medicine onneuronalspinesinthevisualcortexofadultmiceshowsthatthisspinemechanismcanholdlong-termmemories:"Saya10-year-oldkiduses1,000connectionstostoreapieceof information.Whenheis80,one-quarteroftheconnectionswillstillbethere,nomatterhowthingschange.That'swhyyoucanstill rememberyourchildhoodexperiences."Ganalsoexplains,"Our ideawas thatyouactuallydon'tneedtomakemanynewsynapsesandgetridofoldoneswhenyoulearn,memorize.Youjustneedtomodify the strength of the preexisting synapses for short-term learning and memory. However, it's

likelythatafewsynapsesaremadeoreliminatedtoachievelong-termmemory."59Thereasonmemoriescanremainintactevenifthreequartersoftheconnectionshavedisappearedisthatthecodingmethodusedappearstohavepropertiessimilartothoseofahologram.Inahologram,informationisstoredinadiffusepatternthroughoutanextensiveregion.Ifyoudestroythreequartersofthe hologram, the entire image remains intact, although with only one quarter of the resolution.ResearchbyPenttiKanerva,aneuroscientistatRedwoodNeuroscienceInstitute,supportstheideathatmemories are dynamically distributed throughout a region of neurons. This explains why oldermemoriespersistbutnonethelessappearto"fade,"becausetheirresolutionhasdiminished.NeuronModelsResearchersarealsodiscoveringthatspecificneuronsperformspecialrecognitiontasks.Anexperimentwith chickens identified brain-stem neurons that detect particular delays as sounds arrive at the twoears.60 Different neurons respond to different amounts of delay. Although there are many complexirregularities in how these neurons (and the networks they rely on) work, what they are actuallyaccomplishing is easy to describe and would be simple to replicate. According to University ofCalifornia at San Diego neuroscientist Scott Makeig, "Recent neurobiological results suggest animportantroleofpreciselysynchronizedneuralinputsinlearningandmemory."61ElectronicNeurons.A recent experiment at theUniversityofCalifornia atSanDiego's Institute forNonlinear Science demonstrates the potential for electronic neurons to precisely emulate biologicalones.Neurons(biologicalorotherwise)areaprimeexampleofwhatisoftencalledchaoticcomputing.Eachneuronactsinanessentiallyunpredictablefashion.Whenanentirenetworkofneuronsreceivesinput(fromtheoutsideworldorfromothernetworksofneurons),thesignalingamongthemappearsatfirsttobefrenziedandrandom.Overtime,typicallyafractionofasecondorso,thechaoticinterplayoftheneuronsdiesdownandastablepatternoffiringemerges.Thispatternrepresentsthe"decision"ofthe neural network. If the neural network is performing a pattern-recognition task (and such tasksconstitutethebulkoftheactivityinthehumanbrain),theemergentpatternrepresentstheappropriaterecognition..SothequestionaddressedbytheSanDiegoresearcherswas:couldelectronicneuronsengageinthischaoticdancealongsidebiologicalones?Theyconnectedartificialneuronswithrealneuronsfromspinylobstersinasinglenetwork,andtheirhybridbiological-nonbiologicalnetworkperformedinthesameway(thatis,chaoticinterplayfollowedbyastableemergentpattern)andwiththesametypeofresultsasanall-biologicalnetofneurons.Essentially,thebiologicalneuronsacceptedtheirelectronicpeers.Thisindicatesthatthechaoticmathematicalmodeloftheseneuronswasreasonablyaccurate.BrainPlasticityIn1861FrenchneurosurgeonPaulBrocacorrelatedinjuredorsurgicallyaffectedregionsofthebrainwithcertainlostskills,suchasfinemotorskillsorlanguageability.Formorethanacenturyscientistsbelievedtheseregionswerehardwiredforspecifictasks.Althoughcertainbrainareasdotendtobeusedforparticulartypesofskills,wenowunderstandthatsuchassignmentscanbechangedinresponsetobrain injury such as a stroke. In a classic 1965 study, D. H. Hubel and T. N.Wiesel showed thatextensive and far-reaching reorganization of the brain could take place after damage to the nervoussystem,suchasfromastroke.62 Moreover, thedetailed arrangementof connections and synapses in a given region is a directproduct of how extensively that region is used. As brain scanning has attained sufficiently highresolution to detect dendritic-spine growth and the formation of new synapses,we can see our braingrowandadapttoliterallyfollowourthoughts.ThisgivesnewshadesofmeaningtoDescartes'dictum"IthinkthereforeIam."

In one experiment conducted byMichaelMerzenich and his colleagues at theUniversity ofCalifornia at San Francisco, monkeys' food was placed in such a position that the animals had todexterouslymanipulateonefingertoobtainit.Brainscansbeforeandafterrevealeddramaticgrowthinthe interneuronal connections and synapses in the regionof thebrain responsible for controlling thatfinger. Edward Taub at the University of Alabama studied the region of the cortex responsible forevaluating the tactile input from the fingers. Comparing nonmusicians to experienced players ofstringedinstruments,hefoundnodifferenceinthebrainregionsdevotedtothefingersoftherighthandbutahugedifferencefor the fingersof the lefthand. Ifwedrewapictureof thehandsbasedon theamountofbraintissuedevotedtoanalyzingtouch,themusicians'fingersontheirlefthand(whichareusedtocontrolthestrings)wouldbehuge.Althoughthedifferencewasgreaterforthosemusicianswhobeganmusical trainingwitha stringed instrumentaschildren, "even ifyou takeup theviolinat40,"Taubcommented,"youstillgetbrainreorganization."63Asimilarfindingcomesfromanevaluationofasoftwareprogram,developedbyPaulaTallalandSteveMiller atRutgersUniversity, calledFast ForWord, that assists dyslexic students.The program

readstexttochildren,slowingdownstaccatophonemessuchas"b"and"p,"basedontheobservationthatmanydyslexic studentsareunable toperceive these soundswhenspokenquickly.Being read towith thismodified form of speech has been shown to help such children learn to read.Using fMRIscanningJohnGabrieliofStanfordUniversityfoundthattheleftprefrontalregionofthebrain,anareaassociatedwithlanguageprocessing,hadindeedgrownandshowedgreateractivityindyslexicstudentsusingtheprogram.SaysTallal,"Youcreateyourbrainfromtheinputyouget."Itisnotevennecessarytoexpressone'sthoughtsinphysicalactiontoprovokethebraintorewireitself.Dr.AlvaroPascual-LeoneatHarvardUniversityscannedthebrainsofvolunteersbeforeandaftertheypracticed a simplepianoexercise.Thebrainmotor cortexof thevolunteers changedas adirectresultoftheirpractice.Hethenhadasecondgroupjustthinkaboutdoingthepianoexercisebutwithoutactually moving any muscles. This produced an equally pronounced change in the motor-cortexnetwork.64RecentfMRIstudiesoflearningvisual-spatialrelationshipsfoundthatinterneuronalconnectionsareabletochangerapidlyduringthecourseofasinglelearningsession.Researchersfoundchangesinthe connectionsbetweenposterior parietal-cortex cells inwhat is called the "dorsal"pathway (whichcontains information about location and spatial properties of visual stimuli) and posterior inferior-temporalcortexcellsinthe"ventral"pathway(whichcontainsrecognizedinvariantfeaturesofvaryinglevels of abstraction);65 significantly, that rate of change was directly proportional to the rate oflearning.66ResearchersattheUniversityofCaliforniaatSanDiegoreportedakeyinsightintothedifferenceintheformationofshort-termandlong-termmemories.Usingahigh-resolutionscanningmethod,thescientists were able to see chemical changes within synapses in the hippocampus, the brain regionassociated with the formation of long-termmemories.67 They discovered that when a cell was firststimulated,actin,aneurochemical,movedtowardtheneuronstowhichthesynapsewasconnected.Thisalso stimulated the actin in neighboring cells to move away from the activated cell. These changeslasted only a few minutes, however. If the stimulations were sufficiently repeated, then a moresignificantandpermanentchangetookplace."Theshort-termchangesarejustpartofthenormalwaythenervecellstalktoeachother,"leadauthorMichaelA.Colicossaid.

The long-termchanges in theneuronsoccuronlyafter theneuronsare stimulated four timesover the course of an hour. The synapse will actually split and new synapses will form,producingapermanentchangethatwillpresumablylastfortherestofyourlife.Theanalogytohumanmemoryisthatwhenyouseeorhearsomethingonce,itmightstickinyourmindforafewminutes.Ifit'snotimportant,itfadesawayandyouforgetit10minuteslater.Butifyouseeorhearitagainandthiskeepshappeningoverthenexthour,youaregoingtorememberitfor amuch longer time.And things that are repeatedmany timescanbe remembered for anentire lifetime.Onceyou takeanaxonandformtwonewconnections, thoseconnectionsareverystableandthere'snoreasontobelievethatthey'llgoaway.That’sthekindofchangeonewouldenvisionlastingawholelifetime.

"It's like apiano lesson," says coauthor andprofessorofbiologyYukikoGoda. "Ifyouplayamusical scoreoverandoveragain, itbecomes ingrained inyourmemory."Similarly, inanarticle inScience neuroscientists S. Lowel and W. Singer report having found evidence for rapid dynamicformation of new interneuronal connections in the visual cortex, which they described with DonaldHebb'sphrase"Whatfirestogetherwirestogether."68AnotherinsightintomemoryformationisreportedinastudypublishedinCell.Researchersfoundthat theCPEBproteinactuallychanges itsshape insynapses torecordmernories.69ThesurprisewasthatCPEBperformsthismemoryfunctionwhileinaprionstate."Forawhilewe'veknownquiteabitabouthowmemoryworks,butwe'vehadnoclearconceptof

whatthekeystoragedeviceis,"saidcoauthorandWhiteheadInstituteforBiomedicalResearchdirectorSusan Lindquist. "This study suggests what the storage device might be—but it's such a surprisingsuggestion to find that a prion-like activity may be involved....It ... indicates that prions aren't justoddballsofnaturebutmightparticipateinfundamentalprocesses."AsIreportedinchapter3,humanengineersarealsofindingprionstobeapowerfulmeansofbuildingelectronicmemories. Brain-scanning studies are also revealing mechanisms to inhibit unneeded and undesirablememories, a finding thatwouldgratifySigmundFreud.70Using fMRI,StanfordUniversity scientistsasked study subjects to attempt to forget information that they had earlier memorized. During thisactivity,regionsinthefrontalcortexthathavebeenassociatedwithmemoryrepressionshowedahighlevel of activity, while the hippocampus, the region normally associated with remembering, wasrelativelyinactive.Thesefindings"confirmtheexistenceofanactiveforgettingprocessandestablishaneurobiological model for guiding inquiry into motivated forgetting," wrote Stanford psychologyprofessor John Gabrieli and his colleagues. Gabrieli also commented, "The big news is that we'veshownhowthehumanbrainblocksanunwantedmemory,thatthereissuchamechanism,andithasabiologicalbasis.Itgetsyoupast thepossibilitythat there'snothinginthebrainthatwouldsuppressamemory—thatitwasallamisunderstoodfiction." Inadditiontogeneratingnewconnectionsbetweenneurons, thebrainalsomakesnewneuronsfrom neural stem cells, which replicate to maintain a reservoir of themselves. In the course ofreproducing,someoftheneuralstemcellsbecome"neuralprecursor"cells,whichinturnmatureintotwotypesofsupportcellscalledastrocytesandoligodendrocytes,aswellasneurons.Thecellsfurtherevolveintospecifictypesofneurons.However,thisdifferentiationcannottakeplaceunlesstheneuralstemcellsmoveawayfromtheiroriginalsourceinthebrain'sventricles.Onlyabouthalfoftheneuralcellssuccessfullymakethejourney,whichissimilartotheprocessduringgestationandearlychildhoodinwhichonlyaportionoftheearlybrain'sdevelopingneuronssurvive.Scientistshopetobypassthisneuralmigrationprocessbyinjectingneuralstemcellsdirectlyintotargetregions,aswellastocreatedrugs that promote this process of neurogenesis (creatingnewneurons) to repair brain damage frominjuryordisease.71AnexperimentbygeneticsresearchersFredGage,G.Kempermann,andHenriettevanPraagattheSalkInstituteforBiologicalStudiesshowedthatneurogenesisisactuallystimulatedbyourexperience.Movingmicefromasterile,uninterestingcagetoastimulatingoneapproximatelydoubledthenumberofdividingcellsintheirhippocampusregions.72

ModelingRegionsoftheBrain

Mostprobablythehumanbrainis,inthemain,composedoflargenumbersofrelativelysmalldistributedsystems,arrangedbyembryology intoacomplexsociety that iscontrolled inpart(butonlyinpart)byserial,symbolicsystemsthatareaddedlater.Butthesubsymbolicsystemsthatdomostoftheworkfromunderneathmust,bytheirverycharacter,blockalltheotherpartsofthebrainfromknowingmuchabouthowtheywork.Andthis,itself,couldhelpexplainhowpeople do somany things yet have such incomplete ideas on how those things are actuallydone.

—MARVINMINSKYANDSEYMOURPAPERT73Commonsenseisnotasimplething.Instead,itisanimmensesocietyofhard-earnedpracticalideas—ofmultitudesoflife-learnedrulesandexceptions,dispositionsandtendencies,balancesandchecks.

—MARVINMINSKY

In addition to new insights into the plasticity of organization of each brain region, researchers arerapidly creating detailedmodels of particular regions of the brain. These neuromorphicmodels andsimulations lagonly slightlybehind theavailabilityof the informationonwhich theyarebased.Therapid success of turning the detailed data from studies of neurons and the interconnection data fromneuralscanningintoeffectivemodelsandworkingsimulationsbeliesoften-statedskepticismaboutourinherentcapabilityofunderstandingourownbrains. Modelinghuman-brain functionalityonanonlinearity-by-nonlinearityand synapse-by-synapsebasis is generally not necessary. Simulations of regions that storememories and skills in individualneuronsandconnections(forexample,thecerebellum)domakeuseofdetailedcellularmodels.Evenfortheseregions,however,simulationsrequirefarlesscomputationthanisimpliedbyalloftheneuralcomponents.Thisistrueofthecerebellumsimulationdescribedbelow.Althoughthereisagreatdealofdetailedcomplexityandnonlinearityinthesubneuralpartsofeachneuron,aswellasachaotic,semirandomwiringpatternunderlyingthe trillionsofconnections in thebrain, significantprogresshasbeenmadeover thepast twentyyears in themathematicsofmodelingsuchadaptivenonlinearsystems.Preservingtheexactshapeofeverydendriteandtheprecise"squiggle"of every interneuronal connection is generally not necessary. We can understand the principles ofoperation of extensive regions of the brain by examining their dynamics at the appropriate level ofanalysis. Wehavealreadyhadsignificantsuccess increatingmodelsandsimulationsofextensivebrainregions. Applying tests to these simulations and comparing the data to that obtained frompsychophysical experiments on actual human brains have produced impressive results. Given therelativecrudenessofourscanningandsensingtoolstodate,thesuccessinmodeling,asillustratedbythefollowingworksinprogress,demonstratestheabilitytoextracttherightinsightsfromthemassofdatabeinggathered.Thefollowingareonlyafewexamplesofsuccessfulmodelsofbrainregions,allworksinprogress.ANeuromorphicModel:TheCerebellumAquestionIexaminedinTheAgeofSpiritualMachinesis:howdoesaten-year-oldmanagetocatchaflyball?74Allthatachildcanseeistheball'strajectoryfromhispositionintheoutfield.Toactuallyinfer the path of the ball in three-dimensional space would require solving difficult simultaneousdifferentialequations.Additionalequationswouldneedtobesolvedtopredictthefuturecourseoftheball, and more equations to translate these results into what was required of the player's ownmovements.Howdoesayoungoutfielderaccomplishallofthisinafewsecondswithnocomputerandnotrainingindifferentialequations?Clearly,heisnotsolvingequationsconsciously,buthowdoeshisbrainsolvetheproblem?SinceASMwaspublished,wehaveadvancedconsiderablyinunderstandingthisbasicprocessofskillformation.AsIhadhypothesized,theproblemisnotsolvedbybuildingamentalmodelofthree-dimensionalmotion.Rather,theproblemiscollapsedbydirectlytranslatingtheobservedmovementsoftheballintotheappropriatemovementoftheplayerandchangesintheconfigurationofhisarmsandlegs. Alexandre Pouget of the University of Rochester and Lawrence H. Snyder of WashingtonUniversityhavedescribedmathematical"basisfunctions"thatcanrepresent thisdirect transformationof perceived movement in the visual field to required movements of the muscles.75 Furthermore,analysis of recently developed models of the functioning of the cerebellum demonstrate that ourcerebellarneuralcircuitsareindeedcapableoflearningandthenapplyingtherequisitebasisfunctionsto implement these sensorimotor transformations.When we engage in the trial-and-error process oflearning to perform a sensorimotor task, such as catching a fly ball, we are training the synapticpotentialsofthecerebellarsynapsestolearntheappropriatebasisfunctions.Thecerebellumperformstwotypesoftransformationswiththesebasisfunctions:goingfromadesiredresulttoanaction(called"inverse internalmodels")andgoing fromapossible setofactions toananticipated result ("forward

internalmodels").TomasoPoggiohaspointedoutthattheideaofbasisfunctionsmaydescribelearningprocessesinthebrainthatgobeyondmotorcontrol.76 Thegrayandwhite,baseball-sized,bean-shapedbrain regioncalled thecerebellumsitson thebrain stem and comprisesmore than half of the brain's neurons. It provides awide range of criticalfunctions, includingsensorimotorcoordination,balance,controlofmovementtasks,andtheabilitytoanticipate the resultsofactions (ourownaswellas thoseofotherobjectsandpersons).77Despite itsdiversity of functions and tasks, its synaptic and cell organization is extremely consistent, involvingonlyseveraltypesofneurons.Thereappearstobeaspecifictypeofcomputationthatitaccomplishes.78Despitetheuniformityofthecerebellum'sinformationprocessing,thebroadrangeofitsfunctionscanbeunderstoodintermsofthevarietyofinputsitreceivesfromthecerebralcortex(viathebrain-stemnucleiandthenthroughthecerebellum'smossyfibercells)andfromotherregions(particularlythe"inferior olive" region of the brain via the cerebellum's climbing fiber cells). The cerebellum isresponsibleforourunderstandingofthetimingandsequencingofsensoryinputsaswellascontrollingourphysicalmovements. Thecerebellumisalsoanexampleofhowthebrain'sconsiderablecapacitygreatlyexceedsitscompactgenome.Mostof thegenome that isdevoted to thebraindescribes thedetailed structureofeachtypeofneuralcell(includingitsdendrites,spines,andsynapses)andhowthesestructuresrespondtostimulationandchange.Relativelylittlegenomiccodeisresponsiblefortheactual"wiring."Inthecerebellum,thebasicwiringmethodisrepeatedbillionsoftimes.It isclearthatthegenomedoesnotprovidespecificinformationabouteachrepetitionofthiscerebellarstructurebutratherspecifiescertainconstraintsastohowthisstructureisrepeated(justasthegenomedoesnotspecifytheexactlocationofcellsinotherorgans).

Someoftheoutputsofthecerebellumgotoabouttwohundredthousandalphamotorneurons,

whichdeterminethefinalsignalstothebody'sapproximatelysixhundredmuscles.Inputstothealphamotorneuronsdonotdirectlyspecifythemovementsofeachofthesemusclesbutarecodedinamorecompact, asyetpoorlyunderstood, fashion.The final signals to themuscles aredeterminedat lowerlevels of the nervous system, specifically in the brain stem and spinal cord.79 Interestingly, thisorganizationistakentoanextremeintheoctopus,thecentralnervoussystemofwhichapparentlysendsveryhigh-levelcommandstoeachofitsarms(suchas"graspthatobjectandbringitcloser"),leavingituptoanindependentperipheralnervoussystemineacharmtocarryoutthemission.80 Agreatdealhasbeenlearnedinrecentyearsabout theroleof thecerebellum's threeprincipalnervetypes.Neuronscalled"climbingfibers"appeartoprovidesignalstotrainthecerebellum.Mostoftheoutputof thecerebellumcomes from the largePurkinjecells (named for JohannesPurkinje,whoidentified the cell in 1837), each of which receives about two hundred thousand inputs (synapses),comparedtotheaverageofaboutonethousandforatypicalneuron.Theinputscomelargelyfromthegranulecells,whicharethesmallestneurons,packedaboutsixmillionpersquaremillimeter.Studiesofthe role of the cerebellum during the learning of handwritingmovements by children show that thePurkinje cells actually sample the sequence of movements, with each one sensitive to a specificsample.81Obviously,thecerebellumrequirescontinualperceptualguidancefromthevisualcortex.Theresearcherswereabletolinkthestructureofcerebellumcellstotheobservationthatthereisaninverserelationship between curvature and speed when doing handwriting that is, you can write faster bydrawingstraightlinesinsteadofdetailedcurvesforeachletter.

Detailed cell studies and animal studies have provided us with impressive mathematicaldescriptionsofthephysiologyandorganizationofthesynapsesofthecerebellurn,82aswellasofthecodingofinformationinitsinputsandoutputs,andofthetransformationsperforrned.83Gatheringdatafrommultiple studies, JavierF.Medina,MichaelD.Mauk, and their colleagues at theUniversity ofTexasMedicalSchooldevisedadetailedbottom-upsimulationofthecerebellum.Itfeaturesmorethantenthousandsimulatedneuronsandthreehundredthousandsynapses,anditincludesalloftheprincipaltypesofcerebellumcells.84Theconnectionsof thecellsandsynapsesaredeterminedbyacomputer,which "wires" the simulated cerebellar region by following constraints and rules, similar to thestochastic (randomwithin restrictions)method used towire the actual human brain from its geneticcode.85 Itwould not be difficult to expand theUniversity of Texas cerebellar simulation to a largernumberofsynapsesandcells.TheTexasresearchersappliedaclassicallearningexperimenttotheirsimulationandcomparedtheresults to many similar experiments on actual human conditioning. In the human studies, the taskinvolvedassociatinganauditorytonewithapuffofairappliedtotheeyelid,whichcausestheeyelidtoclose.If thepuffofairandthetonearepresentedtogetherforonehundredtotwohundredtrials, thesubjectwilllearntheassociationandclosethesubject'seyeuponmerelyhearingthetone.Ifthetoneisthen presentedmany timeswithout the air puff, the subject ultimately learns to disassociate the twostimuli (to "extinguish" the response), so the learning is bidirectional. After tuning a variety ofparameters, thesimulationprovideda reasonablematch toexperimental resultsonhumanandanimalcerebellar conditioning. Interestingly, the researchers found that if they created simulated cerebellarlesions (by removing portions of the simulated cerebellar network), they got results similar to thoseobtainedinexperimentsonrabbitsthathadreceivedactualcerebellarlesions.86 Onaccountof theuniformityof this largeregionof thebrainand therelativesimplicityof itsinterneuronalwiring,itsinput-outputtransformationsarerelativelywellunderstood,comparedtothoseof other brain regions. Although the relevant equations still require refinement, this bottom-upsimulationhasprovedquiteimpressive.

AnotherExample:Watts'sModeloftheAuditoryRegions

Ibelievethatthewaytocreateabrain-likeintelligenceistobuildareal-timeworkingmodelsystem, accurate in sufficientdetail to express the essenceof eachcomputation that isbeingperformed,andverifyitscorrectoperationagainstmeasurementsoftherealsystem.Themodelmust run in real-time so thatwewill be forced to dealwith inconvenient and complex real-world inputs thatwemight not otherwise think to present to it. Themodelmust operate atsufficientresolution tobecomparable to therealsystem,so thatwebuild theright intuitionsaboutwhatinformationisrepresentedateachstage.FollowingMead,87themodeldevelopmentnecessarilybeginsat theboundariesof thesystem(i.e., thesensors)where therealsystemiswell-understood, and then can advance into the less-understood regions....In this way, themodelcancontributefundamentallytoouradvancingunderstandingofthesystem,ratherthansimply mirroring the existing understanding. In the context of such great complexity, it ispossiblethattheonlypracticalwaytounderstandtherealsystemistobuildaworkingmodel,fromthesensors inward,buildingonournewlyenabledability tovisualize thecomplexityofthesystemasweadvanceintoit.Suchanapproachcouldbecalledreverse-engineeringofthebrain....Note that I amnot advocating ablind copyingof structureswhosepurposewedon'tunderstand,likethelegendaryIcaruswhonaivelyattemptedtobuildwingsoutoffeathersandwax.Rather,Iamadvocatingthatwerespectthecomplexityandrichnessthatisalreadywell-understoodatlowlevels,beforeproceedingtohigherlevels.

—LLOYDWATTS88Amajorexampleofneuromorphicmodelingofaregionofthebrainisthecomprehensivereplicaofasignificant portion of the human auditory-processing system developed by Lloyd Watts and hiscolleagues.89Itisbasedonneurobiologicalstudiesofspecificneurontypesaswellasoninformationregarding interneuronal connection. The model, which has many of the same properties as humanhearingandcan locateand identifysounds,hasfiveparallelpathsofprocessingauditory informationand includes the actual intermediate representations of this information at each stage of neuralprocessing.Wattshasimplementedhismodelasreal-timecomputersoftwarewhich,thoughaworkinprogress,illustratesthefeasibilityofconvertingneurobiologicalmodelsandbrainconnectiondataintoworkingsimulations.Thesoftwareisnotbasedonreproducingeachindividualneuronandconnection,asisthecerebellummodeldescribedabove,butratherthetransformationsperformedbyeachregion. Watts's software is capable of matching the intricacies that have been revealed in subtleexperimentsonhumanhearingandauditorydiscrimination.Wattshasusedhismodelasapreprocessor(frontend)inspeech-recognitionsystemsandhasdemonstrateditsabilitytopickoutonespeakerfrombackgroundsounds(the"cocktailpartyeffect").Thisisanimpressivefeatofwhichhumansarecapablebutupuntilnowhadnotbeenfeasibleinautomatedspeech-recognitionsystems.90Likehumanhearing,Watts'scochleamodelisendowedwithspectralsensitivity(wehearbetteratcertain frequencies), temporal responses (we are sensitive to the timing of sounds, which create thesensation of their spatial locations), masking, nonlinear frequency-dependent amplitude compression(whichallowsforgreaterdynamicrange—theabilitytohearbothloudandquietsounds),gaincontrol(amplification),andothersubtlefeatures.Theresultsitobtainsaredirectlyverifiablebybiologicalandpsychophysicaldata. The next segment of themodel is the cochlear nucleus, whichYaleUniversity professor ofneuroscience andneurobiologyGordonM.Shepherd91 has described as "one of the best understoodregionsofthebrain."92Watts'ssimulationofthecochlearnucleusisbasedonworkbyE.Youngthatdescribes in detail "the essential cell types responsible for detecting spectral energy, broadbandtransients, fine tuning in spectral channels, enhancing sensitivity to temporary envelope in spectralchannels, and spectral edges andnotches, allwhile adjustinggain foroptimumsensitivitywithin the

limiteddynamicrangeofthespikingneuralcode.93 TheWatts model captures many other details, such as the interaural time difference (lTD)computedby themedial superiorolive cells.94 It also represents the interaural-leveldifference (ILD)computedby the lateral superiorolivecellsandnormalizationsandadjustmentsmadeby the inferiorcolliculuscells.95

TheVisualSystemWe'vemadeenoughprogressinunderstandingthecodingofvisualinformationthatexperimentalretinaimplantshavebeendevelopedandsurgically installed inpatients.97However,becauseof the relativecomplexityofthevisualsystem,ourunderstandingoftheprocessingofvisualinformationlagsbehindourknowledgeoftheauditoryregions.Wehavepreliminarymodelsofthetransformationsperformedbytwovisualareas(calledVIandMT),althoughnotattheindividualneuronlevel.Therearethirty-six

othervisualareas,andwewillneedtobeabletoscanthesedeeperregionsatveryhighresolutionorplaceprecisesensorstoascertaintheirfunctions.ApioneerinunderstandingvisualprocessingisMIT'sTomasoPoggio,whohasdistinguisheditstwotasksasidentificationandcategorization.98Theformerisrelativelyeasytounderstand,accordingto Poggio, andwe have already designed experimental and commercial systems that are reasonablysuccessfulinidentifyingfaces.99Theseareusedaspartofsecuritysystemstocontrolentryofpersonnelandinbankmachines.Categorization—theabilitytodifferentiate,forexample,betweenapersonandacarorbetweenadogandacatisamorecomplexmatter,althoughrecentlyprogresshasbeenmade.100 Early (in terms of evolution) layers of the visual system are largely a feedforward (lackingfeedback) system in which increasingly sophisticated features are detected. Poggio andMaximilianRiesenhuberwritethat"singleneuronsinthemacaqueposteriorinferotemporalcortexmaybetunedto...adictionaryof thousandsofcomplexshapes."Evidence thatvisual recognitionusesa feedforwardsystemduring recognition includesMEGstudies that show thehumanvisual system takesabout150milliseconds to detect an object. This matches the latency of feature-detection cells in theinferotemporal cortex, so there does not appear to be time for feedback to playa role in these earlydecisions. Recent experiments have used a hierarchical approach in which features are detected to heanalyzed by later layers of the system.101 From studies on macaque monkeys, neurons in theinferotemporal cortex appear to respond to complex features of objects on which the animals aretrained.Whilemost of theneurons respondonly to a particular viewof theobject, someare able torespondregardlessofperspective.Otherresearchonthevisualsystemofthemacaquemonkeyincludesstudiesonmanyspecifictypesofcells,connectivitypatterns,andhigh-leveldescriptionsofinformationflow.102ExtensiveliteraturesupportstheuseofwhatIcall"hypothesisandtest"inmorecomplexpattern-recognition tasks.Thecortexmakesaguessaboutwhat it is seeingand thendetermineswhether thefeaturesofwhatisactuallyinthefieldofviewmatchitshypothesis.103We’reoftenmorefocusedonthehypothesisthantheactualtest,whichexplainswhypeopleoftenseeandhearwhattheyexpecttoperceive rather than what is actually there. "Hypothesis and test" is also a useful strategy in ourcomputer-basedpattern-recognitionsystems.Althoughwehavetheillusionofreceivinghigh-resolutionimagesfromoureyes,whattheopticnerveactuallysendstothebrainisjustoutlinesandcluesaboutpointsofinterestinourvisualfield.Wethenessentiallyhallucinate theworldfromcorticalmemories that interpretaseriesofextremelylow-resolution movies that arrive in parallel channels. In a 2001 study published in Nature, Frank S.Werblin, professor of molecular and cell biology at the University of California at Berkeley, anddoctoralstudentBotonRoska,M.D.,showedthattheopticnervecarriestentotwelveoutputchannels,eachofwhichcarriesonlyminimalinformationaboutagivenscene.104Onegroupofwhatarecalledganglioncellssendsinformationonlyaboutedges(changesincontrast).Another .groupdetectsonlylargeareasofuniformcolor,whereasathirdgroupissensitiveonlytothebackgroundsbehindfiguresofinterest.

"Eventhoughwethinkweseetheworldsofully,whatwearereceivingisreallyjusthints,edgesinspaceandtime,"saysWerblin."These12picturesoftheworldconstitutealltheinformationwewilleverhaveaboutwhat'sout there,andfromthese12pictures,whicharesosparse,wereconstruct therichnessofthevisualworld.I'mcurioushownatureselectedthese12simplemoviesandhowitcanbethattheyaresufficienttoprovideuswithalltheinformationweseemtoneed."Suchfindingspromisetobeamajor advance indevelopinganartificial system that could replace theeye, retina, andearlyoptic-nerveprocessing. In chapter 3, Imentioned theworkof robotics pioneerHansMoravec,whohas been reverseengineering the imageprocessingdoneby the retinaandearlyvisual-processing regions in thebrain.FormorethanthirtyyearsMoravechasbeenconstructingsystemstoemulatetheabilityofourvisualsystemtobuildrepresentationsoftheworld.Ithasonlybeenrecentlythatsufficientprocessingpowerhasbeenavailable inmicroprocessors to replicate thishuman-level featuredetection,andMoravec isapplyinghiscomputersimulationstoanewgenerationofrobotsthatcannavigateunplanned,complexenvironmentswithhuman-levelvision.105 CarverMeadhasbeenpioneeringtheuseofspecialneuralchipsthatutilizetransistors intheirnative analog mode, which can provide very efficient emulation of the analog nature of neuralprocessing. Mead has demonstrated a chip that performs the functions of the retina and earlytransformationsintheopticnerveusingthisapproach.106

Aspecialtypeofvisualrecognitionisdetectingmotion,oneofthefocusareasoftheMaxPlanckInstituteofBiologyinTubingen,Germany.Thebasicresearchmodelissimple:comparethesignalatonereceptorwithatime-delayedsignalattheadjacentreceptor.107Thismodelworksforcertainspeedsbutleadstothesurprisingresult thataboveacertainspeed,increasesintheIvelocityofanobservedobjectwill decrease the response of thismotion detector.Experimental results on animals (based onbehaviorandanalysisofI,neuronaloutputs)andhumans(basedonreportedperceptions)havecloselymatchedthemodel.OtherWorksinProgress:AnArtificialHippocampusandanArtificialOlivocerebellarRegionThehippocampusisvitalforlearningnewinformationandlong-termstorageofmemories.TedBergerandhiscolleaguesattheUniversityofSouthernCaliforniamappedthesignalpatternsofthisregionbystimulatingslicesofrathippocampuswithelectricalsignalsmillionsoftimestodeterminewhichinputproduced a corresponding output.108 They then developed a real-time mathematical model of thetransformations performedby layers of the hippocampus andprogrammed themodel onto a chip.109Theirplanistotestthechipinanimalsbyfirstdisablingthecorrespondinghippocampusregion,notingthe resultingmemory failure, and then determiningwhether thatmental function can be restored byinstallingtheirhippocampalchipinplaceofthedisabledregion.Ultimately,thisapproachcouldbeusedtoreplacethehippocampusinpatientsaffectedbystrokes,epilepsy,orAlzheimer'sdisease.Thechipwouldbelocatedonapatient'sskull,ratherthaninsidethebrain,andwouldcommunicatewiththebrainviatwoarraysofelectrodes,placedoneithersideofthedamaged hippocampal section.Onewould record the electrical activity coming from the rest of thebrain,whiletheotherwouldsendthenecessaryinstructionsbacktothebrain. Another brain region being modeled and simulated is the olivocerebellar region, which isresponsibleforbalanceandcoordinatingthemovementoflimbs.Thegoaloftheinternationalresearchgroupinvolvedinthiseffortistoapplytheirartificialolivocerebellarcircuittomilitaryrobotsaswellastorobotsthatcouldassistthedisabled.110Oneoftheirreasonsforselectingthisparticularbrainregionwas that "it's present in all vertebrates—it's verymuch the same from themost simple to themostcomplex brains," explains Rodolfo Llinas, one of the researchers and a neuroscientist at NewYorkUniversityMedicalSchool."Theassumptionisthatitisconserved[inevolution]becauseitembodiesavery intelligent solution. As the system is involved in motor coordination—and we want to have amachinethathassophisticatedmotorcontrol—thenthechoice[ofthecircuittomimic]waseasy." One of the unique aspects of their simulator is that it uses analog circuits. Similar toMead'spioneering work on analog emulation of brain regions, the researchers found substantially greaterperformancewithfarfewercomponentsbyusingtransistorsintheirnativeanalogmode. One of the team's researchers, Ferdinando Mussa-Ivaldi, a neuroscientist at NorthwesternUniversity, commented on the applications of an artificial olivocerebellar circuit for the disabled:"Thinkofaparalyzedpatient.Itispossibletoimaginethatmanyordinarytasks—suchasgettingaglassofwater,dressing,undressing,transferringtoawheelchair—couldbecarriedoutbyroboticassistants,thusprovidingthepatientwithmoreindependence."UnderstandingHigher-LevelFunctions:Imitation,Prediction,andEmotion

Operationsofthoughtarelikecavalrychargesinabattle—theyarestrictlylimitedinnumber,theyrequirefreshhorses,andmustonlybemadeatdecisivemoments.

—ALFREDNORTHWHITEHEADBut thebigfeatureofhuman-level intelligence isnotwhat itdoeswhenitworksbutwhat it

doeswhenit'sstuck.

—MARVINMINSKYIfloveistheanswer,couldyoupleaserephrasethequestion?

—LILYTOMLINBecause it sits at the top of the neural hierarchy, the part of the brain least well understood is thecerebral cortex. This region,which consists of six thin layers in the outermost areas of the cerebralhemispheres, contains billions of neurons.According to ThomasM.Bartol Jr. of theComputationalNeurobiology Laboratory of the Salk Institute of Biological Studies, "A single cubic millimeter ofcerebral cortexmaycontainon theorderof5billion ... synapsesofdifferent shapes and sizes.”Thecortex is responsible for perception, planning, decision making and most of what we regard asconsciousthinking.Ourabilitytouselanguage,anotheruniqueattributeofourspecies,appearstobelocatedinthisregion.Anintriguinghintabouttheoriginoflanguageandakeyevolutionarychangethatenabledtheformationofthisdistinguishingskillistheobservationthatonlyafewprimates,includinghumansandmonkeys,areabletousean(actual)mirrortomasterskills.TheoristsGiacomoRizzolattiandMichaelArbib hypothesized that language emerged frommanual gestures (which monkeys—and, of course,humans—are capable of). Performing manual gestures requires the ability to mentally correlate theperformanceandobservationofone'sownhandmovements.111Their"mirrorsystemhypothesis"isthatthekey to theevolutionof language isapropertycalled"parity,"which is theunderstanding that thegesture(orutterance)hasthesamemeaningforthepartymakingthegestureasforthepartyreceivingit;thatis,theunderstandingthatwhatyouseeinamirroristhesame(althoughreversedleft-to-right)aswhat is seen by someone elsewatching you.Other animals are unable to understand the image in amirrorinthisfashion,anditisbelievedthattheyaremissingthiskeyabilitytodeployparity. Acloselyrelatedconceptisthattheabilitytoimitatethemovements(or, inthecaseofhumanbabies, vocal sounds) of others is critical to developing language.112 Imitation requires the ability tobreakdownanobservedpresentationintoparts,eachofwhichcanthenbemasteredthroughrecursiveanditerativerefinement. Recursion is the key capability identified in a new theory of linguistic competence. InNoamChomsky'searlytheoriesoflanguageinhumans,hecitedmanycommonattributesthataccountforthesimilarities in human languages. In a 2002 paper byMarc Hauser, Noam Chomsky, and TecumsehFitch,theauthorscitethesingleattributionof"recursion"asaccountingfortheuniquelanguagefacultyofthehumanspecies.113Recursionistheabilitytoputtogethersmallpartsintoalargerchunk,andthenusethatchunkasapartinyetanotherstructureandtocontinuethisprocessiteratively.Inthisway,weareabletobuildtheelaboratestructuresofsentencesandparagraphsfromalimitedsetofwords.Anotherkeyfeatureofthehumanbrainistheabilitytomakepredictions,includingpredictionsabouttheresultsofitsowndecisionsandactions.Somescientistsbelievethatpredictionistheprimaryfunction of the cerebral cortex, although the cerebellum also plays amajor role in the prediction ofmovement.Interestingly,weareabletopredictoranticipateourowndecisions.WorkbyphysiologyprofessorBenjaminLibetattheUniversityofCaliforniaatDavisshowsthatneuralactivitytoinitiateanactionactuallyoccursaboutathirdofasecondbeforethebrainhasmadethedecisiontotaketheaction.Theimplication,accordingtoLibet,isthatthedecisionisreallyanillusion,that"consciousnessisoutoftheloop."The cognitive scientist and philosopherDanielDennett describes the phenomenon as follows:"The action is originally precipitated in some part of the brain, and off fly the signals to muscles,pausingenroutetotellyou,theconsciousagent,whatisgoingon(butlikeallgoodofficialslettingyou,thebumblingpresident,maintaintheillusionthatyoustarteditall)."114Arelatedexperimentwasconductedrecentlyinwhichneurophysiologistselectronicallystimulated

points in thebrain to induceparticular emotional feelings.The subjects immediately cameupwith arationaleforexperiencingthoseemotions.Ithasbeenknownformanyyearsthatinpatientswhoseleftandrightbrainsarenolongerconnected,onesideofthebrain(usuallythemoreverballeftside)willcreateelaborateexplanations("confabulations")foractionsinitiatedbytheotherside,asiftheleftsidewerethepublic-relationsagentfortherightside.Themostcomplexcapabilityofthehumanbrain—whatIwouldregardasitscuttingedge—isouremotionalintelligence.Sittinguneasilyatthetopofourbrain'scomplexandinterconnectedhierarchyisourabilitytoperceiveandrespondappropriatelytoemotion, tointeract insocialsituations, tohaveamoral sense, to get the joke, and to respond emotionally to art and music, among other high-levelfunctions.Obviously, lower-levelfunctionsofperceptionandanalysisfeed intoourbrain'semotionalprocessing,butwearebeginningtounderstandtheregionsofthebrainandeventomodelthespecifictypesofneuronsthathandlesuchissues.Theserecentinsightshavebeentheresultofourattemptstounderstandhowhumanbrainsdifferfromthoseofothermammals.Theansweristhatthedifferencesareslightbutcritical,andtheyhelpusdiscern how the brain processes emotion and related feelings.One difference is that humans have alarger cortex, reflecting our stronger capability for planning, decision making, and other forms ofanalyticthinking.Anotherkeydistinguishingfeatureisthatemotionallychargedsituationsappeartobehandled by special cells called spindle cells, which are found only in humans and some great apes.Theseneuralcellsare large,with longneural filamentscalledapicaldendrites thatconnectextensivesignalsfrommanyotherbrainregions.Thistypeof"deep"interconnectedness,inwhichcertainneuronsprovide connections across numerous regions, is a feature that occurs increasingly as we go up theevolutionaryladder.Itisnotsurprisingthatthespindlecells,involvedastheyareinhandlingemotionandmoral judgment,would have this form of deep interconnectedness, given the complexity of ouremotionalreactions.Whatisstartling,however,ishowfewspindlecellsthereareinthistinyregion:onlyabout80,000in the human brain (about 45,000 in the right hemisphere and 35,000 in the left hemisphere). Thisdisparityappears toaccount for theperception thatemotional intelligence is theprovinceof the rightbrain,although thedisproportion ismodest.Gorillashaveabout16,000of thesecells,bonobosabout2,100,andchimpanzeesabout1,800.Othermammalslackthemcompletely.

Dr. Arthur Craig of the Barrow Neurological Institute in Phoenix has recently provided adescriptionof thearchitectureof thespindlecells.115 Inputs from thebody (estimatedathundredsofmegabitspersecond),includingnervesfromtheskin,muscles,organs,andotherareas,streamintotheupperspinalcord.Thesecarrymessagesabouttouch,temperature,acidlevels(forexample,lacticacidin muscles), the movement of food through the gastrointestinal tract, and many other types ofinformation.This data is processed through the brain stem andmidbrain.Key cells calledLamina 1neuronscreateamapof thebodyrepresenting itscurrentstate,notunlike thedisplaysusedbyflightcontrollerstotrackairplanes. Theinformationthenflowsthroughanut-sizeregioncalledtheposteriorventromedialnucleus(VMpo),which apparently computes complex reactions to bodily states such as "this tastes terrible,""whatastench,"or"thatlighttouchisstimulating."Theincreasinglysophisticatedinformationendsupattworegionsofthecortexcalledtheinsula.Thesestructures,thesizeofsmallfingers,arelocatedontheleftandrightsidesofthecortex.CraigdescribestheVMpoandthetwoinsularegionsas"asystemthatrepresentsthematerialme."Althoughthemechanismsarenotyetunderstood,theseregionsarecriticaltoself-awarenessandcomplicatedemotions.Theyarealsomuchsmallerinotheranimals.Forexample,theVMpoisaboutthesizeofagrainofsandinmacaquemonkeysandevensmallerinlower-levelanimals.Thesefindingsareconsistentwithagrowingconsensusthatouremotionsarecloselylinkedtoareasofthebrainthatcontainmapsofthebody,aviewpromotedbyDr.AntonioDamasioattheUniversityofIowa.116Theyare also consistent with the view that a great deal of our thinking is directed toward our bodies:protectingandenhancingthem,aswellasattendingtotheirmyriadneedsanddesires.Veryrecentlyyetanotherlevelofprocessingofwhatstartedoutassensoryinformationfromthebodyhasbeendiscovered.Datafromthetwoinsularegionsgoesontoa tinyareaat thefrontof theright insula called the frontoinsular cortex.This is the region containing the spindle cells, and tMRIscanshaverevealedthatitisparticularlyactivewhenapersonisdealingwithhigh-levelemotionssuchas love, anger, sadness, and sexual desire. Situations that strongly activate the spindle cells includewhenasubjectlooksatherromanticpartnerorhearsherchildcrying.Anthropologistsbelievethatspindlecellsmadetheirfirstappearancetentofifteenmillionyearsagointheas-yet-undiscoveredcommonancestortoapesandearlyhominids(thefamilyofhumans)andrapidlyincreasedinnumbersaroundonehundredthousandyearsago.Interestingly,spindlecellsdonotexist in newborn humans but begin to appear only at around the age of four months and increasesignificantly from ages one to three. Children's ability to deal with moral issues and perceive suchhigher-levelemotionsaslovedevelopduringthissametimeperiod.Thespindlecellsgaintheirpowerfromthedeepinterconnectednessoftheirlongapicaldendriteswith many other brain regions. The high-level emotions that the spindle cells process are affected,thereby,byallofourperceptualandcognitiveregions.Itwillbedifficult,therefore,toreverseengineertheexactmethodsofthespindlecellsuntilwehavebettermodelsofthemanyotherregionstowhichtheyconnect.However,itisremarkablehowfewneuronsappeartobeexclusivelyinvolvedwiththeseemotions.Wehavefiftybillionneuronsinthecerebellumthatdealwithskillformation,billionsinthecortex that perform the transformations for perception and rational planning, but only about eightythousandspindlecellsdealingwithhigh-levelemotions.Itisimportanttopointoutthatthespindlecellsarenotdoingrationalproblemsolving,whichiswhywedon'thaverationalcontroloverourresponsestomusicoroverfalling in love.Therestof thebrain isheavilyengaged,however, in tryingtomakesenseofourmysterioushigh-levelemotions.InterfacingtheBrainandMachines

Iwanttodosomethingwithmylife;Iwanttobeacyborg.

—KEVINWARWICKUnderstanding the methods of the human brain will help us to design similar biologically inspiredmachines.Anotherimportantapplicationwillbetoactuallyinterfaceourbrainswithcomputers,whichIbelievewillbecomeanincreasinglyintimatemergerinthedecadesahead. Already theDefenseAdvancedResearchProjectsAgency isspending$24millionperyearoninvestigatingdirect interfacesbetweenbrainandcomputer.Asdescribedabove (see the section"TheVisualSystem"onp.185),TomasoPoggioandJamesDiCarloatMIT,alongwithChristofKochattheCalifornia Institute of Technology (Caltech), are attempting to developmodels of the recognition ofvisualobjectsandhowthisinformationisencoded.Thesecouldeventuallybeusedtotransmitimagesdirectlyintoourbrains. Miguel Nicolelis and his colleagues at Duke University implanted sensors in the brains ofmonkeys,enablingtheanimalstocontrolarobotthroughthoughtalone.Thefirststepintheexperimentinvolvedteachingthemonkeystocontrolacursoronascreenwithajoystick.Thescientistscollectedapattern of signals from EEGs (brain sensors) and subsequently caused the cursor to respond to theappropriatepatternsratherthanphysicalmovementsofthejoystick.Themonkeysquicklylearnedthatthe joystick was no longer operative and that they could control the cursor just by thinking. This"thoughtdetection"systemwasthenhookeduptoarobot,andthemonkeyswereabletolearnhowtocontrol the robot's movements with their thoughts alone. By getting visual feedback on the robot'sperformance, themonkeyswereable toperfect their thought controlover the robot.Thegoalof thisresearchistoprovideasimilarsystemforparalyzedhumansthatwillenablethemtocontroltheirlimbsandenvironment.Akeychallengeinconnectingneuralimplantstobiologicalneuronsisthattheneuronsgenerateglial cells,which surround a "foreign" object in an attempt to protect the brain. TedBerger and hiscolleaguesaredevelopingspecialcoatingsthatwillappeartobebiologicalandthereforeattractratherthanrepelnearbyneurons. Another approachbeingpursuedby theMaxPlanck Institute forHumanCognitiveandBrainSciences inMunich is directly interfacing nerves and electronic devices.A chip created by Infineonallowsneuronstogrowonaspecialsubstratethatprovidesdirectcontactbetweennervesandelectronicsensors and stimulators. Similar work on a "neurochip" at Caltech has demonstrated two-way,noninvasivecommunicationbetweenneuronsandelectronics.117Wehavealreadylearnedhowtointerfacesurgicallyinstalledneuralimplants.Incochlear(inner-ear) implants it has been found that the auditory nerve reorganizes itself to correctly interpret themultichannel signal from the implant. A similar process appears to take place with the deep-brainstimulation implantusedforParkinson'spatients.Thebiologicalneurons in thevicinityof thisFDA-approved brain implant receive signals from the electronic device and respond just as if they hadreceived signals from the biological neurons that were once functional. Recent versions of theParkinson's-disease implantprovide theability todownloadupgradedsoftwaredirectly to the implantfromoutsidethepatient.TheAcceleratingPaceofReverseEngineeringtheBrain

Homosapiens,thefirsttrulyfreespecies,isabouttodecommissionnaturalselection,theforcethatmadeus....[S]oonwemustlookdeepwithinourselvesanddecidewhatwewishtobecome.

—E.O.WILSON,CONSILIENCE:THEUNITYOFKNOWLEDGE,1998Weknowwhatweare,butknownotwhatwemaybe.

—WILLIAMSHAKESPEARE

Themostimportantthingisthis:Tobeableatanymomenttosacrificewhatweareforwhatwecouldbecome.

—CHARLESDUBOISSomeobservershaveexpressedconcernthataswedevelopmodels,simulations,andextensionstothehuman brain we risk not really understandingwhat we are tinkering with and the delicate balancesinvolved.AuthorW.FrenchAndersonwrites:

We may be like the young boy who loves to take things apart. He is bright enough todisassembleawatch,andmaybeevenbrightenoughtogetitbacktogethersothatitworks.Butwhat if he tries to "improve" it? ... The boy can understand what is visible, but he cannotunderstandthepreciseengineeringcalculationsthatdetermineexactlyhowstrongeachspringshouldbe....Attemptsonhisparttoimprovethewatchwillprobablyonlyharmit....Ifear...we,too,donotreallyunderstandwhatmakesthe[lives]wearetinkeringwithtick.118

Anderson'sconcern,however,doesnotreflectthescopeofthebroadandpainstakingeffortbytensof thousands of brain and computer scientists tomethodically test out the limits and capabilities ofmodelsandsimulationsbeforetakingthemtothenextstep.Wearenotattemptingtodisassembleandreconfigure the brain's trillions of parts without a detailed analysis at each stage. The process ofunderstanding the principles of operation of the brain is proceeding through a series of increasinglysophisticatedmodelsderivedfromincreasinglyaccurateandhigh-resolutiondata. As thecomputationalpower toemulate thehumanbrainapproaches—we'realmost therewithsupercomputers—the efforts to scan and sense the human brain and to build working models andsimulationsofitareaccelerating.Aswitheveryotherprojectioninthisbook,itiscriticaltounderstandtheexponentialnatureofprogressinthisfield.Ifrequentlyencountercolleagueswhoarguethatitwillbeacenturyor longerbeforewecanunderstandindetail themethodsof thebrain.Aswithsomanylong-termscientificprojections,thisoneisbasedonalinearviewofthefutureandignorestheinherentaccelerationofprogress,aswellastheexponentialgrowthofeachunderlyingtechnology.Suchoverlyconservative views are also frequently based on an underestimation of the breadth of contemporaryaccomplishments,evenbypractitionersinthefield. Scanningandsensingtoolsaredoublingtheiroverallspatialandtemporalresolutioneachyear.Scanning-bandwidth, price-performance, and image-reconstruction times are also seeing comparableexponentialgrowth.Thesetrendsholdtrueforalloftheformsofscanning:fullynoninvasivescanning,in vivo scanning with an exposed skull, and destructive scanning. Databases of brain-scanninginformationandmodelbuildingarealsodoublinginsizeaboutonceperyear. We have demonstrated that our ability to build detailedmodels andworking simulations ofsubcellularportions,neurons,andextensiveneuralregionsfollowscloselyupontheavailabilityoftherequisitetoolsanddata.Theperformanceofneuronsandsubcellularportionsofneuronsofteninvolvessubstantialcomplexityandnumerousnonlinearities,buttheperformanceofneuralclustersandneuronalregionsisoftensimplerthantheirconstituentparts.Wehaveincreasinglypowerfulmathematicaltools,implementedineffectivecomputersoftware,thatareabletoaccuratelymodelthesetypesofcomplexhierarchical, adaptive, semirandom, self-organizing, highly nonlinear systems.Our success to date ineffectivelymodelingseveralimportantregionsofthebrainshowstheeffectivenessofthisapproach.Thegenerationofscanningtoolsnowemergingwillforthefirsttimeprovidespatialandtemporalresolution capable of observing in real time the performance of individual dendrites, spines, andsynapses. These tools will quickly lead to a new generation of higher-resolution models andsimulations.Oncethenanoboteraarrivesinthe2020swewillbeabletoobservealloftherelevantfeaturesofneuralperformancewithveryhighresolutionfrominsidethebrainitself.Sendingbillionsofnanobotsthroughitscapillarieswillenableustononinvasivelyscananentireworkingbraininrealtime.Wehavealready created effective (although still incomplete) models of extensive regions of the brain with

today's relatively crude tools. Within twenty years, we will have at least a millionfold increase incomputational power and vastly improved scanning resolution and bandwidth. So we can haveconfidencethatwewillhavethedata-gatheringandcomputationaltoolsneededbythe2020stomodeland simulate the entire brain,whichwillmake it possible to combine the principles of operation ofhumanintelligencewiththeformsofintelligentinformationprocessingthatwehavederivedfromotherAI research.Wewill also benefit from the inherent strength ofmachines in storing, retrieving, andquickly sharingmassive amounts of information.We will then be in a position to implement thesepowerfulhybridsystemsoncomputationalplatformsthatgreatlyexceedthecapabilitiesofthehumanbrain'srelativelyfixedarchitecture.The Scalability of Human Intelligence. In response to Hofstadter's concern as to whether humanintelligence is just above or below the threshold necessary for "self-understanding," the acceleratingpace of brain reverse engineeringmakes it clear that there are no limits to our ability to understandourselves—or anything else, for thatmatter. The key to the scalability of human intelligence is ourabilitytobuildmodelsofrealityinourmind.Thesemodelscanberecursive,meaningthatonemodelcanincludeothermodels,whichcanincludeyetfinermodels,withoutlimit.Forexample,amodelofabiological cell can includemodels of thenucleus, ribosomes, andother cellular systems. In turn, themodel of the ribosomemay include models of its submolecular components, and then down to theatomsandsubatomicparticlesandforcesthatitcomprises.Ourabilitytounderstandcomplexsystemsisnotnecessarilyhierarchical.Acomplexsystemlikeacellorthehumanbraincannotbeunderstoodsimplybybreakingitdownintoconstituentsubsystemsandtheircomponents.Wehaveincreasinglysophisticatedmathematicaltoolsforunderstandingsystemsthat combine both order and chaos—and there is plenty of both in a cell and in the brain—and forunderstandingthecomplexinteractionsthatdefylogicalbreakdown. Our computers,which are themselves accelerating, havebeen a critical tool in enablingus tohandleincreasinglycomplexmodels,whichwewouldotherwisebeunabletoenvisionwithourbrainsalone.Clearly,Hofstadter'sconcernwouldbecorrect ifwewerelimitedjust tomodels thatwecouldkeep in ourmindswithout technology to assist us. That our intelligence is just above the thresholdnecessary to understand itself results from our native ability, combined with the tools of our ownmaking, to envision, refine, extend, and alter abstract—and increasingly subtle—models of our ownobservations.UploadingtheHumanBrain

Tobecomeafigmentofyourcomputer'simagination.

—DAVID VICTOR DE TRANSCEND, GODLING'S GLOSSARY,DEFINITIONOF"UPLOAD"

Amorecontroversialapplicationthanthescanning-the-brain-to-understand-itscenarioisscanningthebrain to upload it. Uploading a human brain means scanning all of its salient details and thenreinstantiating those details into a suitably powerful computational substrate. This process wouldcaptureaperson'sentirepersonality,memory,skills,andhistory.Ifwearetrulycapturingaparticularperson'smentalprocesses,thenthereinstantiatedmindwillneed a body, since somuchof our thinking is directed towardphysical needs anddesires.As Iwilldiscussinchapter5,bythetimewehavethetoolstocaptureandre-createahumanbrainwithallofitssubtleties,wewillhaveplentyofoptionsfortwenty-first-centurybodiesforbothnonbiologicalhumansandbiologicalhumanswhoavailthemselvesofextensionstoourintelligence.Thehumanbodyversion2.0 will include virtual bodies in completely realistic virtual environments, nanotechnology-basedphysicalbodies,andmore.Inchapter3Idiscussedmyestimatesforthememoryandcomputationalrequirementstosimulate

the human brain. Although I estimated that 1016 cps of computation and 1013 bits of memory aresufficienttoemulatehumanlevelsofintelligence,myestimatesfortherequirementsofuploadingwerehigher:1019cpsand1018bits,respectively.Thereasonforthehigherestimatesisthattheloweronesarebasedontherequirementstore-createregionsofthebrainathumanlevelsofperformance,whereasthehigheronesarebasedoncapturingthesalientdetailsofeachofourapproximately1011neuronsand1014 interneuronalconnections.Onceuploadingisfeasible,wearelikelytofindthathybridsolutionsareadequate.Forexample,wewillprobablyfindthat it issufficient tosimulatecertainbasicsupportfunctionssuchasthesignalprocessingofsensorydataonafunctionalbasis(byplugginginstandardmodules)andreservethecaptureofsubneurondetailsonlyforthoseregionsthataretrulyresponsibleforindividualpersonalityandskills.Nonetheless,wewilluseourhigherestimatesforthisdiscussion. Thebasiccomputational resources (1019cpsand1018 bits)will be available for one thousanddollarsintheearly2030s,aboutadecadelaterthantheresourcesneededforfunctionalsimulation.Thescanning requirements foruploadingarealsomoredaunting than for "merely" re-creating theoverallpowersofhumanintelligence.Intheoryonecoulduploadahumanbrainbycapturingallthenecessarydetailswithoutnecessarilycomprehendingthebrain'soverallplan.Inpractice,however,thisisunlikelytowork.Understanding the principles of operation of the human brainwill revealwhich details areessential and which details are intended to be disordered. We need to know, for example, whichmoleculesintheneurotransmittersarecritical,andwhetherweneedtocaptureoveralllevels,positionandlocation,and/ormolecularshape.AsIdiscussedabove,wearejustlearning,forexample,thatitisthe position of actin molecules and the shape of ePEB molecules in the synapse that are key formemory. It will not be possible to confirm which details are crucial without having confirmed ourunderstanding of the theory of operation. That confirmation will be in the form of a functionalsimulationofhumanintelligencethatpassestheTuringtest,whichIbelievewilltakeplaceby2029.119 Tocapture this levelofdetailwill require scanning fromwithin thebrainusingnanobots, thetechnology forwhichwill be availableby the late2020s.Thus, the early2030s is a reasonable timeframeforthecomputationalperformance,memory,andbrain-scanningprerequisitesofuploading.Likeanyothertechnology,itwilltakesomeiterativerefinementtoperfectthiscapability,sotheendofthe2030sisaconservativeprojectionforsuccessfuluploading.Weshouldpointoutthataperson'spersonalityandskillsdonotresideonlyinthebrain,althoughthat is their principal location. Our nervous system extends throughout the body, and the endocrine(hormonal)systemhasaninfluence,aswell.Thevastmajorityofthecomplexity,however,residesinthebrain,whichisthelocationofthebulkofthenervoussystem.Thebandwidthofinformationfromtheendocrinesystemisquitelow,becausethedeterminingfactorisoveralllevelsofhormones,notthepreciselocationofeachhormonemolecule.Confirmationoftheuploadingmilestonewillbeintheformofa"RayKurzweil"or"JaneSmith"Turingtest,inotherwordsconvincingahumanjudgethattheuploadedre-creationisindistinguishablefromtheoriginalspecificperson.By that timewe'll facesomecomplications indevising therulesofanyTuringtest.SincenonbiologicalintelligencewillhavepassedtheoriginalTuringtestyearsearlier(around 2029), should we allow a nonbiological human equivalent to be a judge? How about anenhancedhuman?Unenhancedhumansmaybecomeincreasinglyhardtofind.Inanyevent,itwillbeaslipperyslopetodefineenhancement,asmanydifferentlevelsofextendingbiologicalintelligencewillbeavailablebythetimewehavepurporteduploads.Anotherissuewillbethatthehumansweseektouploadwillnotbelimitedtotheirbiologicalintelligence.However,uploadingthenonbiologicalportionof intelligencewill be relatively straightforward, since the easeof copyingcomputer intelligencehasalwaysrepresentedoneofthestrengthsofcomputers.Onequestionthatarisesis,Howquicklydoweneedtoscanaperson'snervoussystem?Itclearlycannotbedone instantaneously,andeven ifwedidprovideananobotforeachneuron, itwould taketimetogatherthedata.Onemightthereforeobjectthatbecauseaperson'sstateischangingduringthedata-gatheringprocess, theupload informationdoesnotaccurately reflect thatpersonat an instant intimebutratheroveraperiodoftime,evenifonlyafractionofasecond.120Consider,however,thatthis

issuewillnot interferewithanupload'spassinga "JaneSmith"Turing test.Whenweencounteroneanotheronaday-to-daybasis,weare recognizedasourselveseven though itmayhavebeendaysorweeks since the last such encounter. If anupload is sufficiently accurate to re-create aperson's statewithin theamountofnaturalchange thatapersonundergoes ina fractionofa secondorevena fewminutes, thatwill be sufficient for any conceivable purpose. Some observers have interpretedRogerPenrose'stheoryofthelinkbetweenquantumcomputingandconsciousness(seechapter9)tomeanthatuploadingis impossiblebecauseaperson's"quantumstate"willhavechangedmanytimesduringthescanningperiod.But Iwouldpointout thatmyquantumstatehaschangedmany times in the time ittookmetowritethissentence,andIstillconsidermyselftobethesameperson(andnooneseemstobeobjecting).NobelPrizewinnerGeraldEdelmanpointsoutthatthereisadifferencebetweenacapabilityandadescriptionofthatcapability.Aphotographofapersonisdifferentfromthepersonherself,evenifthe"photograph" isveryhigh resolutionand three-dimensional.However, theconceptofuploadinggoesbeyond the extremely high-resolution scan, which we can consider the "photograph" in Edelman'sanalogy.Thescandoesneedtocaptureallofthesalientdetails,butitalsoneedstobeinstantiatedintoaworking computational medium that has the capabilities of the original (albeit that the newnonbiologicalplatformsarecertaintobefarmorecapable).Theneuraldetailsneedtointeractwithoneanother (and with the outside world) in the same ways that they do in the original. A comparableanalogyisthecomparisonbetweenacomputerprogramthatresidesonacomputerdisk(astaticpicture)andaprogramthatisactivelyrunningonasuitablecomputer(adynamic,interactingentity).Boththedatacaptureandthereinstantiationofadynamicentityconstitutetheuploadingscenario.Perhapsthemostimportantquestionwillbewhetherornotanuploadedhumanbrainisreallyyou.Even if the upload passes a personalizedTuring test and is deemed indistinguishable fromyou, onecouldstillreasonablyaskwhethertheuploadisthesamepersonoranewperson.Afterall,theoriginalpersonmaystillexist.I'lldefertheseessentialquestionsuntilchapter7. In my view the most important element in uploading will be our gradual transfer of ourintelligence,personality,andskillstothenonbiologicalportionofourintelligence.Wealreadyhaveavariety of neural implants. In the 2020s wewill use nanobots to begin augmenting our brains withnonbiological intelligence, starting with the "routine" functions of sensory processing and memory,movingontoskillformation,patternrecognition,andlogicalanalysis.Bythe2030sthenonbiologicalportion of our intelligence will predominate, and by the 2040s, as I pointed out in chapter 3, thenonbiological portion will be billions of times more capable. Although we are likely to retain thebiological portion for a periodof time, itwill becomeof increasingly little consequence.Sowewillhaveeffectivelyuploadedourselves,albeitgradually,neverquitenoticingthetransfer.Therewillbeno"oldRay"and"newRay,"justanincreasinglycapableRay.AlthoughIbelievethatuploadingasinthesuddenscan-and-transferscenariodiscussedinthissectionwillbeafeatureofourfutureworld,itisthisgradual but inexorable progression to vastly superior nonbiological thinking that will profoundlytransformhumancivilization.SIGMUNDFREUD:Whenyoutalkaboutreverseengineeringthehumanbrain,justwhosebrainare

you talking about?Aman's brain?Awoman's?A child's?The brain of a genius?A retardedindividual?An"idiotsavant"?Agiftedartist?Aserialmurderer?

RAY:Ultimately,we'retalkingaboutalloftheabove.Therearebasicprinciplesofoperationthatweneed tounderstandabouthowhuman intelligenceand itsvariedconstituentskillswork.Giventhehumanbrain'splasticity,ourthoughtsliterallycreateourbrainsthroughthegrowthofnewspines,synapses,dendrites,andevenneurons.Asaresult,Einstein'sparietallobes—theregionassociated with visual imagery and mathematical thinking—became greatly enlarged.121However,thereisonlysomuchroominourskulls,soalthoughEinsteinplayedmusichewasnotaworld-classmusician.Picassodidnotwritegreatpoetry,andsoon.Aswere-createthehumanbrain,wewillnotbelimitedinourabilitytodevelopeachskill.Wewillnothavetocompromise'oneareatoenhanceanother.

Wecanalsogaininsightintoourdifferencesandanunderstandingofhumandysfunction.Whatwentwrongwiththeserialmurderer?Itmust,afterall,havesomethingtodowithhisbrain.Thistypeofdisastrousbehaviorisclearlynottheresultofindigestion.

MOLLY2004:Youknow,Idoubtit'sjustthebrainswe'rebornwiththataccountforourdifferences.Whataboutourstrugglesthroughlife,andallthisstuffI'mtryingtolearn?

RAY:Yes,well,that'spartoftheparadigm,too,isn'tit?Wehavebrainsthatcanlearn,startingfromwhenwelearntowalkandtalktowhenwestudycollegechemistry.

MARVINMINSKY: It's truethateducatingourAIswillbeanimportantpartof theprocess,butwecanautomatea lotof thatandgreatlyspeeditup.Also,keepinmindthatwhenoneAI learnssomething,itcanquicklysharethatknowledgewithmanyotherAIs.

RAY:They'llhaveaccesstoallofourexponentiallygrowingknowledgeontheWeb,whichwillincludehabitable,full-immersionvirtual-realityenvironmentswheretheycaninteractwithoneanotherandwithbiologicalhumanswhoareprojectingthemselvesintotheseenvironments.

SIGMUND:TheseAIsdon'thavebodiesyet.Aswehavebothpointedout,humanemotionandmuchofourthinkingaredirectedatourbodiesandtomeetingtheirsensualandsexualneeds.

RAY:Who says theywon't have bodies?As Iwill discuss in the human body version 2.0 section inchapter6,we'llhavethemeansofcreatingnonbiologicalyethumanlikebodies,aswellasvirtualbodiesinvirtualreality.

SIGMUND:Butavirtualbodyisnotarealbody.RAY:Theword"virtual"issomewhatunfortunate.It implies"notreal,"buttherealitywillbethata

virtual body is just as real as a physical body in all the ways that matter. Consider that thetelephoneisauditoryvirtualreality.Noonefeelsthathisvoiceinthisvirtual-realityenvironmentisnota"real"voice.Withmyphysicalbodytoday,Idon'tdirectlyexperiencesomeone'stouchonmyarm.Mybrainreceivesprocessedsignalsinitiatedbynerveendingsinmyarm,whichwindtheirway through the spinal cord, through thebrain stem,andup to the insula regions. Ifmybrain—or an AI's brain—receives comparable signals of someone's virtual touch on a virtualarm,there'snodiscernibledifference.

MARVIN:KeepinmindthatnotallAIswillneedhumanbodies.RAY:Indeed.Ashumans,despite someplasticity,bothourbodiesandbrainshavearelatively fixed

architecture.MOLLY2004:Yes,it'scalledbeinghuman,somethingyouseemtohaveaproblemwith.RAY:Actually,Ioftendohaveaproblemwithallthelimitationsandmaintenancethatmyversion1.0

bodyrequires,nottomentionallthelimitationsofmybrain.ButIdoappreciatethejoysofthehumanbody.MypointisthatAIscanandwillhavetheequivalentofhumanbodiesinbothrealandvirtual-realityenvironments.AsMarvinpointsout,however,theywillnotbelimitedjusttothis.

MOLLY2104:It'snotjustAIsthatwillbeliberatedfromthelimitationsofversion1.abodies.Humansofbiologicaloriginwillhavethesamefreedominbothrealandvirtualreality.

GEORGE2048:Keepinmind,therewon'tbeacleardistinctionbetweenAIsandhumans.MOLLY2104:Yes,exceptfortheMOSHs(MostlyOriginalSubstrateHumans)ofcourse.

T

CHAPTERFIVE

GNRThreeOverlappingRevolutions

Therearefewthingsofwhichthepresentgenerationismorejustlyproudthanthewonderfulimprovementswhich are daily taking place in all sorts ofmechanical appliances....Butwhatwould happen if technology continued to evolve somuchmore rapidly than the animal andvegetable kingdoms?Would it displace us in the supremacy of earth? Just as the vegetablekingdomwasslowlydevelopedfromthemineral,andasinlikemannertheanimalsuperveneduponthevegetable,sonowintheselastfewagesanentirelynewkingdomhassprungup,ofwhichweasyethaveonlyseenwhatwillonedaybeconsideredtheantediluvianprototypesoftherace....Wearedailygiving[machines]greaterpowerandsupplyingbyallsortsofingeniouscontrivances that self-regulating, self-acting powerwhichwill be to themwhat intellect hasbeentothehumanrace.

—SAMUELBUTLER, 1863 (FOURYEARSAFTER PUBLICATIONOFDARWIN'STHEORIGINOFSPECIES

Whowill beman's successor? Towhich the answer is:We are ourselves creating our ownsuccessors. Man will become to the machine what the horse and the dog are to man; theconclusionbeingthatmachinesare,orarebecoming,animate.

—SAMUEL BUTLER, 1863 LETTER, "DARWIN AMONG THEMACHINES"1

he first half of the twenty-first centurywill be characterized by three overlapping revolutions—inGenetics,Nanotechnology,andRobotics.ThesewillusherinwhatIreferredtoearlierasEpochFive,the beginning of the Singularity. We are in the early stages of the "G" revolution today. Byunderstanding the information processes underlying life,we are starting to learn to reprogramour

biologytoachievethevirtualeliminationofdisease,dramaticexpansionofhumanpotential,andradicallife extension.HansMoravec points out, however, that nomatter how successfullywe fine-tune ourDNA-basedbiology,humanswillremain"second-classrobots,"meaningthatbiologywillneverbeabletomatchwhatwewillbeabletoengineeroncewefullyunderstandbiology'sprinciplesofoperation.2The"N"revolutionwillenableustoredesignandrebuild—moleculebymolecule—ourbodiesandbrains and theworldwithwhichwe interact, going far beyond the limitations of biology. Themostpowerfulimpendingrevolutionis"R":human-levelrobotswiththeirintelligencederivedfromourownbut redesigned to far exceed human capabilities. R represents the most significant transformation,becauseintelligenceisthemostpowerful"force"intheuniverse.Intelligence,ifsufficientlyadvanced,is,well,smartenoughtoanticipateandovercomeanyobstaclesthatstandinitspath.Whileeachrevolutionwillsolvetheproblemsfromearliertransformations,itwillalsointroducenewperils.Gwillovercometheage-olddifficultiesofdiseaseandagingbutestablishthepotentialfornew bioengineered viral threats. Once N is fully developed we will be able to apply it to protectourselves from all biological hazards, but it will create the possibility of its own self-replicating

dangers,whichwillbefarmorepowerfulthananythingbiological.WecanprotectourselvesfromthesehazardswithfullydevelopedR,butwhatwillprotectusfrompathologicalintelligencethatexceedsourown?Idohaveastrategyfordealingwiththeseissues,whichIdiscussattheendofchapter8.Inthischapter, however, wewill examine how the Singularity will unfold through these three overlappingrevolutions:G,N,andR.

Genetics:TheIntersectionofInformationandBiology

Ithasnotescapedournoticethatthespecificpairingwehavepostulatedimmediatelysuggestsapossiblecopyingmechanismforthegeneticmaterial.

—JAMESWATSONANDFRANCISCRICK3

Afterthreebillionyearsofevolution,wehavebeforeustheinstructionsetthatcarrieseachofusfromtheone-celleggthroughadulthoodtothegrave.

—DR. ROBERT WATERSON, INTERNATIONAL HUMAN GENOMESEQUENCECONSORTIUM4

Underlying all of the wonders of life and misery of disease are information processes, essentiallysoftwareprograms,thataresurprisinglycompact.Theentirehumangenomeisasequentialbinarycodecontaining only about eight hundred million bytes of information. As I mentioned earlier, when itsmassive redundanciesare removedusingconventional compression techniques,weare leftwithonlythirty to one hundred million bytes, equivalent to the size of an average contemporary softwareprogram.5 This code is supported by a set of biochemicalmachines that translate these linear (one-dimensional) sequences of DNA "letters" into strings of simple building blocks called amino acids,which are in turn folded into three-dimensional proteins, which make up all living creatures frombacteria to humans. (Viruses occupy a niche in between living and nonliving matter but are alsocomposed of fragments ofDNAorRNA.)Thismachinery is essentially a self-replicating nanoscalereplicator that builds the elaborate hierarchy of structures and increasingly complex systems that alivingcreaturecomprises.

Life'sComputer

Intheveryearlystagesofevolutioninformationwasencodedinthestructuresofincreasinglycomplexorganicmoleculesbasedoncarbon.AfterbillionsofyearsbiologyevolveditsowncomputerforstoringandmanipulatingdigitaldatabasedontheDNAmolecule.ThechemicalstructureoftheDNAmoleculewasfirstdescribedbyJ.D.WatsonandF.H.C.Crickin1953asadoublehelixconsistingofapairofstrandsofpolynucleotides.6We finished transcribing the genetic code at the beginning of this century.We are now beginning to understand thedetailedchemistryofthecommunicationandcontrolprocessesbywhichDNAcommandsreproductionthroughsuchothercomplexmoleculesandcellularstructuresasmessengerRNA(mRNA),transferRNA(tRNA),andribosomes.Atthelevelofinformationstoragethemechanismissurprisinglysimple.Supportedbyatwistingsugar-phosphatebackbone,theDNAmoleculecontainsup toseveralmillion rungs, eachofwhich is codedwithone letter drawn froma four-letteralphabet; each rung is thuscoding twobitsofdata inaone-dimensionaldigitalcode.Thealphabetconsistsof the fourbasepairs:adenine-thymine, thymine-adenine,cytosine-guanine,andguanine-cytosine. Specialenzymescancopythe informationoneachrungbysplittingeachbasepairandassemblingtwoidenticalDNAmoleculesbyrematchingthebrokenbasepairs.Otherenzymesactuallycheckthevalidityofthecopybycheckingtheintegrityofthebase-pairmatching.With these copying and validation steps, this chemical data-processing system makes only about one error in ten billion base-paircombinations.7Furtherredundancyanderror-correctioncodesarebuiltintothedigitaldataitself,someaningfulmutationsresultingfrombase-pairreplicationerrorsarerare.Mostoftheerrorsresultingfromtheone-in-ten-billionerrorratewillresultsintheequivalentofa"parity"error,whichcanbedetectedandcorrectedbyother levelsof thesystem, includingmatchingagainst thecorrespondingchromosome,whichcanpreventtheincorrectbitfromcausinganysignificantdamage.8RecentresearchhasshownthatthegeneticmechanismdetectssucherrorsintranscriptionofthemaleYchromosomebymatchingeachYchromosomegeneagainstacopyonthesamechromosome.9Onceinalongwhileatranscriptionerrorwillresultinabeneficialchangethatevolutionwillcometofavor.Inaprocesstechnicallycalledtranslation,anotherseriesofchemicalsputthiselaboratedigitalprogramintoactionbybuildingproteins.Itis the protein chains that give each cell its structure, behavior, and intelligence. Special enzymes unwind a region of DNA for building aparticularprotein.AstrandofmRNAiscreatedbycopyingtheexposedsequenceofbases.ThemRNAessentiallyhasacopyofaportionoftheDNAlettersequence.ThemRNAtravelsoutofthenucleusandintothecellbody.ThemRNAcodearethenreadbyaribosomemolecule,whichrepresentsthecentralmolecularplayerinthedramaofbiologicalreproduction.Oneportionof theribosomeacts likeatape-recorderhead, "reading" thesequenceofdataencoded in themRNAbasesequence.The "letters" (bases)aregrouped intowordsof three letterscalledcodons,withonecodonforeachofthetwentypossibleaminoacids,thebasicbuildingblocksofprotein.AribosomereadsthecodonsfromthemRNAandthen,usingtRNA,assemblesaproteinchainoneaminoacidatatime.

Thenotablefinalstepinthisprocessisthefoldingoftheone-dimensionalchainofaminoacid"beads"intoathree-dimensional protein. Simulating this process has not yet been feasible because of the enormous complexity of theinteracting forces from all the atoms involved. Supercomputers scheduled to come online around the time of thepublicationofthisbook(2005)areexpectedtohavethecomputationalcapacitytosimulateproteinfolding,aswellastheinteractionofonethree-dimensionalproteinwithanother.Proteinfolding,alongwithcelldivision,isoneofnature'sremarkableandintricatedancesinthecreationandre-creation of life. Specialized "chaperone" molecules protect and guide the amine-acid strands as they assume theirprecise three-dimensional protein configurations. As many as one third of formed protein molecules are foldedimproperly.Thesedisfiguredproteinsmustimmediatelybedestroyedortheywillrapidlyaccumulate,disruptingcellularfunctionsonmanylevels.Undernormalcircumstances,assoonasamisfoldedproteinisformed,itistaggedbyacarriermolecule,ubiquitin,andescortedtoaspecializedproteosome,whereit isbrokenbackdowninto itscomponentaminoacidsforrecyclingintonew(correctlyfolded)proteins.Ascellsage,however,theyproducelessoftheenergyneededforoptimalfunctionof this mechanism. Accumulation of these misformed proteins aggregate into particles called protofibrils, which arethoughtounderliediseaseprocessesleadingtoAlzheimer'sdiseaseandotherafflictions.10Theabilitytosimulatethethree-dimensionalwaltzofatomic-levelinteractionswillgreatlyaccelerateourknowledgeofhowDNAsequencescontrollifeanddisease.Wewillthenbeinapositiontorapidlysimulatedrugsthatinterveneinanyof the steps in this process, therebyhasteningdrugdevelopment and the creation of highly targeted drugs thatminimizeunwantedsideeffects. It is the jobof theassembledproteins tocarryout the functionsof thecell,andbyextension theorganism.Amoleculeofhemoglobin,forexample,whichhasthejobofcarryingoxygenfromthelungstobodytissues, iscreatedfivehundredtrilliontimeseachsecondinthehumanbody.Withmorethanfivehundredaminoacidsineachmoleculeofhemoglobin,thatcomesto1.5Î1019(fifteenbillionbillion)"read"operationseveryminutebytheribosomesjustforthemanufactureofhemoglobin.Insomewaysthebiochemicalmechanismoflifeisremarkablycomplexandintricate.Inotherwaysitisremarkablysimple.Only fourbasepairsprovide thedigitalstorageforallof thecomplexityofhuman lifeandallother lifeasweknowit.Theribosomesbuildproteinchainsbygrouping together tripletsofbasepairs toselectsequencesfromonlytwentyaminoacids.Theamineacidsthemselvesarerelativelysimple,consistingofacarbonatomwithitsfourbondslinkedtoonehydrogenatom,oneamino(-NH2)group,onecarboxylicacid(-COOH)group,andoneorganicgroupthatisdifferent foreachaminoacid.Theorganicgroupforalanine, forexample,hasonlyfouratoms(CH3-) fora totalof

thirteenatoms.Oneofthemorecomplexaminoacids,arginine(whichplaysavitalroleinthehealthoftheendothelialcellsinourarteries)hasonlyseventeenatomsinitsorganicgroupforatotaloftwenty-sixatoms.Thesetwentysimplemolecularfragmentsarethebuildingblocksofallife.Theproteinchainsthencontroleverythingelse:thestructureofbonecells,theabilityofmusclecellstoflexandactinconcertwithothermusclecells,allofthecomplexbiochemicalinteractionsthattakeplaceinthebloodstream,and,ofcourse,thestructureandfunctioningofthebrain.11DesignerBabyBoomersSufficientinformationalreadyexiststodaytoslowdowndiseaseandagingprocessestothepointthatbaby boomers likemyself can remain in good health until the full blossoming of the biotechnologyrevolution,whichwill itselfbeabridgetothenanotechnologyrevolution(seeResourcesandContactInformation,p.489).InFantasticVoyage:LiveLongEnoughtoLiveForever,whichIcoauthoredwithTerry Grossman, M.D., a leading longevity expert, we discuss these three bridges to radical lifeextension(today'sknowledge,biotechnology,andnanotechnology).12Iwrotethere:"Whereassomeofmycontemporariesmaybesatisfiedtoembraceaginggracefullyaspartofthecycleoflife,thatisnotmy view. It may be 'natural,' but I don't see anything positive in losingmymental agility, sensoryacuity,physicallimberness,sexualdesire,oranyotherhumanability.Iviewdiseaseanddeathatanyageasacalamity,asproblemstobeovercome." Bridgeoneinvolvesaggressivelyapplyingtheknowledgewenowpossesstodramaticallyslowdown aging and reverse themost important disease processes, such as heart disease, cancer, type 2diabetes, and stroke. You can, in effect, reprogram your biochemistry, for we have the knowledgetoday, if aggressively applied, to overcome our genetic heritage in the vast majority of cases. "It'smostlyinyourgenes"isonlytrueifyoutaketheusualpassiveattitudetowardhealthandaging.Myownstoryisinstructive.MorethantwentyyearsagoIwasdiagnosedwithtype2diabetes.Theconventional treatment made my condition worse, so I approached this health challenge from myperspective as an inventor. I immersedmyself in the scientific literature and came upwith a uniqueprogramthatsuccessfullyreversedmydiabetes.In1993Iwroteahealthbook(The10%SolutionforaHealthyLife)aboutthisexperience,andIcontinuetodaytobefreeofanyindicationorcomplicationofthisdisease.13Inaddition,whenIwastwenty-two,myfatherdiedofheartdiseaseattheageoffifty-eight,andIhaveinheritedhisgenespredisposingmetothisillness.Twentyyearsago,despitefollowingthepublicguidelinesoftheAmericanHeartAssociation,mycholesterolwasinthehigh200s(itshouldbewellbelow180),myHDL(high-density lipoprotein, the"good"cholesterol)below30(itshouldbeabove50),andmyhomocysteine(ameasureofthehealthofabiochemicalprocesscalledmethylation)wasanunhealthy 11 (it should be below 7.5). By following a longevity program that Grossman and Ideveloped,mycurrentcholesterollevelis130,myHDLis55,myhomocysteineis6.2,myC-reactiveprotein(ameasureofinflammationinthebody)isaveryhealthy0.01,andallofmyotherindexes(forheartdisease,diabetes,andotherconditions)areatideallevels.14 When Iwas forty,mybiological agewas around thirty-eight.Although I amnow fifty-six, acomprehensive test of my biological aging (measuring various sensory sensitivities, lung capacity,reaction times, memory, and related tests) conducted at Grossman's longevity clinic measured mybiologicalageatforty.15Althoughthereisnotyetaconsensusonhowtomeasurebiologicalage,myscoresonthesetestsmatchedpopulationnormsforthisage.So,accordingtothissetoftests,Ihavenotagedverymuchinthelastsixteenyears,whichisconfirmedbythemanybloodtestsItake,aswellasthewayIfeel. These results are not accidental; I have been very aggressive about reprogramming mybiochemistry.Itake250supplements(pills)adayandreceiveahalf-dozenintravenoustherapieseachweek(basicallynutritionalsupplementsdelivereddirectlyintomybloodstream,therebybypassingmyGI tract). As a result, themetabolic reactions inmy body are completely different than theywould

otherwise be.16 Approaching this as an engineer, I measure dozens of levels of nutrients (such asvitamins,minerals,andfats),hormones,andmetabolicby-productsinmybloodandotherbodysamples(suchashairandsaliva).Overall,mylevelsarewhereIwantthemtobe,althoughIcontinuallyfine-tunemyprogrambasedon the research that I conductwithGrossman.17Althoughmyprogrammayseemextreme, it is actually conservative—andoptimal (based onmy current knowledge).Grossmanand I have extensively researched each of the several hundred therapies that I use for safety andefficacy. I stay away from ideas that are unproven or appear to be risky (the use of human-growthhormone,forexample).Weconsidertheprocessofreversingandovercomingthedangerousprogressionofdiseaseasawar.Asinanywarit isimportanttomobilizeallthemeansofintelligenceandweaponrythatcanbeharnessed,throwingeverythingwehaveattheenemy.Forthisreasonweadvocatethatkeydangers—suchasheartdisease,cancer,diabetes,stroke,andaging—beattackedonmultiplefronts.Forexample,ourstrategyforpreventingheartdiseaseistoadopttendifferentheart-disease-preventiontherapiesthatattackeachoftheknownriskfactors.Byadoptingsuchmultiprongedstrategiesforeachdiseaseprocessandeachagingprocess,evenbaby boomers likemyself can remain in good health until the full blossoming of the biotechnologyrevolution(whichwecall"bridgetwo"),whichisalreadyinitsearlystagesandwillreachitspeakintheseconddecadeofthiscentury.Biotechnologywillprovidethemeanstoactuallychangeyourgenes:notjustdesignerbabieswillbefeasiblebutdesignerbabyboomers.We'llalsobeable torejuvenateallofyourbody's tissuesandorgans by transformingyour skin cells into youthful versions of every other cell type.Already, newdrugdevelopment isprecisely targetingkeysteps in theprocessofatherosclerosis(thecauseofheartdisease), cancerous tumor formation, and themetabolic processes underlying eachmajor disease andagingprocess.CanWeReallyLiveForever?AnenergeticandinsightfuladvocateofstoppingtheagingprocessbychangingtheinformationprocessesunderlyingbiologyisAubreydeGrey,ascientistinthedepartmentofgeneticsatCambridgeUniversity.DeGreyusesthemetaphorofmaintainingahouse.Howlongdoesahouselast?Theanswerobviouslydependsonhowwellyoutakecareofit.Ifyoudonothing,theroofwill spring a leak before long, water and the elements will invade, and eventually the house willdisintegrate.But ifyouproactively takecareof thestructure, repairalldamage,confrontalldangers,and rebuildor renovateparts from time to timeusingnewmaterials and technologies, the lifeof thehousecanessentiallybeextendedwithoutlimit. The same holds true for our bodies and brains. The only difference is that, while we fullyunderstandthemethodsunderlyingthemaintenanceofahouse,wedonotyetfullyunderstandallofthebiological principles of life. But with our rapidly increasing comprehension of the biochemicalprocesses and pathways of biology, we are quickly gaining that knowledge. We are beginning tounderstandaging,notasasingleinexorableprogressionbutasagroupofrelatedprocesses.Strategiesare emerging for fully reversing each of these aging progressions, using different combinations ofbiotechnologytechniques.DeGreydescribeshisgoalas"engineerednegligiblesenescence"—stoppingthebodyandbrainfrom becoming more frail and disease-prone as it grows older.18 As he explains, "All the coreknowledgeneededtodevelopengineerednegligiblesenescenceisalreadyinourpossession—itmainlyjustneedstobepiecedtogether."19DeGreybelieveswe'lldemonstrate"robustlyrejuvenated"mice—mice thatarefunctionallyyounger thanbeforebeing treatedandwith the lifeextension toprove it—withintenyears,andhepointsoutthatthisachievementwillhaveadramaticeffectonpublicopinion.Demonstratingthatwecanreversetheagingprocessinananimalthatshares99percentofourgeneswill profoundly challenge the common wisdom that aging and death are inevitable. Once robustrejuvenationisconfirmedinananimal,therewillbeenormouscompetitivepressuretotranslatetheseresultsintohumantherapies,whichshouldappearfivetotenyearslater.Thediversefieldofbiotechnologyisfueledbyouracceleratingprogressinreverseengineeringthe

information processes underlying biology and by a growing arsenal of tools that can modify theseprocesses.For example, drugdiscoverywasonce amatter of finding substances that produced somebeneficial result without excessive side effects. This process was similar to early humans' tooldiscovery,whichwaslimitedtosimplyfindingrocksandothernaturalimplementsthatcouldbeusedforhelpfulpurposes.Todaywearelearningtheprecisebiochemicalpathwaysthatunderliebothdiseaseandagingprocessesandareabletodesigndrugstocarryoutprecisemissionsat themolecularlevel.Thescopeandscaleoftheseeffortsarevast. Anotherpowerfulapproach is to startwithbiology's informationbackbone: thegenome.Withrecentlydevelopedgene technologieswe're on thevergeof being able to control howgenes expressthemselves.Geneexpression is theprocessbywhich specific cellular components (specificallyRNAandtheribosomes)produceproteinsaccordingtoaspecificgeneticblueprint.Whileeveryhumancellhasthefullcomplementofthebody'sgenes,aspecificcell,suchasaskincellorapancreaticisletcell,getsitscharacteristicsfromonlythesmallfractionofgeneticinformationrelevanttothatparticularcelltype.20Thetherapeuticcontrolofthisprocesscantakeplaceoutsidethecellnucleus,soitiseasiertoimplementthantherapiesthatrequireaccessinsideit.Geneexpressioniscontrolledbypeptides(moleculesmadeupofsequencesofuptoonehundredaminoacids)andshortRNAstrands.Wearenowbeginningtolearnhowtheseprocesseswork.21Manynew therapiesnow indevelopment and testing arebasedonmanipulating themeither to turnoff theexpressionofdisease-causinggenesortoturnondesirablegenesthatmayotherwisenotbeexpressedinaparticulartypeofcell.RNAi(RNAInterference).ApowerfulnewtoolcalledRNAinterference(RNAi)iscapableofturningoff specificgenesbyblocking theirmRNA, thuspreventing them fromcreatingproteins.Sinceviraldiseases,cancer,andmanyotherdiseasesusegeneexpressionatsomecrucialpointintheirlifecycle,this promises to be a breakthrough technology. Researchers construct short, double-stranded DNAsegmentsthatmatchandlockontoportionsoftheRNAthataretranscribedfromatargetedgene.Withtheirabilitytocreateproteinsblocked,thegeneiseffectivelysilenced.Inmanygeneticdiseasesonlyonecopyof agivengene isdefective.Sinceweget twocopiesof eachgene,one fromeachparent,blockingthedisease-causinggeneleavesonehealthygenetomakethenecessaryprotein.Ifbothgenesaredefective,RNAicouldsilencethemboth,butthenahealthygenewouldhavetobeinserted.22CellTherapies.Another important lineofattack is to regrowourowncells, tissues,andevenwholeorgans and introduce them into our bodies without surgery. One major benefit of this "therapeuticcloning"techniqueisthatwewillbeabletocreatethesenewtissuesandorgansfromversionsofourcellsthathavealsobeenmadeyoungerviatheemergingfieldofrejuvenationmedicine.Forexample,wewillbeabletocreatenewheartcellsfromskincellsandintroducethemintothesystemthroughthebloodstream.Overtime,existingheartcellswillbereplacedwiththesenewcells,andtheresultwillbea rejuvenated "young" heart manufactured using a person's own DNA. I discuss this approach toregrowingourbodiesbelow.GeneChips.Newtherapiesareonlyonewaythatthegrowingknowledgebaseofgeneexpressionwilldramaticallyimpactourhealth.Sincethe1990smicroarrays,orchipsnolargerthanadime,havebeenused to study and compare expression patterns of thousands of genes at a time.23 The possibleapplicationsofthetechnologyaresovariedandthetechnologicalbarriershavebeenreducedsogreatlythathugedatabasesarenowdevotedtotheresultsfrom"do-it-yourselfgenewatching."24Geneticprofilingisnowbeingusedto:

·Revolutionizetheprocessesofdrugscreeninganddiscovery.Microarrayscan"notonlyconfirmthemechanismof action of a compound" but "discriminate between compounds acting at differentstepsinthesamemetabolicpathway."25

·Improve cancer classifications. One study reported in Science demonstrated the feasibility ofclassifyingsomeleukemias"solelyongeneexpressionmonitoring."Theauthorsalsopointedtoacaseinwhichexpressionprofilingresultedinthecorrectionofamisdiagnosis.26

·Identifythegenes,cells,andpathwaysinvolvedinaprocess,suchasagingortumorigenesis.Forexample,bycorrelatingthepresenceofacutemyeloblasticleukemiaandincreasedexpressionofcertain genes involved with programmed cell death, a study helped identify new therapeutictargets.27

·Determinetheeffectivenessofaninnovativetherapy.OnestudyrecentlyreportedinBonelookedattheeffectofgrowth-hormonereplacementontheexpressionofinsulinlikegrowthfactors(IGFs)andbonemetabolismmarkers.28

·Test the toxicity of compounds in food additives, cosmetics, and industrial products quickly andwithoutusinganimals.Suchtestscanshow,forexample,thedegreetowhicheachgenehasbeenturnedonoroffbyatestedsubstance.29

Somatic Gene Therapy (gene therapy for nonreproductive cells). This is the holy grail ofbioengineering,whichwill enable us to effectively changegenes inside the nucleus by "infecting" itwith newDNA, essentially creating new genes.30 The concept of controlling the geneticmakeup ofhumans is often associated with the idea of influencing new generations in the form of "designerbabies."But the real promise of gene therapy is to actually change our adult genes.31 These can bedesignedtoeitherblockundesirabledisease-encouraginggenesorintroducenewonesthatslowdownandevenreverseagingprocesses.Animalstudiesthatbeganinthe1970sand1980shavebeenresponsibleforproducingarangeoftransgenicanimals,suchascattle,chickens,rabbits,andseaurchins.Thefirstattemptsathumangenetherapywereundertakenin1990.ThechallengeistotransfertherapeuticDNAintotargetcellsthatwillthenbeexpressedattherightlevelandattherighttime. Consider the challenge involved in effecting a gene transfer.Viruses are often the vehicle ofchoice.Longagoviruseslearnedhowtodelivertheirgeneticmaterialtohumancellsand,asaresult,causedisease.Researchersnowsimplyswitch thematerialavirusunloads intocellsby removing itsgenesandinsertingtherapeuticones.Althoughtheapproachitselfisrelativelyeasy,thegenesaretoolargetopassintomanytypesofcells(suchasbraincells).TheprocessisalsolimitedinthelengthofDNAitcancarry,anditmaycauseanimmuneresponse.AndpreciselywherethenewDNAintegratesintothecell'sDNAhasbeenalargelyuncontrollableprocess.32 Physical injection (microinjection) ofDNA into cells is possible but prohibitively expensive.Exciting advances have recently been made, however, in other means of transfer. For example,liposomes—fatty sphereswith awatery core—can be used as a "molecular Trojan horse" to delivergenes to brain cells, thereby opening the door to treatment of disorders such as Parkinson's andepilepsy.33 Electric pulses can also be employed to deliver a range of molecules (including drugproteins,RNA,andDNA)tocells.34YetanotheroptionistopackDNAintoultratiny"nanoballs"formaximumimpact.35 Themajor hurdle thatmust be overcome for gene therapy to be applied in humans is properpositioningofageneonaDNAstrandandmonitoringofthegene'sexpression.Onepossiblesolutionistodeliveranimagingreportergenealongwiththetherapeuticgene.Theimagesignalswouldallowforclosesupervisionofbothplacementandlevelofexpression.36Evenfacedwiththeseobstaclesgenetherapyisstartingtoworkinhumanapplications.Ateamledby University of Glasgow research doctor Andrew H. Baker has successfully used adenoviruses to"infect" specificorgansandeven specific regionswithinorgans.Forexample, thegroupwasable todirectgene therapyprecisely at the endothelial cells,which line the insideofbloodvessels.AnotherapproachisbeingdevelopedbyCeleraGenomics,acompanyfoundedbyCraigVenter(theheadofthe

private effort to transcribe thehumangenome).Celerahas alreadydemonstrated theability to createsynthetic viruses from genetic information and plans to apply these biodesigned viruses to genetherapy.37OneofthecompaniesIhelptodirect,UnitedTherapeutics,hasbegunhumantrialsofdeliveringDNAintocells through thenovelmechanismofautologous (thepatient'sown)stemcells,whicharecapturedfromafewvialsoftheirblood.DNAthatdirectsthegrowthofnewpulmonarybloodvesselsis inserted into thestemcellgenes,and thecellsare reinjected into thepatient.When thegeneticallyengineered stem cells reach the tiny pulmonary blood vessels near the lung's alveoli, they begin toexpress growth factors for new blood vessels. In animal studies this has safely reversed pulmonaryhypertension,afatalandpresentlyincurabledisease.Basedonthesuccessandsafetyofthesestudies,theCanadiangovernmentgavepermissionforhumanteststocommenceinearly2005.ReversingDegenerativeDiseaseDegenerative (progressive)diseases—heartdisease, stroke, cancer, type2diabetes, liverdisease, andkidneydisease—account for about 90 percent of the deaths in our society.Our understandingof theprincipalcomponentsofdegenerativediseaseandhumanagingisgrowingrapidly,andstrategieshavebeenidentifiedtohaltandevenreverseeachoftheseprocesses.InFantasticVoyage,GrossmanandIdescribeawiderangeoftherapiesnowinthetestingpipelinethathavealreadydemonstratedsignificantresultsinattackingthekeybiochemicalstepsunderlyingtheprogressofsuchdiseases.CombatingHeartDisease.As one ofmany examples, exciting research is being conducted with asyntheticformofHDLcholesterolcalledrecombinantApo-A-IMilano(AAIM).InanimaltrialsAAIMwasresponsibleforarapidanddramaticregressionofatheroscleroticplaque.38Inaphase1FDAtrial,whichincludedforty-sevenhumansubjects,administeringAAIMbyintravenousinfusionresultedinasignificantreduction(anaverage4.2percentdecrease)inplaqueafterjustfiveweeklytreatments.Nootherdrughasevershowntheabilitytoreduceatherosclerosisthisquickly.39 Another exciting drug for reversing atherosclerosis now in phase 3 FDA trials is Pfizer'sTorcetrapib.40Thisdrugboosts levelsofHDLbyblockinganenzyme thatnormallybreaks itdown.Pfizer is spending a recordonebilliondollars to test thedrug andplans to combine itwith its best-selling"statin"(cholesterol-lowering)drug,Lipitor.OvercomingCancer.Many strategies are being intensely pursued to overcome cancer. Particularlypromisingarecancervaccinesdesigned tostimulate the immunesystemtoattackcancercells.Thesevaccinescouldbeusedasaprophylaxistopreventcancer,asafirst-linetreatment,ortomopupcancercellsafterothertreatments.41Thefirstreportedattemptstoactivateapatient'simmuneresponsewereundertakenmorethanonehundredyearsago,with littlesuccess.42Morerecenteffortsfocusonencouragingdendriticcells, thesentinelsoftheimmunesystem,totriggeranormalimmuneresponse.Manyformsofcancerhaveanopportunitytoproliferatebecausetheysomehowdonottriggerthatresponse.Dendriticcellsplayakeyrolebecausetheyroamthebody,collectingforeignpeptidesandcellfragmentsanddeliveringthemtothe lymph nodes, which in response produce an army of T cells primed to eliminate the flaggedpeptides. Someresearchersarealteringcancer-cellgenes toattractTcells,with theassumption that thestimulatedTcellswouldthenrecognizeothercancercellstheyencounter.43Othersareexperimentingwith vaccines for exposing the dendritic cells to antigens, unique proteins found on the surfaces ofcancer cells. One group used electrical pulses to fuse tumor and immune cells to create an"individualizedvaccine."44Oneof theobstacles todevelopingeffectivevaccines is that currentlywehavenotyetidentifiedmanyofthecancerantigensweneedtodeveloppotenttargetedvaccines.45

Blockingangiogenesis—thecreationofnewbloodvessels—isanotherstrategy.Thisprocessusesdrugstodiscourageblood-vesseldevelopment,whichanemergentcancerneedstogrowbeyondasmallsize. Interest in angiogenesis has skyrocketed since 1997, when doctors at the Dana Farber CancerCenterinBostonreportedthatrepeatedcyclesofendostatin,anangiogenesisinhibitor,hadresultedincompleteregressionoftumors.46Therearenowmanyantiangiogenicdrugsinclinicaltrials,includingavastinandatrasentan.47 Akey issue forcanceraswellas foragingconcerns telomere"beads," repeatingsequencesofDNAfoundattheendofchromosomes.Eachtimeacellreproduces,onebeaddropsoff.Onceacellhasreproducedtothepointthatallofitstelomerebeadshavebeenexpended,thatcellisnolongerableto divide andwill die. Ifwe could reverse this process, cells could survive indefinitely. Fortunately,recentresearchhasfoundthatonlyasingleenzyme(telomerase)isneededtoachievethis.48Thetrickypartistoadministertelomeraseinsuchawayasnottocausecancer.Cancercellspossessagenethatproducestelomerase,whicheffectivelyenablesthemtobecomeimmortalbyreproducingindefinitely.A key cancer-fighting strategy, therefore, involves blocking the ability of cancer cells to generatetelomerase.Thismayseemtocontradicttheideaofextendingthetelomeresinnormalcellstocombatthis sourceof aging, but attacking the telomerase of the cancer cells in an emerging tumor couldbedone without necessarily compromising an orderly telomere-extending therapy for normal cells.However,toavoidcomplications,suchtherapiescouldbehaltedduringaperiodofcancertherapy.ReversingAgingItislogicaltoassumethatearlyintheevolutionofourspecies(andprecursorstoourspecies)survivalwouldnothavebeenaided—indeed,itwouldhavebeencompromised—byindividualslivinglongpasttheirchild-rearingyears.Recentresearch,however,supportstheso-calledgrandmahypothesis,whichsuggests a countereffect. University of Michigan anthropologist Rachel Caspari and University ofCaliforniaatRiverside'sSan-HeeLeefoundevidencethattheproportionofhumanslivingtobecomegrandparents(whoinprimitivesocietieswereoftenasyoungasthirty)increasedsteadilyoverthepasttwo million years, with a fivefold increase occurring in the Upper Paleolithic era (around thirtythousandyearsago).Thisresearchhasbeencitedtosupportthehypothesisthatthesurvivalofhumansocietieswasaidedbygrandmothers,whonotonlyassistedinraisingextendedfamiliesbutalsopassedontheaccumulatedwisdomofelders.Sucheffectsmaybeareasonableinterpretationofthedata,butthe overall increase in longevity also reflects an ongoing trend toward longer life expectancy thatcontinuestothisday.Likewise,onlyamodestnumberofgrandmas(andafewgrandpas)wouldhavebeen needed to account for the societal effects that proponents of this theory have claimed, so thehypothesis does not appreciably challenge the conclusion that genes that supported significant lifeextensionwerenotselectedfor.Agingisnotasingleprocessbutinvolvesamultiplicityofchanges.DeGreydescribessevenkeyagingprocessesthatencouragesenescence,andhehasidentifiedstrategiesforreversingeachone.DNAMutations.49GenerallymutationstonuclearDNA(theDNAinthechromosomesinthenucleus)resultinadefectivecellthat'squicklyeliminatedoracellthatsimplydoesn'tfunctionoptimally.Thetype ofmutation that is of primary concern (as it leads to increased death rates) is one that affectsorderly cellular reproduction, resulting in cancer. This means that if we can cure cancer using thestrategiesdescribedabove,nuclearmutationsshouldlargelyberenderedharmless.DeGrey'sproposedstrategyforcancerispreemptive:itinvolvesusinggenetherapytoremovefromallourcellsthegenesthatcancersneedtoturnoninordertomaintaintheirtelomereswhentheydivide.Thiswillcauseanypotential cancer tumors towither away before they grow large enough to cause harm.Strategies fordeletingandsuppressinggenesarealreadyavailableandarebeingrapidlyimproved.ToxicCells.Occasionallycellsreachastateinwhichthey'renotcancerous,butitwouldstillbebestforthebodyiftheydidnotsurvive.Cellsenescenceisanexample,asishavingtoomanyfatcells.Inthese

cases,itiseasiertokillthesecellsthantoattempttorevertthemtoahealthystate.Methodsarebeingdeveloped to target "suicidegenes" to such cells and also to tag these cells in away that directs theimmunesystemtodestroythem.MitochrondrialMutations. Another aging process is the accumulation ofmutations in the thirteengenes in the mitochondria, the energy factories for the cell.50 These few genes are critical to theefficientfunctioningofourcellsandundergomutationatahigherratethangenesinthenucleus.Oncewemastersomaticgenetherapy,wecouldputmultiplecopiesofthesegenesinthecellnucleus,therebyprovidingredundancy(backup)forsuchvitalgeneticinformation.Themechanismalreadyexistsinthecelltoallownucleus-encodedproteinstobeimportedintothemitochondria,soitisnotnecessaryfortheseproteinstobeproducedinthemitochondriathemselves.Infact,mostoftheproteinsneededformitochondrial function are already coded by the nuclear DNA. Researchers have already beensuccessfulintransferringmitochondrialgenesintothenucleusincellcultures.Intracellular Aggregates. Toxins are produced both inside and outside cells. De Grey describesstrategies using somatic gene therapy to introduce new genes that will break down what he calls"intracellularaggregates"—toxinswithincells.Proteinshavebeenidentifiedthatcandestroyvirtuallyanytoxin,usingbacteriathatcandigestanddestroydangerousmaterialsrangingfromTNTtodioxin. Akeystrategybeingpursuedbyvariousgroupsforcombatingtoxicmaterialsoutside thecell,includingmisformedproteinsandamyloidplaque(seeninAlzheimer'sdiseaseandotherdegenerativeconditions),istocreatevaccinesthatactagainsttheirconstituentmolecules.51Althoughthisapproachmayresultinthetoxicmaterial'sbeingingestedbyimmunesystemcells,wecanthenusethestrategiesforcombatingintracellularaggregatesdescribedabovetodisposeofit.ExtracellularAggregates. AGEs (advanced glycation end-products) result from undesirable cross-linkingofusefulmoleculesasasideeffectofexcesssugar.Thesecross-linksinterferewiththenormalfunctioningofproteinsandarekeycontributorstotheagingprocess.AnexperimentaldrugcalledALT-711 (phenacyldimenthylthiazolium chloride) can dissolve these cross-links without damaging theoriginaltissue.52Othermoleculeswiththiscapabilityhavealsobeenidentified.CellLossandAtrophy.Ourbody'stissueshavethemeanstoreplaceworn-outcells,butthisabilityislimited in certain organs. For example, as we get older, the heart is unable to replace its cells at asufficientrate,soitcompensatesbymakingsurvivingcellsbiggerusingfibrousmaterial.Overtimethiscausesthehearttobecomelesssuppleandresponsive.Aprimarystrategyhereistodeploytherapeuticcloningofourowncells,asdescribedbelow.Progressincombatingallofthesesourcesofagingismovingrapidlyinanimalmodels,andtranslationintohumantherapieswillfollow.Evidencefromthegenomeprojectindicatesthatnomorethanafewhundredgenesareinvolvedintheagingprocess.Bymanipulatingthesegenes,radicallifeextensionhasalreadybeenachievedinsimpleranimals.Forexample,bymodifyinggenesintheC.eleganswormthatcontrolitsinsulinandsex-hormonelevels,thelifespanofthetestanimalswasexpandedsixfold,totheequivalentofafive-hundred-yearlifespanforahuman.53Ahybridscenarioinvolvingbothbio-andnanotechnologycontemplatesturningbiologicalcellsinto computers. These "enhanced intelligence" cells can then detect and destroy cancer cells andpathogens or even regrowhuman body parts. Princeton biochemistRonWeiss hasmodified cells toincorporate a variety of logic functions that are used for basic computation.54 Boston University'sTimothyGardnerhasdevelopedacellular logicswitch,anotherbasicbuildingblockfor turningcellsintocomputers.55ScientistsattheMITMediaLabhavedevelopedwaystousewirelesscommunicationtosendmessages,includingintricatesequencesofinstructions,tothecomputersinsidemodifiedcells.56Weisspointsoutthat"onceyouhavetheabilitytoprogramcells,youdon'thavetobeconstrainedby

whatthecellsknowhowtodoalready.Youcanprogramthemtodonewthings,innewpatterns."HumanCloning:TheLeastInterestingApplicationofCloningTechnologyOne of the most powerful methods of applying life's machinery involves harnessing biology's ownreproductivemechanisms in the formof cloning.Cloningwill be a key technology—not for cloningactualhumansbutforlife-extensionpurposes,intheformof"therapeuticcloning."Thisprocesscreatesnew tissues with "young" telomere-extended and DNA-corrected cells to replace without surgerydefectivetissuesororgans. All responsible ethicists, includingmyself, consider human cloning at the present time to beunethical.Thereasons,however,formehavelittletodowiththeslippery-slopeissuesofmanipulatinghumanlife.Rather, the technologytodaysimplydoesnotyetworkreliably.Thecurrent techniqueoffusingacellnucleusfromadonortoaneggcellusinganelectricsparksimplycausesahighlevelofgeneticerrors.57Thisistheprimaryreasonthatmostofthefetusescreatedbythismethoddonotmakeit to term. Even those that do make it have genetic defects. Dolly the Sheep developed an obesityproblem in adulthood, and themajority of cloned animals produced thus far have had unpredictablehealthproblems.58Scientistshaveanumberofideasforperfectingcloning,includingalternativewaysoffusingthenucleus and egg cell without use of a destructive electrical spark, but until the technology isdemonstrablysafe, itwouldbeunethical tocreateahumanlifewithsuchahigh likelihoodofseverehealthproblems.There isnodoubt thathumancloningwill occur, andoccur soon,drivenbyall theusual reasons, ranging from itspublicityvalue to itsutilityasaveryweak formof immortality.Themethods that are demonstrable in advanced animals will work quite well in humans. Once thetechnologyisperfectedintermsofsafety,theethicalbarrierswillbefeebleiftheyexistatan.Cloningisasignificanttechnology,butthecloningofhumansisnotitsmostnoteworthyusage.Let'sfirstaddressitsmostvaluableapplicationsandthenreturntoitsmostcontroversialone.WhyIsCloningImportant?Themostimmediateuseforcloningisimprovedbreedingbyofferingtheability to directly reproduce an animalwith a desirable set of genetic traits. A powerful example isreproducing animals from transgenic embryos (embryos with foreign genes) for pharmaceuticalproduction. A case in point: a promising anticancer treatment is an antiangiogenesis drug calledaaATIII,whichisproducedinthemilkoftransgenicgoats.59Preserving Endangered Species and Restoring Extinct Ones. Another exciting application is re-creatinganimalsfromendangeredspecies.Bycryopreservingcellsfromthesespecies,theyneverneedbecomeextinct.Itwilleventuallybepossibletore-createanimalsfromrecentlyextinctspecies.In2001scientistswereabletosynthesizeDNAfortheTasmaniantiger,whichhadthenbeenextinctforsixty-fiveyears,withthehopeofbringingthisspeciesbacktolife.60Asforlong-extinctspecies(forexample,dinosaurs),itishighlydoubtfulthatwewillfindthefullyintactDNArequiredinasinglepreservedcell(as they did in the movie Jurassic Park). It is likely, however, that we will eventually be able tosynthesize the necessary DNA by patching together the information derived from multiple inactivefragments.TherapeuticCloning.Perhapsthemostvaluableemergingapplicationis therapeuticcloningofone'sown organs. By starting with germ-line cells (inherited from the eggs or sperm and passed on tooffspring), genetic engineers can trigger differentiation into diverse types of cells. Becausedifferentiation takes place during the prefetal stage (that is, prior to implantation of a fetus), mostethicists believe this process does not raise concerns, although the issue has remained highlycontentious.61

Human Somatic-Cell Engineering. This even more promising approach, which bypasses thecontroversyofusingfetalstemcellsentirely,iscalledtransdifferentiation;itcreatesnewtissueswithapatient'sownDNAbyconvertingonetypeofcell(suchasaskincell)intoanother(suchasapancreaticisletcelloraheartcell).62ScientistsfromtheUnitedStatesandNorwayhaverecentlybeensuccessfulinreprogramminglivercellsintobecomingpancreascells.Inanotherseriesofexperiments,humanskincellsweretransformedtotakeonmanyofthecharacteristicsofimmune-systemcellsandnervecells.63Considerthequestion,Whatisthedifferencebetweenaskincellandanyothertypeofcellinthebody?After all, they all have the sameDNA.As noted above, the differences are found in proteinsignaling factors,which include shortRNA fragments andpeptides,whichwe are nowbeginning tounderstand.64Bymanipulatingtheseproteins,wecaninfluencegeneexpressionandtrickonetypeofcellintobecominganother. Perfectingthis technologywouldnotonlydefuseasensitiveethicalandpolitical issuebutalsoofferanidealsolutionfromascientificperspective.Ifyouneedpancreaticisletcellsorkidneytissues—orevenawholenewheart—toavoidautoimmunereactions,youwouldstronglyprefertoobtainthesewith your own DNA rather than the DNA from someone else's germ-line cells. In addition, thisapproachusesplentifulskincells(ofthepatient)ratherthanrareandpreciousstemcells. Transdifferentiation will directly grow an organ with your genetic makeup. Perhaps mostimportant,theneworgancanhaveitstelomeresfullyextendedtotheiroriginalyouthfullength,sothattheneworganiseffectivelyyoungagain.65WecanalsocorrectaccumulatedDNAerrorsbyselectingtheappropriateskincells(thatis,oneswithoutDNAerrors)priortotransdifferentiationintoothertypesofcells.Usingthismethodaneighty-year-oldmancouldhavehisheartreplacedwiththesamehearthehadwhenhewas,say,twenty-five.Currenttreatmentsfortype1diabetesrequirestrongantirejectiondrugsthatcanhavedangeroussideeffects.66Withsomatic-cellengineering,type1diabeticswillbeabletomakepancreaticisletcellsfromtheirowncells,eitherfromskincells(transdifferentiation)orfromadultstemcells.TheywouldbeusingtheirownDNA,anddrawinguponarelativelyinexhaustiblesupplyofcells,sonoantirejectiondrugs would be required. (But to fully cure type 1 diabetes, we would also have to overcome thepatient'sautoimmunedisorder,whichcauseshisbodytodestroyisletcells.) Evenmore exciting is the prospect of replacing one's organs and tissueswith their "young"replacements without surgery. Introducing cloned, telomere-extended, DNA-corrected cells into anorganwillallowthemtointegratethemselveswiththeoldercells.Byrepeatedtreatmentsofthiskindoveraperiodoftime,theorganwillendupbeingdominatedbytheyoungercells.Wenormallyreplaceourowncellsonaregularbasisanyway,sowhynotdosowithyouthfulrejuvenatedcellsratherthantelomere-shortenederror-filledones?There'snoreasonwhywecouldn't repeat thisprocess foreveryorganandtissueinourbody,enablingustogrowprogressivelyyounger.SolvingWorldHunger.Cloningtechnologiesevenofferapossiblesolutionforworldhunger:creatingmeatandotherproteinsourcesinafactorywithoutanimalsbycloninganimalmuscletissue.Benefitswould include extremely low cost, avoidanceof pesticides andhormones that occur in naturalmeat,greatlyreducedenvironmentalimpact(comparedtofactoryfarming),improvednutritionalprofile,andnoanimalsuffering.Aswiththerapeuticcloning,wewouldnotbecreatingtheentireanimalbutratherdirectlyproducingthedesiredanimalpartsorflesh.Essentially,allofthemeat—billionsofpoundsofit—wouldbederivedfromasingleanimal. Thereareotherbenefits tothisprocessbesidesendinghunger.Bycreatingmeatinthisway,itbecomessubjecttothelawofacceleratingreturns—theexponentialimprovementsinprice-performanceof information-based technologies over time—and will thus become extremely inexpensive. Eventhoughhungerintheworldtodayiscertainlyexacerbatedbypoliticalissuesandconflicts,meatcouldbecomesoinexpensivethatitwouldhaveaprofoundeffectontheaffordabilityoffood. Theadventofanimal-lessmeatwillalsoeliminateanimalsuffering.Theeconomicsoffactoryfarmingplaceaverylowpriorityonthecomfortofanimals,whicharetreatedascogsinamachine.Themeatproducedinthismanner,althoughnormalinallotherrespects,wouldnotbepartofananimalwith

anervoussystem,whichisgenerallyregardedasanecessaryelementforsufferingtooccur,atleastinabiologicalanimal.Wecouldusethesameapproachtoproducesuchanimalby-productsasleatherandfur.Othermajoradvantageswouldbetoeliminatetheenormousecologicalandenvironmentaldamagecreatedbyfactoryfarmingaswellastheriskofprion-baseddiseases,suchasmad-cowdiseaseanditshumancounterpart,vCJD.67HumanCloningRevisited.Thisbringsusagaintohumancloning.Ipredictthatoncethetechnologyisperfected,neithertheacutedilemmasseenbyethicistsnortheprofoundpromiseheraldedbyenthusiastswillpredominate.Sowhatifwehavegenetictwinsseparatedbyoneormoregenerations?Cloningislikely to prove to be like other reproductive technologies thatwere briefly controversial but rapidlyaccepted.Physicalcloningisfardifferentfrommentalcloning,inwhichaperson'sentirepersonality,memory, skills, and history will ultimately be downloaded into a different, and most likely morepowerful,thinkingmedium.There'snoissueofphilosophicalidentitywithgeneticcloning,sincesuchcloneswouldbedifferentpeople,evenmoresothanconventionaltwinsaretoday. Ifweconsider the full conceptofcloning, fromcell toorganisms, itsbenefitshaveenormoussynergywiththeotherrevolutionsoccurringinbiologyaswellasincomputertechnology.Aswelearntounderstandthegenomeandproteome(theexpressionof thegenomeintoproteins)ofbothhumansand animals, and as we develop powerful new means of harnessing genetic information, cloningprovidesthemeanstoreplicateanimals,organs,andcells.Andthathasprofoundimplicationsforthehealthandwell-beingofbothourselvesandourevolutionarycousinsintheanimalkingdom.NEDLUDD: If everyonecanchange theirgenes, theneveryonewill choose tobe "perfect" in every

way,sothere'llbenodiversityandexcellingwillbecomemeaningless.RAY: Not exactly. Genes are obviously important, but our nature—skills, knowledge, memory,

personality—reflectsthedesigninformationinourgenes,asourbodiesandbrainsself-organizethroughour experience.This isalso readily evident inourhealth. I personallyhaveageneticdispositiontotype2diabetes,havingactuallybeendiagnosedwiththatdiseasemorethantwentyyearsago.ButIdon'thaveanyindicationofdiabetestodaybecauseI'veovercomethisgeneticdisposition as a result of reprogramming my biochemistry through lifestyle choices such asnutrition, exercise, and aggressive supplementation. With regard to our brains, we all havevarious aptitudes, but our actual talents are a functionofwhatwe've learned, developed, andexperienced.Ourgenesreflectdispositionsonly.Wecanseehowthisworksinthedevelopmentof the brain. The genes describe certain rules and constraints for patterns of interneuronalconnections, but the actual connections we have as adults are the result of a self-organizingprocessbasedonourlearning.Thefinalresult—whoweare—isdeeplyinfluencedbybothnature(genes)andnurture(experience).

Sowhenwegaintheopportunitytochangeourgenesasadults,wewon'twipeouttheinfluenceofourearliergenes.Experiencespriortothegenetherapywillhavebeentranslatedthroughthepretherapy genes, so one's character and personality would still be shaped primarily by theoriginalgenes.Forexample, ifsomeoneaddedgenesformusicalaptitudetohisbrainthroughgenetherapy,hewouldnotsuddenlybecomeamusicgenius.

NED:Okay,Iunderstandthatdesignerbabyboomerscan'tgetawaycompletelyfromtheirpredesignergenes,butwithdesignerbabiesthey'llhavethegenesandthetimetoexpressthem.

RAY:The"designerbaby"revolutionisgoingtobeaveryslowone;itwon'tbeasignificantfactorinthiscentury.Otherrevolutionswillovertakeit.Wewon'thavethetechnologyfordesignerbabiesforanothertentotwentyyears.Totheextentthatitisused,itwouldbeadoptedgradually,andthen itwill take thosegenerationsanother twenty years to reachmaturity.By that time,we'reapproaching theSingularity,with the real revolutionbeing thepredominanceofnonbiologicalintelligence.Thatwillgofarbeyondthecapabilitiesofanydesignergenes.Theideaofdesignerbabiesandbabyboomersisjustthereprogrammingoftheinformationprocessesinbiology.Butit'sstillbiology,withallitsprofoundlimitations.

NED:You'remissingsomething.Biologicaliswhatweare.Ithinkmostpeoplewouldagreethatbeingbiologicalisthequintessentialattributeofbeinghuman..

RAY:That'scertainlytruetoday.NED:AndIplantokeepitthatway.RAY:Well, if you're speaking for yourself, that's finewith 'me. But if you stay biological and don't

reprogramyourgenes,youwon'tbearoundforverylongtoinfluencethedebate.

Nanotechnology:TheIntersectionofInformationandthePhysicalWorld

Theroleoftheinfinitelysmallisinfinitelylarge.

—LOUISPASTEURButIamnotafraidtoconsiderthefinalquestionastowhether,ultimately,inthegreatfuture,wecanarrangetheatomsthewaywewant;theveryatoms,allthewaydown!

—RICHARDFEYNMANNanotechnology has the potential to enhance human performance, to bring sustainabledevelopment formaterials,water, energy, and food, toprotect againstunknownbacteria andviruses, and even to diminish the reasons for breaking the peace [by creating universalabundance].

—NATIONALSCIENCEFOUNDATIONNANOTECHNOLOGYREPORTNanotechnology promises the tools to rebuild the physicalworld—our bodies and brains included—molecularfragmentbymolecularfragment,potentiallyatombyatom.Weareshrinkingthekeyfeaturesize of technology, in accordance with the law of accelerating returns, at the exponential rate ofapproximatelyafactoroffourperlineardimensionperdecade.68Atthisratethekeyfeaturesizesformost electronic and many mechanical technologies will be in the nanotechnology range—generallyconsideredtobeunderonehundrednanometers—bythe2020s.(Electronicshasalreadydippedbelowthisthreshold,althoughnotyetinthree-dimensionalstructuresandnotyetself-assembling.)Meanwhilerapid progress has been made, particularly in the last several years, in preparing the conceptualframeworkanddesignideasforthecomingageofnanotechnology.Asimportantasthebiotechnologyrevolutiondiscussedabovewillbe,onceitsmethodsarefullymature,limitswillbeencounteredinbiologyitself.Althoughbiologicalsystemsareremarkableintheircleverness,wehavealsodiscoveredthattheyaredramaticallysuboptimal.I'vementionedtheextremelyslowspeedofcommunicationinthebrain,andasIdiscussbelow(seep.253),roboticreplacementsforour red blood cells could be thousands of times more efficient than their biological counterparts.69Biologywillneverbeabletomatchwhatwewillbecapableofengineeringoncewefullyunderstandbiology'sprinciplesofoperation. The revolution innanotechnology,however,willultimatelyenableus to redesignand rebuild,molecule by molecule, our bodies and brains and the world with which we interact.70 These tworevolutions areoverlapping,but the full realizationofnanotechnology lagsbehind thebiotechnologyrevolutionbyaboutonedecade.MostnanotechnologyhistoriansdatetheconceptualbirthofnanotechnologytophysicistRichardFeynman'sseminalspeechin1959,"There'sPlentyofRoomattheBottom,"inwhichhedescribedtheinevitabilityandprofoundimplicationsofengineeringmachinesatthelevelofatoms:

The principles of physics, as far as I can see, do not speak against the possibility ofmaneuvering things atom by atom. It would be, in principle, possible ... for a physicist tosynthesizeanychemicalsubstancethatthechemistwritesdown....How?Puttheatomsdownwhere the chemist says, and so you make the substance. The problems of chemistry andbiologycanbegreatlyhelpedifourability toseewhatwearedoing,andtodothingsonanatomiclevel,isultimatelydeveloped—adevelopmentwhichIthinkcannotbeavoided.71

Anevenearlier conceptual foundation fornanotechnologywas formulatedby the informationtheoristJohnvonNeumannin theearly1950swithhismodelofaself-replicatingsystembasedona

universal constructor, combined with a universal computer.72 In this proposal the computer runs aprogramthatdirectstheconstructor,whichinturnconstructsacopyofboththecomputer(includingitsself-replicationprogram) and the constructor.At this level of descriptionvonNeumann's proposal isquiteabstract—thecomputerandconstructorcouldbemadeinagreatvarietyofways,aswellasfromdiversematerials,andcouldevenbea theoreticalmathematicalconstruction.Buthetooktheconceptonestepfurtherandproposeda"kinematicconstructor":arobotwithatleastonemanipulator(arm)thatwouldbuildareplicaofitselffroma"seaofparts"initsmidst.73 It was left to EricDrexler to found themodern field of nanotechnology,with a draft of hislandmark Ph.D. thesis in the mid-1980s, in which he essentially combined these two intriguingsuggestions.DrexlerdescribedavonNeumannkinematicconstructor,which for its seaofpartsusedatoms andmolecular fragments, as suggested inFeynman's speech.Drexler's vision cut acrossmanydisciplinaryboundariesandwassofar-reachingthatnoonewasdaringenoughtobehisthesisadviserexceptformyownmentor,MarvinMinsky.Drexler'sdissertation(whichbecamehisbookEnginesofCreationin1986andwasarticulatedtechnicallyinhis1992book,Nanosystems)laidoutthefoundationofnanotechnologyandprovidedtheroadmapstillbeingfollowedtoday.74 Drexler's"molecularassembler"willbeabletomakealmostanythingintheworld.Ithasbeenreferred toasa "universalassembler,"butDrexlerandothernanotechnology theoristsdonotuse theword"universal"because theproductsofsuchasystemnecessarilyhave tobesubject to the lawsofphysicsandchemistry,soonlyatomicallystablestructureswouldbeviable.Furthermore,anyspecificassemblerwouldberestrictedtobuildingproductsfromitsseaofparts,althoughthefeasibilityofusingindividualatomshasbeenshown.Nevertheless,suchanassemblercouldmakejustaboutanyphysicaldevicewewouldwant,includinghighlyefficientcomputersandsubsystemsforotherassemblers.AlthoughDrexlerdidnotprovideadetaileddesignforanassembler—suchadesignhasstillnotbeen fully specified—his thesis did provide extensive feasibility arguments for each of the principalcomponentsofamolecularassembler,whichincludethefollowingsubsystems:

· Thecomputer: to provide the intelligence to control the assembly process.Aswith all of thedevice's subsystems, the computer needs to be small and simple. As I described in chapter 3,Drexlerprovidesan intriguingconceptualdescriptionofamechanicalcomputerwithmolecular"locks" instead of transistor gates. Each lock would require only sixteen cubic nanometers ofspaceandcouldswitchtenbilliontimespersecond.Thisproposalremainsmorecompetitivethanany known electronic technology, although electronic computers built from three-dimensionalarrays of carbon nanotubes appear to provide even higher densities of computation (that is,calculationspersecondpergram).75

· Theinstructionarchitecture:DrexlerandhiscolleagueRalphMerklehaveproposedanSIMD(single instruction multiple data) architecture in which a single data store would record theinstructionsandtransmitthemtotrillionsofmolecular-sizedassemblers(eachwithitsownsimplecomputer)simultaneously.IdiscussedsomeofthelimitationsoftheSIMDarchitectureinchapter3, but this design (which is easier to implement than the more flexible multiple-instructionmultiple-data approach) is sufficient for the computer in auniversalnanotechnologyassembler.With this approach each assemblerwould not have to store the entire program for creating thedesired product. A "broadcast" architecture also addresses a key safety concern: the self-replication process could be shut down, if it got out of control, by terminating the centralizedsourceofthereplicationinstructions.

However,asDrexlerpointsout,ananoscaleassemblerdoesnotnecessarilyhavetobeself-replicating.76Giventheinherentdangersinself-replication,theethicalstandardsproposedbytheForesight Institute (a think tank founded by Eric Drexler and Christine Peterson) containprohibitionsagainstunrestrictedself-replication,especiallyinanaturalenvironment.

AsIwilldiscussinchapter8,thisapproachshouldbereasonablyeffectiveagainstinadvertentdangers,althoughitcouldbecircumventedbyadeterminedandknowledgeableadversary.

·Instructiontransmission:Transmissionoftheinstructionsfromthecentralizeddatastoretoeach

of themany assemblers would be accomplished electronically if the computer is electronic orthroughmechanicalvibrationsifDrexler'sconceptofamechanicalcomputerwereused.

· Theconstructionrobot:Theconstructorwouldbeasimplemolecular robotwithasinglearm,similartovonNeumann'skinematicconstructorbutonatinyscale.Therearealreadyexamplesofexperimentalmolecular-scalesystemsthatcanactasmotorsandrobotlegs,asIdiscussbelow.

·Therobotarmtip:Drexler'sNanosystemsprovidedanumberoffeasiblechemistriesforthetipoftherobotarmtomakeitcapableofgrasping(usingappropriateatomic-forcefields)amolecularfragment,orevenasingleatom,andthendepositingitinadesiredlocation.Inthechemical-vapordeposition process used to construct artificial diamonds, individual carbon atoms, as well asmolecularfragments,aremovedtootherlocationsthroughchemicalreactionsatthetip.Buildingartificialdiamondsisachaoticprocessinvolvingtrillionsofatoms,butconceptualproposalsbyRobert Freitas and RalphMerkle contemplate robot arm tips that can remove hydrogen atomsfroma sourcematerial anddeposit thematdesired locations in theconstructionof amolecularmachine.Inthisproposal,thetinymachinesarebuiltoutofadiamondoidmaterial.Inadditiontohaving great strength, thematerial can be dopedwith impurities in a precise fashion to createelectronic components such as transistors. Simulations have shown that such molecular-scalegears,levers,motors,andothermechanicalsystemswouldoperateproperlyasintended.77Morerecentlyattentionhasbeenfocusedoncarbonnanotubes,comprisinghexagonalarraysofcarbonatomsassembled in threedimensions,whicharealsocapableofprovidingbothmechanicalandelectronicfunctionsatthemolecularlevel.Iprovideexamplesbelowofmolecular-scalemachinesthathavealreadybeenbuilt.

·Theassembler'sinternalenvironmentneedstopreventenvironmentalimpuritiesfrominterferingwiththedelicateassemblyprocess.Drexler'sproposalistomaintainanearvacuumandbuildtheassembler walls out of the same diamondoid material that the assembler itself is capable ofmaking.

· The energy required for the assembly process can be provided either through electricity orthroughchemicalenergy.Drexlerproposedachemicalprocesswith thefuel interlacedwith theraw building material. More recent proposals use nanoengineered fuel cells incorporatinghydrogenandoxygenorglucoseandoxygen,oracousticpoweratultrasonicfrequencies.78

Althoughmanyconfigurationshavebeenproposed,thetypicalassemblerhasbeendescribedasatabletopunitthatcanmanufacturealmostanyphysicallypossibleproductforwhichwehaveasoftwaredescription,rangingfromcomputers,clothes,andworksofarttocookedmeals.79Largerproducts,suchas furniture, cars, or even houses, can be built in amodular fashion or using larger assemblers. Ofparticular importance is the fact that an assembler can create copies of itself, unless its designspecifically prohibits this (to avoid potentially dangerous self-replication), The incremental cost ofcreating any physical product, including the assemblers themselves, would be pennies per pound—basically the cost of the raw materials. Drexler estimates total manufacturing cost for a molecular-manufacturingprocess in the rangeof tencents to fiftycentsperkilogram, regardlessofwhether themanufactured productwere clothing,massively parallel supercomputers, or additionalmanufacturingsystems.80Therealcost,ofcourse,wouldbethevalueoftheinformationdescribingeachtypeofproduct—thatis, thesoftwarethatcontrolstheassemblyprocess.Inotherwords, thevalueofeverythingintheworld,includingphysicalobjects,wouldbebasedessentiallyoninformation.Wearenotthatfarfromthis situation today, since the information content of products is rapidly increasing, graduallyapproachinganasymptoteof100percentoftheirvalue. The design of the software controlling molecular-manufacturing systems would itself beextensivelyautomated,muchaschipdesignistoday.Chipdesignersdon'tspecifythelocationofeachofthebillionsofwiresandcomponentsbutratherthespecificfunctionsandfeatures,whichcomputer-aideddesign(CAD)systemstranslateintoactualchiplayouts.Similarly,CADsystemswouldproducethemolecular-manufacturing control software from high-level specifications. Thiswould include the

abilitytoreverseengineeraProductbyscanningitinthreedimensionsandthengeneratingthesoftwareneededtoreplicateitsoverallcapabilities.Inoperation,thecentralizeddatastorewouldsendoutcommandssimultaneouslytomanytrillions(someestimatesashighas1018)ofrobots inanassembler,eachreceivingthesameinstructionat thesametime.Theassemblerwouldcreatethesemolecularrobotsbystartingwithasmallnumberandthenusingtheserobotstocreateadditionalonesinaniterativefashion,untiltherequisitenumberhadbeencreated.Eachrobotwouldhavealocaldatastoragethatspecifiesthetypeofmechanismit'sbuilding.Thisstoragewouldbeusedtomasktheglobalinstructionsbeingsentfromthecentralizeddatastoresothatcertaininstructionsareblockedandlocalparametersarefilledin.Inthisway,eventhoughalloftheassemblersarereceivingthesamesequenceofinstructions,thereisalevelofcustomizationtothepartbeingbuiltbyeachmolecularrobot.Thisprocessisanalogoustogeneexpressioninbiologicalsystems.Althougheverycellhaseverygene,only thosegenes relevant toaparticularcell typeareexpressed.Each robot extracts the rawmaterials and fuel it needs, which include individual carbon atoms andmolecularfragments,fromthesourcematerial.TheBiologicalAssembler

Nature shows thatmoleculescan serveasmachinesbecause living thingsworkbymeansofsuchmachinery.Enzymesaremolecularmachines thatmake,break,andrearrange thebondsholdingothermolecules together.Musclesaredrivenbymolecularmachines thathaul fiberspast one another. DNA serves as a data-storage system, transmitting digital instructions tomolecular machines, the ribosomes, that manufacture protein molecules. And these proteinmolecules,inturn,makeupmostofthemolecularmachinery.

—ERICDREXLERThe ultimate existence proof of the feasibility of amolecular assembler is life itself. Indeed, as wedeepenourunderstandingoftheinformationbasisoflifeprocesses,wearediscoveringspecificideasthat are applicable to the design requirements of a generalized molecular assembler. For example,proposalshavebeenmadetouseamolecularenergysourceofglucoseandATP,similartothatusedbybiologicalcells.ConsiderhowbiologysolveseachofthedesignchallengesofaDrexlerassembler.Theribosomerepresentsboththecomputerandtheconstructionrobot.Lifedoesnotusecentralizeddatastoragebutprovidestheentirecodetoeverycell.Theabilitytorestrictthelocaldatastorageofananoengineeredrobottoonlyasmallpartoftheassemblycode(usingthe"broadcast"architecture),particularlywhendoingself-replication,isonecriticalwaynanotechnologycanbeengineeredtobesaferthanbiology. Life's local data storage is, of course, the DNA strands, broken into specific genes on thechromosomes.Thetaskofinstructionmasking(blockinggenesthatdonotcontributetoaparticularcelltype)iscontrolledbytheshortRNAmoleculesandpeptidesthatgoverngeneexpression.Theinternalenvironment in which the ribosome is able to function is the particular chemical environmentmaintainedinsidethecell,whichincludesaparticularacid-alkalineequilibrium(pHaround7inhumancells) and other chemical balances. The cell membrane is responsible for protecting this internalenvironmentfromdisturbance.Upgrading the Cell Nucleus with a Nanocomputer and Nanobot. Here's a conceptually simpleproposaltoovercomeallbiologicalpathogensexceptforprions(self-replicatingpathologicalproteins).With the advent of full-scale nanotechnology in the 2020s we will have the potential to replacebiology'sgenetic-information repository in thecellnucleuswithananoengineered system thatwouldmaintain thegenetic code and simulate the actionsofRNA, the ribosome, andother elementsof thecomputerinbiology'sassembler.Ananocomputerwouldmaintainthegeneticcodeandimplementthegene-expression algorithms. A nanobot would then construct the amino-acid sequences for the

expressedgenes. There would be significant benefits in adopting such a mechanism.We could eliminate theaccumulationofDNAtranscriptionerrors,onemajorsourceoftheagingprocess.WecouldintroduceDNA changes to essentially reprogram our genes (something we'll be able to do long before thisscenario, using gene-therapy techniques). We would also be able to defeat biological pathogens(bacteria,viruses,andcancercells)byblockinganyunwantedreplicationofgeneticinformation.

Withsuchananoengineeredsystemtherecommendedbroadcastarchitecturewouldenableustoturn off unwanted replication, thereby defeating cancer, autoimmune reactions, and other diseaseprocesses. Although most of these disease processes will already have been vanquished by thebiotechnology methods described in the previous section, reengineering the computer of life usingnanotechnologycouldeliminateanyremainingobstaclesandcreatealevelofdurabilityandflexibilitythatgoesbeyondtheinherentcapabilitiesofbiology.Therobotarmtipwouldusetheribosome'sabilitytoimplementenzymaticreactionstobreakoffanindividualaminoacid,eachofwhichisboundtoaspecifictRNA,andtoconnectittoitsadjoiningaminoacidusingapeptidebond.Thus,suchasystemcouldutilizeportionsoftheribosomeitself,sincethisbiologicalmachineiscapableofconstructingtherequisitestringofaminoacids.However,thegoalofmolecularmanufacturingisnotmerelytoreplicatethemolecular-assemblycapabilities of biology. Biological systems are limited to building systems from protein, which hasprofoundlimitationsinstrengthandspeed.Althoughbiologicalproteinsarethree-dimensional,biologyisrestrictedtothatclassofchemicalsthatcanbefoldedfromaone-dimensionalstringofaminoacids.Nanobotsbuilt fromdiamondoidgears and rotors can alsobe thousandsof times faster and strongerthanbiologicalcells. The comparison is evenmore dramatic with regard to computation: the switching speed ofnanotube-based computation would be millions of times faster than the extremely slow transactionspeedoftheelectrochemicalswitchingusedinmammalianinterneuronalconnections. The concept of a diamondoid assembler described aboveuses a consistent inputmaterial (forconstructionandfuel),whichrepresentsoneofseveralprotectionsagainstmolecular-scalereplication

ofrobotsinanuncontrolledfashionintheoutsideworld.Biology'sreplicationrobot,theribosome,alsorequirescarefullycontrolledsourceandfuelmaterials,whichareprovidedbyourdigestivesystem.ABnanobasedreplicatorsbecomemoresophisticated,morecapableofextractingcarbonatomsandcarbon-basedmolecular fragments from lesswell-controlled sourcematerials, and able to operate outside ofcontrolledreplicatorenclosuressuchasinthebiologicalworld,theywillhavethepotentialtopresentagravethreat to thatworld.This isparticularly trueinviewof thevastlygreaterstrengthandspeedofnanobased replicators over any biological system. That ability is, of course, the source of greatcontroversy,whichIdiscussinchapter8. In thedecade sincepublicationofDrexler'sNanosystems, eachaspectofDrexler's conceptualdesignshasbeenvalidatedthroughadditionaldesignproposals.81supercomputersimulations,and,mostimportant, actual construction of relatedmolecularmachines.BostonCollege chemistry professorT.RossKellyreportedthatheconstructedachemicallypowerednanomotoroutofseventy-eightatoms.82

A biomolecular research group headed by Carlo Montemagno created an ATP-fueled nanomotor.83Another molecule-sized motor fueled by solar energy was created out of fifty-eight atoms by BenFeringaattheUniversityofGroningenintheNetherlands.84Similarprogresshasbeenmadeonothermolecular-scalemechanical components such asgears, rotors, and levers.Systemsdemonstrating theuseofchemicalenergyandacousticenergy (asoriginallydescribedbyDrexler)havebeendesigned,simulated, and actually constructed. Substantial progress has also been made in developing varioustypes of electronic components from molecular-scale devices, particularly in the area of carbonnanotubes,anareathatRichardSmalleyhaspioneered. Nanotubes are also proving to be very versatile as a structural component. A conveyor beltconstructedout of nanotubeswasdemonstrated recently by scientists atLawrenceBerkeleyNationalLaboratory.85Thenanoscaleconveyorbeltwasusedtotransporttinyindiumparticlesfromonelocationto another, although the techniquecouldbe adapted tomoveavarietyofmolecule-sizedobjects.Bycontrollinganelectricalcurrentapplied to thedevice, thedirectionandvelocityofmovementcanbemodulated. "It's theequivalentof turningaknob ... and takingmacroscalecontrolofnanoscalemasstransport," saidChris Regan, one of the designers. "And it's reversible:we can change the current'spolarityanddrivetheindiumbacktoitsoriginalposition."Theabilitytorapidlyshuttlemolecule-sizedbuildingblockstopreciselocationsisakeysteptowardbuildingmolecularassemblylines. A study conducted forNASAbyGeneralDynamics has demonstrated the feasibility of self-replicatingnanoscalemachines.86Usingcomputersimulations,theresearchersshowedthatmolecularlyprecise robots calledkinematic cellular automata,built from reconfigurablemolecularmodules,werecapableofreproducingthemselves.Thedesignsalsousedthebroadcastarchitecture,whichestablishedthefeasibilityofthissaferformofself-replication.DNAisprovingtobeasversatileasnanotubesforbuildingmolecularstructures.DNA'sproclivitytolinkupwithitselfmakesitausefulstructuralcomponent.Futuredesignsmaycombinethisattributeaswellasitscapacityforstoringinformation.BothnanotubesandDNAhaveoutstandingpropertiesforinformationstorageandlogicalcontrol,aswellasforbuildingstrongthree-dimensionalstructures.AresearchteamatLudwigMaximiliansUniversityinMunichhasbuilta"DNAhand"thatcanselect one of several proteins, bind to it, and then release it upon command.87 Important steps increating a DNA assembler mechanism akin to the ribosome were demonstrated recently bynanotechnologyresearchersShipingLiaoandNadrianSeeman.88Graspingandlettinggoofmolecularobjects inacontrolledmanner isanother importantenablingcapability formolecularnanotechnologyassembly.ScientistsattheScrippsResearchInstitutedemonstratedtheabilitytocreateDNAbuildingblocksby generating many copies of a 1,669-nucleotide strand of DNA that had carefully placed self-complementary regions.89 The strands self-assembled spontaneously into rigid octahedrons, whichcouldbeusedasblocksforelaboratethree-dimensionalstructures.Anotherapplicationofthisprocesscouldbetoemploytheoctahedronsascompartmentstodeliverproteins,whichGeraldF.Joyce,oneofthe Scripps researchers, called a "virus in reverse."Viruses, which are also self-assembling, usually

haveoutershellsofproteinwithDNA(orRNA)ontheinside."Withthis,"Joycepointsout,"youcouldinprinciplehaveDNAontheoutsideandproteinsontheinside." Aparticularly impressivedemonstrationofananoscaledeviceconstructedfromDNAisa tinybipedrobotthatcanwalkonlegsthataretennanometerslong.90Boththelegsandthewalkingtrackarebuilt from DNA, again chosen for the molecule's ability to attach and detach itself in a controlledmanner. The nanorobot, a project of chemistry professorsNadrian Seeman andWilliamSherman ofNewYorkUniversity,walksbydetachingitslegsfromthetrack,movingdownit,andthenreattachingits legs to the track. The project is another impressive demonstration of the ability of nanoscalemachinestoexecuteprecisemaneuvers. Analternatemethodofdesigningnanobots is to learn fromnature.NanotechnologistMichaelSimpsonofOakRidgeNationalLaboratorydescribes thepossibilityofexploitingbacteria"as ready-mademachine[s]."Bacteria,whicharenaturalnanobot-sizeobjects,areabletomove,swim,andpumpliquids.91LindaTurner,ascientistattheRowlandInstituteatHarvard,hasfocusedontheirthread-sizearms, called fimbriae, which are able to perform a wide variety of tasks, including carrying othernanoscaleobjectsandmixingfluids.Anotherapproachistouseonlypartsofbacteria.AresearchgroupheadedbyViolaVogelat theUniversityofWashingtonbuiltasystemusingjust the limbsofE.colibacteriathatwasabletosortoutnanoscalebeadsofdifferentsizes.Sincebacteriaarenaturalnanoscalesystemsthatcanperformawidevarietyoffunctions,theultimategoalofthisresearchwillbetoreverseengineerthebacteriasothatthesamedesignprinciplescanbeappliedtoourownnanobotdesigns.FatandStickyFingersInthewakeoftherapidlyexpandingdevelopmentofeachfacetoffuturenanotechnologysystems,noseriousflawinDrexler'snanoassemblerconcepthasbeendescribed.Ahighlypublicizedobjectionin2001byNobelistRichardSmalley inScientificAmericanwasbasedonadistorteddescriptionof theDrexlerproposal;92itdidnotaddresstheextensivebodyofworkthathasbeencarriedoutinthepastdecade.AsapioneerofcarbonnanotubesSmalleyhasbeenenthusiasticaboutavarietyofapplicationsof nanotechnology, having written that "nanotechnology holds the answer, to the extent there areanswers,tomostofourpressingmaterialneedsinenergy,health,communication,transportation,food,water,"butheremainsskepticalaboutmolecularnanotechnologyassembly.SmalleydescribesDrexler'sassemblerasconsistingoffivetoten"fingers"(manipulatorarms)tohold,move,andplaceeachatominthemachinebeingconstructed.Hethengoesontopointout thatthereisn'troomforsomanyfingersinthecrampedspaceinwhichamolecular-assemblynanorobothastowork(whichhecallsthe"fatfingers"problem)andthatthesefingerswouldhavedifficultylettinggo'of their atomic cargo because ofmolecular attraction forces (the "sticky fingers" problem). Smalleyalsopointsoutthatan"intricatethree-dimensionalwaltz...iscarriedout"byfivetofifteenatomsinatypicalchemicalreaction. In fact,Drexler's proposal doesn't look anything like the straw-man description that Smalleycriticizes.Drexler'sproposal,andmostof those thathavefollowed,usesasingle"finger."Moreover,there have been extensive descriptions and analyses of viable tip chemistries that do not involvegraspingandplacingatomsasiftheyweremechanicalpiecestobedepositedinplace.Inadditiontotheexamples I provided above (for example, theDNAhand), the feasibility ofmoving hydrogen atomsusingDrexler's"propynylhydrogenabstraction"tiphasbeenextensivelyconfirmedintheinterveningyears.93Theabilityofthescanning-probemicroscope(SPM),developedatIBMin1981,andthemoresophisticatedatomic-forcemicroscope(AFM)toplaceindividualatomsthroughspecificreactionsofatip with a molecular-scale structure provides additional proof of the concept. Recently, scientists atOsakaUniversityusedanAFMtomoveindividualnonconductiveatomsusingamechanicalratherthanelectrical technique.94 The ability tomove both conductive and nonconductive atoms andmoleculeswillbeneededforfuturemolecularnanotechnology.95

Indeed,ifSmalley'scritiquewerevalid,noneofuswouldbeheretodiscussit,becauselifeitselfwouldbeimpossible,giventhatbiology'sassemblerdoesexactlywhatSmalleysaysisimpossible. Smalley also objects that, despite "working furiously, ... generating even a tiny amount of aproductwouldtake[ananobot] ...millionsofyears."Smalleyiscorrect,ofcourse, thatanassemblerwithonlyonenanobotwouldn't produceanyappreciablequantitiesof aproduct.However, thebasicconceptofnanotechnologyisthatwewillusetrillionsofnanobotstoaccomplishmeaningfulresults—afactorthatisalsothesourceofthesafetyconcernsthathavereceivedsomuchattention.Creatingthismanynanobotsat reasonablecostwill require self-replicationat some level,whichwhile solving theeconomicissuewillintroducepotentiallygravedangers,aconcernIwilladdressinchapter8.Biologyusesthesamesolutiontocreateorganismswithtrillionsofcells,andindeedwefindthatvirtuallyalldiseasesderivefrombiology'sself-replicationprocessgoneawry.Earlierchallengestotheconceptsunderlyingnanotechnologyhavealsobeeneffectivelyaddressed.Critics pointed out that nanobots would be subject to bombardment by thermal vibration of nuclei,atoms, andmolecules. This is one reason conceptual designers of nanotechnology have emphasizedbuilding structural components from diamondoid or carbon nanotubes. Increasing the strength orstiffnessofasystemreducesitssusceptibilitytothermaleffects.Analysisofthesedesignshasshownthemtobethousandsoftimesmorestableinthepresenceofthermaleffectsthanarebiologicalsystems,sotheycanoperateinafarwidertemperaturerange.96Similarchallengesweremaderegardingpositionaluncertaintyfromquantumeffects,basedontheextremelysmallfeaturesizeofnanoengineereddevices.Quantumeffectsaresignificantforanelectron,butasinglecarbonatomnucleusismorethantwentythousandtimesmoremassivethananelectron.Ananobotwillbeconstructedfrommillionstobillionsofcarbonandotheratoms,makingituptotrillionsof timesmoremassive thananelectron.Plugging this ratio in the fundamentalequation forquantumpositionaluncertaintyshowsittobeaninsignificantfactor.97Powerhasrepresentedanotherchallenge.Proposalsinvolvingglucose-oxygenfuelcellshaveheldupwellinfeasibilitystudiesbyFreitasandothers.98Anadvantageoftheglucose-oxygenapproachisthatnanomedicineapplicationscanharness theglucose,oxygen,andATPresourcesalreadyprovidedbythehumandigestivesystem.AnanoscalemotorwasrecentlycreatedusingpropellersmadeofnickelandpoweredbyanATP-basedenzyme.99However,recentprogressinimplementingMEMS-scaleandeven nanoscale hydrogen-oxygen fuel cells has provided an alternative approach, which I report onbelow.TheDebateHeatsUpInApril 2003Drexler challengedSmalley'sScientificAmerican articlewith an open letter.100 Citingtwentyyearsof researchbyhimselfandothers, the letter respondedspecifically toSmalley's fat-andsticky-fingersobjections.AsIdiscussedabove,molecularassemblerswereneverdescribedashavingfingers at all but rather relyingonprecisepositioningof reactivemolecules.Drexler citedbiologicalenzymesandribosomesasexamplesofprecisemolecularassemblyinthenaturalworld.DrexlerclosedbyquotingSmalley'sownobservation,"Whenascientistsayssomethingispossible, they'reprobablyunderestimatinghowlongitwilltake.Butiftheysayit'simpossible,they'reprobablywrong."Threemoreroundsofthisdebateoccurredin2003.SmalleyrespondedtoDrexler'sopenletterbybackingoffofhisfat-andsticky-fingersobjectionsandacknowledgingthatenzymesandribosomesdoindeed engage in the precisemolecular assembly that Smalley had earlier indicatedwas impossible.Smalleythenarguedthatbiologicalenzymesworkonlyinwaterandthatsuchwater-basedchemistryislimited to biological structures such as "wood, flesh and bone." As Drexler has stated, this, too, iserroneous.101Manyenzymes,eventhosethatordinarilyworkinwater,canalsofunctioninanhydrousorganic solvents, and some enzymes can operate on substrates in the vapor phase,with no liquid atall.102

Smalleygoesontostate(withoutanyderivationorcitations)thatenzymatic-likereactionscantakeplaceonlywithbiologicalenzymesand inchemical reactions involvingwater.This isalsomistaken.MIT professor of chemistry and biological engineering Alexander Klibanov demonstrated suchnonaqueous (not involving water) enzyme catalysis in 1984. Klibanov writes in 2003, "Clearly[Smalley's]statementsaboutnonaqueousenzymecatalysisareincorrect.Therehavebeenhundredsandperhaps thousands of papers published about nonaqueous enzyme catalysis since our first paperwaspublished20yearsago."103It'seasytoseewhybiologicalevolutionadoptedwater-basedchemistry.Waterisaveryabundantsubstance on our planet, and constitutes 70 to 90 percent of our bodies, our food, and indeed of allorganicmatter.The three-dimensionalelectricalpropertiesofwaterarequitepowerfulandcanbreakapartthestrongchemicalbondsofothercompounds.Waterisconsidered"theuniversalsolvent,"andbecauseitisinvolvedinmostofthebiochemicalpathwaysinourbodieswecanregardthechemistryoflifeonourplanetprimarilyaswaterchemistry.However,theprimarythrustofourtechnologyhasbeento develop systems that are not limited to the restrictions of biological evolution,which exclusivelyadoptedwater-based chemistry and proteins as its foundation.Biological systems can fly, but if youwanttoflyatthirtythousandfeetandathundredsorthousandsofmilesperhour,youwoulduseourmoderntechnology,notproteins.Biologicalsystemssuchashumanbrainscanrememberthingsanddocalculations,butifyouwanttododataminingonbillionsofitemsofinformation,youwouldwanttouseelectronictechnology,notunassistedhumanbrains. Smalley is ignoring thepastdecadeof researchonalternativemeansofpositioningmolecularfragments using precisely guided molecular reactions. Precisely controlled synthesis of diamondoidmaterialhasbeenextensivelystudied, including theability to removeasinglehydrogenatomfromahydrogenated diamond surface104 and the ability to add one or more carbon atoms to a diamondsurface.105 Related research supporting the feasibility of hydrogen abstraction and precisely guideddiamondoidsynthesishasbeenconductedattheMaterialsandProcessSimulationCenteratCaltech;thedepartment ofmaterials science and engineering atNorthCarolina StateUniversity; the Institute forMolecularManufacturingattheUniversityofKentucky;theU.S.NavalAcademy;andtheXeroxPaloAltoResearchCenter.106Smalleyalsoavoidsmentioningthewell-establishedSPMmentionedabove,whichusespreciselycontrolledmolecular reactions. Building on these concepts, RalphMerkle has described possible tipreactions that could involve up to four reactants.107 There is an extensive literature on site-specificreactionsthathavethepotentialtobepreciselyguidedandthuscouldbefeasibleforthetipchemistryinamolecularassembler.108Recently,many tools that gobeyondSPMsare emerging that can reliablymanipulateatomsandmolecularfragments. OnSeptember3,2003,DrexlerrespondedtoSmalley'sresponsetohis initial letterbyalludingonceagaintotheextensivebodyofliteraturethatSmalleyfailstoaddress.109Hecitedtheanalogytoamodern factory, only at a nanoscale. He cited analyses of transition-state theory indicating thatpositionalcontrolwouldbefeasibleatmegahertzfrequenciesforappropriatelyselectedreactants.Smalleyagainrespondedwithaletterthatisshortonspecificcitationsandcurrentresearchandlongonimprecisemetaphors.110Hewrites,forexample,that"muchlikeyoucan'tmakeaboyandagirlfallinlovewitheachothersimplybypushingthemtogether,youcannotmakeprecisechemistryoccurasdesiredbetweentwomolecularobjectswithsimplemechanicalmotion....[It]cannotbedonesimplyby mushing two molecular objects together." He again acknowledges that enzymes do in factaccomplish this but refuses to accept that such reactions could take place outside of a biology-likesystem:"ThisiswhyIledyou... totalkaboutrealchemistrywithrealenzymes....[A]nysuchsystemwillneedaliquidmedium.Fortheenzymesweknowabout,thatliquidwillhavetobewater,andthetypesofthingsthatcanbesynthesizedwithwateraroundcannotbemuchbroaderthanmeatandboneofbiology." Smalley's argument is of the form "We don't haveX today, thereforeX is impossible."Weencounter this class of argument repeatedly in the area of artificial intelligence. Criticswill cite the

limitationsoftoday'ssystemsasproofthatsuchlimitationsareinherentandcanneverbeovercome.Forexample,suchcriticsdisregardtheextensivelistofcontemporaryexamplesofAI(seethesection"ANarrowAISampler"onp.279)thatrepresentcommerciallyavailableworkingsystemsthatwereonlyresearchprogramsadecadeago.Thoseofuswhoattempttoprojectintothefuturebasedonwell-groundedmethodologiesareatadisadvantage.Certainfuturerealitiesmaybeinevitable,buttheyarenotyetmanifest,sotheyareeasytodeny.Asmallbodyofthoughtatthebeginningofthetwentiethcenturyinsistedthatheavier-than-airflightwasfeasible,butmainstreamskepticscouldsimplypointoutthatifitwassofeasible,whyhaditneverbeendemonstrated?Smalleyrevealsatleastpartofhismotivesattheendofhismostrecentletterwhenhewrites:AfewweeksagoIgaveatalkonnanotechnologyandenergytitled"BeaScientist,SavetheWorld"toabout700middleandhighschoolstudentsintheSpringBranchISO,alargepublicschoolsystemhereintheHoustonarea.Leadinguptomyvisitthestudentswereaskedtowriteanessayon"whyIamaNanogeek". Hundreds responded, and I had the privilege of reading the top 30 essays, picking myfavorite top5.Of theessays I read,nearlyhalfassumed that self-replicatingnanobotswerepossible,andmostweredeeplyworriedaboutwhatwouldhappenintheirfutureasthesenanobotsspreadaroundtheworld.IdidwhatIcouldtoallaytheirfears,butthereisnoquestionthatmanyoftheseyoungstershavebeentoldabedtimestorythatisdeeplytroubling.Youandpeoplearoundyouhavescaredourchildren. IwouldpointouttoSmalleythatearliercriticsalsoexpressedskepticismthateitherworldwidecommunicationnetworksorsoftwarevirusesthatwouldspreadacrossthemwerefeasible.Today,wehaveboththebenefitsandthevulnerabilitiesfromthesecapabilities.However,alongwiththedangerofsoftware viruses has emerged a technological immune system.We are obtaining farmore gain thanharmfromthislatestexampleofintertwinedpromiseandperil.Smalley'sapproachtoreassuringthepublicaboutthepotentialabuseofthisfuturetechnologyisnottherightstrategy.Bydenyingthefeasibilityofnanotechnology-basedassembly,heisalsodenyingitspotential.Denyingboththepromiseandtheperilofmolecularassemblywillultimatelybackfireandwillfail toguideresearchintheneededconstructivedirection.Bythe2020smolecularassemblywillprovidetoolstoeffectivelycombatpoverty,cleanupourenvironment,overcomedisease,extendhumanlongevity, and many other worthwhile pursuits. Like every other technology that humankind hascreated,itcanalsobeusedtoamplifyandenableourdestructiveside.It'simportantthatweapproachthistechnologyinaknowledgeablemannertogaintheprofoundbenefitsitpromises,whileavoidingitsdangers.EarlyAdoptersAlthough Drexler's concept of nanotechnology dealt primarily with precise molecular control ofmanufacturing, it has expanded to include any technology inwhich key features aremeasured by amodestnumberofnanometers(generallylessthanonehundred).Justascontemporaryelectronicshasalready quietly slipped into this realm, the area of biological and medical applications has alreadyenteredtheeraofnanoparticles,inwhichnanoscaleobjectsarebeingdevelopedtocreatemoreeffectivetestsandtreatments.Althoughnanoparticlesarecreatedusingstatisticalmanufacturingmethodsratherthanassemblers, theynonetheless relyon theiratomic-scaleproperties for theireffects.Forexample,nanoparticlesarebeingemployedinexperimentalbiologicaltestsastagsandlabelstogreatlyenhancesensitivityindetectingsubstancessuchasproteins.Magneticnanotags,forexample,canbeusedtobindwithantibodies,whichcanthenbereadusingmagneticprobeswhilestill insidethebody.Successfulexperiments have been conductedwith gold nanoparticles that are bound toDNA segments and canrapidlytestforspecificDNAsequencesinasample.Smallnanoscalebeadscalledquantumdotscanbeprogrammed with specific codes combining multiple colors, similar to a color bar code, which can

facilitatetrackingofsubstancesthroughthebody.Emergingmicrofluidicdevices,whichincorporatenanoscalechannels,canrunhundredsoftestssimultaneously on tiny samples of a given substance. These deviceswill allow extensive tests to beconductedonnearlyinvisiblesamplesofblood,forexample.Nanoscalescaffoldshavebeenusedtogrowbiologicaltissuessuchasskin.Futuretherapiescouldusethesetinyscaffoldstogrowanytypeoftissueneededforrepairsinsidethebody.Aparticularlyexcitingapplicationistoharnessnanoparticlestodelivertreatmentstospecificsitesinthebody.Nanoparticlescanguidedrugsintocellwallsandthroughtheblood-brainbarrier.ScientistsatMcGillUniversity inMontrealdemonstratedananopillwith structures in the25- to45-nanometerrange.111Thenanopillissmallenoughtopassthroughthecellwallanddeliversmedicationsdirectlytotargetedstructuresinsidethecell. Japanese scientists have created nanocages of 110 amino-acidmolecules, each holding drugmolecules.Adheredtothesurfaceofeachnanocageisapeptidethatbindstotargetsitesinthehumanbody. In one experiment scientists used a peptide that binds to a specific receptor on human livercells.112MicroCHIPSofBedford,Massachusetts,hasdevelopedacomputerizeddevicethatisimplantedundertheskinanddeliversprecisemixturesofmedicinesfromhundredsofnanoscalewellsinsidethedevice.113Futureversionsofthedeviceareexpectedtobeabletomeasurebloodlevelsofsubstancessuchasglucose.Thesystemcouldbeusedasanartificialpancreas,releasingpreciseamountsofinsulinbasedonbloodglucoseresponse.Itwouldalsobecapableofsimulatinganyotherhormone-producingorgan.Iftrialsgosmoothly,thesystemcouldbeonthemarketby2008. Anotherinnovativeproposalistoguidegoldnanoparticlestoatumorsite,thenheatthemwithinfraredbeamstodestroythecancercells.Nanoscalepackagescanbedesignedtocontaindrugs,protectthemthroughtheGItract,guidethemtospecificlocations,andthenreleasetheminsophisticatedways,including allowing them to receive instructions fromoutside thebody.Nanotherapeutics inAlachua,Florida, has developed a biodegradable polymer only several nanometers thick that uses thisapproach.114PoweringtheSingularityWeproduceabout14trillion(about1013)wattsofpowertodayintheworld.Ofthisenergyabout33percent comes from oil, 25 percent from coal, 20 percent from gas, 7 percent from nuclear fissionreactors,15percentfrombiomassandhydroelectricsources,andonly0.5percentfromrenewablesolar,wind, and geothermal technologies.115Most air pollution and significant contributions to water andotherformsofpollutionresultfromtheextraction,transportation,processing,andusesofthe78percentofourenergythatcomesfromfossilfuels.Theenergyobtainedfromoilalsocontributestogeopoliticaltensions,andthere'sthesmallmatterofits$2trillionperyearpricetagforallofthisenergy.Althoughthe industrial-era energy sources that dominate energy production today will becomemore efficientwithnewnanotechnology-basedmethodsofextraction,conversion,andtransmission,it'stherenewablecategorythatwillneedtosupportthebulkoffutureenergygrowth.By2030theprice-performanceofcomputationandcommunicationwillincreasebyafactoroftentoonehundredmillioncomparedtotoday.Othertechnologieswillalsoundergoenormousincreasesincapacity and efficiency. Energy requirements will grow far more slowly than the capacity oftechnologies,however,becauseofgreatly increasedefficiencies in theuseofenergy,whichIdiscussbelow.Aprimary implicationof thenanotechnologyrevolution is thatphysical technologies, suchasmanufacturingandenergy,willbecomegovernedbythe lawofacceleratingreturns.All technologieswillessentiallybecomeinformationtechnologies,includingenergy.Worldwideenergyrequirementshavebeenestimatedtodoubleby2030,farlessthananticipatedeconomic growth, let alone the expected growth in the capability of technology.116 The bulk of the

additional energy needed is likely to come from new nanoscale solar, wind, and geothermaltechnologies. It's important to recognize thatmostenergysources today represent solarpower inoneformoranother.Fossilfuelsrepresentstoredenergyfromtheconversionofsolarenergybyanimalsandplantsandrelated processes overmillions of years (although the theory that fossil fuels originated from livingorganismshasrecentlybeenchallenged).Buttheextractionofoilfromhigh-gradeoilwellsisatapeak,and some experts believewemay have already passed that peak. It's clear, in any case, thatwe arerapidly depleting easily accessible fossil fuels.We do have far larger fossil-fuel resources that willrequiremoresophisticated technologies toextractcleanlyandefficiently (suchascoalandshaleoil),andtheywillbepartofthefutureofenergy.Abillion-dollardemonstrationplantcalledFutureGen,nowbeing constructed, is expected to be the world's first zero-emissions energy plant based on fossilfuels.117Ratherthansimplyburncoal,asisdonetoday,the275-million-wattplantwillconvertthecoalto a synthetic gas comprising hydrogen and carbon monoxide, which will then react with steam toproducediscretestreamsofhydrogenandcarbondioxide,whichwillbesequestered.Thehydrogencanthenbeusedinfuelcellsorelseconvertedintoelectricityandwater.Keytotheplant'sdesignarenewmaterialsformembranesthatseparatehydrogenandcarbondioxide.Ourprimaryfocus,however,willbeonthedevelopmentofclean,renewable,distributed,andsafeenergy technologies made possible by nanotechnology. For the past several decades energytechnologies have been on the slow slope of the industrial era S-curve (the late stage of a specifictechnology paradigm, when the capability slowly approaches an asymptote or limit). Although thenanotechnologyrevolutionwillrequirenewenergyresources,itwillalsointroducemajornewS-curvesineveryaspectofenergy—production,storage,transmission,andutilization—bythe2020s. Let's deal with these energy requirements in reverse, starting with utilization. Because ofnanotechnology's ability to manipulate matter and energy at the extremely fine scale of atoms andmolecularfragments,theefficiencyofusingenergywillbefargreater,whichwilltranslateintolowerenergy requirements.Over the next several decades computingwillmake the transition to reversiblecomputing.(See"TheLimitsofComputation"inchapter3.)AsIdiscussed, theprimaryenergyneedfor computing with reversible logic gates is to correct occasional errors from quantum and thermaleffects.Asaresultreversiblecomputinghasthepotentialtocutenergyneedsbyasmuchasafactorofa billion, compared to nonreversible computing.Moreover, the logic gates andmemory bitswill besmaller, by at least a factor of ten in each dimension, reducing energy requirements by anotherthousand.Fullydevelopednanotechnology,therefore,willenabletheenergyrequirementsforeachbitswitchtobereducedbyaboutatrillion.Ofcourse,we'llbeincreasingtheamountofcomputationbyeven more than this, but this substantially augmented energy efficiency will largely offset thoseincreases.Manufacturingusingmolecularnanotechnologyfabricationwillalsobefarmoreenergyefficientthan contemporary manufacturing, which moves bulk materials from place to place in a relativelywasteful manner. Manufacturing today also devotes enormous energy resources to producing basicmaterials, such as steel. A typical nanofactory will be a tabletop device that can produce productsranging from computers to clothing. Larger products (such as vehicles, homes, and even additionalnanofactories)will be produced asmodular subsystems that larger robots can then assemble.Wasteheat,whichaccountsfortheprimaryenergyrequirementfornanomanufacturing,willbecapturedandrecycled. The energy requirements for nanofactories are negligible. Drexler estimates that molecularmanufacturingwillbeanenergygeneratorratherthananenergyconsumer.AccordingtoDrexler,"Amolecular manufacturing process can be driven by the chemical energy content of the feedstockmaterials, producing electrical energy as a by-product (if only to reduce the heat dissipationburden)....Using typical organic feedstock, and assuming oxidation of surplus hydrogen, reasonablyefficientmolecularmanufacturingprocessesarenetenergyproducers."118 Products can bemade from new nanotube-based and nanocompositematerials, avoiding theenormous energy used today to manufacture steel, titanium, and aluminum. Nanotechnology-based

lightingwillusesmall,cool, light-emittingdiodes,quantumdots,orother innovative lightsources toreplacehot,inefficientincandescentandfluorescentbulbs. Although the functionality and value of manufactured products will rise, product size willgenerally not increase (and in some cases, such asmost electronics, productswill get smaller). Thehigher value of manufactured goods will largely be the result of the expanding value of theirinformationcontent.Althoughtheroughly50percentdeflationrateforinformation-basedproductsandserviceswill continue throughout this period, the amount of valuable informationwill increase at anevengreater,morethanoffsettingpace. I discussed the lawof accelerating returns as applied to the communicationof information inchapter2.Theamountofinformationbeingcommunicatedwillcontinuetogrowexponentially,buttheefficiencyofcommunicationwillgrowalmostas fast,so theenergyrequirements forcommunicationwillexpandslowly.Transmissionofenergywillalsobemadefarmoreefficient.Agreatdealofenergytodayislostintransmissionduetotheheatcreatedinpowerlinesandinefficienciesinthetransportationoffuel,whichalso represent a primary environmental assault. Smalley, despite his critique of molecularnanomanufacturing,hasneverthelessbeenastrongadvocateofnewnanotechnology-basedparadigmsfor creating and transmitting energy. He describes new power-transmission lines based on carbonnanotubeswoven into longwires thatwill be far stronger, lighter, and,most important,muchmoreenergy efficient than conventional copper ones.119He also envisions using superconductingwires toreplacealuminumandcopperwiresinelectricmotorstoprovidegreaterefficiency.Smalley'svisionofananoenabledenergyfutureincludesapanoplyofnewnanotechnology-enabledcapabilities:120

·Photovoltaics:droppingthecostofsolarpanelsbyafactoroftentoonehundred.· Productionof hydrogen: new technologies for efficientlyproducinghydrogen fromwater andsunlight.

·Hydrogenstorage:light,strongmaterialsforstoringhydrogenforfuelcells.·Fuelcells:droppingthecostoffuelcellsbyafactoroftentoonehundred.· Batteriesandsupercapacitorstostoreenergy:improvingenergystoragedensitiesbyafactoroftentoonehundred.

· Improving the efficiency of vehicles such as cars and planes through strong and lightnanomaterials.

·Strong,lightnanomaterialsforcreatinglarge-scaleenergy-harvestingsystemsinspace,includingonthemoon.

· Robotsusingnanoscaleelectronicswithartificialintelligencetoautomaticallyproduceenergy-generatingstructuresinspaceandonthemoon.

·Newnanomaterialcoatingstogreatlyreducethecostofdeepdrilling.·Nanocatalyststoobtaingreaterenergyyieldsfromcoal,atveryhightemperatures.· Nanofilters to capture the soot created from high-energy coal extraction. The soot is mostlycarbon,whichisabasicbuildingblockformostnanotechnologydesigns.

· Newmaterials to enable hot, dry rock geothermal-energy sources (converting the heat of theEarth'shotcoreintoenergy).

Another option for energy transmission iswireless transmission bymicrowaves.Thismethodwouldbeespeciallywellsuited toefficientlybeamenergycreatedinspacebygiantsolarpanels(seebelow).121 The Millennium Project of the American Council for the United Nations Universityenvisionsmicrowaveenergytransmissionasakeyaspectof"aclean,abundantenergyfuture."122 Energystorage today ishighlycentralized,which representsakeyvulnerability in that liquid-natural-gastanksandotherstoragefacilitiesaresubjecttoterroristattacks,withpotentiallycatastrophiceffects.Oiltrucksandshipsareequallyexposed.Theemergingparadigmforenergystoragewillbefuelcells,whichwillultimatelybewidelydistributedthroughoutourinfrastructure,anotherexampleofthetrendfrominefficientandvulnerablecentralizedfacilitiestoanefficientandstabledistributedsystem.

Hydrogen-oxygen fuel cells, with hydrogen provided by methanol and other safe forms ofhydrogen-richfuel,havemadesubstantialprogressinrecentyears.AsmallcompanyinMassachusetts,IntegratedFuelCellTechnologies,hasdemonstratedaMEMS(MicroElectronicMechanicalSystem)-based fuel cell.123 Each postage-stamp-size device contains thousands of microscopic fuel cells andincludesthefuellinesandelectroniccontrols.NECplanstointroducefuelcellsbasedonnanotubesinthe near future for notebook computers and other portable electronics.124 It claims its small powersourceswillrundevicesforuptofortyhoursatatime.Toshibaisalsopreparingfuelcellsforportableelectronicdevices.125Largerfuelcellsforpoweringappliances,vehicles,andevenhomesarealsomakingimpressiveadvances. A 2004 report by the U.S. Department of Energy concluded that nanobased technologiescould facilitate every aspectof ahydrogen fuel cell-poweredcar.126 For example, hydrogenmust bestoredinstrongbutlighttanksthatcanwithstandveryhighpressure.Nanomaterialssuchasnanotubesandnanocompositescouldprovidetherequisitematerialforsuchcontainers.Thereportenvisionsfuelcellsthatproducepowertwiceasefficientlyasgasoline-basedengines,producingonlywateraswaste. Manycontemporaryfuel-celldesignsusemethanol toprovidehydrogen,which thencombineswiththeoxygenintheairtoproducewaterandenergy.Methanol(woodalcohol),however,isdifficulttohandle,andintroducessafetyconcernsbecauseofitstoxicityandflammability.ResearchersfromSt.Louis University have demonstrated a stable fuel cell that uses ordinary ethanol (drinkable grainalcohol).127 This device employs an enzyme called dehydrogenase that removes hydrogen ions fromalcohol, which subsequently reactwith the oxygen in the air to produce power. The cell apparentlyworkswith almost any formof drinkable alcohol. "Wehave run it on various types," reportedNickAkers, a graduate studentwhohasworkedon theproject. "It didn't like carbonatedbeer anddoesn'tseemfondofwine,butanyotherworksfine." Scientists at the University of Texas have developed a nanobot-size fuel cell that produceselectricity directly from the glucose-oxygen reaction in human blood.128 Called a "vampire bot" bycommentators, the cell produces electricity sufficient topower conventional electronics andcouldbeusedforfutureblood-bornenanobots.Japanesescientistspursuingasimilarprojectestimatedthattheirsystem had the theoretical potential to produce a peak of one hundredwatts from the blood of oneperson, although implantable deviceswould use far less. (A newspaper in Sydney observed that theprojectprovidedabasisforthepremiseintheMatrixmoviesofusinghumansasbatteries.)129AnotherapproachtoconvertingtheabundantsugarfoundinthenaturalworldintoelectricityhasbeendemonstratedbySwadesK.ChaudhuriandDerekR.LovleyattheUniversityofMassachusetts.Theirfuelcell,whichincorporatesactualmicrobes(theRhodoferaxferrireducensbacterium),boastsaremarkable81percentefficiencyandusesalmostnoenergy in its idlingmode.Thebacteriaproduceelectricitydirectly fromglucosewithnounstable intermediaryby-products.Thebacteriaalsouse thesugarfueltoreproduce,therebyreplenishingthemselves,resultinginstableandcontinuousproductionofelectricalenergy.Experimentswithothertypesofsugarssuchasfructose,sucrose,andxylosewereequallysuccessful.Fuelcellsbasedon this researchcouldutilize theactualbacteriaor,alternatively,directly apply the chemical reactions that the bacteria facilitate. In addition to powering nanobots insugar-rich blood, these devices have the potential to produce energy from industrial and agriculturalwasteproducts.Nanotubeshavealsodemonstratedthepromiseofstoringenergyasnanoscalebatteries,whichmaycompetewithnanoengineeredfuelcells.130Thisextendsfurthertheremarkableversatilityofnanotubes,which have already revealed their prowess in providing extremely efficient computation,communicationof information,and transmissionofelectricalpower,aswellas increatingextremelystrongstructuralmaterials.Themostpromisingapproachtonanomaterials-enabledenergyisfromsolarpower,whichhasthepotentialtoprovidethebulkofourfutureenergyneedsinacompletelyrenewable,emission-free,anddistributed manner. The sunlight input to a solar panel is free. At about 1017 watts, or about tenthousand timesmore energy than the1013watts currently consumed by human civilization, the total

energy from sunlight falling on the Earth is more than sufficient to provide for our needs.131 Asmentioned above, despite the enormous increases in computation and communication over the nextquarter century and the resulting economic growth, the far greater energy efficiencies ofnanotechnology imply that energy requirements will increase onlymodestly to around thirty trillionwatts(3Î1013)by2030.Wecouldmeetthisentireenergyneedwithsolarpoweraloneifwecapturedonly0.0003(threeten-thousandths)ofthesun'senergyasithitstheEarth. It's interesting to compare these figures to the total metabolic energy output of all humans,estimatedbyRobertFreitasat1012watts,andthatofallvegetationonEarth,at1014watts.Freitasalsoestimates that the amount of energywe could produce and usewithout disrupting the global energybalance required to maintain current biological ecology (referred to by climatologists as the"hypsithermallimit")isaround1015watts.Thiswouldallowaverysubstantialnumberofnanobotsperperson for intelligence enhancement and medical purposes, as well as other applications, such asprovidingenergyandcleaninguptheenvironment.Estimatingaglobalpopulationofaroundtenbillion(1010)humans,Freitasestimatesaround1016(tenthousandtrillion)nanobotsforeachhumanwouldbeacceptablewithin this limit.132Wewould need only 1011 nanobots (tenmillionths of this limit) perpersontoplaceoneineveryneuron. Bythetimewehavetechnologyofthisscale,wewillalsobeabletoapplynanotechnologytorecycleenergybycapturingat leastasignificantportionof theheatgeneratedbynanobotsandothernanomachinery and converting that heat back into energy. Themost effectiveway to do thiswouldprobably be to build the energy recycling into the nanobot itself.133 This is similar to the idea ofreversible logic gates in computation, in which each logic gate essentially immediately recycles theenergyitusedforitslastcomputation.Wecouldalsopullcarbondioxideoutoftheatmospheretoprovidethecarbonfornanomachinery,which would reverse the increase in carbon dioxide resulting from our current industrial-eratechnologies.Wemight,however,wanttobeparticularlycautiousaboutdoingmorethanreversingtheincreaseoverthepastseveraldecades,lestwereplaceglobalwarmingwithglobalcooling.Solarpanelshavetodatebeenrelativelyinefficientandexpensive,butthetechnologyisrapidlyimproving. The efficiency of converting solar energy to electricity has steadily advanced for siliconphotovoltaiccellsfromaround4percentin1952to24percentin1992.134Currentmultilayercellsnowprovide around 34 percent efficiency. A recent analysis of applying nanocrystals to solar-energyconversionindicatesthatefficienciesabove60percentappeartobefeasible.135

Today solar power costs an estimated $2.75 per watt.136 Several companies are developingnanoscale solar cells and hope to bring the cost of solar power below that of other energy sources.Industry sources indicate thatonce solarpower fallsbelow$1.00perwatt, itwillbecompetitive fordirectlysupplyingelectricitytothenation'spowergrid.Nanosolarhasadesignbasedontitaniumoxidenanoparticlesthatcanbemass-producedonverythinflexiblefilms.CEOMartinRoscheisenestimatesthathistechnologyhasthepotentialtobringdownsolar-powercoststoaroundfiftycentsperwattby2006, lower than that of natural gas.137 CompetitorsNanosys andKonarka have similar projections.Whether or not these business plans panout, oncewehaveMNT (molecular nanotechnology)-basedmanufacturing, we will be able to produce solar panels (and almost everything else) extremelyinexpensively,essentiallyatthecostofrawmaterials,ofwhichinexpensivecarbonistheprimaryone.Atanestimatedthicknessofseveralmicrons,solarpanelscouldultimatelybeasinexpensiveasapennypersquaremeter.Wecouldplaceefficientsolarpanelsonthemajorityofhuman-madesurfaces,suchasbuildingsandvehicles,andevenincorporatethemintoclothingforpoweringmobiledevices.A0.0003conversionrateforsolarenergyshouldbequitefeasible,therefore,andrelativelyinexpensive. Terrestrial surfaces could be augmented by huge solar panels in space.ASpaceSolarPowersatellitealreadydesignedbyNASAcouldconvertsunlightinIspacetoelectricityandbeamittoEarthbymicrowave. Each such satellite could provide billions of watts of electricity, enough for tens ofthousandsofhomes.138Withcirca-2029MNTmanufacturing,wecouldproduce solarpanelsofvast

sizedirectlyinorbitaroundtheEarth,requiringonlytheshipmentoftherawmaterialstospacestations,possibly via the planned Space Elevator, a thin ribbon, extending from a shipborne anchor to acounterweight well beyond geosynchronous orbit, made out of a material called carbon nanotubecomposite.139 Desktop fusion also remains a possibility. Scientists atOakRidgeNational Laboratory usedultrasonicsoundwaves toshakea liquidsolvent,causinggasbubbles tobecomesocompressed theyachievedtemperaturesofmillionsofdegrees,resultinginthenuclearfusionofhydrogenatomsandthecreationofenergy.140Despitethebroadskepticismovertheoriginalreportsofcoldfusionin1989,thisultrasonicmethodhasbeenwarmlyreceivedbysomepeerreviewers.141However,notenoughisknownabout the practicality of the technique, so its future role in energy production remains a matter ofspeculation.ApplicationsofNanotechnologytotheEnvironmentEmergingnanotechnologycapabilitiespromiseaprofoundimpactontheenvironment.Thisincludesthecreation of newmanufacturing and processing technologies thatwill dramatically reduce undesirableemissions,aswellasremediatingthepriorimpactofindustrial-agepollution.Ofcourse,providingforourenergyneedswithnanotechnology-enabledrenewable,cleanresourcessuchasnanosolarpanels,asIdiscussedabove,willclearlybealeadingeffortinthisdirection. Bybuildingparticles anddevices at themolecular scale, not only is sizegreatly reduced andsurface area increased, but new electrical, chemical, and biological properties are introduced.Nanotechnology will eventually provide us with a vastly expanded toolkit for improved catalysis,chemicalandatomicbonding,sensing,andmechanicalmanipulation,nottomentionintelligentcontrolthroughenhancedmicroelectronics.Ultimatelywewillredesignallofourindustrialprocessestoachievetheirintendedresultswithminimalconsequences,suchasunwantedby-productsandtheirintroductionintotheenvironment.Wediscussed in the previous section a comparable trend in biotechnology: intelligently designedpharmaceuticalagentsthatperformhighlytargetedbiochemicalinterventionswithgreatlycurtailedsideeffects.Indeed,thecreationofdesignedmoleculesthroughnanotechnologywillitselfgreatlyacceleratethebiotechnologyrevolution. Contemporary nanotechnology research and development involves relatively simple "devices"such as nanoparticles, molecules created through nanolayers, and nanotubes. Nanoparticles, whichcomprise between tens and thousands of atoms, are generally crystalline in nature and use crystal-growing techniques, since we do not yet have the means for precise nanomolecular manufacturing.Nanostructuresconsistofmultiplelayersthatself-assemble.Suchstructuresaretypicallyheldtogetherwithhydrogenorcarbonbondingandotheratomicforces.BiologicalstructuressuchascellmembranesandDNAitselfarenaturalexamplesofmultilayernanostructures.Aswithallnewtechnologies,thereisadownsidetonanoparticles:theintroductionofnewformsoftoxinsandotherunanticipatedinteractionswiththeenvironmentandlife.Manytoxicmaterials,suchasgalliumarsenide,arealreadyenteringtheecosystemthroughdiscardedelectronicproducts.Thesamepropertiesthatenablenanoparticlesandnanolayerstodeliverhighlytargetedbeneficialresultscanalsoleadtounforeseenreactions,particularlywithbiologicalsystemssuchasourfoodsupplyandourownbodies.Althoughexistingregulationsmayinmanycasesbeeffectiveincontrollingthem,theoverridingconcernisourlackofknowledgeaboutawiderangeofunexploredinteractions. Nonetheless,hundredsofprojectshavebegunapplyingnanotechnologytoenhancingindustrialprocessesandexplicitlyaddressexistingformsofpollution.Afewexamples:

· There is extensive investigation of the use of nanoparticles for treating, deactivating, andremovingawidevarietyofenvironmentaltoxins.Thenanoparticleformsofoxidants,reductants,and other active materials have shown the ability to transform a wide range of undesirable

substances. Nanoparticles activated by light (for example, forms of titanium dioxide and zincoxide) are able to bind and remove organic toxins and have low toxicity themselves.142 Inparticular, zinc oxide nanoparticles provide a particularly powerful catalyst for detoxifyingchlorinatedphenols.Thesenanoparticlesactasbothsensorsandcatalystsandcanbedesignedtotransformonlytargetedcontaminants.

·Nanofiltrationmembranesforwaterpurificationprovidedramaticallyimprovedremovaloffine-particle contaminants, compared to conventional methods of using sedimentation basins andwastewater clarifiers. Nanoparticles with designed catalysis are capable of absorbing andremoving impurities. By usingmagnetic separation, these nanomaterials can be reused,whichprevents them from becoming contaminants themselves. As one of many examples, considernanoscale aluminosilicate molecular sieves called zeolites, which are being developed forcontrolled oxidation of hydrocarbons (for example, converting toluene to nontoxicbenzaldehyde).143 This method requires less energy and reduces the volume of inefficientphotoreactionsandwasteproducts.

·Extensiveresearchisunderwaytodevelopnanoproducedcrystallinematerialsforcatalystsandcatalystsupportsinthechemicalindustry.Thesecatalystshavethepotentialtoimprovechemicalyields,reducetoxicby-products,andremovecontaminants.144Forexample,thematerialMCM-41isnowusedbytheoilindustrytoremoveultrafinecontaminantsthatotherpollution-reductionmethodsmiss.

·It'sestimatedthatthewidespreaduseofnanocompositesforstructuralmaterialinautomobileswould reduce gasoline consumption by 1.5 billion liters per year,which in turnwould reducecarbondioxideemissionsbyfivebillionkilogramsperyear,amongotherenvironmentalbenefits.

· Nanorobotics canbeused to assistwithnuclear-wastemanagement.Nanofilters can separateisotopes when processing nuclear fuel. Nanofluids can improve the effectiveness of coolingnuclearreactors.

· Applying nanotechnology to home and industrial lighting could reduce both the need forelectricityandanestimatedtwohundredmilliontonsofcarbonemissionsperyear.145

·Self-assemblingelectronicdevices(forexample,self-organizingbiopolymers),ifperfected,willrequire less energy to manufacture and use and will produce fewer toxic by-products thanconventionalsemiconductor-manufacturingmethods.

· New computer displays using nanotube-based field-emission displays (FEDs) will providesuperiordisplayspecificationswhileeliminatingtheheavymetalsandothertoxicmaterialsusedinconventionaldisplays.

·Bimetallicnanoparticles(suchasiron/palladiumoriron/silver)canserveaseffectivereductantsandcatalystsforPCBs,pesticides,andhalogenatedorganicsolvents.146

·Nanotubesappeartobeeffectiveabsorbentsfordioxinsandhaveperformedsignificantlybetteratthisthantraditionalactivatedcarbon.147

Thisisasmallsampleofcontemporaryresearchonnanotechnologyapplicationswithpotentiallybeneficialimpactontheenvironment.Oncewecangobeyondsimplenanoparticlesandnanolayersandcreatemore complex systems through precisely controlledmolecular nanoassembly,wewill be in apositiontocreatemassivenumbersoftinyintelligentdevicescapableofcarryingoutrelativelycomplextasks.Cleaninguptheenvironmentwillcertainlybeoneofthosemissions.NanobotsintheBloodstream

Nanotechnologyhasgivenusthetools ... toplaywiththeultimatetoyboxofnature—atomsand molecules. Everything is made from it....The possibilities to create new things appearlimitless.

—NOBELISTHORSTSTORMERTheneteffectofthesenanomedicalinterventionswillbethecontinuingarrestofallbiologicalaging, alongwith the reduction of current biological age towhatever new biological age isdeemeddesirablebythepatient,severingforeverthelinkbetweencalendartimeandbiologicalhealth. Such interventions may become commonplace several decades from today. Usingannualcheckupsandcleanouts,andsomeoccasionalmajorrepairs,yourbiologicalagecouldbe restored once a year to themore or less constant physiological age that you select. Youmightstilleventuallydieofaccidentalcauses,butyou'llliveatleasttentimeslongerthanyoudonow.

—ROBERTA.FREITASJR.148A prime example of the application of precise molecular control in manufacturing will be thedeploymentofbillionsor trillionsofnanobots: small robots thesizeofhumanbloodcellsor smallerthatcantravelinsidethebloodstream.Thisnotionisnotasfuturisticasitmaysound;successfulanimalexperimentshavebeenconductedusing thisconcept, andmanysuchmicro scaledevicesarealreadyworking in animals. At least four major conferences on BioMEMS (Biological Micro ElectronicMechanicalSystems)dealwithdevicestobeusedinthehumanbloodstream.149 Consider several examples of nanobot technology,which, based onminiaturization and cost-reduction trends,will be feasiblewithin about twenty-five years. In addition to scanning the humanbrain to facilitate its reverse engineering, these nanobots will be able to perform a broad variety ofdiagnosticandtherapeuticfunctions. Robert A. Freitas Jr.—a pioneering nanotechnology theorist and leading proponent ofnanomedicine (reconfiguring our biological systems through engineering on a molecular scale), andauthor of a book with that title150—has designed robotic replacements for human blood cells thatperform hundreds or thousands of times more effectively than their biological counterparts. WithFreitas'srespirocytes(roboticredbloodcells)arunnercoulddoanOlympicsprintforfifteenminuteswithout taking a breath.151 Freitas's robotic macrophages, called "microbivores," will be far moreeffectivethanourwhitebloodcellsatcombatingpathogens.152HisDNA-repairrobotwouldbeabletomendDNAtranscriptionerrorsandevenimplementneededDNAchanges.Othermedicalrobotshehasdesignedcanserveascleaners, removingunwanteddebrisandchemicals (suchasprions,malformedproteins,andprotofibrils)fromindividualhumancells. Freitasprovidesdetailedconceptualdesigns for awide rangeofmedicalnanorobots (Freitas'spreferredterm)aswellasareviewofnumeroussolutionstothevarieddesignchallengesinvolvedincreating them.Forexample,heprovidesaboutadozenapproaches todirectedandguidedmotion.153somebasedonbiologicaldesignssuchaspropulsivecilia.Idiscusstheseapplicationsinmoredetailinthenextchapter. GeorgeWhitesidescomplained inScientificAmerican that "for nanoscaleobjects, even if onecould fabricate a propeller, a new and serious problem would emerge: random jarring by watermolecules.Thesewatermoleculeswouldbesmaller thanananosubmarinebutnotmuchsmaller."154Whitesides's analysis is based on misconceptions. All medical nanobot designs, including those ofFreitas, are at least ten thousand times larger than awatermolecule.Analyses byFreitas and othersshowtheimpactof theBrownianmotionofadjacentmolecules tobeinsignificant.Indeed,nanoscalemedicalrobotswillbethousandsoftimesmorestableandprecisethanbloodcellsorbacteria.155Itshouldalsobepointedoutthatmedicalnanobotswillnotrequiremuchoftheextensiveoverheadbiological cells need tomaintainmetabolic processes such as digestion and respiration.Nor do theyneedtosupportbiologicalreproductivesystems. AlthoughFreitas's conceptual designs are a couple of decades away, substantial progress hasalreadybeenmadeonbloodstream-baseddevices.Forexample,aresearcherattheUniversityofIllinois

atChicagohascuredtype1diabetesinratswithananoengineereddevicethatincorporatespancreaticisletcells.156Thedevicehasseven-nanometerporesthatletinsulinoutbutwon'tletintheantibodiesthatdestroythesecells.Therearemanyotherinnovativeprojectsofthistypealreadyunderway.MOLLY2004:Okay,soI'llhaveallthesenanobotsinmybloodstream.Asidefrombeingabletositat

thebottomofmypoolforhours,whatelseisthisgoingtodoforme?RAY:Itwill keepyouhealthy.They'lldestroypathogens suchasbacteria, viruses,andcancercells,

and they won't be subject to the various pitfalls of the immune system, such as autoimmunereactions.Unlikeyourbiologicalimmunesystem,ifyoudon'tlikewhatthenanobotsaredoing,youcantellthemtodosomethingdifferent.

MOLLY2004: Youmean, sendmy nanobots an e-mail? Like,Hey, nanobots, stop destroying thosebacteriainmyintestinesbecausethey'reactuallygoodformydigestion?

RAY: Yes, good example. The nanobots will be under our control. They'll communicate with oneanotherandwiththeInternet.Eventodaywehaveneuralimplants(forexample,forParkinson'sdisease)thatallowthepatienttodownloadnewsoftwareintothem.

MOLLY2004:Thatkindofmakesthesoftware-virusissuealotmoreserious,doesn'tit?Rightnow,ifIget hit with a bad software virus, I may have to run a virus-cleansing program and loadmybackupfiles,butifnanobotsinmybloodstreamgetaroguemessage,theymaystartdestroyingmybloodcells.

RAY:Well,that'sanotherreasonyou'llprobablywantroboticbloodcells,butyourpointiswelltaken.However, it'snotanewissue.Evenin2004,wealreadyhavemission-criticalsoftwaresystemsthatrunintensive-careunits,manage911emergencysystems,controlnuclear-powerplants,landairplanes,andguidecruisemissiles.Sosoftwareintegrityisalreadyofcriticalimportance.

MOLLY2004:True,buttheideaofsoftwarerunninginmybodyandbrainseemsmoredaunting.Onmypersonal computer, I getmore thanonehundred spammessagesaday, at least severalofwhich containmalicious software viruses. I'm not real comfortablewith nanobots inmy bodygettingsoftwareviruses.

RAY: You're thinking in terms of conventional Internet access. With VPNs (private networks), wealready have the means today to create secure firewalls—otherwise, contemporary mission-critical systems would be impossible. They do work reasonably well, and Internet securitytechnologywillcontinuetoevolve.

MOLLY2004:Ithinksomepeoplewouldtakeissuewithyourconfidenceinfirewalls.RAY:They'renotperfect,true,andtheyneverwillbe,butwehaveanothercoupledecadesbeforewe'll

haveextensivesoftwarerunninginourbodiesandbrains.MOLLY2004:Okay,buttheviruswriterswillbeimprovingtheircraftaswell.RAY:It'sgoingtobeanervousstandoff,noquestionaboutit.Butthebenefittodayclearlyoutweighs

thedamage.MOLLY2004:Howclearisthat?RAY:Well,nooneisseriouslyarguingweshoulddoawaywiththeInternetbecausesoftwareviruses

aresuchabigproblem.MOLLY2004:I'llgiveyouthat.RAY:When nanotechnology is mature, it's going to solve the problems of biology by overcoming

biological pathogens, removing toxins, correctingDNA errors, and reversing other sources ofaging.Wewill then have to contend with new dangers that it introduces, just as the Internetintroducedthedangerofsoftwareviruses.Thesenewpitfallswill includethepotential forself-replicating nanotechnology getting out of control, as well as the integrity of the softwarecontrollingthesepowerful,distributednanobots.

MOLLY2004:Didyousayreverseaging?RAY:Iseeyou'realreadypickinguponakeybenefit.MOLLY2004:Sohowarethenanobotsgoingtodothat?RAY:We'llactuallyaccomplishmostofthatwithbiotechnology,methodssuchasRNAinterferencefor

turningoffdestructivegenes,genetherapyforchangingyourgeneticcode, therapeuticcloningforregeneratingyourcellsandtissues,smartdrugstoreprogramyourmetabolicpathways,andmany other emerging techniques. But whatever biotechnology doesn't get around toaccomplishing,we'llhavethemeanstodowithnanotechnology.

MOLLY2004:Suchas?RAY:Nanobotswillbeable to travel through thebloodstream, thengo inandaroundourcellsand

performvariousservices,suchasremovingtoxins,sweepingoutdebris,correctingDNAerrors,repairing and restoring cell membranes, reversing atherosclerosis, modifying the levels ofhormones,neurotransmitters,andothermetabolicchemicals,andamyriadofother tasks.Foreachagingprocess,wecandescribeameans fornanobots toreverse theprocess,downto thelevelofindividualcells,cellcomponents,andmolecules.

MOLLY2004:SoI'llstayyoungindefinitely?RAY:That'stheidea.MOLLY2004:WhendidyousayIcouldgetthese?RAY:Ithoughtyouwereworriedaboutnanobotfirewalls.MOLLY2004:Yeah,well,I'vegottimetoworryaboutthat.Sowhatwasthattimeframeagain?RAY:Abouttwentytotwenty-fiveyears.MOLLY2004:I'mtwenty-fivenow,soI'llagetoaboutforty-fiveandthenstaythere?RAY:No,that'snotexactlytheidea.Youcanslowdownagingtoacrawlrightnowbyadoptingthe

knowledgewealreadyhave.Withintentotwentyyears,thebiotechnologyrevolutionwillprovidefarmorepowerfulmeanstostopandinmanycasesreverseeachdiseaseandagingprocess.Andit's not like nothing is going to happen in themeantime.Each year,we'll havemorepowerfultechniques,andtheprocesswillaccelerate.Thennanotechnologywillfinishthejob.

MOLLY 2004: Yes, of course, it's hard for you to get out a sentence without using the word"accelerate."SowhatbiologicalageamIgoingtogetto?

RAY:Ithinkyou'llsettlesomewhereinyourthirtiesandstaythereforawhile.MOLLY2004:Thirties sounds pretty good. I think a slightlymoremature age than twenty-five is a

goodideaanyway.Butwhatdoyoumean"forawhile"?RAY:Stoppingandreversingagingisonlythebeginning.Usingnanobotsforhealthandlongevityis

justtheearlyadoptionphaseofintroducingnanotechnologyandintelligentcomputationintoourbodiesandbrains.Themoreprofoundimplicationis thatwe'llaugmentourthinkingprocesseswith nanobots that communicate with one another and with our biological neurons. Oncenonbiologicalintelligencegetsafoothold,sotospeak,inourbrains,itwillbesubjecttothelawofacceleratingreturnsandexpandexponentially.Ourbiologicalthinking,ontheotherhand,isbasicallystuck.

MOLLY2004:Thereyougoagainwiththingsaccelerating,butwhenthisreallygetsgoing,thinkingwithbiologicalneuronswillbeprettytrivialincomparison.

RAY:That'safairstatement.MOLLY2004:So,MissMollyofthefuture,whendidIdropmybiologicalbodyandbrain?MOLLY2104:Well,youdon'treallywantmetospelloutyourfuture,doyou?Andanywayit'sactually

notastraightforwardquestion.MOLLY2004:How'sthat?MOLLY2104: In the 2040s we developed themeans to instantly create new portions of ourselves,

either biological or nonbiological. It became apparent that our true nature was a pattern ofinformation,butwestillneededtomanifestourselvesinsomephysicalform.However,wecouldquicklychangethatphysicalform.

MOLLY2004:By?MOLLY2104: By applying new high-speed MNT manufacturing. So we could readily and rapidly

redesign our physical instantiation. So I could have a biological body at one time and not atanother,thenhaveitagain,thenchangeit,andsoon.

MOLLY2004:IthinkI'mfollowingthis.MOLLY2104:ThepointisthatIcouldhavemybiologicalbrainand/orbodyornothaveit.It'snota

matterofdroppinganything,becausewecanalwaysgetbacksomethingwedrop.MOLLY2004:Soyou'restilldoingthis?MOLLY 2104: Some people still do this, but now in 2104 it's a bit anachronistic. I mean, the

simulations of biology are totally indistinguishable from actual biology, so why bother withphysicalinstantiations?MOLLY2004:Yeah,it'smessyisn'tit?

MOLLY2104:I'llsay.MOLLY2004:Idohavetosaythatitseemsstrangetobeabletochangeyourphysicalembodiment.I

mean,where'syour—my—continuity?MOLLY2104: It's the same as your continuity in 2004. You're changing your particles all the time

also.It'sjustyourpatternofinformationthathascontinuity.MOLLY2004:Butin2104you'reabletochangeyourpatternofinformationquicklyalso.Ican'tdo

thatyet.MOLLY2104:It'sreallynotthatdifferent.Youchangeyourpattern-yourmemory,skills,experiences,

evenpersonalityovertime—butthereisacontinuity,acorethatchangesonlygradually.MOLLY2004:But I thought you could change yourappearanceandpersonality dramatically in an

instant?MOLLY2104:Yes,but that's justasurfacemanifestation.My truecorechangesonlygradually,justlikewhenIwasyouin2004.

MOLLY2004:Well, there are lots of times when I'd be delighted to instantly change my surfaceappearance.

Robotics:StrongAI

ConsideranotherargumentputforthbyTuring.Sofarwehaveconstructedonlyfairlysimpleand predictable artifacts. When we increase the complexity of our machines, there may,perhaps,besurprises instore forus.Hedrawsaparallelwitha fissionpile.Belowacertain"critical" size,nothingmuchhappens:but above thecritical size, the sparksbegin to fly.Sotoo, perhaps, with brains andmachines.Most brains and all machines are, at present "sub-critical"—theyreact to incomingstimuli inastodgyanduninterestingway,haveno ideasoftheirown,canproduceonlystockresponses—butafewbrainsatpresent,andpossiblysomemachines in the future, are super-critical, and scintillate on their own account. Turing issuggestingthatitisonlyamatterofcomplexity,andthataboveacertainlevelofcomplexityaqualitativedifferenceappears,sothat"super-critical"machineswillbequiteunlikethesimpleoneshithertoenvisaged.

—J.R.LUCAS,OXFORDPHILOSOPHER,INHIS1961ESSAY"MINDS,MACHINES,ANDGÖDEL"157

Given that superintelligence will one day be technologically feasible, will people choose todevelop it? This question can pretty confidently be answered in the affirmative. Associatedwith every step along the road to superintelligence are enormous economic payoffs. Thecomputer industry investshugesums in thenextgenerationofhardwareandsoftware,and itwillcontinuedoingsoaslongasthereisacompetitivepressureandprofitstobemade.Peoplewant better computers and smarter software, and theywant the benefits thesemachines canhelp produce. Better medical drugs; relief for humans from the need to perform boring ordangerousjobs;entertainment—thereisnoendtothelistofconsumer-benefits.Thereisalsoastrongmilitarymotivetodevelopartificialintelligence.Andnowhereonthepathisthereanynaturalstoppingpointwheretechnophobicscouldplausiblyargue"hitherbutnotfurther."

—NICK BOSTROM, “HOW LONG BEFORE SUPERINTELLIGENCE?”1997

Itishardtothinkofanyproblemthatasuperintelligencecouldnoteithersolveoratleasthelpussolve.Disease,poverty,environmentaldestruction,unnecessarysufferingofallkinds:thesearethingsthatasuperintelligenceequippedwithadvancednanotechnologywouldbecapableofeliminating. Additionally, a superintelligence could give us indefinite lifespan, either bystoppingandreversingtheagingprocessthroughtheuseofnanomedicine,orbyofferingustheoptiontouploadourselves.Asuperintelligencecouldalsocreateopportunitiesforustovastlyincrease our own intellectual and emotional capabilities, and it could assist us in creating ahighly appealing experiential world in which we could live lives devoted to joyfulgameplaying,relatingtoeachother,experiencing,personalgrowth,andtolivingclosertoourideals.

—NICK BOSTROM, “ETHICAL ISSUES IN ADVANCED ARTIFICIALINTELLIGENCE,"2003

Willrobotsinherittheearth?Yes,buttheywillbeourchildren.

—MARVINMINSKY,1995Of the three primary revolutions underlying the Singularity (G,N, andR), themost profound is R,whichreferstothecreationofnonbiologicalintelligencethatexceedsthatofunenhancedhumans.Amoreintelligentprocesswillinherentlyoutcompeteonethatislessintelligent,makingintelligencethemostpowerfulforceintheuniverse. While theR inGNR stands for robotics, the real issue involved here is strongAI (artificialintelligence that exceeds human intelligence). The standard reason for emphasizing robotics in thisformulationisthatintelligenceneedsanembodiment,aphysicalpresence,toaffecttheworld.Idisagreewiththeemphasisonphysicalpresence,however,forIbelievethatthecentralconcernisintelligence.Intelligencewill inherently find away to influence theworld, including creating its ownmeans forembodimentandphysicalmanipulation.Furthermore,wecanincludephysicalskillsasafundamentalpartofintelligence;alargeportionofthehumanbrain(thecerebellum,comprisingmorethanhalfourneurons),forexample,isdevotedtocoordinatingourskillsandmuscles. Artificial intelligence at human levelswill necessarily greatly exceed human intelligence forseveralreasons.AsIpointedoutearlier,machinescanreadilyshare theirknowledge.Asunenhancedhumans we do not have the means of sharing the vast patterns of interneuronal connections andneurotransmitter-concentration levels that comprise our learning, knowledge, and skills, other thanthroughslow,language-basedcommunication.Ofcourse,eventhismethodofcommunicationhasbeenverybeneficial, as it has distinguishedus fromother animals andhas been an enabling factor in thecreationoftechnology.Humanskillsareabletodeveloponlyinwaysthathavebeenevolutionarilyencouraged.Thoseskills, which are primarily based on massively parallel pattern recognition, provide proficiency forcertain tasks, such as distinguishing faces, identifyingobjects, and recognizing language sounds.Butthey'renotsuitedformanyothers,suchasdeterminingpatternsinfinancialdata.Oncewefullymasterpattern-recognitionparadigms,machinemethodscanapplythesetechniquestoanytypeofpattern.158 Machinescanpooltheirresourcesinwaysthathumanscannot.Althoughteamsofhumanscanaccomplishbothphysicalandmentalfeatsthatindividualhumanscannotachieve,machinescanmoreeasilyandreadilyaggregatetheircomputational,memory,andcommunicationsresources.Asdiscussedearlier, the Internet is evolving into a worldwide grid of computing resources that can instantly bebroughttogethertoformmassivesupercomputers. Machines have exacting memories. Contemporary computers can master billions of factsaccurately,acapabilitythatisdoublingeveryyear.159Theunderlyingspeedandprice-performanceofcomputingitselfisdoublingeveryyear,andtherateofdoublingisitselfaccelerating. As human knowledgemigrates to theWeb, machines will be able to read, understand, andsynthesize all human-machine information. The last time a biological human was able to grasp all

humanscientificknowledgewashundredsofyearsago.Anotheradvantageofmachineintelligenceisthatitcanconsistentlyperformatpeaklevelsandcan combine peak skills. Among humans one personmay have mastered music composition, whileanothermayhavemasteredtransistordesign,butgiventhefixedarchitectureofourbrainswedonothave the capacity (or the time) to develop and utilize the highest level of skill in every increasinglyspecialized area.Humans also vary a great deal in a particular skill, so thatwhenwe speak, say, ofhuman levels of composing music, do we mean Beethoven, or do we mean the average person?Nonbiologicalintelligencewillbeabletomatchandexceedpeakhumanskillsineacharea.Forthesereasons,onceacomputerisabletomatchthesubtletyandrangeofhumanintelligence,itwillnecessarilysoarpastitandthencontinueitsdouble-exponentialascent. A key question regarding the Singularity is whether the "chicken" (strong AI) or the "egg"(nanotechnology)willcomefirst.Inotherwords,willstrongAIleadtofullnanotechnology(molecular-manufacturingassemblersthatcanturninformationintophysicalproducts),orwillfullnanotechnologyleadtostrongAI?ThelogicofthefirstpremiseisthatstrongAIwouldimplysuperhumanAIforthereasonsjustcited,andsuperhumanAIwouldbeinapositiontosolveanyremainingdesignproblemsrequiredtoimplementfullnanotechnology.ThesecondpremiseisbasedontherealizationthatthehardwarerequirementsforstrongAIwillbemetbynanotechnology-basedcomputation.Likewise the software requirementswill be facilitatedbynanobots that could createhighlydetailed scansofhumanbrain functioning and thereby achieve thecompletionofreverseengineeringthehumanbrain.Bothpremisesarelogical;it'sclearthateithertechnologycanassisttheother.Therealityisthatprogress in both areas will necessarily use our most advanced tools, so advances in each field willsimultaneouslyfacilitatetheother.However,IdoexpectthatfullMNTwillemergepriortostrongAI,butonlybyafewyears(around2025fornanotechnology,around2029forstrongAI).Asrevolutionaryasnanotechnologywillbe,strongAIwillhavefarmoreprofoundconsequences.Nanotechnology is powerful but not necessarily intelligent.We candeviseways of at least trying tomanagetheenormouspowersofnanotechnology,butsuperintelligenceinnatelycannotbecontrolled.RunawayAI.OncestrongAIisachieved,itcanreadilybeadvancedanditspowersmultiplied,asthatisthefundamentalnatureofmachineabilities.AsonestrongAIimmediatelybegetsmanystrongAls,thelatteraccesstheirowndesign,understandandimproveit,andtherebyveryrapidlyevolveintoayetmore capable, more intelligent AI, with the cycle repeating itself indefinitely. Each cycle not onlycreatesamoreintelligentAIbuttakeslesstimethanthecyclebeforeit,asisthenatureoftechnologicalevolution (or any evolutionary process). The premise is that once strong AI is achieved, it willimmediatelybecomearunawayphenomenonofrapidlyescalatingsuperintelligence.160 Myownview isonlyslightlydifferent.The logicof runawayAI isvalid,butwestillneed toconsider the timing. Achieving human levels in a machine will not immediately cause a runawayphenomenon.Consider that a human level of intelligence has limitations.We have examples of thistoday—aboutsixbillionofthem.Considerascenarioinwhichyoutookonehundredhumansfrom,say,ashoppingmall.Thisgroupwouldconstituteexamplesofreasonablywell-educatedhumans.Yetifthisgroupwaspresentedwith the taskof improvinghuman intelligence, itwouldn't get very far, even ifprovidedwiththetemplatesofhumanintelligence.Itwouldprobablyhaveahardtimecreatingasimplecomputer.Speedingupthethinkingandexpandingthememorycapacitiesoftheseonehundredhumanswouldnotimmediatelysolvethisproblem.Ipointedoutabovethatmachineswillmatch(andquicklyexceed)peakhumanskillsineachareaofskill.Soinstead,let'stakeonehundredscientistsandengineers.Agroupoftechnicallytrainedpeoplewith the right backgroundswould be capable of improving accessible designs. If amachine attainedequivalencetoonehundred(andeventuallyonethousand,thenonemillion)technicallytrainedhumans,each operating much faster than a biological human, a rapid acceleration of intelligence wouldultimatelyfollow.However,thisaccelerationwon'thappenimmediatelywhenacomputerpassestheTuringtest.The

Turingtestiscomparabletomatchingthecapabilitiesofanaverage,educatedhumanandthusiscloserto theexampleofhumans froma shoppingmall. Itwill take time for computers tomaster allof therequisiteskillsandtomarrytheseskillswithallthenecessaryknowledgebases. Oncewe've succeeded in creating amachine that canpass theTuring test (around2029), thesucceedingperiodwillbeaneraofconsolidationinwhichnonbiological intelligencewillmakerapidgains. However, the extraordinary expansion contemplated for the Singularity, in which humanintelligenceismultipliedbybillions,won'ttakeplaceuntilthemid-2040s(asdiscussedinchapter3).TheAIWinter

There'sthisstupidmythouttherethatA.I.hasfailed,butA.I.iseverywherearoundyoueverysecond of the day. People just don't notice it. You've got A.I. systems in cars, tuning theparametersofthefuelinjectionsystems.Whenyoulandinanairplane,yourgategetschosenbyanA.I.schedulingsystem.EverytimeyouuseapieceofMicrosoftsoftware,you'vegotanA.I. system trying to figure outwhat you're doing, likewriting a letter, and it does a prettydamnedgoodjob.Everytimeyouseeamoviewithcomputer-generatedcharacters,they'realllittleA.I. characters behaving as a group. Every time you playa video game, you're playingagainstanA.I.system.

—RODNEYBROOKS,DIRECTOROFTHEMITAILAB161Istill run intopeoplewhoclaimthatartificial intelligencewithered in the1980s,anargument that iscomparabletoinsistingthattheInternetdiedinthedot-combustoftheearly2000s.162Thebandwidthandprice-performanceofInternettechnologies,thenumberofnodes(servers),andthedollarvolumeofe-commerceallacceleratedsmoothly through theboomaswellas thebustand theperiodsince.ThesamehasbeentrueforAI. The technology hype cycle for a paradigm shift—railroads,AI, Internet, telecommunications,possiblynownanotechnology—typicallystartswithaperiodofunrealisticexpectationsbasedonalackof understanding of all the enabling factors required.Although utilization of the newparadigmdoesincreaseexponentially,earlygrowthisslowuntilthekneeoftheexponential-growthcurveisrealized.While thewidespread expectations for revolutionary change are accurate, they are incorrectly timed.Whentheprospectsdonotquicklypanout,aperiodofdisillusionmentsetsin.Neverthelessexponentialgrowthcontinuesunabated,andyearslateramorematureandmorerealistictransformationdoesoccur.Wesawthisintherailroadfrenzyofthenineteenthcentury,whichwasfollowedbywidespreadbankruptcies. (I have some of these early unpaid railroad bonds in my collection of historicaldocuments.)Andwe are still feeling the effects of the e-commerce and telecommunicationsbusts ofseveralyearsago,whichhelpedfuelarecessionfromwhichwearenowrecovering.AIexperiencedasimilarprematureoptimisminthewakeofprogramssuchasthe1957GeneralProblemSolvercreatedbyAllenNewell,J.C.Shaw,andHerbertSimon,whichwasabletofindproofsfortheoremsthathadstumpedmathematicianssuchasBertrandRussell,andearlyprogramsfromtheMITArtificialIntelligenceLaboratory,whichcouldanswerSATquestions(suchasanalogiesandstoryproblems)atthelevelofcollegestudents.163ArashofAIcompaniesoccurredinthe1970s,butwhenprofitsdidnotmaterialize therewasanAI"bust" in the1980s,whichhasbecomeknownas the"AIwinter."ManyobserversstillthinkthattheAIwinterwastheendofthestoryandthatnothinghassincecomeoftheAIfield.YettodaymanythousandsofAIapplicationsaredeeplyembeddedintheinfrastructureofeveryindustry.Most of these applicationswere researchprojects ten to fifteenyears ago;Peoplewho ask,"WhateverhappenedtoAI?"remindmeoftravelerstotherainforestwhowonder,"Whereareallthemany species that are supposed to live here?" when hundreds of species of flora and fauna areflourishingonlyafewdozenmetersaway,deeplyintegratedintothelocalecology. Wearewell into theeraof"narrowAI,"which refers toartificial intelligence thatperformsauseful and specific function thatonce requiredhuman intelligence toperform,anddoes soathuman

levelsorbetter.OftennarrowAIsystemsgreatlyexceed thespeedofhumans,aswellasprovide theability tomanage and consider thousands of variables simultaneously. I describe a broad variety ofnarrowAIexamplesbelow.ThesetimeframesforAI'stechnologycycle(acoupleofdecadesofgrowingenthusiasm,adecadeofdisillusionment,thenadecadeandahalfofsolidadvanceinadoption)mayseemlengthy,comparedto the relatively rapid phases of the Internet and telecommunications cycles (measured in years, notdecades),but two factorsmustbeconsidered.First, the Internet and telecommunicationscycleswererelativelyrecent,sotheyaremoreaffectedbytheaccelerationofparadigmshift(asdiscussedinchapter1).Sorecentadoptioncycles(boom,bust,andrecovery)willbemuchfasterthanonesthatstartedfortyyearsago.Second,theAIrevolutionisthemostprofoundtransformationthathumancivilizationwillexperience,so itwill take longer tomature than lesscomplex technologies. It ischaracterizedby themasteryof themost importantandmostpowerfulattributeofhumancivilization, indeedof theentiresweepofevolutiononourplanet:intelligence. It's the nature of technology to understand a phenomenon and then engineer systems thatconcentrateandfocusthatphenomenontogreatlyamplifyit.Forexample,scientistsdiscoveredasubtlepropertyofcurvedsurfacesknownasBernoulli'sprinciple:agas(suchasair)travelsmorequicklyoveracurvedsurfacethanoveraflatsurface.Thus,airpressureoveracurvedsurfaceislowerthanoveraflatsurface.Byunderstanding,focusing,andamplifyingtheimplicationsofthissubtleobservation,ourengineeringcreatedallofaviation.Onceweunderstand theprinciplesof intelligence,wewillhaveasimilaropportunitytofocus,concentrate,andamplifyitspowers.Aswereviewedinchapter4,everyaspectofunderstanding,modeling,andsimulatingthehumanbrainisaccelerating: theprice-performanceandtemporalandspatialresolutionofbrainscanning, theamountofdataandknowledgeavailableaboutbrainfunction,andthesophisticationofthemodelsandsimulationsofthebrain'svariedregions.WealreadyhaveasetofpowerfultoolsthatemergedfromAIresearchandthathavebeenrefinedandimprovedoverseveraldecadesofdevelopment.Thebrainreverse-engineeringprojectwillgreatlyaugment this toolkit by also providing a panoply of new, biologically inspired, self-organizingtechniques. We will ultimately be able to apply engineering's ability to focus and amplify humanintelligencevastlybeyondthehundredtrillionextremelyslowinterneuronalconnectionsthateachofusstruggleswithtoday.Intelligencewillthenbefullysubjecttothelawofacceleratingreturns,whichiscurrentlydoublingthepowerofinformationtechnologieseveryyear.AnunderlyingproblemwithartificialintelligencethatIhavepersonallyexperiencedinmyfortyyearsinthisareaisthatassoonasanAItechniqueworks,it'snolongerconsideredAIandisspunoffasitsownfield(forexample,characterrecognition,speechrecognition,machinevision,robotics,datamining,medicalinformatics,automatedinvesting).ComputerscientistElaineRichdefinesAIas"thestudyofhowtomakecomputersdothingsatwhich,atthemoment,peoplearebetter."RodneyBrooks,directoroftheMITAILab,putsitadifferentway: "Every time we figure out a piece of it, it stops being magical; we say, Oh, that's just acomputation."IamalsoremindedofWatson'sremarktoSherlockHolmes,"Ithoughtatfirstthatyouhad done something clever, but I see that there was nothing in it after all."164 That has been ourexperienceasAIscientists.Theenchantmentofintelligenceseemstobereducedto"nothing"whenwefullyunderstanditsmethods.Themysterythatisleftistheintrigueinspiredbytheremaining,notasyetunderstoodmethodsofintelligence.AI'sToolkit

AI is thestudyof techniques for solvingexponentiallyhardproblems inpolynomial timebyexploitingknowledgeabouttheproblemdomain.

—ELAINERICH

As Imentioned in chapter 4, it's only recently thatwehave been able to obtain sufficiently detailedmodelsof howhumanbrain regions function to influenceAIdesign.Prior to that, in the absenceoftoolsthatcouldpeerintothebrainwithsufficientresolution,AIscientistsandengineersdevelopedtheirowntechniques.Justasaviationengineersdidnotmodeltheabilitytoflyontheflightofbirds,theseearlyAImethodswerenotbasedonreverseengineeringnaturalintelligence. Asmallsampleoftheseapproachesisreviewedhere.Sincetheiradoption,theyhavegrowninsophistication,which has enabled the creation of practical products that avoid the fragility and higherrorratesofearliersystems.ExpertSystems. In the 1970sAIwas often equatedwith one specificmethod: expert systems.Thisinvolvesthedevelopmentofspecificlogicalrulestosimulatethedecision-makingprocessesofhumanexperts.Akeypartoftheprocedureentailsknowledgeengineersinterviewingdomainexpertssuchasdoctorsandengineerstocodifytheirdecision-makingrules.Therewereearlysuccessesinthisarea,suchasmedicaldiagnosticsystemsthatcomparedwelltohumanphysicians,atleastinlimitedtests.Forexample,asystemcalledMYCIN,whichwasdesignedto diagnose and recommend remedial treatment for infectious diseases, was developed through the1970s. In 1979 a team of expert evaluators compared diagnosis and treatment recommendations byMYCIN to those of human doctors and found thatMYCIN did aswell as or better than any of thephysicians.165 It became apparent from this research that human decisionmaking typically is based not ondefinitive logic rulesbut ratheron"softer" typesofevidence.Adarkspotonamedical imaging testmay suggest cancer, but other factors such as its exact shape, location, and contrast are likely toinfluence a diagnosis. The hunches of human decisionmaking are usually influenced by combiningmany pieces of evidence from prior experience, none definitive by itself. Often we are not evenconsciouslyawareofmanyoftherulesthatweuse. Bythelate1980sexpertsystemswereincorporatingtheideaofuncertaintyandcouldcombinemany sources of probabilistic evidence to make a decision. The MYCIN system pioneered thisapproach.AtypicalMYCIN"rule"reads:

Iftheinfectionwhichrequirestherapyismeningitis,andthetypeoftheinfectionisfungal,andorganismswerenotseenonthestainoftheculture,andthepatientisnotacompromisedhost,and thepatienthasbeen toanarea that is endemic forcoccidiomycoses, and the raceof thepatientisBlack,Asian,orIndian,andthecryptococcalantigeninthecsftestwasnotpositive,THEN there is a 50 percent chance that cryptococcus is not one of the organismswhich iscausingtheinfection.

Althoughasingleprobabilisticrulesuchasthiswouldnotbesufficientbyitselftomakeausefulstatement,bycombiningthousandsofsuchrulestheevidencecanbemarshaledandcombinedtomakereliabledecisions.Probablythelongest-runningexpertsystemprojectisCYC(forenCYClopedic),createdbyDougLenatandhiscolleaguesatCycorp.Initiatedin1984,CYChasbeencodingcommonsenseknowledgetoprovidemachineswithanability tounderstand theunspokenassumptionsunderlyinghuman ideasand reasoning. The project has evolved from hard-coded logical rules to probabilistic ones and nowincludesmeansofextractingknowledgefromwrittensources (withhumansupervision).Theoriginalgoalwastogenerateonemillionrules,whichreflectsonlyasmallportionofwhattheaveragehumanknowsabouttheworld.Lenat'slatestgoalisforCYCtomaster"100millionthings,aboutthenumberatypicalpersonknowsabouttheworld,by2007."166 Another ambitious expert system is being pursued by Darryl Macer, associate professor ofbiologicalsciencesattheUniversityofTsukubainJapan.Heplanstodevelopasystemincorporatingallhumanideas.167Oneapplicationwouldbetoinformpolicymakersofwhichideasareheldbywhichcommunity.

Bayesian Nets. Over the last decade a technique called Bayesian logic has created a robustmathematicalfoundationforcombiningthousandsorevenmillionsofsuchprobabilisticrulesinwhatare called "belief networks" orBayesian nets.Originally devised byEnglishmathematicianThomasBayes and published posthumously in 1763, the approach is intended to determine the likelihood offuture events based on similar occurrences in the past.168 Many expert systems based on Bayesiantechniques gather data from experience in an ongoing fashion, thereby continually learning andimprovingtheirdecisionmaking.Themostpromisingtypeofspamfiltersarebasedonthismethod.IpersonallyuseaspamfiltercalledSpamBayes,whichtrainsitselfone-mailthatyouhaveidentifiedaseither"spam"or"okay."169Youstartoutbypresentingafolderofeachtothefilter.IttrainsitsBayesianbeliefnetworkonthesetwofilesandanalyzes thepatternsofeach, thusenabling it toautomaticallymovesubsequente-mailinto the proper category. It continues to train itself on every subsequent e-mail, especiallywhen it'scorrectedbytheuser.Thisfilterhasmadethespamsituationmanageableforme,whichissayingalot,asitweedsouttwohundredtothreehundredspammessageseachday,lettingmorethanonehundred"good"messages through. Only about 1 percent of themessages it identifies as "okay" are actuallyspam;italmostnevermarksagoodmessageasspam.ThesystemisalmostasaccurateasIwouldbeandmuchfaster.MarkovModels.Anothermethodthatisgoodatapplyingprobabilisticnetworkstocomplexsequencesof information involves Markov models.170 Andrei Andreyevich Markov (1856–1922), a renownedmathematician,establishedatheoryof"Markovchains,"whichwasrefinedbyNorbertWiener(1894–1964)in1923.Thetheoryprovidedamethodtoevaluatethelikelihoodthatacertainsequenceofeventswouldoccur.Ithasbeenpopular,forexample,inspeechrecognition,inwhichthesequentialeventsarephonemes (partsof speech).TheMarkovmodelsused in speech recognitioncode the likelihood thatspecific patterns of sound are found in each phoneme, how the phonemes influence each other, andlikely orders of phonemes. The system can also include probability networks on higher levels oflanguage,suchastheorderofwords.Theactualprobabilitiesinthemodelsaretrainedonactualspeechandlanguagedata,sothemethodisself-organizing.MarkovmodelingwasoneofthemethodsmycolleaguesandIusedinourownspeech-recognitiondevelopment.171 Unlike phonetic approaches, in which specific rules about phoneme sequences areexplicitlycodedbyhumanlinguists,wedidnottellthesystemthatthereareapproximatelyforty-fourphonemesinEnglish,nordidwetellitwhatsequencesofphonemesweremorelikelythanothers.Welet the systemdiscover these "rules" for itself from thousands of hours of transcribedhuman speechdata. The advantage of this approach over hand-coded rules is that the models develop subtleprobabilisticrulesofwhichhumanexpertsarenotnecessarilyaware.NeuralNets.Anotherpopularself-organizingmethodthathasalsobeenusedinspeechrecognitionandawide variety of other pattern-recognition tasks is neural nets.This technique involves simulating asimplifiedmodelofneuronsand interneuronalconnections.Onebasicapproach toneuralnetscanbedescribedasfollows.Eachpointofagiveninput(forspeech,eachpointrepresentstwodimensions,onebeing frequency and the other time; for images, each point would be a pixel in a two-dimensionalimage)israndomlyconnectedtotheinputsofthefirstlayerofsimulatedneurons.Everyconnectionhasan associated synaptic strength,which represents its importance andwhich is set at a randomvalue.Eachneuronaddsupthesignalscomingintoit.Ifthecombinedsignalexceedsaparticularthreshold,theneuronfiresandsendsasignaltoitsoutputconnection;ifthecombinedinputsignaldoesnotexceedthe threshold, theneurondoesnotfire,anditsoutput iszero.Theoutputofeachneuronis randomlyconnected to the inputsof theneurons in thenext layer.Therearemultiple layers (generally threeormore),andthelayersmaybeorganizedinavarietyofconfigurations.Forexample,onelayermayfeedback toanearlier layer.At the top layer, theoutputofoneormoreneurons,also randomlyselected,providestheanswer.(Foranalgorithmicdescriptionofneuralnets,seethisnote:172)

Since theneural-netwiringand synapticweights are initially set randomly, the answersof anuntrainedneuralnetwillberandom.Thekeytoaneuralnet,therefore,isthatitmustlearnitssubjectmatter.Likethemammalianbrainsonwhichit'slooselymodeled,aneuralnetstartsoutignorant.Theneural net's teacher—whichmaybe ahuman, a computer program,orperhaps another,morematureneural net that has already learned its lessons—rewards the student neural netwhen it generates therightoutputandpunishesitwhenitdoesnot.Thisfeedbackisinturnusedbythestudentneuralnettoadjust thestrengthsofeach interneuronalconnection.Connections thatwereconsistentwith the rightansweraremadestronger.Thosethatadvocatedawronganswerareweakened.Overtime,theneuralnetorganizesitselftoprovidetherightanswerswithoutcoaching.Experimentshaveshownthatneuralnetscanlearntheirsubjectmatterevenwithunreliableteachers.Iftheteacheriscorrectonly60percentofthetime,thestudentneuralnetwillstilllearnitslessons. A powerful, well-taught neural net can emulate a wide range of human pattern-recognitionfaculties. Systems using multilayer neural nets have shown impressive results in a wide variety ofpattern-recognition tasks, including recognizing handwriting, human faces, fraud in commercialtransactionssuchascredit-cardcharges,andmanyothers.Inmyownexperienceinusingneuralnetsinsuchcontexts,themostchallengingengineeringtaskisnotcodingthenetsbutinprovidingautomatedlessonsforthemtolearntheirsubjectmatter.Thecurrenttrendinneuralnetsistotakeadvantageofmorerealisticandmorecomplexmodelsofhow actual biological neural nets work, now that we are developing detailed models of neuralfunctioningfrombrainreverseNEARengineering.173Sincewedohaveseveraldecadesofexperienceinusingself-organizingparadigms,newinsightsfrombrainstudiescanquicklybeadaptedtoneural-netexperiments.Neuralnetsarealsonaturallyamenabletoparallelprocessing,sincethatishowthebrainworks.Thehumanbraindoesnothaveacentralprocessorthatsimulateseachneuron.Rather,wecanconsidereachneuronandeach interneuronalconnection tobean individualslowprocessor.Extensivework isunderway todevelop specialized chips that implementneural-net architectures inparallel toprovidesubstantiallygreaterthroughput.174Genetic Algorithms (GAs). Another self-organizing paradigm inspired by nature is genetic, orevolutionary,algorithms,whichemulateevolution,includingsexualreproductionandmutations.Hereis a simplified description of how theywork. First, determine away to code possible solutions to agivenproblem.Iftheproblemisoptimizingthedesignparametersforajetengine,definealistoftheparameters (with a specific number of bits assigned to each parameter). This list is regarded as thegeneticcodeinthegeneticalgorithm.Thenrandomlygeneratethousandsormoregeneticcodes.Eachsuchgeneticcode(whichrepresentsonesetofdesignparameters)isconsideredasimulated"solution"organism.Nowevaluateeachsimulatedorganisminasimulatedenvironmentbyusingadefinedmethodtoevaluateeachsetofparameters.Thisevaluationisakeytothesuccessofageneticalgorithm.Inourexample, we would apply each solution organism to a jet-engine simulation and determine howsuccessful that set of parameters is, according to whatever criteria we are interested in (fuelconsumption,speed,andsoon).Thebestsolutionorganisms(thebestdesigns)areallowedtosurvive,andtherestareeliminated.Nowhaveeachofthesurvivorsmultiplythemselvesuntiltheyreachthesamenumberofsolutioncreatures.Thisisdonebysimulatingsexualreproduction.Inotherwords,eachnewoffspringsolutiondraws part of its genetic code from one parent and another part from a second parent. Usually nodistinctionismadebetweenmaleorfemaleorganisms;it'ssufficienttogenerateanoffspringfromtwoarbitraryparents.Astheymultiply,allowsomemutation(randomchange)inthechromosomestooccur. We've now defined one generation of simulated evolution; now repeat these steps for eachsubsequentgeneration.Attheendofeachgenerationdeterminehowmuchthedesignshaveimproved.When the improvement in the evaluation of the design creatures from one generation to the nextbecomesverysmall,westopthisiterativecycleofimprovementandusethebestdesign(s)inthelast

generation.(Foranalgorithmicdescriptionofgeneticalgorithms,seethisnote.175)ThekeytoaGAisthatthehumandesignersdon'tdirectlyprogramasolution;rather,theyletoneemerge through an iterative process of simulated competition and improvement. As we discussed,biological evolution is smart but slow, so to enhance its intelligencewe retain its discernmentwhilegreatlyspeedingupitsponderouspace.Thecomputerisfastenoughtosimulatemanygenerationsinamatterofhoursordaysorweeks.Butwehavetogothroughthisiterativeprocessonlyonce;oncewehave let thissimulatedevolutionrun itscourse,wecanapply theevolvedandhighly refinedrules torealproblemsinarapidfashion.LikeneuralnetsGAsareawaytoharnessthesubtlebutprofoundpatternsthatexistinchaoticdata. A key requirement for their success is a valid way of evaluating each possible solution. Thisevaluationneeds tobe fastbecause itmust takeaccountofmany thousandsofpossible solutions foreachgenerationofsimulatedevolution.GAsareadeptathandlingproblemswithtoomanyvariablestocomputepreciseanalyticsolutions.The design of a jet engine, for example, involves more than one hundred variables and requiressatisfyingdozensofconstraints.GAsusedbyresearchersatGeneralElectricwereabletocomeupwithenginedesignsthatmettheconstraintsmorepreciselythanconventionalmethods.WhenusingGAsyoumust,however,becarefulwhatyouaskfor.UniversityofSussexresearcherJonBirdusedaGAtooptimallydesignanoscillatorcircuit.Severalattemptsgeneratedconventionaldesignsusinga smallnumberof transistors,but thewinningdesignwasnotanoscillatoratallbutasimpleradiocircuit.ApparentlytheGAdiscoveredthattheradiocircuitpickedupanoscillatinghumfromanearbycomputer.176TheGA'ssolutionworkedonlyintheexactlocationonthetablewhereitwasaskedtosolvetheproblem.Geneticalgorithms,partofthefieldofchaosorcomplexitytheory,areincreasinglybeingusedtosolve otherwise intractable business problems, such as optimizing complex supply chains. Thisapproach isbeginning tosupplantmoreanalyticmethods throughout industry. (Seeexamplesbelow.)Theparadigmisalsoadeptat recognizingpatterns,and isoftencombinedwithneuralnetsandotherself-organizingmethods. It's also a reasonableway towrite computer software, particularly softwarethatneedstofinddelicatebalancesforcompetingresources. In the novelusr/bin/god, CoryDoctorow, a leading science-fictionwriter, uses an intriguingvariationofaGAtoevolveanAI.TheGAgeneratesa largenumberof intelligentsystemsbasedonvariousintricatecombinationsoftechniques,witheachcombinationcharacterizedbyitsgeneticcode.ThesesystemsthenevolveusingaGA..Theevaluationfunctionworksasfollows:eachsystemlogsontovarioushumanchatroomsandtries to pass for a human, basically a covert Turing test. If one of the humans in a chat room sayssomething like "What are you, a chatterbot?" (chatterbot meaning an automatic program, which attoday'slevelofdevelopmentisexpectedtonotunderstandlanguageatahumanlevel),theevaluationisover,thatsystemendsitsinteractions,andreportsitsscoretotheGA.Thescoreisdeterminedbyhowlong itwasable topass forhumanwithoutbeingchallenged in thisway.TheGAevolvesmoreandmoreintricatecombinationsoftechniquesthatareincreasinglycapableofpassingforhuman.Themaindifficultywiththisideaisthattheevaluationfunctionisfairlyslow,althoughitwilltakeanappreciableamountof timeonlyafter thesystemsarereasonably intelligent.Also, theevaluationscantakeplacelargelyinparallel.It'saninterestingideaandmayactuallybeausefulmethodtofinishthe jobofpassing theTuring test, onceweget to thepointwherewehave sufficiently sophisticatedalgorithmstofeedintosuchaGA,sothatevolvingaTuring-capableAIisfeasible.Recursive Search. Often we need to search through a vast number of combinations of possiblesolutionstosolveagivenproblem.Aclassicexampleis inplayinggamessuchaschess.Asaplayerconsiders her next move, she can list all of her possible moves, and then, for each such move, allpossiblecountermovesbytheopponent,andsoon.Itisdifficult,however,forhumanplayerstokeepahuge"tree"ofmove-countermovesequencesin theirheads,andsotheyrelyonpatternrecognition—recognizingsituationsbasedonpriorexperience—whereasmachinesuselogicalanalysisofmillionsof

movesandcountermoves.Suchalogicaltreeisattheheartofmostgame-playingprograms.Considerhowthisisdone.Weconstruct a program called Pick Best Next Step to select eachmove. Pick Best Next Step starts bylisting all of the possiblemoves from the current state of the board. (If the problemwas solving amathematicaltheorem,ratherthangamemoves,theprogramwouldlistallofthepossiblenextstepsinaproof.)Foreachmovetheprogramconstructsahypotheticalboardthatreflectswhatwouldhappenifwemade thismove. For each such hypothetical board,wenowneed to considerwhat our opponentwoulddo ifwemade thismove.Nowrecursioncomes in,becausePickBestNextStep simplycallsPickBestNextStep(inotherwords,itself)topickthebestmoveforouropponent.Incallingitself,PickBestNextStepthenlistsallofthelegalmovesforouropponent.Theprogramkeepscallingitself,lookingaheadasmanymovesaswehavetimetoconsider,whichresults in the generation of a hugemove-countermove tree. This is another example of exponentialgrowth,becausetolookaheadanadditionalmove(orcountermove)requiresmultiplyingtheamountofavailablecomputationbyaboutfive.Keytothesuccessof therecursiveformulaispruningthishugetreeofpossibilitiesandultimatelystoppingitsgrowth.Inthegamecontext, ifaboardlookshopelessforeitherside,theprogramcanstoptheexpansionofthemove-countermovetreefromthatpoint(calleda"terminalleaf"ofthetree)andconsiderthemostrecentlyconsideredmovetobealikelywinorloss.When all of these nested program calls are completed, the program will have determined the bestpossiblemoveforthecurrentactualboardwithinthelimitsofthedepthofrecursiveexpansionthatithadtimetopursueandthequalityofitspruningalgorithm.(Foranalgorithmicdescriptionofrecursivesearch,seethisnote:177)Therecursiveformulaisofteneffectiveatmathematics.Ratherthangamemoves,the"moves"aretheaxiomsofthefieldofmathbeingaddressed,aswellaspreviouslyprovedtheorems.Theexpansionateachpointis thepossibleaxioms(orpreviouslyprovedtheorems)thatcanbeappliedtoaproofateachstep.(ThiswastheapproachusedbyNewell,Shaw,andSimons'sGeneralProblemSolver.)Fromtheseexamplesitmayappearthatrecursioniswellsuitedonlyforproblemsinwhichwehave crisply defined rules and objectives. But it has also shown promise in computer generation ofartistic creations. For example, a program I designed called Ray Kurzweil's Cybernetic Poet uses arecursive approach.178 The program establishes a set of goals for each word—achieving a certainrhythmic pattern, poem structure, andword choice that is desirable at that point in the poem. If theprogramisunabletofindawordthatmeetsthesecriteria,itbacksupanderasesthepreviouswordithaswritten,reestablishesthecriteriaithadoriginallysetforthewordjusterased,andgoesfromthere.If thatalso leads toadeadend, itbacksupagain, thusmovingbackwardandforward.Eventually, itforcesitselftomakeupitsmindbyrelaxingsomeoftheconstraintsifallpathsleadtodeadends.

Combining Methods. The most powerful approach to building robust AI systems is to combineapproaches,whichishowthehumanbrainworks.Aswediscussed,thebrainisnotonebigneuralnetbutinsteadconsistsofhundredsofregions,eachofwhichisoptimizedforprocessinginformationinadifferent way. None of these regions by itself operates at what wewould consider human levels ofperformance,butdearlybydefinitiontheoverallsystemdoesexactlythat. I've used this approach in my own AI work, especially in pattern recognition. In speechrecognition,forexample,weimplementedanumberofdifferentpattern-recognitionsystemsbasedondifferent paradigms. Somewere specifically programmedwith knowledge of phonetic and linguisticconstraints from experts. Some were based on rules to parse sentences (which involves creatingsentence diagrams showingword usage, similar to the diagrams taught in grade school). Somewerebasedonself-organizingtechniques,suchasMarkovmodels,trainedonextensivelibrariesofrecordedandannotatedhumanspeech.Wethenprogrammedasoftware"expertmanager"tolearnthestrengthsandweaknessesof thedifferent"experts" (recognizers)andcombine their results inoptimalways. Inthis fashion, a particular technique that by itself might produce unreliable results can nonethelesscontributetoincreasingtheoverallaccuracyofthesystem.

TherearemanyintricatewaystocombinethevariedmethodsinAI'stoolbox.Forexample,onecanuseageneticalgorithmtoevolvetheoptimaltopology(organizationofnodesandconnections)foraneural net or aMarkovmodel.The final output of theGA-evolvedneural net can thenbeused tocontrol the parameters of a recursive search algorithm.We can add in powerful signal- and image-processing techniques that have been developed for pattern-processing systems. Each specificapplication calls for a different architecture. Computer science professor and AI entrepreneur BenGoertzel has written a series of books and articles that describe strategies and architectures forcombining the diverse methods underlying intelligence. His Novamente architecture is intended toprovideaframeworkforgeneral-purposeAI.179TheabovebasicdescriptionsprovideonlyaglimpseintohowincreasinglysophisticatedcurrentAIsystemsaredesigned.It'sbeyondthescopeofthisbooktoprovideacomprehensivedescriptionofthetechniquesofAI,andevenadoctoralprogramincomputerscienceisunabletocoverallofthevariedapproachesinusetoday.Manyoftheexamplesofreal-worldnarrowAIsystemsdescribedinthenextsectionuseavarietyofmethodsintegratedandoptimizedforeachparticulartask.NarrowAIisstrengtheningasaresultofseveralconcurrenttrends:continuedexponentialgainsincomputationalresources,extensivereal-worldexperience with thousands of applications, and fresh insights into how the human brain makesintelligentdecisions.ANarrowAISamplerWhenIwrotemyfirstAIbook,TheAgeofIntelligentMachines, inthelate1980s,IhadtoconductextensiveinvestigationstofindafewsuccessfulexamplesofAIinpractice.TheInternetwasnotyetprevalent,soIhadtogotoreallibrariesandvisittheAIresearchcentersintheUnitedStates,Europe,andAsia. I included inmybookprettymuch all of the reasonable examples I could identify. Inmyresearch for this book my experience has been altogether different. I have been inundated withthousandsofcompellingexamples.InourreportingontheKurzweilAI.netWebsite,wefeatureoneormoredramaticsystemsalmosteveryday.180A2003studybyBusinessCommunicationsCompanyprojecteda$21billionmarketby2007forAIapplications,withaverageannualgrowthof12.2percentfrom2002to2007.181Leadingindustriesfor AI applications include business intelligence, customer relations, finance, defense and domesticsecurity,andeducation.HereisasmallsampleofnarrowAIinaction.MilitaryandIntelligence.TheU.S.militaryhasbeenanaviduserofAIsystems.Pattern-recognitionsoftwaresystemsguideautonomousweaponssuchascruisemissiles,whichcanflythousandsofmilestofindaspecificbuildingorevenaspecificwindow.182Althoughtherelevantdetailsoftheterrainthatthemissile flies over are programmed ahead of time, variations inweather, ground cover, and otherfactorsrequireaflexiblelevelofreal-timeimagerecognition. Thearmyhasdevelopedprototypesof self-organizingcommunicationnetworks (called"meshnetworks") to automatically configure many thousands of communication nodes when a platoon isdroppedintoanewlocation.183 ExpertsystemsincorporatingBayesiannetworksandGAsareusedtooptimizecomplexsupplychains that coordinate millions of provisions, supplies, and weapons based on rapidly changingbattlefieldrequirements. AIsystemsare routinelyemployed tosimulate theperformanceofweapons, includingnuclearbombsandmissiles. Advancewarningof theSeptember11, 2001, terrorist attackswas apparentlydetectedby theNational Security Agency's AI-based Echelon system, which analyzes the agency's extensivemonitoring of communications traffic.184 Unfortunately, Echelon's warnings were not reviewed by

humanagentsuntilitwastoolate. The2002militarycampaigninAfghanistansawthedebutofthearmedPredator,anunmannedrobotic flying fighter.Although theair force'sPredatorhadbeenunderdevelopment formanyyears,arming it with army-supplied missiles was a last-minute improvisation that proved remarkablysuccessful.IntheIraqwarthatbeganin2003thearmedPredator(operatedbytheCIA)andotherflyingunmannedaerialvehicles(UAVs)destroyedthousandsofenemytanksandmissilesites. All of the military services are using robots. The army utilizes them to search caves (inAfghanistan) and buildings. The navy uses small robotic ships to protect its aircraft carriers. As Idiscussinthenextchapter,movingsoldiersawayfrombattleisarapidlygrowingtrend.SpaceExploration.NASA isbuilding self-understanding into the software controlling its unmannedspacecraft.BecauseMarsisaboutthreelight-minutesfromEarth,andJupiteraroundfortylight-minutes(dependingontheexactpositionoftheplanets),communicationbetweenspacecraftheadedthereandearthbound controllers is significantly delayed. For this reason it's important that the softwarecontrolling these missions have the capability of performing its own tactical decision making. ToaccomplishthisNASAsoftwareisbeingdesignedtoincludeamodelofthesoftware'sowncapabilitiesandthoseofthespacecraft,aswellasthechallengeseachmissionislikelytoencounter.SuchAI-basedsystems are capable of reasoning through new situations rather than just following preprogrammedrules.ThisapproachenabledthecraftDeepSpace••Onein1999touseitsowntechnicalknowledgetodevise a seriesof original plans toovercomea stuck switch that threatened todestroy itsmissionofexploring an asteroid.185 The AI system's first plan failed to work, but its second plan saved themission. "These systems have a commonsense model of the physics of their internal components,"explainsBrianWilliams,coinventorofDeepSpaceOne'sautonomoussoftwareandnowascientistatMIT'sSpaceSystemsandAIlaboratories."[Thespacecraft]canreasonfromthatmodel todeterminewhatiswrongandtoknowhowtoact." Using a networkof computersNASAusedGAs to evolve an antennadesign for threeSpaceTechnology5satellitesthatwillstudytheEarth'smagneticfield.Millionsofpossibledesignscompetedinthesimulatedevolution.AccordingtoNASAscientistandprojectleaderJasonLohn,"Wearenowusing the [GA] software to design tinymicroscopicmachines, including gyroscopes, for spaceflightnavigation.Thesoftwarealsomayinventdesignsthatnohumandesignerwouldeverthinkof."186 AnotherNASAAI system learned on its own to distinguish stars fromgalaxies in very faintimageswithanaccuracysurpassingthatofhumanastronomers.Newland-basedrobotictelescopesareabletomaketheirowndecisionsonwheretolookandhowto optimize the likelihood of finding desired phenomena. Called "autonomous, semi-intelligentobservatories,"thesystemscanadjusttotheweather,noticeitemsofinterest,anddecideontheirowntotrackthem.Theyareabletodetectverysubtlephenomena,suchasastarblinkingforananosecond,whichmayindicateasmallasteroidintheouterregionsofoursolarsystempassinginfrontofthelightfrom that star.187 One such system, called Moving Object and Transient Event Search System(MOTESS),hasidentifiedonitsown180newasteroidsandseveralcometsduringitsfirsttwoyearsofoperation. "We have an intelligent observing system," explained University of Exeter astronomerAlasdair Allan. "It thinks and reacts for itself, deciding whether something it has discovered isinterestingenoughtoneedmoreobservations.Ifmoreobservationsareneeded,itjustgoesaheadandgetsthem."Similarsystemsareusedbythemilitarytoautomaticallyanalyzedatafromspysatellites.Currentsatellite technology is capable of observing ground-level features about an inch in size and is notaffected by bad weather, clouds, or darkness.188 The massive amount of data continually generatedwould not be manageable without automated image recognition programmed to look for relevantdevelopments.Medicine. If you obtain an electrocardiogram (ECG) your doctor is likely to receive an automateddiagnosis using pattern recognition applied to ECG recordings. My own company (Kurzweil

Technologies) isworkingwithUnitedTherapeutics to develop a new generation of automatedECGanalysis for long-term unobtrusive monitoring (via sensors embedded in clothing and wirelesscommunication using a cell phone) of the early warning signs of heart disease.189 Other pattern-recognitionsystemsareusedtodiagnoseavarietyofimagingdata. EverymajordrugdeveloperisusingAIprogramstodopatternrecognitionandintelligentdatamining in thedevelopmentofnewdrug therapies.For exampleSRI International isbuilding flexibleknowledgebasesthatencodeeverythingweknowaboutadozendiseaseagents,includingtuberculosisand H. pylori (the bacteria that cause ulcers).190 The goal is to apply intelligent datamining tools(software that can search for new relationships in data) to find new ways to kill or disrupt themetabolismsofthesepathogens.Similarsystemsarebeingappliedtoperformingtheautomaticdiscoveryofnewtherapiesforotherdiseases, as well as understanding the function of genes and their roles in disease.191 For exampleAbbottLaboratoriesclaimsthatsixhumanresearchers inoneof itsnewlabsequippedwithAI-basedrobotic and data-analysis systems are able tomatch the results of two hundred scientists in its olderdrug-developmentlabs.192 Menwithelevatedprostate-specificantigen(PSA)levelstypicallyundergosurgicalbiopsy,butabout75percentofthesemendonothaveprostatecancer.Anewtest,basedonpatternrecognitionofproteinsintheblood,wouldreducethisfalsepositiveratetoabout29percent.193ThetestisbasedonanAIprogramdesignedbyCorrelogicSystemsinBethesda,Maryland,andtheaccuracyisexpectedtoimprovefurtherwithcontinueddevelopment. Pattern recognition applied to protein patterns has also been used in the detection of ovariancancer.Thebestcontemporarytest forovariancancer,calledCA-125,employedincombinationwithultrasound,missesalmostallearly-stagetumors."Bythetimeitisnowdiagnosed,ovariancanceristoooften deadly," says Emanuel Petricoin III, codirector of theClinical Proteomics Program run by theFDAandtheNationalCancerInstitute.PetricoinistheleaddeveloperofanewAI-basedtestlookingfor unique patterns of proteins found only in the presence of cancer. In an evaluation involvinghundreds of blood samples, the test was, according to Petricoin, "an astonishing 100% accurate indetectingcancer,evenattheearlieststages."194About10percentofallPap-smearslidesintheUnitedStatesareanalyzedbyaself-learningAIprogramcalledFocalPoint,developedbyTriPathImaging.Thedevelopersstartedoutbyinterviewingpathologists on the criteria they use. The AI system then continued to learn by watching expertpathologists.Onlythebesthumandiagnosticianswereallowedtobeobservedbytheprogram."That'sthe advantage of an expert system," explainsBobSchmidt,Tri-Path's technical productmanager. "Itallowsyoutoreplicateyourverybestpeople."OhioStateUniversityHealthSystemhasdevelopedacomputerizedphysicianorder-entry(CPOE)systembasedonanexpertsystemwithextensiveknowledgeacrossmultiplespecialties.195Thesystemautomatically checks every order for possible allergies in the patient, drug interactions, duplications,drugrestrictions,dosingguidelines,andappropriatenessgiven informationabout thepatient fromthehospital'slaboratoryandradiologydepartments.ScienceandMath.A"robotscientist"hasbeendevelopedattheUniversityofWalesthatcombinesanAI-basedsystemcapableofformulatingoriginaltheories,aroboticsystemthatcanautomaticallycarryoutexperiments,andareasoningenginetoevaluateresults.Theresearchersprovidedtheircreationwitha model of gene expression in yeast. The system "automatically originates hypotheses to explainobservations, devises experiments to test these hypotheses, physically runs the experiments using alaboratoryrobot,interpretstheresultstofalsifyhypothesesinconsistentwiththedata,andthenrepeatsthecycle."196Thesystemiscapableofimprovingitsperformancebylearningfromitsownexperience.Theexperimentsdesignedbytherobotscientistwerethreetimeslessexpensivethanthosedesignedbyhumanscientists.Atestofthemachineagainstagroupofhumanscientistsshowedthatthediscoveriesmadebythemachinewerecomparabletothosemadebythehumans.

MikeYoung,directorofbiologyattheUniversityofWales,wasoneofthehumanscientistswholosttothemachine.Heexplainsthat"therobotdidbeatme,butonlybecauseIhitthewrongkeyatonepoint." Along-standingconjectureinalgebrawasfinallyprovedbyanAIsystematArgonneNationalLaboratory.Humanmathematicianscalledtheproof"creative."Business,Finance,andManufacturing.CompaniesineveryindustryareusingAIsystemstocontrolandoptimizelogistics,detectfraudandmoneylaundering,andperformintelligentdataminingonthehordeofinformationtheygathereachday.Wal-Mart,forexample,gathersvastamountsofinformationfromitstransactionswithshoppers.AI-basedtoolsusingneuralnetsandexpertsystemsreviewthisdatato provide market-research reports for managers. This intelligent data mining allows them to makeremarkablyaccuratepredictionsoftheinventoryrequiredforeachproductineachstoreforeachday.197 AI-basedprogramsareroutinelyusedtodetectfraudinfinancialtransactions.FutureRoute,anEnglishcompany, forexample,offers iHex,basedonAI routinesdevelopedatOxfordUniversity, todetectfraudincredit-cardtransactionsandloanapplications.198Thesystemcontinuouslygeneratesandupdates its own rules based on its experience. First Union Home Equity Bank in Charlotte, NorthCarolina, uses Loan Arranger, a similar AI-based system, to decide whether to approve mortgageapplications.199NASDAQsimilarlyusesalearningprogramcalledtheSecuritiesObservation,NewsAnalysis,andRegulation (SONAR) system to monitor all trades for fraud as well as the possibility of insidertrading.200Asoftheendof2003morethan180incidentshadbeendetectedbySONARandreferredtotheU.S.SecuritiesandExchangeCommissionandDepartmentofJustice.Theseincludedseveralcasesthatlaterreceivedsignificantnewscoverage.AscentTechnology,foundedbyPatrickWinston,whodirectedMIT'sAILabfrom1972through1997, has designed a GA-based system called Smart-Airport Operations Center (SAOe) that canoptimize the complex logistics of an airport, such as balancing work assignments of hundreds ofemployees, making gate and equipment assignments, and managing a myriad of other details.201Winston points out that "figuring out ways to optimize a complicated situation is what geneticalgorithmsdo."SAOChasraisedproductivitybyapproximately30percentintheairportswhereithasbeenimplemented.Ascent'sfirstcontractwastoapplyitsAItechniquestomanagingthelogisticsforthe1991DesertStormcampaigninIraq.DARPAclaimedthatAl-basedlogistic-planningsystems,includingtheAscentsystem, resulted inmore savings than the entire government research investment inAI over severaldecades.ArecenttrendinsoftwareisforAIsystemstomonitoracomplexsoftwaresystem'sperformance,recognize malfunctions, and determine the best way to recover automatically without necessarilyinformingthehumanuser.202Theideastemsfromtherealizationthatassoftwaresystemsbecomemorecomplex, like humans, they will never be perfect, and that eliminating all bugs is impossible. Ashumans,weuse the samestrategy:wedon't expect tobeperfect,butweusually try to recover frominevitablemistakes."Wewanttostandthisnotionofsystemsmanagementonitshead,"saysArmandoFox,theheadofStanfordUniversity'sSoftwareInfrastructuresGroup,whoisworkingonwhatisnowcalled"autonomiccomputing."Foxadds,"Thesystemhastobeabletosetitselfup,ithastooptimizeitself. Ithas to repair itself, and if somethinggoeswrong, ithas toknowhow to respond toexternalthreats." IBM, Microsoft, and other software vendors are all developing systems that incorporateautonomiccapabilities.Manufacturing and Robotics. Computer-integrated manufacturing (CIM) increasingly employs AItechniquestooptimizetheuseofresources,streamlinelogistics,andreduceinventoriesthroughjust-in-timepurchasingofpartsandsupplies.Anew trend inCIMsystems is touse "case-based reasoning"ratherthanhard-coded,rule-basedexpertsystems.Suchreasoningcodesknowledgeas"cases,"which

areexamplesofproblemswithsolutions.Initialcasesareusuallydesignedbytheengineers,butthekeytoasuccessfulcase-based reasoningsystem is itsability togathernewcases fromactualexperience.Thesystemisthenabletoapplythereasoningfromitsstoredcasestonewsituations. Robotsareextensivelyusedinmanufacturing.ThelatestgenerationofrobotsusesflexibleAl-basedmachine-visionsystems—fromcompaniessuchasCognexCorporationinNatick,Massachusetts—thatcanrespondflexiblytovaryingconditions.Thisreducestheneedforprecisesetupfortherobotto operate correctly. Brian Carlisle, CEO of Adept Technologies, a Livermore, California, factory-automationcompany,pointsoutthat"eveniflaborcostswereeliminated[asaconsideration],astrongcasecanstillbemadeforautomatingwithrobotsandotherflexibleautomation.Inadditiontoqualityandthroughput,usersgainbyenablingrapidproductchangeoverandevolution thatcan'tbematchedwithhardtooling."OneofAI'sleadingroboticists,HansMoravec,hasfoundedacompanycalledSeegridtoapplyhismachine-vision technology to applications in manufacturing, materials handling, and militarymissions.203Moravec'ssoftwareenablesadevice(arobotor justamaterial-handlingcart) towalkorroll through an unstructured environment and in a single pass build a reliable "voxel" (three-dimensional pixel)map of the environment. The robot can then use themap and its own reasoningabilitytodetermineanoptimalandobstacle-freepathtocarryoutitsassignedmission. This technology enables autonomous carts to transfer materials throughout a manufacturingprocess without the high degree of preparation required with conventional preprogrammed roboticsystems.Inmilitarysituationsautonomousvehiclescouldcarryoutprecisemissionswhileadjustingtorapidlychangingenvironmentsandbattlefieldconditions. Machine vision is also improving the ability of robots to interactwith humans.Using small,inexpensive cameras, head- and eye-tracking software can sense where a human user is, allowingrobots,aswellasvirtualpersonalitiesonascreen, tomaintaineyecontact,akeyelement fornaturalinteractions.Head-andeye-trackingsystemshavebeendevelopedatCarnegieMellonUniversityandMITandareofferedbysmallcompaniessuchasSeeingMachinesofAustralia. AnimpressivedemonstrationofmachinevisionwasavehiclethatwasdrivenbyanAIsystemwith no human intervention for almost the entire distance fromWashington,D.C., to SanDiego.204Bruce Buchanan, computer-science professor at the University of Pittsburgh and president of theAmericanAssociationofArtificialIntelligence,pointedoutthatthisfeatwouldhavebeen"unheardof10yearsago."PaloAltoResearchCenter(PARC)isdevelopingaswarmofrobotsthatcannavigateincomplexenvironments,suchasadisasterzone,andfinditemsofinterest,suchashumanswhomaybeinjured.InaSeptember2004demonstrationatanAIconferenceinSanJose, theydemonstratedagroupofself-organizing robots on amock but realistic disaster area.205 The robotsmoved over the rough terrain,communicatedwithoneanother,usedpatternrecognitiononimages,anddetectedbodyheattolocatehumans.SpeechandLanguage.Dealingnaturallywithlanguageisthemostchallengingtaskofallforartificialintelligence.Nosimpletricks,shortoffullymasteringtheprinciplesofhumanintelligence,willallowacomputerized system to convincingly emulate human conversation, even if restricted to just textmessages.ThiswasTuring'senduringinsightindesigninghiseponymoustestbasedentirelyonwrittenlanguage.Althoughnotyetathumanlevels,naturallanguage-processingsystemsaremakingsolidprogress.Searchengineshavebecomesopopularthat"Google"hasgonefromapropernountoacommonverb,and its technology has revolutionized research and access to knowledge. Google and other searchengines use Al-based statistical-learning methods and logical inference to determine the ranking oflinks.Themostobvious failingof these searchengines is their inability tounderstand thecontextofwords. Although an experienced user learns how to design a string of keywords to find the mostrelevantsites(forexample,asearchfor"computerchip"islikelytoavoidreferencestopotatochipsthatasearchfor"chip"alonemightturnup),whatwewouldreallyliketobeabletodoisconversewithour

search engines in natural language.Microsoft has developed a natural-language search engine calledAsk MSR (Ask Micro-Soft Research), which actually answers natural-language questions such as"WhenwasMickeyMantle born?"206After the systemparses the sentence to determine the parts ofspeech(subject,verb,object,adjectiveandadverbmodifiers,andsoon),aspecialsearchenginethenfinds matches based on the parsed sentence. The found documents are searched for sentences thatappeartoanswerthequestion,andthepossibleanswersareranked.Atleast75percentofthetime,thecorrectanswerisinthetopthreerankedpositions,andincorrectanswersareusuallyobvious(suchas"MickeyMantlewasbornin3").Theresearchershopetoincludeknowledgebasesthatwilllowertherankofmanyofthenonsensicalanswers. MicrosoftresearcherEricBrill,whohasledresearchonAskMSR,hasalsoattemptedanevenmore difficult task: building a system that provides answers of about fifty words to more complexquestions,suchas,"HowaretherecipientsoftheNobelPrizeselected?"OneofthestrategiesusedbythissystemistofindanappropriateFAQsectionontheWebthatanswersthequery. Natural-language systems combined with large-vocabulary, speaker-independent (that is,responsivetoanyspeaker)speechrecognitionoverthephoneareenteringthemarketplacetoconductroutine transactions.You can talk toBritishAirways' virtual travel agent about anythingyou like aslongas ithas todowithbookingflightsonBritishAirways.207You'realso likely to talk toavirtualperson if you call Verizon for customer service or Charles Schwab and Merrill Lynch to conductfinancialtransactions.Thesesystems,whiletheycanbeannoyingtosomepeople,arereasonablyadeptat respondingappropriately to theoftenambiguousand fragmentedwaypeoplespeak.Microsoftandothercompaniesareofferingsystemsthatallowabusinesstocreatevirtualagentstobookreservationsfortravelandhotelsandconductroutinetransactionsofallkindsthroughtwo-way,reasonablynaturalvoicedialogues.Noteverycallerissatisfiedwiththeabilityofthesevirtualagentstogetthejobdone,butmostsystemsprovideameans togetahumanon the line.Companiesusing thesesystemsreport that theyreducetheneedforhumanserviceagentsupto80percent.Asidefromthemoneysaved,reducingthesizeofcallcentershasamanagementbenefit.Call-centerjobshaveveryhighturnoverratesbecauseoflowjobsatisfaction.It'ssaidthatmenareloathtoaskothersfordirections,butcarvendorsarebettingthatbothmaleandfemaledriverswillbewillingtoasktheirowncarforhelpingettingtotheirdestination.In2005theAcuraRLandHondaOdysseywillbeofferingasystemfromIBMthatallowsusers toconversewiththeircars.208Drivingdirectionswillincludestreetnames(forexample,"turnleftonMainStreet,then right on Second Avenue"). Users can ask such questions as "Where is the nearest Italianrestaurant?"ortheycanenterspecificlocationsbyvoice,askforclarificationsondirections,andgivecommandstothecaritself(suchas"Turnuptheairconditioning").TheAcuraRLwillalsotrackroadconditionsandhighlighttrafficcongestiononitsscreeninrealtime. The speech recognition is claimed to be speaker-independent and to be unaffected by enginesound,wind,andothernoises.Thesystemwillreportedlyrecognize1.7millionstreetandcitynames,inadditiontonearlyonethousandcommands.Computerlanguagetranslationcontinuestoimprovegradually.BecausethisisaTuring-leveltask—thatis,itrequiresfullhuman-levelunderstandingoflanguagetoperformathumanlevels—itwillbeone of the last application areas to competewith human performance. Franz JosefOch, a computerscientist at theUniversityofSouthernCalifornia,hasdevelopeda technique that cangenerateanewlanguage-translationsystembetweenanypairoflanguagesinamatterofhoursordays.209Allheneedsisa"Rosettastone"—thatis,textinonelanguageandthetranslationofthattextintheotherlanguage—although he needs millions of words of such translated text. Using a self-organizing technique, thesystemisabletodevelopitsownstatisticalmodelsofhowtextistranslatedfromonelanguagetotheotheranddevelopsthesemodelsinbothdirections.Thiscontrastswithothertranslationsystems,inwhichlinguistspainstakinglycodegrammarruleswith long lists of exceptions to each rule. Och's system recently received the highest score in acompetitionoftranslationsystemsconductedbytheU.S.CommerceDepartment'sNationalInstituteof

StandardsandTechnology.EntertainmentandSports.InanamusingandintriguingapplicationofGAs,OxfordscientistTorstenReilcreatedanimatedcreatureswithsimulatedjointsandmusclesandaneuralnetforabrain.Hethenassignedthematask:towalk.HeusedaGAtoevolvethiscapability,whichinvolvedsevenhundredparameters. "Ifyou lookat that systemwithyourhumaneyes, there'snowayyoucando itonyourown,becausethesystemisjusttoocomplex,"Reilpointsout."That'swhereevolutioncomesin."210 While some of the evolved creatures walked in a smooth and convincingway, the researchdemonstrated awell-known attribute ofGAs: you getwhat you ask for. Some creatures figured outnovelnewwaysofpassingforwalking.AccordingtoReil,"Wegotsomecreaturesthatdidn'twalkatall,buthadtheseverystrangewaysofmovingforward:crawlingordoingsomersaults."Softwareisbeingdevelopedthatcanautomaticallyextractexcerptsfromavideoofasportsgamethatshowthemoreimportantplays.211AteamatTrinityCollegeinDublinisworkingontable-basedgameslikepool,inwhichsoftwaretracksthelocationofeachballandisprogrammedtoidentifywhena significant shot has beenmade. A team at the University of Florence is working on soccer. Thissoftwaretracksthelocationofeachplayerandcandeterminethetypeofplaybeingmade(suchasfreekickingorattemptingagoal),whenagoalisachieved,whenapenaltyisearned,andotherkeyevents.TheDigitalBiologyInterestGroupatUniversityCollegeinLondonisdesigningFormulaOneracecarsbybreedingthemusingGAs.212TheAIwinterislongsinceover.WearewellintothespringofnarrowAI.Mostoftheexamplesabovewereresearchprojectsjust tentofifteenyearsago.Ifall theAIsystemsintheworldsuddenlystoppedfunctioning,oureconomicinfrastructurewouldgrindtoahalt.Yourbankwouldceasedoingbusiness.Most transportationwould be crippled.Most communicationswould fail. Thiswas not thecase a decade ago. Of course, our AI systems are not smart enough—yet—to organize such aconspiracy.StrongAI

Ifyouunderstandsomethinginonlyoneway,thenyoudon'treallyunderstanditatall.Thisisbecause,ifsomethinggoeswrong,yougetstuckwithathoughtthatjustsitsinyourmindwithnowheretogo.Thesecretofwhatanythingmeanstousdependsonhowwe'veconnectedit toall theotherthingsweknow.Thisiswhy,whensomeonelearns"byrote,"wesaythattheydon'treallyunderstand.However,ifyouhaveseveraldifferentrepresentationsthen,whenoneapproachfailsyoucantryanother.Ofcourse,making too many indiscriminate connections will turn a mind to mush. But well-connected representations letyou turn ideasaround inyourmind, toenvision thingsfrommany perspectives until you find one that works for you.And that's whatwemeanbythinking!

—MARVINMINSKY213

Advancingcomputerperformanceislikewaterslowlyfloodingthelandscape.Ahalfcenturyagoitbegantodrownthelowlands,drivingouthumancalculatorsandrecordclerks,but leavingmostofusdry.Now the floodhas reached the foothills, andouroutpoststherearecontemplatingretreat.Wefeelsafeonourpeaks,but,atthepresentrate,thosetoowillbesubmergedwithinanotherhalfcentury.IproposethatwebuildArksasthatdaynears,andadoptaseafaringlife!Fornow,though,wemustrelyonourrepresentativesinthelowlandstotelluswhatwaterisreallylike.Ourrepresentativesonthefoothillsofchessandtheorem-provingreportsignsofintelligence.Whydidn'twegetsimilarreportsdecadesbefore,fromthelowlands,as

computerssurpassedhumansinarithmeticandrotememorization?Actually,wedid,atthe time.Computers thatcalculated like thousandsofmathematicianswerehailedas"giantbrains,"andinspiredthefirstgenerationofAIresearch.Afterall,themachineswere doing something beyond any animal, that needed human intelligence,concentrationandyearsoftraining.Butitishardtorecapturethatmagicnow. Onereason is thatcomputers'demonstratedstupidity inotherareasbiasesourjudgment.Anotherrelatestoourownineptitude.Wedoarithmeticorkeeprecordssopainstakinglyandexternallythat thesmallmechanicalstepsinalongcalculationareobvious,whilethebigpictureoftenescapesus.LikeDeepBlue'sbuilders,weseetheprocess toomuch from the inside to appreciate the subtlety that itmay have on theoutside.But there isanonobviousness insnowstormsor tornadoes thatemergefromthe repetitivearithmeticofweathersimulations,or in rippling tyrannosaurskin frommovieanimationcalculations.Werarelycallitintelligence,but"artificialreality"maybeanevenmoreprofoundconceptthanartificialintelligence.Thementalstepsunderlyinggoodhumanchessplayingandtheoremprovingarecomplexandhidden,puttingamechanicalinterpretationoutofreach.Thosewhocanfollowtheplaynaturallydescribe it instead inmentalistic language,using terms likestrategy,understandingandcreativity.Whenamachinemanagestobesimultaneouslymeaningful and surprising in the same rich way, it too compels a mentalisticinterpretation.Ofcourse,somewherebehindthescenes,thereareprogrammerswho,inprinciple,haveamechanicalinterpretation.Butevenforthem,thatinterpretationlosesitsgripastheworkingprogramfillsitsmemorywithdetailstoovoluminousforthemtograsp. Astherisingfloodreachesmorepopulatedheights,machineswillbegintodowell in areas a greater number can appreciate. The visceral sense of a thinkingpresenceinmachinerywillbecomeincreasinglywidespread.Whenthehighestpeaksarecovered,therewillbemachinesthatcaninteractasintelligentlyasanyhumanonanysubject.Thepresenceofmindsinmachineswillthenbecomeself-evident.

—HANSMORAVEC214Because of the exponential nature of progress in information-based technologies, performance oftenshiftsquicklyfrompathetictodaunting.Inmanydiverserealms,astheexamplesintheprevioussectionmakeclear,theperformanceofnarrowAIisalreadyimpressive.Therangeofintelligenttasksinwhichmachinescannowcompetewithhumanintelligenceiscontinuallyexpanding.InacartoonIdesignedforTheAgeofSpiritualMachines,adefensive"humanrace"isseenwritingoutsignsthatstatewhatonlypeople(andnotmachines)cando.215Litteredonthefloorarethesignsthehumanracehasalreadydiscardedbecausemachinescannowperformthesefunctions:diagnoseanelectrocardiogram,composeinthestyleofBach,recognizefaces,guideamissile,playPing-Pong,playmasterchess,pickstocks,improvisejazz,proveimportanttheorems,andunderstandcontinuousspeech.Backin1999thesetaskswerenolongersolelytheprovinceofhumanintelligence;machinescoulddothemall.

Onthewallbehindthemansymbolizingthehumanracearesignshehaswrittenoutdescribingthetasks that were still the sole province of humans: have common sense, review a movie, hold pressconferences,translatespeech,cleanahouse,anddrivecars.Ifweweretoredesignthiscartooninafewyears,someofthesesignswouldalsobelikelytoenduponthefloor.WhenCYCreachesonehundredmillion items of commonsense knowledge, perhaps human superiority in the realm of commonsensereasoningwon'tbesoclear.Theeraofhouseholdrobots,althoughstillfairlyprimitivetoday,hasalreadystarted.Tenyearsfromnow, it's likelywewill consider "clean a house" aswithin the capabilities ofmachines.As fordrivingcars,robotswithnohumaninterventionhavealreadydrivennearlyacrosstheUnitedStatesonordinary roads with other normal traffic. We are not yet ready to turn over all steering wheels tomachines,butthereareseriousproposalstocreateelectronichighwaysonwhichcars(withpeopleinthem)willdrivebythemselves.Thethreetasksthathavetodowithhuman-levelunderstandingofnaturallanguage—reviewingamovie, holding a press conference, and translating speech—are themost difficult.Oncewe can takedownthesesigns,we'llhaveTuring-levelmachines,andtheeraofstrongAIwillhavestarted.Thiserawillcreepuponus.Aslongasthereareanydiscrepanciesbetweenhumanandmachineperformance—areas in which humans outperformmachines—strong AI skeptics will seize on thesedifferences.ButourexperienceineachareaofskillandknowledgeislikelytofollowthatofKasparov.Our perceptions of performance will shift quickly from pathetic to daunting as the knee of theexponentialcurveisreachedforeachhumancapability. Howwill strongAI be achieved?Most of thematerial in this book is intended to layout thefundamental requirements forbothhardwareand softwareandexplainwhywecanbeconfident thattheserequirementswillbemetinnonbiologicalsystems.Thecontinuationoftheexponentialgrowthoftheprice-performanceofcomputationtoachievehardwarecapableofemulatinghumanintelligencewasstill controversial in 1999.Therehasbeen somuchprogress in developing the technology for three-dimensionalcomputingoverthepastfiveyearsthatrelativelyfewknowledgeableobserversnowdoubtthat thiswillhappen.Even just taking thesemiconductor industry'spublishedITRSroadmap,whichrunsto2018,wecanprojecthuman-levelhardwareatreasonablecostbythatyear.216I'vestatedthecaseinchapter4ofwhywecanhaveconfidencethatwewillhavedetailedmodelsandsimulationsofallregionsofthehumanbrainbythelate2020s.Untilrecently,ourtoolsforpeeringinto the brain did not have the spatial and temporal resolution, bandwidth, or price-performance toproduce adequate data to create sufficiently detailed models. This is now changing. The emerginggenerationofscanningandsensingtoolscananalyzeanddetectneuronsandneuralcomponentswithexquisiteaccuracy,whileoperatinginrealtime.Futuretoolswillprovidefargreaterresolutionandcapacity.Bythe2020s,wewillbeabletosendscanningandsensingnanobotsintothecapillariesofthebraintoscanitfrominside.We'veshowntheability to translate the data from diverse sources of brain scanning and sensing into models andcomputer simulations that hold up well to experimental comparison with the performance of thebiologicalversionsof these regions.Wealreadyhavecompellingmodels and simulations for severalimportantbrainregions.AsIarguedinchapter4,it'saconservativeprojectiontoexpectdetailedandrealisticmodelsofallbrainregionsbythelate2020s.OnesimplestatementofthestrongAIscenarioisthat.wewilllearntheprinciplesofoperationofhumanintelligencefromreverseengineeringallthebrain'sregions,andwewillapplytheseprinciplesto the brain-capable computing platforms thatwill exist in the 2020s.We already have an effectivetoolkit for narrow AI. Through the ongoing refinement of these methods, the development of newalgorithms, and the trend toward combiningmultiplemethods into intricate architectures, narrowAIwill continue to become less narrow. That is, AI applications will have broader domains, and theirperformancewillbecomemore flexible.AI systemswilldevelopmultiplewaysof approachingeachproblem, just as humans do. Most important, the new insights and paradigms resulting from theaccelerationofbrainreverseengineeringwillgreatlyenrichthissetoftoolsonanongoingbasis.Thisprocessiswellunderway.

It'softensaidthatthebrainworksdifferentlyfromacomputer,sowecannotapplyourinsightsabout brain function intoworkable nonbiological systems.This view completely ignores the field ofself-organizingsystems,forwhichwehaveasetofincreasinglysophisticatedmathematicaltools.AsIdiscussedinthepreviouschapter, thebraindiffersinanumberofimportantwaysfromconventional,contemporarycomputers.IfyouopenupyourPalmPilotandcutawire,there'sagoodchanceyouwillbreakthemachine.Yetweroutinelylosemanyneuronsandinterneuronalconnectionswithnoilleffect,becausethebrainisself-organizingandreliesondistributedpatternsinwhichmanyspecificdetailsarenotimportant.Whenwegettothemid-tolate2020s,wewillhaveaccesstoagenerationofextremelydetailedbrain-region models. Ultimately the toolkit will be greatly enriched with these new models andsimulationsandwillencompassa fullknowledgeofhowthebrainworks.Asweapply the toolkit tointelligenttasks,wewilldrawupontheentirerangeoftools,somederiveddirectlyfrombrainreverseengineering,somemerelyinspiredbywhatweknowaboutthebrain,andsomenotbasedonthebrainatallbutondecadesofAIresearch.Partofthebrain'sstrategyistolearninformation,ratherthanhavingknowledgehard-codedfromthestart.("Instinct"isthetermweusetorefertosuchinnateknowledge.)Learningwillbeanimportantaspect of AI, as well. In my experience in developing pattern-recognition systems in characterrecognition, speech recognition, and financial analysis, providing for the AI's education is the mostchallenging and important part of the engineering. With the accumulated knowledge of humancivilization increasingly accessible online, future Als will have the opportunity to conduct theireducationbyaccessingthisvastbodyofinformation.TheeducationofAIswillbemuchfasterthanthatofunenhancedhumans.Thetwenty-yeartimespanrequiredtoprovideabasiceducationtobiologicalhumanscouldbecompressedintoamatterofweeks or less. Also, because nonbiological intelligence can share its patterns of learning andknowledge, only oneAI has tomaster each particular skill.As I pointed out,we trained one set ofresearchcomputers tounderstandspeech,but thenthehundredsof thousandsofpeoplewhoacquiredourspeech-recognitionsoftwarehadtoloadonlythealreadytrainedpatternsintotheircomputers. Oneof themanyskills thatnonbiological intelligencewill achievewith thecompletionof thehumanbrainreverse-engineeringprojectissufficientmasteryoflanguageandsharedhumanknowledgetopasstheTuringtest.TheTuringtestisimportantnotsomuchforitspracticalsignificancebutratherbecauseitwilldemarcateacrucialthreshold.AsIhavepointedout,thereisnosimplemeanstopassaTuring test, other than to convincingly emulate the flexibility, subtlety, and suppleness of humanintelligence.Havingcapturedthatcapabilityinourtechnology,itwillthenbesubjecttoengineering'sabilitytoconcentrate,focus,andamplifyit. Variationsof theTuring testhavebeenproposed.TheannualLoebnerPrize contest awardsabronze prize to the chatterbot (conversational bot) best able to convince human judges that it'shuman.217ThecriteriaforwinningthesilverprizeisbasedonTuring'soriginal test,anditobviouslyhasyettobeawarded.Thegoldprizeisbasedonvisualandauditorycommunication.Inotherwords,theAImusthaveaconvincingfaceandvoice,astransmittedoveraterminal,andthusitmustappeartothehumanjudgeasifheorsheisinteractingwitharealpersonoveravideophone.Onthefaceofit,thegoldprizesoundsmoredifficult.I'vearguedthatitmayactuallybeeasier,becausejudgesmaypaylessattention to the text portion of the language being communicated and could be distracted by aconvincingfacialandvoiceanimation.Infact,wealreadyhavereal-timefacialanimation,andwhileitis not quite up to thesemodifiedTuring standards, it's reasonably close.We also have very natural-sounding voice synthesis,which is often confusedwith recordings of human speech, althoughmoreworkisneededonprosodies(intonation).We'relikelytoachievesatisfactoryfacialanimationandvoiceproductionsoonerthantheTuring-levellanguageandknowledgecapabilities.Turingwascarefullyimpreciseinsettingtherulesforhistest,andsignificantliteraturehasbeendevoted to thesubtletiesofestablishingtheexactproceduresfordetermininghowtoassesswhentheTuringtesthasbeenpassed.218In2002InegotiatedtherulesforaTuring-testwagerwithMitchKaporontheLongNowWebsite.219Thequestionunderlyingourtwenty-thousand-dollarbet,theproceedsof

whichgoto thecharityof thewinner'schoice,was,"Will theTuringtestbepassedbyamachineby2029?" I saidyes,andKaporsaidno. It tookusmonthsofdialogue toarriveat the intricate rules toimplementourwager.Simplydefining"machine"and"human,"forexample,wasnotastraightforwardmatter. Is thehumanjudgeallowed tohaveanynonbiological thinkingprocesses inhisorherbrain?Conversely,canthemachinehaveanybiologicalaspects? Because thedefinitionof theTuring testwill vary fromperson toperson,Turing test-capablemachineswillnotarriveonasingleday,andtherewillbeaperiodduringwhichwewillhearclaimsthat machines have passed the threshold. Invariably, these early claims will be debunked byknowledgeableobservers,probably includingmyself.By the time there isabroadconsensus that theTuringtesthasbeenpassed,theactualthresholdwillhavelongsincebeenachieved.EdwardFeigenbaumproposesavariationoftheTuringtest,whichassessesnotamachine'sabilitytopassforhumanincasual,everydaydialoguebutitsabilitytopassforascientificexpertinaspecificfield.220TheFeigenbaum test (FT)maybemore significant than theTuring testbecauseFT-capablemachines, being technically proficient,will be capable of improving their own designs. Feigenbaumdescribeshistestinthisway:

TwoplayersplaytheFTgame.Oneplayerischosenfromamongtheelitepractitionersineachofthreepre-selectedfieldsofnaturalscience,engineering,ormedicine.(Thenumber could be larger, but for this challenge not greater than ten). Let's say wechoose the fields from among those covered in the U.S. National Academy....Forexample,wecouldchooseastrophysics,computerscience,andmolecularbiology.Ineachroundofthegame,thebehaviorofthetwoplayers(elitescientistandcomputer)isjudgedbyanotherAcademymemberinthatparticulardomainofdiscourse,e.g.,anastrophysicistjudgingastrophysicsbehavior.

OfcoursetheidentityoftheplayersishiddenfromthejudgeasitisintheTuringtest.Thejudgeposes problems, asks questions, asks for explanations, theories, and so on—as onemight dowith acolleague.Can the human judge choose, at better than chance level,which is hisNationalAcademycolleagueandwhichisthecomputer?OfcourseFeigenbaumoverlooksthepossibilitythatthecomputermightalreadybeaNationalAcademycolleague,butheisobviouslyassumingthatmachineswillnotyethaveinvadedinstitutionsthat todaycompriseexclusivelybiologicalhumans.WhileitmayappearthattheFTismoredifficultthantheTuringtest,theentirehistoryofAIrevealsthatmachinesstartedwiththeskillsofprofessionalsandonlygraduallymovedtowardthelanguageskillsofachild.EarlyAIsystems demonstrated their prowess initially in professional fields such as proving mathematicaltheorems anddiagnosingmedical conditions.These early systemswouldnot be able to pass theFT,however,becausetheydonothavethelanguageskillsandtheflexibleabilitytomodelknowledgefromdifferentperspectivesthatareneededtoengageintheprofessionaldialogueinherentintheFT.ThislanguageabilityisessentiallythesameabilityneededintheTuringtest.Reasoninginmanytechnicalfields isnotnecessarilymoredifficult thanthecommonsensereasoningengagedinbymosthumanadults. Iwouldexpect thatmachineswillpass theFT,at least insomedisciplines,aroundthesametimeastheypasstheTuringtest.PassingtheFTinalldisciplinesislikelytotakelonger,however.This iswhy I see the2030sasaperiodofconsolidation, asmachine intelligence rapidlyexpands itsskillsandincorporatesthevastknowledgebasesofourbiologicalhumanandmachinecivilization.Bythe 2040s we will have the opportunity to apply the accumulated knowledge and skills of ourcivilizationtocomputationalplatformsthatarebillionsoftimesmorecapablethanunassistedbiologicalhumanintelligence. TheadventofstrongAI is themost important transformation thiscenturywillsee. Indeed, it'scomparable in importance to the advent of biology itself. Itwillmean that a creationofbiologyhasfinally mastered its own intelligence and discovered means to overcome its limitations. Once theprinciplesofoperationofhumanintelligenceareunderstood,expandingitsabilitieswillbeconductedbyhumanscientistsandengineerswhoseownbiologicalintelligencewillhavebeengreatlyamplified

throughan intimatemergerwithnonbiological intelligence.Over time, thenonbiologicalportionwillpredominate.We'vediscussedaspectsoftheimpactofthistransformationthroughoutthisbook,whichIfocuson in the next chapter. Intelligence is the ability to solve problemswith limited resources, includinglimitationsof time.TheSingularitywill be characterizedby the rapid cycleof human intelligence—increasinglynonbiological—capableofcomprehendingandleveragingitsownpowers.FRIENDOFFUTURISTBACTERIUM,2BILLIONB.C.Sotellmeagainabouttheseideasyouhave

aboutthefuture.FUTURISTBACTERIUM,2BILLIONB.C.:Well,Iseebacteriagettingtogetherintosocieties,with

thewholebandofcellsbasicallyactinglikeonebigcomplicatedorganismwithgreatlyenhancedcapabilities.

FRIENDOFFUTURISTBACTERIUM:Whatgivesyouthatidea?FUTURISTBACTERIUM: Well already, some of our fellow Daptobacters have gone inside other

largerbacteriatoformalittleduo.221It'sinevitablethatourfellowcellswillbandtogethersothat each cell can specialize its function. As it is now, we each have to do everything byourselves:findfood,digestit,excreteby-products.

FRIENDOFFUTURISTBACTERIUM:Andthenwhat?FUTURISTBACTERIUM:All thesecellswilldevelopwaysofcommunicatingwithoneanotherthat

gobeyondjusttheswappingofchemicalgradientsthatyouandIcando.FRIEND OF FUTURIST BACTERIUM: Okay, now tell me again the part about that future

superassemblyoftentrillioncells.FUTURISTBACTERIUM:Yes,well,accordingtomymodels,inabouttwobillionyearsabigsociety

oftentrillioncellswillmakeupasingleorganismandincludetensofbillionsofspecialcellsthatcancommunicatewithoneanotherinverycomplicatedpatterns.

FRIENDOFFUTURISTBACTERIUM:Whatsortofpatterns?FUTURISTBACTERIUM:Well,"music,"foronething.Thesehugebandsofcellswillcreatemusical

patternsandcommunicatethemtoalltheotherbandsofcells.FRIENDOFFUTURISTBACTERIUM:Music?FUTURISTBACTERIUM:Yes,patternsofsound.FRIENDOFFUTURISTBACTERIUM:Sound?FUTURISTBACTERIUM: Okay, look at it this way. These supercell societies will be complicated

enough to understand their own organization. Theywill be able to improve their own design,gettingbetterandbetter,fasterandfaster.Theywillreshapetherestoftheworldintheirimage.

FRIEND OF FUTURIST BACTERIUM: Now, wait a second. Sounds like we'll lose our basicbacteriumity.

FUTURISTBACTERIUM:Oh,buttherewillbenoloss.FRIENDOFFUTURISTBACTERIUM:Iknowyoukeepsayingthat,but...FUTURISTBACTERIUM:Itwillbeagreatstepforward.It'sourdestinyasbacteria.And,anyway,

therewillstillbelittlebacterialikeusfloatingaround.FRIENDOFFUTURISTBACTERIUM:Okay,butwhataboutthedownside?Imean,howmuchharm

canourfellowDaptobacterandBdellovibriobacteriado?Butthesefuturecellassociationswiththeirvastreachmaydestroyeverything.

FUTURISTBACTERIUM:It'snotcertain,butIthinkwe'llmakeitthrough.FRIENDOFFUTURISTBACTERIUM:Youalwayswereanoptimist.FUTURISTBACTERIUM: Look, we won't have to worry about the downside for a couple billion

years.FRIENDOFFUTURISTBACTERIUM:Okay,then,let'sgetlunch.MEANWHILE,TWOBILLIONYEARSLATER...

NEDLUDD:ThesefutureintelligenceswillbeworsethanthetextilemachinesIfoughtbackin1812.Back thenwehad toworryaboutonlyonemanwithamachinedoing theworkof twelve.Butyou'retalkingaboutamarble-sizemachineoutperformingallofhumanity.

RAY:Itwillonlyoutperformthebiologicalpartofhumanity.Inanyevent,thatmarbleisstillhuman,evenifnotbiological.

NED:Thesesuperintelligenceswon'teat food.Theywon'tbreatheair.Theywon'treproducethroughsex....Sojusthowaretheyhuman?

RAY:We'regoingtomergewithourtechnology.We'realreadystartingtodothatin2004,evenifmostof themachinesarenotyet insideourbodiesandbrains.Ourmachinesnonethelessextendthereachofourintelligence.Extendingourreachhasalwaysbeenthenatureofbeinghuman.

NED:Look,sayingthatthesesuperintelligentnonbiologicalentitiesarehumanislikesayingthatwe'rebasicallybacteria.Afterall,we'reevolvedfromthemalso.

RAY: It's true that a contemporary human is a collection of cells, and that we are a product ofevolution, indeed its cutting edge. But extending our intelligence by reverse engineering it,modeling it, simulating it, reinstantiating it on more capable substrates, and modifying andextendingitisthenextstepinitsevolution.Itwasthefateofbacteriatoevolveintoatechnology-creatingspecies.Andit'sourdestinynowtoevolveintothevastintelligenceoftheSingularity.

A

CHAPTERSIX

TheImpact...

Thefutureentersintousinordertotransformitselfinuslongbeforeithappens.

—RAINERMARIARILKEOneofthebiggestflawsinthecommonconceptionofthefutureisthatthefutureissomethingthathappenstous,notsomethingwecreate.

—MICHAELANISSIMOV"Playing God" is actually the highest expression of human nature. The urges to improveourselves, tomaster our environment, and to set our childrenon thebest pathpossible havebeenthefundamentaldrivingforcesofallofhumanhistory.Withouttheseurgesto"playGod,"theworldasweknowitwouldn'texisttoday.Afewmillionhumanswouldliveinsavannahsandforests,eekingoutahunter-gathererexistence,withoutwritingorhistoryormathematicsoranappreciationoftheintricaciesoftheirownuniverseandtheirowninnerworkings.

—RAMEZNAAM

Panoply of Impacts. What will be the nature of human experience once nonbiologicalintelligencepredominates?Whataretheimplicationsforthehuman-machinecivilizationwhenstrongAIandnanotechnologycancreateanyproduct,anysituation,anyenvironmentthatwecanimagineatwill?Istresstheroleofimaginationherebecausewewillstillbeconstrainedin

our creations to what we can imagine. But our tools for bringing imagination to life are growingexponentiallymorepowerful.AstheSingularityapproacheswewillhavetoreconsiderourideasaboutthenatureofhumanlifeand redesign our human institutions. We will explore a few of these ideas and institutions in thischapter.Forexample,theintertwinedrevolutionsofG,N,andRwilltransformourfrailversion1.0humanbodiesintotheirfarmoredurableandcapableversion2.0counterparts.Billionsofnanobotswilltravelthrough thebloodstreaminourbodiesandbrains. Inourbodies, theywilldestroypathogens,correctDNAerrors,eliminatetoxins,andperformmanyothertaskstoenhanceourphysicalwell-being.Asaresult,wewillbeabletoliveindefinitelywithoutaging. Inourbrains,themassivelydistributednanobotswillinteractwithourbiologicalneurons.Thiswill provide full-immersion virtual reality incorporating all of the senses, as well as neurologicalcorrelatesofouremotions, fromwithin thenervoussystem.More important, this intimateconnectionbetween our biological thinking and the nonbiological intelligence we are creating will profoundlyexpandhumanintelligence.Warfarewillmovetowardnanobot-basedweapons,aswellascyberweapons.Learningwillfirstmoveonline,butonceourbrainsareonlinewewillbeabletodownloadnewknowledgeandskills.Theroleofworkwillbetocreateknowledgeofallkinds,frommusicandarttomathandscience.Theroleofplaywillbe,well,tocreateknowledge,sotherewon'tbeacleardistinctionbetweenworkandplay.

Intelligenceonandaround theEarthwill continue toexpandexponentiallyuntilwe reach thelimitsofmatterandenergytosupportintelligentcomputation.Asweapproachthislimitinourcornerof the galaxy, the intelligence of our civilization will expand outward into the rest of the universe,quicklyreachingthefastestspeedpossible.Weunderstandthatspeedtobethespeedoflight,butthereare suggestions that wemay be able to circumvent this apparent limit (possibly by taking shortcutsthroughwormholes,forexample)....ontheHumanBody

Somanydifferentpeopletobe.

—DONOVANI1Cosmeticbaby,plugintomeAndnever,everfindanother.AndIrealizenoone'swiseTomyplasticfantasticlover.

—JEFFERSONAIRPLANE,"PLASTICFANTASTICLOVER"Our machines will become much more like us, and we will become much more like ourmachines.

—RODNEYBROOKSOnceoutofnatureIshallnevertakeMybodilyformfromanynaturalthing,ButsuchaformasGreciangoldsmithsmakeOfhammeredgoldandgoldenamelling.

—WILLIAMBUTLERYEATS,"SAILINGTOBYZANTIUM"A radical upgrading of our bodies' physical and mental systems is already under way, usingbiotechnology and emerging genetic-engineering technologies.Beyond the next twodecadeswewillusenanoengineeredmethodssuchasnanobotstoaugmentandultimatelyreplaceourorgans.ANewWayofEating.Sexhaslargelybeenseparatedfromitsbiologicalfunction.Forthemostpart,we engage in sexual activity for intimate communication and sensual pleasure, not reproduction.Conversely,we have devisedmultiplemethods for creating babieswithout physical sex, albeitmostreproductiondoesstillderivefromthesexact.Thisdisentanglementofsexfromitsbiologicalfunctionis not condoned by all sectors of society, but it has been readily, even eagerly, adopted by themainstreaminthedevelopedworld.Sowhydon'tweprovidethesameextricationofpurposefrombiologyforanotheractivitythatalsoprovidesbothsocialintimacyandsensualpleasure—namely,eating?Theoriginalbiologicalpurposeofconsumingfoodwas toprovide thebloodstreamwithnutrients,whichwere thendelivered toeachofour trillions of cells. These nutrients include caloric (energy-bearing) substances such as glucose(mainlyfromcarbohydrates),proteins,fats,andamyriadoftracemolecules,suchasvitamins,minerals,andphytochemicalsthatprovidebuildingblocksandenzymesfordiversemetabolicprocesses.Likeanyothermajorhumanbiologicalsystem,digestionisastonishinginitsintricacy,enablingourbodiestoextractthecomplexresourcesneededtosurvive,despitesharplyvaryingconditions,whileat the same time filtering out a multiplicity of toxins. Our knowledge of the complex pathwaysunderlyingdigestionisrapidlyexpanding,althoughthereisstillagreatdealwedonotfullyunderstand. Butwedoknow that our digestive processes, in particular, are optimized for a period in ourevolutionarydevelopmentthatisdramaticallydissimilartotheoneinwhichwenowfindourselves.Formostofourhistorywefacedahighlikelihoodthatthenextforagingorhuntingseason(andforabrief,

relatively recent period, the next planting season) might be catastrophically lean. It made sense,therefore, for our bodies to hold on to every possible calorie we consumed. Today that biologicalstrategy iscounterproductiveandhasbecome theoutdatedmetabolicprogramming thatunderliesourcontemporary epidemic of obesity and fuels pathological processes of degenerative disease, such ascoronaryarterydiseaseandTypeIIdiabetes. Consider the reasons that the designs of our digestive and other bodily systems are far fromoptimalforcurrentconditions.Untilrecently(onanevolutionarytimescale)itwasnotintheinterestofthespeciesforoldpeoplelikemyself(Iwasbornin1948)touseupthelimitedresourcesoftheclan.Evolutionfavoredashortlifespan—lifeexpectancywasthirty-sevenyearsasrecentlyastwocenturiesago—toallowrestricted reserves tobedevoted to theyoung, thosecaringfor them,and thosestrongenoughtoperformintensephysicalwork.Asdiscussedearlier,theso-calledgrandmahypothesis(whichsuggests that a small number of "wise" elderly members of the tribe were beneficial to the humanspecies)doesnotappreciablychallengetheobservationthattherewasnostrongselectivepressureforgenesthatsignificantlyextendedhumanlongevity.Wenowliveinaneraofgreatmaterialabundance,atleastintechnologicallyadvancednations.Mostworkrequiresmentaleffort rather thanphysicalexertion.Acenturyago30percentof theU.S.workforcewasemployedonfarms,withanother30percentinfactories.Bothofthesefiguresarenowunder3percent.Manyoftoday'sjobcategories,rangingfromflightcontrollertoWebdesigner,simplydidn't exist a century ago. Circa 2004 we have the opportunity to continue to contribute to ourcivilization'sexponentiallygrowingknowledgebase—which is, incidentally,auniqueattributeofourspecies—wellpastourchild-rearingdays.(Asababyboomermyself,thatiscertainlymyview.) Our species has already augmented our natural lifespan through our technology: drugs,supplements,replacementpartsforvirtuallyallbodilysystems,andmanyotherinterventions.Wehavedevices to replace our hips, knees, shoulders, elbows, wrists, jaws, teeth, skin, arteries, veins, heartvalves,arms, legs, feet, fingers,and toes,andsystems toreplacemorecomplexorgans(forexample,ourhearts)arebeginningtobeintroduced.Aswelearntheoperatingprinciplesofthehumanbodyandbrain,wewillsoonbeinapositiontodesignvastlysuperiorsystemsthatwilllastlongerandperformbetter,withoutsusceptibilitytobreakdown,disease,andaging.Oneexampleofaconceptualdesignforsuchasystem,calledPrimoPosthuman,wascreatedbyartistandculturalcatalystNatashaVita-More.3Herdesignisintendedtooptimizemobility,flexibility,andsuperlongevity.Itenvisionsfeaturessuchasametabrainforglobal-netconnectionwithaprostheticneocortexofAIinterwovenwithnanobots,solar-protectedsmartskinthathasbiosensorsfortoneandtexturechangeability,andhigh-acuitysenses.Althoughversion2.0ofthehumanbodyisanongoinggrandprojectthatwillultimatelyresultintheradicalupgradingofallourphysicalandmentalsystems,wewill implement itonesmall,benignstep at a time.Based on our current knowledge,we can describe themeans for accomplishing eachaspectofthisvision.Redesigning the Digestive System. From this perspective, let's return to a consideration of thedigestivesystem.Wealreadyhaveacomprehensivepictureofthecomponentsofthefoodweeat.Weknowhow to enable peoplewho cannot eat to survive, using intravenous nutrition.However, this isclearly not a desirable alternative, since our technologies for getting substances in and out of thebloodstreamarecurrentlyquitelimited.Thenextphaseofimprovementinthisareawillbelargelybiochemical,intheformofdrugsandsupplements thatwillpreventexcesscaloricabsorptionandotherwise reprogrammetabolicpathwaysforoptimalhealth.ResearchbyDr.RonKahnattheJoslinDiabetesCenterhasalreadyidentifiedthe"fatinsulinreceptor"(FIR)gene,whichcontrolsaccumulationoffatbythefatcells.Byblockingtheexpressionofthissinglegeneinthefatcellsofmice,Dr.Kahn'spioneeringresearchhasdemonstratedthattheanimalswereabletoeatwithoutrestrictionyetremainleanandhealthy.Althoughtheyatefarmore than the control mice, the "FIR knockout" mice actually lived 18 percent longer and hadsubstantially lower ratesofheartdiseaseanddiabetes. It'snosurprise thatpharmaceuticalcompanies

arehardatworktoapplythesefindingstothehumanFIRgene.Inanintermediatephasenanobotsinthedigestivetractandbloodstreamwillintelligentlyextracttheprecisenutrientsweneed,orderadditionalnutrientsandsupplementsthroughourpersonalwirelesslocal-areanetwork,andsendtherestofthematterontobeeliminated.Ifthisseemsfuturistic,keepinmindthatintelligentmachinesarealreadymakingtheirwayintoourbloodstream.Therearedozensofprojectsunderwaytocreatebloodstream-basedBioMEMSforawide range of diagnostic and therapeutic applications.4 As mentioned, there are several majorconferencesdevotedtotheseprojects.5BioMEMSdevicesarebeingdesignedtointelligentlyscoutoutpathogensanddelivermedicationsinverypreciseways.Forexample,nanoengineeredblood-bornedevicesthatdeliverhormonessuchasinsulinhavebeendemonstratedinanimals.6SimilarsystemscouldpreciselydeliverdopaminetothebrainforParkinson'spatients,provideblood-clottingfactorsforpatientswithhemophilia,anddelivercancerdrugsdirectlytotumorsites.Onenewdesignprovidesuptotwentysubstance-containingreservoirsthatcanreleasetheircargoatprogrammedtimesandlocationsinthebody.7KensallWise,aprofessorofelectricalengineeringattheUniversityofMichigan,hasdevelopedatinyneuralprobe thatcanprovideprecisemonitoringof theelectricalactivityofpatientswithneuraldiseases.8Futuredesignsarealsoexpected todeliverdrugs toprecise locations in thebrain.KazushiIshiyamaatTohokuUniversity inJapanhasdevelopedmicromachines thatusemicroscopicspinningscrewstodeliverdrugstosmallcancertumors.9AparticularlyinnovativemicromachinedevelopedbySandiaNationalLaboratorieshasmicroteethwithajawthatopensandclosestotrapindividualcellsandthenimplantthemwithsubstancessuchasDNA,proteins,ordrugs.10Manyapproachesarebeingdevelopedformicro-andnanoscalemachinestogointothebodyandbloodstream. Ultimatelywewill be able to determine the precise nutrients (including all the hundreds ofphytochemicals) necessary for the optimal health of each individual. These will be freely andinexpensivelyavailable,sowewon'tneedtobotherwithextractingnutrientsfromfoodatall. Nutrientswillbeintroduceddirectlyintothebloodstreambyspecialmetabolicnanobots,whilesensorsinourbloodstreamandbody,usingwirelesscommunication,willprovidedynamicinformationonthenutrientsneededateachpointintime.Thistechnologyshouldbereasonablymaturebythelate2020s. Akeyquestion indesigning such systemswill be,Howwill nanobotsbe introduced into andremovedfromthebody?Thetechnologieswehavetoday,suchasintravenouscatheters,leavemuchtobedesired.Unlikedrugsandnutritionalsupplements,however,nanobotshaveameasureofintelligenceandcankeeptrackoftheirowninventoriesandintelligentlyslipinandoutofourbodiesincleverways.Onescenario is thatwewouldwearaspecialnutrientdevice inabeltorundershirt,whichwouldbeloadedwithnutrient-bearingnanobotsthatcouldenterthebodythroughtheskinorotherbodycavities. Atthatstageoftechnologicaldevelopment,wewillbeabletoeatwhateverwewant,whatevergivesuspleasureandgastronomicfulfillment,exploringtheculinaryartsfortheirtastes,textures,andaromaswhilehavinganoptimal flowofnutrients toourbloodstream.Onepossibility toachieve thiswould be to have all the food we eat pass through a modified digestive tract that doesn't allowabsorptionintothebloodstream.Butthiswouldplaceaburdenonourcolonandbowelfunctions,soamorerefinedapproachwouldbe todispensewith theconventional functionofelimination.Wecouldaccomplish that by using special elimination nanobots that act like tiny garbage compactors.As thenutrientnanobotsmaketheirwayintoourbodies,theeliminationnanobotsgotheotherway.Suchaninnovationwouldalsoenableustooutgrowtheneedfortheorgansthatfilterthebloodforimpurities,suchasthekidneys.Ultimatelywewon'tneedtobotherwithspecialgarmentsorexplicitnutritionalresources.Justascomputation will be ubiquitous, the basic metabolic nanobot resources we need will be embeddedthroughout our environment. But it will also be important tomaintain ample reserves of all neededresourcesinside thebody.Ourversion1.0bodiesdo this toonlyavery limitedextent—forexample,

storingafewminutes'worthofoxygeninourbloodandafewdays'worthofcaloricenergyinglycogenandotherreserves.Version2.0willprovidesubstantiallygreaterreserves,enablingustobeseparatedfrommetabolicresourcesforgreatlyextendedperiodsoftime. Of course, most of us won't do away with our old-fashioned digestive process when thesetechnologies are first introduced.After all, people didn't throw away their typewriterswhen the firstgeneration of word processors was introduced. However, these new technologies will in due coursedominate.Fewpeopletodaystilluseatypewriter,ahorseandbuggy,awood-burningstove,orotherdisplacedtechnologies(other thanasdeliberateexperiences inantiquity).Thesamephenomenonwillhappenwith our reengineered bodies. Oncewe'veworked out the inevitable complications that willarisewitharadicallyreengineeredgastrointestinalsystem,we'llbegintorelyonitmoreandmore.Ananobot-based digestive system can be introduced gradually, first augmenting our digestive tract,replacingitonlyaftermanyiterations.ProgrammableBlood. One pervasive system that has already been the subject of a comprehensiveconceptual redesign based on reverse engineering is our blood. I mentioned earlier Rob Freitas'snanotechnology-based designs to replace our red blood cells, platelets, andwhite blood cells.11 Likemostofourbiologicalsystemsourredbloodcellsperformtheiroxygenatingfunctionveryinefficiently,soFreitas has redesigned them for optimal performance.Because his respirocytes (robotic red bloodcells)wouldenableonetogohourswithoutoxygen,12itwillbeinterestingtoseehowthisdevelopmentis dealt with in athletic contests. Presumably the use of respirocytes and similar systems will beprohibited in events like the Olympics, but then we will face the prospect of teenagers (whosebloodstreamswilllikelycontainrespirocyte-enrichedblood)routinelyoutperformingOlympicathletes.Althoughprototypesarestillonetotwodecadesinthefuture,theirphysicalandchemicalrequirementshavebeenworkedoutinimpressivedetail.AnalysesshowthatFreitas'sdesignswouldbehundredsorthousandsoftimesmorecapableofstoringandtransportingoxygenthanourbiologicalblood. Freitas also envisionsmicron-size artificial platelets that could achievehomeostasis (bleedingcontrol) up to one thousand times faster than biological platelets do,13 as well as nanorobotic"microbivores" (white-blood-cell replacements) that will download software to destroy specificinfectionshundredsoftimesfasterthanantibioticsandwillbeeffectiveagainstallbacterial,viral,andfungalinfections,aswellascancer,withnolimitationsofdrugresistance.14HaveaHeart,orNot.Thenextorganonourlistforenhancementistheheart,which,whileanintricateandimpressivemachine,hasanumberofsevereproblems.Itissubjecttoamyriadoffailuremodesandrepresents a fundamental weakness in our potential longevity. The heart usually breaks down longbeforetherestofthebody,oftenveryprematurely.Althoughartificialheartsarebeginningtobefeasiblereplacements,amoreeffectiveapproachwillbe togetridof theheartaltogether.AmongFreitas'sdesignsarenanoroboticbloodcells thatprovidetheirownmobility.Ifthebloodmovesautonomously,theengineeringissuesoftheextremepressuresrequired for centralized pumping can be eliminated.Aswe perfectways to transfer nanobots to andfrom the blood supply, we will eventually be able to continuously replace them. Freitas has alsopublished a design for a complex five-hundred-trillion-nanorobot system, called a "vasculoid," thatreplacestheentirehumanbloodstreamwithnonfluid-baseddeliveryofessentialnutrientsandcells.15Energyforthebodywillalsobeprovidedbymicroscopicfuelcells,usingeitherhydrogenorthebody'sownfuel,ATP.AsIdescribedinthelastchapter,substantialprogresshasbeenmaderecentlywithbothMEMS-scaleandnanoscalefuelcells, includingsomethatuse thebody'sownglucoseandATPenergysources.16 Withtherespirocytesprovidinggreatlyimprovedoxygenation,wewillbeabletoeliminatethelungsbyusingnanobotstoprovideoxygenandremovecarbondioxide.Aswithothersystems,wewillgothroughintermediatestageswherethesetechnologiessimplyaugmentournaturalprocesses,sowecan have the best of both worlds. Eventually, though, there will be no reason to continue with the

complicationsofactualbreathingandtheburdensomerequirementofbreathableaireverywherewego.Ifwefindbreathingitselfpleasurable,wecandevelopvirtualwaysofhavingthissensualexperience.Intimewealsowon'tneedthevariousorgansthatproducechemicals,hormones,andenzymesthatflow into the blood and othermetabolic pathways.We can now synthesize bio-identical versions ofmanyof thesesubstances,andwithinone to twodecadeswewillbeable to routinelycreate thevastmajorityofbiochemicallyrelevantsubstances.Wearealreadycreatingartificialhormoneorgans.Forexample, theLawrenceLivermoreNationalLaboratoryandCalifornia-basedMedtronicMiniMedaredevelopinganartificialpancreas tobe implantedunder the skin. Itwillmonitorbloodglucose levelsand release precise amounts of insulin, using a computer program to function like our biologicalpancreaticisletcells.17Inhumanbodyversion2.0hormonesandrelatedsubstances(totheextentthatwestillneedthem)willbedeliveredviananobots,controlledby intelligentbiofeedbacksystems tomaintainandbalancerequiredlevels.Sincewewillbeeliminatingmostofourbiologicalorgans,manyofthesesubstancesmaynolongerbeneededandwillbereplacedbyotherresourcesrequiredbythenanoroboticsystems.SoWhat'sLeft?Let'sconsiderwhereweare,circaearly2030s.We'veeliminatedtheheart,lungs,redand white blood cells, platelets, pancreas, thyroid and all the hormone-producing organs, kidneys,bladder, liver, loweresophagus, stomach,small intestines, large intestines,andbowel.Whatwehaveleft at this point is the skeleton, skin, sex organs, sensory organs,mouth and upper esophagus, andbrain.Theskeletonisastablestructure,andwealreadyhaveareasonableunderstandingofhowitworks.Wecannowreplacepartsofit(forexample,artificialhipsandjoints),althoughtheprocedurerequirespainfulsurgery,andourcurrenttechnologyfordoingsohasseriouslimitations.Interlinkingnanobotswillonedayprovidetheabilitytoaugmentandultimatelyreplacetheskeletonthroughagradualandnoninvasiveprocess.Thehumanskeletonversion2.0willbeverystrong,stable,andself-repairing.Wewillnotnoticetheabsenceofmanyofourorgans,suchastheliverandpancreas,sincewedonot directly experience their operation. But the skin,which includes our primary and secondary sexorgans,mayprovetobeanorganwewillactuallywanttokeep,orwemayatleastwanttomaintainitsvitalfunctionsofcommunicationandpleasure.However,wewillultimatelybeabletoimproveontheskinwithnewnanoengineeredsupplematerials thatwillprovidegreaterprotectionfromphysicalandthermal environmental effects while enhancing our capacity for intimate communication. The sameobservation holds for themouth and upper esophagus,which constitute the remaining aspects of thedigestivesystemthatweusetoexperiencetheactofeating.Redesigning the Human Brain. As we discussed earlier, the process of reverse engineering andredesign will also encompass themost important system in our bodies: the brain.We already haveimplants based on "neuromorphic" modeling (reverse engineering of the human brain and nervoussystem)forarapidlygrowinglistofbrainregions.18ResearchersatMITandHarvardaredevelopingneural implants to replace damaged retinas.19 Implants are available for Parkinson's patients thatcommunicatedirectlywiththeventralposteriornucleusandsubthalmicnucleusregionsofthebraintoreversethemostdevastatingsymptomsofthisdisease.20Animplantforpeoplewithcerebralpalsyandmultiplesclerosiscommunicateswiththeventrallateralthalamusandhasbeeneffectiveincontrollingtremors.21 "Rather than treat the brain like soup, adding chemicals that enhance or suppress certainneurotransmitters,"saysRickTrosch,anAmericanphysicianhelpingtopioneerthesetherapies,"we'renowtreatingitlikecircuitry."Avarietyoftechniquesisalsobeingdevelopedtoprovidethecommunicationsbridgebetweenthewetanalogworldofbiologicalinformationprocessinganddigitalelectronics.ResearchersatGermany'sMaxPlanckInstitutehavedevelopednoninvasivedevicesthatcancommunicatewithneuronsinbothdirections.22Theydemonstratedtheir"neurontransistor"bycontrollingthemovementsofalivingleechfromapersonalcomputer.Similartechnologyhasbeenusedtoreconnectleechneuronsandcoaxthem

toperformsimplelogicalandarithmeticproblems. Scientists are also experimenting with "quantum dots," tiny chips comprising crystals ofphotoconductive(reactivetolight)semiconductormaterialthatcanbecoatedwithpeptidesthatbindtospecificlocationsonneuroncellsurfaces.Thesecouldallowresearcherstouseprecisewavelengthsoflighttoremotelyactivatespecificneurons(fordrugdelivery,forexample),replacinginvasiveexternalelectrodes.23Suchdevelopmentsalsoprovidethepromiseofreconnectingbrokenneuralpathwaysforpeoplewithnervedamageandspinal-cord injuries. Ithad longbeen thought that re-creating thesepathwayswouldbefeasibleonlyforrecentlyinjuredpatients,becausenervesgraduallydeterioratewhenunused.Arecentdiscovery,however,showsthefeasibilityofaneuroprostheticsystemforpatientswithlong-standing spinal-cord injuries. Researchers at the University of Utah asked a group of long-termquadriplegicpatientstomovetheirlimbsinavarietyofwaysandthenobservedtheresponseoftheirbrains,usingmagneticresonanceimaging(MRI).Althoughtheneuralpathwaystotheirlimbshadbeeninactive formanyyears, thepatternsof theirbrainactivitywhenattempting tomove their limbswasveryclosetothoseobservedinnondisabledpersons.24Wewillalsobeabletoplacesensorsinthebrainofaparalyzedpersonthatwillbeprogrammedtorecognize the brain patterns associated with intendedmovements and then stimulate the appropriatesequence ofmuscle actions. For those patients whosemuscles no longer function, there are alreadydesignsfor"nanoelectromechanical"systems(NEMS)thatcanexpandandcontracttoreplacedamagedmusclesandthatcanbeactivatedbyeitherrealorartificialnerves.WeAreBecomingCyborgs. The humanbodyversion 2.0 scenario represents the continuation of along-standing trend inwhichwe growmore intimatewith our technology.Computers started out aslarge,remotemachinesinair-conditionedroomstendedbywhite-coatedtechnicians.Theymovedontoourdesks, thenunderour arms, andnow intoourpockets.Soon,we'll routinelyput them insideourbodiesandbrains.Bythe2030swewillbecomemorenonbiologicalthanbiological.AsIdiscussedinchapter 3, by the 2040s nonbiological intelligence will be billions of times more capable than ourbiologicalintelligence. The compelling benefits of overcoming profound diseases and disabilities will keep thesetechnologiesonarapidcourse,butmedicalapplicationsrepresentonlytheearly-adoptionphase.Asthetechnologiesbecomeestablished,therewillbenobarrierstousingthemforvastexpansionofhumanpotential.StephenHawkingrecentlycommentedintheGermanmagazineFocusthatcomputerintelligencewill surpass that of humanswithin a few decades. He advocated that we "urgently need to developdirect connections to the brain, so that computers can add to human intelligence, rather than be inopposition."25Hawking can take comfort that the development programhe is recommending iswellunderway.Therewillbemanyvariationsofhumanbodyversion2.0,andeachorganandbodysystemwillhave its own course of development and refinement.Biological evolution is only capable ofwhat iscalled "local optimization,"meaning that it can improve a design but onlywithin the constraints ofdesign"decisions" thatbiologyarrivedat longago.Forexample,biological evolution is restricted tobuilding everything fromavery limited class ofmaterials—namely, proteins,which are folded fromone-dimensionalstringsofaminoacids.Itisrestrictedtothinkingprocesses(patternrecognition,logicalanalysis,skill formation,andothercognitiveskills) thatuseextremelyslowchemicalswitching.Andbiological evolution itselfworks very slowly, only incrementally improving designs that continue toapply thesebasic concepts. It is incapableof suddenly changing, for example, to structuralmaterialsmadeofdiamondoidortonanotube-basedlogicalswitching.However,thereisawayaroundthisinherentlimitation.Biologicalevolutiondidcreateaspeciesthat could think andmanipulate its environment. That species is now succeeding in accessing—andimproving—itsowndesignandiscapableofreconsideringandalteringthesebasictenetsofbiology.

Human Body Version 3.0. I envision human body 3.0—in the 2030s and 2040s—as a morefundamental redesign. Rather than reformulating each subsystem, we (both the biological andnonbiological portions of our thinking, working together) will have the opportunity to revamp ourbodiesbasedonourexperiencewithversion2.0.Aswiththetransitionfrom1.0to2.0,thetransitionto3.0willbegradualandwillinvolvemanycompetingideas.OneattributeIenvisionforversion3.0istheabilitytochangeourbodies.We'llbeabletodothatveryeasilyinvirtual-realityenvironments(seethenextsection),butwewillalsoacquirethemeanstodo this in real reality.Wewill incorporateMNT-based fabrication intoourselves, sowe'll beable torapidlyalterourphysicalmanifestationatwill.Evenwithourmostlynonbiologicalbrainswe'relikelytokeeptheaestheticsandemotionalimportofhumanbodies,giventheinfluencethisaesthetichasonthehumanbrain.(Evenwhenextended,thenonbiological portion of our intelligence will still have been derived from biological humanintelligence.)That is, humanbodyversion 3.0 is likely still to look humanby today's standards, butgiventhegreatlyexpandedplasticitythatourbodieswillhave,ideasofwhatconstitutesbeautywillbeexpandeduponovertime.Already,peopleaugmenttheirbodieswithbodypiercing,tattoos,andplasticsurgery, and social acceptance of these changes has rapidly increased. Since we'll be able to makechangesthatarereadilyreversible,thereislikelytobefargreaterexperimentation.J.StorrsHallhasdescribednanobotdesignshecalls"foglets"thatareabletolinktogethertoforma great variety of structures and that can quickly change their structural organization. They're called"foglets"becauseifthere'sasufficientdensityoftheminanarea,theycancontrolsoundandlighttoformvariablesoundsandimages.Theyareessentiallycreatingvirtual-realityenvironmentsexternally(thatis,inthephysicalworld)ratherthaninternally(inthenervoussystem).Usingthemapersoncanmodifyhisbodyorhisenvironment,thoughsomeofthesechangeswillactuallybeillusions,sincethefoglets can control sound and images.26 Hall's foglets are one conceptual design for creating realmorphablebodiestocompetewiththoseinvirtualreality.BILL (ANENVIRONMENTALIST):On this humanbody version 2.0 stuff, aren't you throwing the

baby out—quite literally—with the bathwater? You're suggesting replacing the entire humanbodyandbrainwithmachines.There'snohumanbeingleft.

RAY:We don't agree on the definition of human, but just where do you suggest drawing the line?Augmentingthehumanbodyandbrainwithbiologicalornonbiologicalinterventionsishardlyanewconcept.There'sstillalotofhumansuffering.

BILL:Ihavenoobjectiontoalleviatinghumansuffering.Butreplacingahumanbodywithamachinetoexceedhumanperformanceleavesyouwith,well,amachine.Wehavecarsthatcantravelonthegroundfasterthanahuman,butwedon'tconsiderthemtobehuman.

RAY:Theproblemherehasalottodowiththeword"machine."Yourconceptionofamachineisofsomething that is much less valued—less complex, less creative, less intelligent, lessknowledgeable, less subtle and supple—thana human.That's reasonable for today'smachinesbecauseallthemachineswe'veevermet—likecars—arelikethis.Thewholepointofmythesis,ofthecomingSingularityrevolution,isthatthisnotionofamachine—ofnonbiologicalintelligence—willfundamentallychange.

BILL:Well,that'sexactlymyproblem.Partofourhumannessisourlimitations.Wedon'tclaimtobethe fastestentitypossible, tohavememorieswith thebiggestcapacitypossible,andsoon.Butthere is an indefinable, spiritual quality to being human that a machine inherently doesn'tpossess.

RAY:Again,where do you draw the line?Humans are already replacing parts of their bodies andbrainswithnonbiologicalreplacementsthatworkbetteratperformingtheir"human"functions.

BILL: Better only in the sense of replacing diseased or disabled organs and systems. But you'rereplacing essentially all of our humanness to enhance human ability, and that's inherentlyinhuman.

RAY:Thenperhapsourbasicdisagreementisoverthenatureofbeinghuman.Tome,theessenceof

beinghumanisnotourlimitations—althoughwedohavemany—it'sourabilitytoreachbeyondour limitations.We didn't stay on the ground.We didn't even stay on the planet. Andwe arealreadynotsettlingforthelimitationsofourbiology.

BILL:We have to use these technological powerswith great discretion. Past a certain point, we'relosingsomeineffablequalitythatgiveslifemeaning.

RAY:Ithinkwe'reinagreementthatweneedtorecognizewhat'simportantinourhumanity.Butthereisnoreasontocelebrateourlimitations.

...ontheHumanBrain

Isallwhatweseeorseem,butadreamwithinadream?

—EDGARALLENPOEThe computer programmer is a creator of universes forwhich he alone is the lawgiver. Noplaywright,nostagedirector,noemperor,howeverpowerful,haseverexercisedsuchabsoluteauthority to arrange a stage or a field of battle and to command such unswervingly dutifulactorsortroops.

—JOSEPHWEIZENBAUMOnewindy day twomonkswere arguing about a flapping banner. The first said, "I say thebannerismoving,notthewind."Thesecondsaid,"Isaythewindismoving,notthebanner."Athirdmonkpassedbyandsaid,"Thewindisnotmoving.Thebannerisnotmoving.Yourmindsaremoving."

—ZENPARABLESuppose someone were to say, "Imagine this butterfly exactly as it is, but ugly instead ofbeautiful."

—LUDWIGWITTGENSTEINThe2010Scenario.Computers arriving at thebeginningof thenext decadewill becomeessentiallyinvisible:wovenintoourclothing,embeddedinourfurnitureandenvironment.Theywilltapintotheworldwide mesh (what the World Wide Web will become once all of its linked devices becomecommunicatingWebservers, therebyformingvastsupercomputersandmemorybanks)ofhigh-speedcommunications and computational resources. We'll have very high-bandwidth, wirelesscommunicationtotheInternetatalltimes.Displayswillbebuiltintooureyeglassesandcontactlensesandimagesprojecteddirectlyontoourretinas.TheDepartmentofDefenseisalreadyusingtechnologyalong these lines to create virtual-reality environments in which to train soldiers.27 An impressiveimmersivevirtualrealitysystemalreadydemonstratedbythearmy'sInstituteforCreativeTechnologiesincludesvirtualhumansthatrespondappropriatelytotheuser'sactions.Similartinydeviceswillprojectauditoryenvironments.Cellphonesarealreadybeingintroducedinclothingthatprojectssoundtotheears.28Andthere'sanMP3playerthatvibratesyourskulltoplaymusicthatonlyyoucanhear.29Thearmyhasalsopioneeredtransmittingsoundthroughtheskullfromasoldier'shelmet.Therearealsosystemsthatcanprojectfromadistancesoundthatonlyaspecificpersoncanhear,atechnologythatwasdramatizedbythepersonalizedtalkingstreetadsinthemovieMinorityReport.TheHypersonic Sound technology and the Audio Spotlight systems achieve this bymodulating thesoundonultrasonicbeams,whichcanbepreciselyaimed.Soundisgeneratedbythebeamsinteracting

withair,which restores sound in theaudible range.By focusingmultiple setsofbeamsonawallorothersurface,anewkindofpersonalizedsurroundsoundwithoutspeakersisalsopossible.30Theseresourceswillprovidehigh-resolution,full-immersionvisual-auditoryvirtualrealityatanytime.Wewillalsohaveaugmentedrealitywithdisplaysoverlayingtherealworldtoprovidereal-timeguidanceandexplanations.Forexample,yourretinaldisplaymightremindus,"That'sDr.JohnSmith,directoroftheABCInstitute—youlastsawhimsixmonthsagoattheXYZconference"or,"That'stheTime-LifeBuilding—yourmeetingisonthetenthfloor."We'llhavereal-timetranslationofforeignlanguages,essentiallysubtitlesontheworld,andaccessto many forms of online information integrated into our daily activities. Virtual personalities thatoverlay the realworldwill helpuswith information retrieval andour chores and transactions.Thesevirtual assistantswon't alwayswait for questions and directives butwill step forward if they see usstrugglingtofindapieceofinformation.(Aswewonderabout"Thatactress...whoplayedtheprincess,orwas it the queen ... in thatmoviewith the robot," our virtual assistantmaywhisper in our ear ordisplayinourvisualfieldofview:"NataliePortmanasQueenAmidalainStarWars,episodes1,2,and3.")The 2030 Scenario. Nanobot technology will provide fully immersive, totally convincing virtualreality.Nanobotswill takeuppositions inclosephysicalproximity toeveryinterneuronalconnectioncomingfromoursenses.Wealreadyhavethetechnologyforelectronicdevices tocommunicatewithneurons in both directions, yet requiring no direct physical contact with the neurons. For example,scientistsattheMaxPlanckInstitutehavedeveloped"neurontransistors"thatcandetectthefiringofanearby neuron, or alternatively can cause a nearby neuron to fire or suppress it from firing.31 Thisamounts to two-waycommunicationbetweenneuronsand theelectronic-basedneuron transistors.Asmentioned above, quantum dots have also shown the ability to provide noninvasive communicationbetweenneuronsandelectronics.32Ifwewanttoexperiencerealreality,thenanobotsjuststayinposition(inthecapillaries)anddonothing.Ifwewanttoentervirtualreality,theysuppressalloftheinputscomingfromouractualsensesandreplacethemwiththesignalsthatwouldbeappropriateforthevirtualenvironment.33Yourbrainexperiences these signals as if they came from your physical body. After all, the brain does notexperience the body directly. As I discussed in chapter 4, inputs from the body—comprising a fewhundred megabits per second—representing information about touch, temperature, acid levels, themovementoffood,andotherphysicalevents,streamthroughtheLamina1neurons,thenthroughtheposterior ventromedial nucleus, ending up in the two insula regions of cortex. If these are codedcorrectly—andwewillknowhowtodothatfromthebrainreverse-engineeringeffort—yourbrainwillexperiencethesyntheticsignalsjustasitwouldrealones.Youcoulddecidetocauseyourmusclesandlimbs tomove as you normallywould, but the nanobotswould intercept these interneuronal signals,suppress your real limbs frommoving, and instead cause your virtual limbs tomove, appropriatelyadjusting your vestibular system and providing the appropriate movement and reorientation in thevirtualenvironment.TheWebwillprovideapanoplyofvirtualenvironmentstoexplore.Somewillbere-creationsofrealplaces;otherswillbefancifulenvironmentsthathavenocounterpartinthephysicalworld.Some,indeed,wouldbeimpossible,perhapsbecausetheyviolatethelawsofphysics.Wewillbeabletovisitthesevirtualplacesandhaveanykindof interactionwithother real,aswellassimulated,people (ofcourse, ultimately there won't be a clear distinction between the two), ranging from businessnegotiationstosensualencounters."Virtual-realityenvironmentdesigner"willbeanewjobdescriptionandanewartform.BecomeSomeoneElse.Invirtualrealitywewon'tberestrictedtoasinglepersonality,sincewewillbeabletochangeourappearanceandeffectivelybecomeotherpeople.Withoutalteringourphysicalbody(in real reality) we will be able to readily transform our projected body in these three-dimensionalvirtual environments.We can select different bodies at the same time for different people. So your

parentsmayseeyouasoneperson,whileyourgirlfriendwillexperienceyouasanother.However,theotherpersonmaychoosetooverrideyourselections,preferringtoseeyoudifferentlythanthebodyyouhavechosenforyourself.Youcouldpickdifferentbodyprojectionsfordifferentpeople:BenFranklinforawiseuncle,aclownforanannoyingcoworker.Romanticcouplescanchoosewhomtheywishtobe,eventobecomeeachother.Thesearealleasilychangeabledecisions.Ihadtheopportunitytoexperiencewhatitisliketoprojectmyselfasanotherpersonainavirtual-realitydemonstrationatthe2001TED(technology,entertainment,design)conferenceinMonterey.Bymeansofmagnetic sensors inmyclothinga computerwas able to trackall ofmymovements.Withultrahigh-speedanimationthecomputercreatedalife-size,nearphotorealisticimageofayoungwoman—Ramona—whofollowedmymovementsinrealtime.Usingsignal-processingtechnology,myvoicewas transformed into a woman's voice and also controlled the movements of Ramona's lips. So itappearedtotheTEDaudienceasifRamonaherselfweregivingthepresentation.34Tomaketheconceptunderstandable,theaudiencecouldseemeandseeRamonaatthesametime,both moving simultaneously in exactly the same way. A band came onstage, and I—Ramona—performedJeffersonAirplane's"WhiteRabbit,"aswellasanoriginalsong.Mydaughter,thenfourteen,alsoequippedwithmagneticsensors,joinedme,andherdancemovementsweretransformedintothoseofamalebackupdancer—whohappenedtobeavirtualRichardSaulWurman, theimpresarioof theTEDconference.ThehitofthepresentationwasseeingWurman—notknownforhiship-hopmoves—convincinglydoingmydaughter'sdancesteps.Present in theaudiencewas thecreative leadershipofWarnerBros.,whothenwentoffandcreatedthemovieSimone,inwhichthecharacterplayedbyAIPacinotransformshimselfintoSimoneinessentiallythesameway. Theexperiencewasaprofoundandmovingoneforme.WhenIlookedinthe"cybermirror"(adisplayshowingmewhat theaudiencewasseeing),IsawmyselfasRamonarather thanthepersonIusually see in the mirror. I experienced the emotional force—and not just the intellectual idea—oftransformingmyselfintosomeoneelse.People'sidentitiesarefrequentlycloselytiedtotheirbodies("I'mapersonwithabignose,""I'mskinny,""I'mabigguy,"andsoon).Ifoundtheopportunitytobecomeadifferentpersonliberating.Allofushaveavarietyofpersonalitiesthatwearecapableofconveyingbutgenerallysuppressthemsincewe have no readily available means of expressing them. Today we have very limited technologiesavailable—such as fashion,makeup, and hairstyle—to changewhowe are for different relationshipsand occasions, but our palette of personalities will greatly expand in future full-immersion virtual-realityenvironments. Inadditiontoencompassingallofthesenses,thesesharedenvironmentscanincludeemotionaloverlays. Nanobots will be capable of generating the neurological correlates of emotions, sexualpleasure, and other derivatives of our sensory experience and mental reactions. Experiments duringopenbrain surgeryhavedemonstrated that stimulatingcertain specificpoints in thebrain can triggeremotional experiences (for example, the girl who found everything funny when stimulated in aparticular spot of her brain, as I reported inThe Age of SpiritualMachines).35 Some emotions andsecondary reactions involveapatternof activity in thebrain rather than the stimulationof a specificneuron,butwithmassivelydistributednanobots,stimulatingthesepatternswillalsobefeasible.ExperienceBeamers."Experiencebeamers"willsendtheentireflowoftheirsensoryexperiencesaswellastheneurologicalcorrelatesoftheiremotionalreactionsoutontotheWeb,justaspeopletodaybeamtheirbedroomimagesfromtheirWebcams.Apopularpastimewillbetoplugintosomeoneelse'ssensory-emotionalbeamandexperiencewhatit'sliketobethatperson,alathepremiseofthemovieBeingJohnMalkovich.Therewillalsobeavastselectionofarchivedexperiencestochoosefrom,withvirtual-experiencedesignanothernewartform.ExpandYourMind.Themostimportantapplicationofcirca-2030nanobotswillbeliterallytoexpandourminds through themerger of biological andnonbiological intelligence.The first stagewill be toaugmentourhundredtrillionveryslowinterneuronalconnectionswithhigh-speedvirtualconnections

viananorobotcommunication.36Thiswillprovideuswiththeopportunitytogreatlyboostourpattern-recognition abilities, memories, and overall thinking capacity, as well as to directly interface withpowerfulformsofnonbiologicalintelligence.Thetechnologywillalsoprovidewirelesscommunicationfromonebraintoanother. It is important topointout thatwellbeforetheendof thefirsthalfof thetwenty-firstcentury,thinkingvianonbiological substrateswillpredominate.As I reviewed inchapter3,biologicalhumanthinking is limited to 1016 calculations per second (cps) per human brain (based on neuromorphicmodelingofbrainregions)andabout1026cpsforallhumanbrains.Thesefigureswillnotappreciablychange,evenwithbioengineeringadjustmentstoourgenome.Theprocessingcapacityofnonbiologicalintelligence,incontrast,isgrowingatanexponentialrate(withtherateitselfincreasing)andwillvastlyexceedbiologicalintelligencebythemid-2040s. By that timewewillhavemovedbeyond just theparadigmofnanobots inabiologicalbrain.Nonbiologicalintelligencewillbebillionsoftimesmorepowerful,soitwillpredominate.Wewillhaveversion3.0humanbodies,whichwewillbeabletomodifyandreinstantiateintonewformsatwill.Wewillbeabletoquicklychangeourbodiesinfull-immersionvisual-auditoryvirtualenvironmentsintheseconddecadeof this century; in full-immersionvirtual-reality environments incorporating all of thesensesduringthe2020s;andinrealrealityinthe2040s.Nonbiologicalintelligenceshouldstillbeconsideredhuman,sinceitisfullyderivedfromhuman-machine civilization andwill be based, at least in part, on reverse engineering human intelligence. Iaddress this important philosophical issue in the next chapter. The merger of these two worlds ofintelligence is not merely a merger of biological and non biological thinking mediums, but moreimportant,oneofmethodandorganizationof thinking,one thatwillbeable toexpandourminds invirtuallyanyimaginableway. Ourbrains todayare relatively fixed indesign.Althoughwedo addpatternsof interneuronalconnections andneurotransmitter concentrations as anormalpart of the learningprocess, the currentoverallcapacityofthehumanbrainishighlyconstrained.Asthenonbiologicalportionofourthinkingbeginstopredominatebytheendofthe2030s,wewillbeabletomovebeyondthebasicarchitectureofthe brain's neural regions. Brain implants based on massively distributed intelligent nanobots willgreatlyexpandourmemoriesandotherwisevastlyimproveallofoursensory,pattern-recognition,andcognitive abilities. Since the nanobotswill be communicatingwith one another, theywill be able tocreate any set of new neural connections, break existing connections (by suppressing neural firing),create newhybrid biological-nonbiological networks, and add completely nonbiological networks, aswellasinterfaceintimatelywithnewnonbiologicalformsofintelligence.Theuseofnanobotsasbrainextenderswillbeasignificantimprovementoversurgicallyinstalledneural implants,whicharebeginning tobeused today.Nanobotswillbe introducedwithout surgery,throughthebloodstream,andifnecessarycanallbedirectedtoleave,sotheprocessiseasilyreversible.They are programmable, in that they can provide virtual reality one minute and a variety of brainextensions the next. They can change their configuration and can alter their software. Perhapsmostimportant,theyaremassivelydistributedandthereforecantakeupbillionsofpositionsthroughoutthebrain, whereas a surgically introduced neural implant can be placed only in one or at most a fewlocations.MOLLY2004: Full-immersion virtual reality doesn't seem very inviting. Imean, all those nanobots

runningaroundinmyhead,likelittlebugs.RAY:Oh,youwon'tfeelthem,anymorethanyoufeeltheneuronsinyourheadorthebacteriainyour

GItract.MOLLY2004:Actually,thatIcanfeel.ButIcanhavefullimmersionwithmyfriendsrightnow,just

by,youknow,gettingtogetherphysically.SIGMUNDFREUD:Hmmm, that's what they used to say about the telephone when I was young.

Peoplewouldsay,"Whoneedstotalktosomeonehundredsofmilesawaywhenyoucanjustgettogether?"

RAY:Exactly,thetelephoneisauditoryvirtualreality.Sofull-immersionVRis,basically,afull-bodytelephone.Youcangettogetherwithanyoneanytimebutdomorethanjusttalk.

GEORGE2048:It's certainly beena boon for sexworkers; they never have to leave their homes. Itbecame so impossible to draw anymeaningful lines that the authorities had no choice but tolegalizevirtualprostitutionin2033.

MOLLY2004:Veryinterestingbutactuallynotveryappealing.GEORGE2048:Okay,butconsiderthatyoucanbewithyourfavoriteentertainmentstar.MOLLY2004:IcandothatinmyimaginationanytimeIwant.RAY:Imaginationisnice,buttherealthing—or,rather,thevirtualthing—issomuchmore,well,real.MOLLY2004:Yeah,butwhatifmy"favorite"celebrityisbusy?RAY:That'sanotherbenefitofvirtualrealitycirca2029;youhaveyourchoiceofmillionsofartificial

people.MOLLY2104:Iunderstandthatyou'rebackin2004,butwekindofgotridofthatterminologyback

whentheNonbiologicalPersonsActwaspassedin2052.Imean,we'realotmorerealthan...umm,letmerephrasethat.

MOLLY2004:Yes,maybeyoushould.MOLLY2104:Let'sjustsaythatyoudon'thavetohaveexplicitbiologicalstructurestobe—GEORGE2048:—passionate?MOLLY2104:Iguessyoushouldknow.TIMOTHYLEARY:Whatifyouhaveabadtrip?RAY:Youmean,somethinggoesawrywithavirtual-realityexperience?TIMOTHY:Exactly.RAY:Well,youcanleave.It'slikehanginguponaphonecall.MOLLY2004:Assumingyoustillhavecontroloverthesoftware.RAY:Yes,wedoneedtobeconcernedwiththat.SIGMUND:Icanseesomerealtherapeuticpotentialhere.RAY:Yes,youcanbewhomeveryouwanttobeinvirtualreality.SIGMUND:Excellent,theopportunitytoexpresssuppressedlongings...RAY:Andnotonlytobewiththepersonyouwanttobewith,buttobecomethatperson.SIGMUND: Exactly.We create the objects of our libido in our subconscious anyway. Just think, a

couplecouldbothchangetheirgenders.Theycouldeachbecometheother.MOLLY2004:Justasatherapeuticinterlude,Ipresume?SIGMUND:Ofcourse.Iwouldonlysuggestthisundermycarefulsupervision.MOLLY2004:Naturally.MOLLY2104:Hey,George,rememberwhenweeachbecamealloftheoppositegendercharactersin

theAllenKurzweilnovelsatthesametimeit?GEORGE2048:Ha, I liked you best as that eighteenth-century French inventor, the onewhomade

eroticpocketwatches!MOLLY2004:Okay,nowrunthisvirtualsexbymeagain.Howdoesitworkexactly?RAY:You'reusingyourvirtualbody,whichissimulated.Nanobotsinandaroundyournervoussystem

generate the appropriate encoded signals for all of your senses: visual, auditory, tactile ofcourse,evenolfactory.Fromtheperspectiveofyourbrain,it'srealbecausethesignalsarejustasrealas ifyoursenseswereproducing themfromrealexperiences.Thesimulation invirtualreality would generally follow the laws of physics, although that would depend on theenvironment you selected. If you go there with another person or persons, then these otherintelligences,whetherofpeoplewithbiologicalbodiesorotherwise,wouldalsohavebodies inthisvirtualenvironment.Yourbodyinvirtualrealitydoesnotneedtomatchyourbodyinrealreality.Infact,thebodyyouchooseforyourselfinthevirtualenvironmentmaybedifferentfromthebodythatyourpartnerchoosesforyouatthesametime.Thecomputersgeneratingthevirtualenvironment,virtualbodies,andassociatednervesignalswouldcooperatesothatyouractionsaffectthevirtualexperienceoftheothersandviceversa.

MOLLY2004:So Iwould experience sexual pleasure even though I'm not actually, you know,with

someone?RAY:Well,youwouldbewithsomeone,justnotinrealreality,and,ofcourse,thesomeonemaynot

even exist in real reality. Sexual pleasure is not a direct sensory experience, it's akin to anemotion.It'sasensationgeneratedinyourbrain,whichisreflectingonwhatyou'redoingandthinking,justlikethesensationofhumororanger.

MOLLY 2004: Like the girl you mentioned who found everything hilarious when the surgeonsstimulatedaparticularspotinherbrain?

RAY:Exactly.Thereareneurologicalcorrelatesofallofourexperiences, sensations,andemotions.Somearelocalizedwhereassomereflectapatternofactivity.Ineithercasewe'llbeabletoshapeandenhanceouremotionalreactionsaspartofourvirtual-realityexperiences.

MOLLY 2004: That could work out quite well. I think I'll enhance my funniness reaction in myromanticinterludes.Thatwillfitjustaboutright.Ormaybemyabsurdityresponse—Ikindoflikethatone,too.

NEDLUDD:Icanseethisgettingoutofhand.Peoplearegoingtostartspendingmostoftheirtimeinvirtualreality.

MOLLY2004:Oh,Ithinkmyten-year-oldnephewisalreadythere,withhisvideogames.RAY:They'renotfullimmersionyet.MOLLY2004:That'strue.Wecanseehim,butI'mnotsurehenoticesus.Butwhenwegettothepoint

whenhisgamesarefullimmersion,we'llneverseehim.GEORGE2048:Icanseeyourconcernifyou'rethinkingintermsofthethinvirtualworldsof2004,but

it'snotaproblemwithour2048virtualworlds.They'resomuchmorecompellingthantherealworld.

MOLLY2004:Yeah,howwouldyouknowsinceyou'veneverbeeninrealreality?GEORGE2048:Ihearaboutitquiteabit.Anyway,wecansimulateit.MOLLY2104:Well, Icanhavearealbodyany timeIwant,reallynotabigdeal. Ihave tosay it's

ratherliberatingtonotbedependentonaparticularbody,letaloneabiologicalone.Canyouimagine,beingalltiedupwithitsendlesslimitationsandburdens?

MOLLY2004:Yes,Icanseewhereyou'recomingfrom....onHumanLongevity

Itisoneofthemostremarkablethingsthatinallofthebiologicalsciencesthereisnoclueastothe necessity of death. If you say we want to make perpetual motion, we have discoveredenoughlawsaswestudiedphysicstoseethatitiseitherabsolutelyimpossibleorelsethelawsarewrong.But there isnothing inbiologyyet found that indicates the inevitabilityofdeath.Thissuggeststomethatitisnotatallinevitableandthatitisonlyamatteroftimebeforethebiologists discover what it is that is causing us the trouble and that this terrible universaldiseaseortemporarinessofthehuman'sbodywillbecured.

—RICHARDFEYNMANNevergive in,nevergive in,never,never,never,never—innothing,greator small, largeorpetty—nevergivein.

—WINSTONCHURCHILLImmortalityfirst!Everythingelsecanwait.

—CORWYNPRATERInvoluntarydeathisacornerstoneofbiologicalevolution,butthatfactdoesnotmakeitagood

thing.

—MICHAELANISSIMOVSupposeyou'reascientist200yearsagowhohas figuredouthowtodrastically lower infantmortalitywithbetterhygiene.Yougiveatalkonthis,andsomeonestandsupinbackandsays,"hang on, if we do that we're going to have a population explosion!" If you reply, "No,everythingwillbefinebecausewe'llallweartheseabsurdrubberthingswhenwehavesex,"nobodywouldhavetakenyouseriously.Yetthat'sjustwhathappened—barriercontraceptionwaswidelyadopted[aroundthetimethatinfantmortalitydropped].

—AUBREYDEGREY,GERONTOLOGISTWehaveadutytodie.

—DICKLAMM,FORMERGOVERNOROFCOLORADOSomeofusthinkthisisratherapity.

—BERTRAND RUSSEL, 1955, COMMENTING ON THE STATISTICTHAT ABOUT ONE HUNDRED THOUSAND PEOPLE DIE OF AGE-RELATEDCAUSESEVERYDAY38

Evolution,theprocessthatproducedhumanity,possessesonlyonegoal:creategenemachinesmaximally capable of producing copies of themselves. In retrospect, this is the only waycomplexstructuressuchaslifecouldpossiblyariseinanunintelligentuniverse.Butthisgoaloftencomesintoconflictwithhumaninterests,causingdeath,suffering,andshort lifespans.Thepastprogressofhumanityhasbeenahistoryofshatteringevolutionaryconstraints.

—MICHAELANISSIMOV

MostofthereadersofthisbookarelikelytobearoundtoexperiencetheSingularity.Aswereviewedinthepreviouschapter,acceleratingprogressinbiotechnologywillenableustoreprogramourgenesandmetabolicprocessestoturnoffdiseaseandagingprocesses.Thisprogresswillincluderapidadvancesingenomics(influencinggenes),proteomics(understandingandinfluencingtheroleofproteins),genetherapy (suppressing gene expressionwith such technologies asRNA interference and inserting newgenes into thenucleus), rationaldrugdesign(formulatingdrugs that targetprecisechanges indiseaseandagingprocesses),and therapeuticcloningof rejuvenated (telomere-extendedandDNA-corrected)versionsofourowncells,tissues,andorgans,andrelateddevelopments. Biotechnologywillextendbiologyandcorrectitsobviousflaws.Theoverlappingrevolutionofnanotechnologywillenableustoexpandbeyondtheseverelimitationsofbiology.AsTerryGrossmanandIarticulated inFantasticVoyage:LiveLongEnough toLiveForever,wearerapidlygaining theknowledgeandthetoolstoindefinitelymaintainandextendthe"house"eachofuscallshisbodyandbrain.Unfortunatelythevastmajorityofourbaby-boomerpeersareunawareofthefactthattheydonothave to suffer and die in the "normal" course of life, as prior generations have done—if they takeaggressive action, action that goes beyond the usual notion of a basically healthy lifestyle (see"ResourcesandContactInformation,"p.489).Historically,theonlymeansforhumanstooutlivealimitedbiologicallifespanhasbeentopassonvalues,beliefs,andknowledgetofuturegenerations.Wearenowapproachingaparadigmshiftinthemeanswewillhaveavailabletopreservethepatternsunderlyingourexistence.Humanlifeexpectancyis itself growing steadily andwill accelerate rapidly, now thatwe are in the early stages of reverseengineering the information processes underlying life and disease. Robert Freitas estimates thateliminating a specific list comprising 50 percent of medically preventable conditions would extendhuman life expectancy to over 150 years.39 By preventing 90 percent of medical problems, lifeexpectancygrowstooverfivehundredyears.At99percent,we'dbeoveronethousandyears.Wecanexpect that the full realizationof thebiotechnologyandnanotechnologyrevolutionswillenableus toeliminatevirtuallyallmedicalcausesofdeath.Aswemovetowardanonbiologicalexistence,wewillgain themeansof "backingourselvesup" (storing thekeypatternsunderlyingourknowledge, skills,andpersonality),therebyeliminatingmostcausesofdeathasweknowit.

TheTransformationtoNonbiologicalExperience

Amindthatstaysatthesamecapacitycannotliveforever;afterafewthousandyearsitwouldlookmore likea repeating tape loop thanaperson.To live indefinitely long, themind itselfmustgrow,...andwhenitbecomesgreatenough,andlooksback...whatfellowfeelingcanithavewiththesoulthatitwasoriginally?Thelaterbeingwouldbeeverythingtheoriginalwas,butvastlymore.

—VERNORVINGETheempiresofthefuturearetheempiresofthemind.

—WINSTONCHURCHILLIreportedonbrainuploadinginchapter4.Thestraightforwardbrain-portingscenarioinvolvesscanninga human brain (most likely fromwithin), capturing all of the salient details, and reinstantiating thebrain'sstateinadifferent—mostlikelymuchmorepowerful—computationalsubstrate.Thiswillbeafeasibleprocedureandwillhappenmostlikelyaroundthelate2030s.ButthisisnottheprimarywaythatIenvisionthetransitiontononbiologicalexperiencetakingplace.Itwillhappen,rather,inthesamewaythatallotherparadigmshiftshappen:gradually(butatanacceleratingpace). AsIpointedoutabove, theshift tononbiological thinkingwillbeaslipperyslope,butoneonwhich we have already started. We will continue to have human bodies, but they will becomemorphableprojectionsofourintelligence.Inotherwords,oncewehaveincorporatedMNTfabricationintoourselves,wewillbeabletocreateandre-createdifferentbodiesatwill.Howeverachieved,willsuchfundamentalshiftsenableustoliveforever?Theanswerdependsonwhatwemeanby"living"and"dying."Considerwhatwedotodaywithourpersonalcomputerfiles.Whenwechangefromanoldercomputertoanewerone,wedon'tthrowallourfilesaway.Rather,wecopythemandreinstallthemonthenewhardware.Althoughoursoftwaredoesnotnecessarilycontinueitsexistenceforever,itslongevityisinessenceindependentofanddisconnectedfromthehardwarethatitrunson. Currently,when our humanhardware crashes, the software of our lives—our personal "mindfile"—dieswithit.However,thiswillnotcontinuetobethecasewhenwehavethemeanstostoreandrestore the thousands of trillions of bytes of information represented in the pattern that we call ourbrains (togetherwith the rest of our nervous system, endocrine system, and other structures that ourmindfilecomprises). At thatpoint thelongevityofone'smindfilewillnotdependonthecontinuedviabilityofanyparticular hardware medium (for example, the survival of a biological body and brain). Ultimately

software-basedhumanswillbevastlyextendedbeyond thesevere limitationsofhumansasweknowthem today. They will live out on the Web, projecting bodies whenever they need or want them,including virtual bodies in diverse realms of virtual reality, holographically projected bodies, foglet-projectedbodies,andphysicalbodiescomprisingnanobotswarmsandotherformsofnanotechnology. Bythemiddleofthetwenty-firstcenturyhumanswillbeabletoexpandtheir thinkingwithoutlimit.Thisisaformofimmortality,althoughit isimportanttopointoutthatdataandinformationdonot necessarily last forever: the longevity of information depends on its relevance, utility, andaccessibility.Ifyou'veevertriedtoretrieveinformationfromanobsoleteformofdatastorageinanold,obscureformat(forexample,areelofmagnetic tapefroma1970minicomputer),youunderstandthechallengesinkeepingsoftwareviable.However,ifwearediligentinmaintainingourmindfile,makingfrequentbackups,andportingtocurrentformatsandmediums,aformofimmortalitycanbeattained,atleastforsoftware-basedhumans.Laterinthiscenturyitwillseemremarkabletopeoplethathumansinanearlierera lived their liveswithoutabackupof theirmostprecious information: that contained intheirbrainsandbodies. Is thisformof immortality thesameconceptasaphysicalhuman,asweknowit today, livingforever?Inonesenseitis,becausetodayone'sselfisnotaconstantcollectionofmatter,either.Recentresearch shows that even our neurons, thought to be relatively long lasting, change all of theirconstituentsubsystems,suchasthetubules,inamatterofweeks.Onlyourpatternofmatterandenergypersists, and even that gradually changes. Similarly, it will be the pattern of a software human thatpersistsanddevelopsandslowlyalters.Butisthatpersonbasedonmymindfile,whomigratesacrossmanycomputationalsubstratesandwhooutlivesanyparticularthinkingmedium,reallyme?Thisconsiderationtakesusbacktothesamequestions of consciousness and identity that have been debated since Plato's dialogues (which weexamineinthenextchapter).Duringthecourseofthetwenty-firstcenturythesewillnotremaintopicsfor polite philosophical debates butwill have to be confronted as vital, practical, political, and legalissues.Arelatedquestion:Isdeathdesirable?The"inevitability"ofdeathisdeeplyingrainedinhumanthinking. If death seems unavoidable, we have little choice but to rationalize it as necessary, evenennobling.ThetechnologyoftheSingularitywillprovidepracticalandaccessiblemeansforhumanstoevolve into somethinggreater, sowewill no longerneed to rationalizedeath as aprimarymeansofgivingmeaningtolife.TheLongevityofInformation

"The horror of thatmoment," theKingwent on, "I shall never, never forget it!" "Youwill,though,"theQueensaid,"ifyoudon'tmakeamemorandumofit."

—LEWISCARROLL,THROUGHTHELOOKING-GLASSTheonlythingsyoucanbesureof,sothesayinggoes,aredeathandtaxes—butdon'tbetoosureaboutdeath.

—JOSEPHSTROUT,NEUROSCIENTISTIdonotknowsire,butwhatevertheywillturnouttobeIamsureyouwilltaxthem.

—MICHAELFARADAY,RESPONDINGTOAQUESTIONFROMTHEBRITISHEXCHEQUERASTOWHATPRACTICALUSECOULDBEMADEOFHISDEMONSTRATIONOFELECTROMAGNETISM

Donotgogentleintothatgoodnight,...Rage,rageagainstthedyingofthelight.

—DYLANTHOMAS

Theopportunitytotranslateour lives,ourhistory,our thoughts,andourskills intoinformationraisestheissueofhowlonginformationlasts.I'vealwaysreveredknowledgeandgatheredinformationofallkindsasachild,aninclinationIsharedwithmyfather.Bywayofbackground,myfatherwasoneofthosepeoplewholikedtostorealltheimagesandsoundsthatdocumentedhislife.Uponhisuntimelydeathattheageoffifty-eightin1970,Iinheritedhisarchives,whichItreasuretothisday.Ihavemyfather's1938doctoraldissertationfromtheUniversityof Vienna, which contains his unique insights into the contributions of Brahms to our musicalvocabulary.TherearealbumsofneatlyarrangednewspaperclippingsofhisacclaimedmusicalconcertsasateenagerinthehillsofAustria.ThereareurgentletterstoandfromtheAmericanmusicpatronwhosponsoredhisflightfromHitler,justbeforeKristallnachtandrelatedhistoricaldevelopmentsinEuropein the late 1930s made such escape impossible. These items are among dozens of aging boxescontaining a myriad of remembrances, including photographs, musical recordings on vinyl andmagnetictape,personalletters,andevenoldbills. Ialso inheritedhispenchant forpreserving the recordsofone's life, soalongwithmyfather'sboxesIhaveseveralhundredboxesofmyownpapersandfiles.Myfather'sproductivity,assistedonlybythetechnologyofhismanualtypewriterandcarbonpaper,cannotcomparewithmyownprolificacy,aidedandabettedbycomputersandhigh-speedprintersthatcanreproducemythoughtsinallkindsofpermutations.Tuckedawayinmyownboxesarealsovariousformsofdigitalmedia:punchcards,paper-tapereels, and digital magnetic tapes and disks of various sizes and formats. I often wonder just howaccessible this information remains. Ironically the ease of approaching this information is inverselyproportionaltothelevelofadvancementofthetechnologyusedtocreateit.Moststraightforwardarethepaperdocuments,whichalthoughshowingsignsofageareeminentlyreadable.Onlyslightlymorechallenging are the vinyl records and analog tape recordings. Although some basic equipment isrequired,itisnotdifficulttofindoruse.Thepunchcardsaresomewhatmorechallenging,butit'sstillpossibletofindpunch-cardreaders,andtheformatsareuncomplicated.Byfarthemostdemandinginformationtoretrieveisthatcontainedonthedigitaldisksandtapes.Considerthechallengesinvolved.ForeachmediumIhavetofigureoutexactlywhichdiskortapedrivewas used, whether an IBM 1620 circa 1960 or a Data General Nova I circa 1973.Then, once I'veassembledtherequisiteequipment,therearelayersofsoftwaretodealwith:theappropriateoperatingsystem,diskinformationdrivers,andapplicationprograms.And,whenIrunintotheinevitablescoresof problems inherent in each layer of hardware and software, just whom am I going to call forassistance?It'shardenoughgettingcontemporarysystemstowork,letalonesystemsforwhichthehelpdesksweredisbandeddecadesago(iftheyeverexisted).EvenattheComputerHistoryMuseummostofthedevicesondisplaystoppedfunctioningmanyyearsago.41AssumingIdoprevailagainstalloftheseobstacles,Ihavetoaccountforthefactthattheactualmagneticdataonthediskshasprobablydecayedandthattheoldcomputerswouldstillgeneratemostlyerrormessages.42Butistheinformationgone?Theansweris,Notentirely.Eventhoughthemagneticspotsmayno longer be readable by theoriginal equipment, the faded regions couldbe enhancedbysuitablysensitiveequipment,viamethodsthatareanalogoustotheimageenhancementoftenappliedtothepagesofoldbookswhentheyarescanned.Theinformationisstillthere,althoughverydifficulttogetat.Withenoughdevotionandhistoricalresearch,onemightactuallyretrieveit.Ifwehadreasontobelieve that one of these disks contained secrets of enormous value,wewould probably succeed inrecoveringtheinformation.Butmerenostalgiaisunlikelytobesufficienttomotivateanyonetoundertakethisformidabletask.IwillsaythatbecauseIdidlargelyanticipatethisdilemma,Ididmakepaperprintoutsofmostoftheseoldfiles.Butkeepingallourinformationonpaperisnottheanswer,ashard-copyarchivespresenttheirownsetofproblems.AlthoughIcanreadilyreadevenacentury-oldpapermanuscriptifI'mholdingit

inmyhand,findingadesireddocumentfromamongthousandsofonlymodestlyorganizedfilefolderscanbeafrustratingandtime-consumingtask.Itcantakeanentireafternoontolocatetherightfolder,not to mention the risk of straining one's back from moving dozens of heavy file boxes. Usingmicrofilm or microfiche may alleviate some of the difficulty, but the matter of locating the rightdocumentremains.Ihavedreamedoftakingthesehundredsofthousandsofrecordsandscanningthemintoamassivepersonaldatabase,whichwouldallowmetoutilizepowerfulcontemporarysearch-and-retrievemethodsonthem.Ievenhaveanameforthisventure—DAISI(DocumentandImageStorageInvention)—andhave been accumulating ideas for it for many years. Computer pioneer Gordon Bell (former chiefengineerofDigitalEquipmentCorporation),DARPA(DefenseAdvancedResearchProjectsAgency),andtheLongNowFoundationarealsoworkingonsystemstoaddressthischallenge.43 DAISI will involve the rather daunting task of scanning and patiently cataloging all thesedocuments.ButtherealchallengetomydreamofDAISIissurprisinglydeep:howcanIpossiblyselectappropriatehardwareandsoftwarelayersthatwillgivemetheassurancethatmyarchiveswillbeviableandaccessibledecadesfromnow?Ofcoursemyownarchivalneedsareonlyamicrocosmoftheexponentiallyexpandingknowledgebase that human civilization is accumulating. It is this shared species-wide knowledge base thatdistinguishesusfromotheranimals.Otheranimalscommunicate,buttheydon'taccumulateanevolvingandgrowingbaseofknowledgetopassdowntothenextgeneration.SincewearewritingourpreciousheritageinwhatmedicalinformaticsexpertBryanBergeroncalls"disappearingink,"ourcivilization'slegacywouldappeartobeatgreatrisk.44Thedangerappearstobegrowingexponentiallyalongwiththegrowthofourknowledgebases.Theproblemisfurtherexacerbatedbytheacceleratingspeedwithwhich we adopt new standards in the many layers of hardware and software we employ to storeinformation.Thereisanothervaluablerepositoryofinformationstoredinourbrains.Ourmemoriesandskills,although they may appear to be fleeting, do represent information, coded in vast patterns ofneurotransmitter concentrations, interneuronal connections, and other relevant neural details. Thisinformationisthemostpreciousofall,whichisonereasondeathissotragic.Aswehavediscussed,wewillultimatelybeabletoaccess,permanentlyarchive,aswellasunderstandthethousandsoftrillionsofbytesofinformationwehavetuckedawayineachofourbrains.Copyingourmindstoothermediumsraisesanumberofphilosophicalissues,whichIwilldiscussinthenextchapter—forexample,"Is thatreallymeorrathersomeoneelsewhojusthappenstohavemastered all my thoughts and knowledge?" Regardless of how we resolve these issues, the idea ofcapturing the informationand informationprocesses inourbrainsseems to imply thatwe(orat leastentitiesthatactverymuchlikewedo)could"liveforever."Butisthatreallytheimplication? Foreonsthelongevityofourmentalsoftwarehasbeeninexorablylinkedtothesurvivalofourbiologicalhardware.Beingabletocaptureandreinstantiateallthedetailsofourinformationprocesseswouldindeedseparatethesetwoaspectsofourmortality.Butaswehaveseen,softwareitselfdoesnotnecessarilysurviveforever,andthereareformidableobstaclestoitsenduringverylongatall. Sowhether informationrepresentsoneman'ssentimentalarchive, theaccumulatingknowledgebaseof thehuman-machinecivilization,or themindfilesstored inourbrains,whatcanweconcludeabouttheultimatelongevityofsoftware?Theanswerissimplythis:Informationlastsonlysolongassomeonecaresaboutit.TheconclusionthatI'vecometowithregardtomyDAISIproject,afterseveraldecades of careful consideration, is that there is no set of hardware and software standards existingtoday,noranylikelytocomealong,thatwillprovideanyreasonablelevelofconfidencethatthestoredinformationwill still be accessible (without unreasonable levels of effort) decades from now.45 Theonly way that my archive (or any other information base) can remain viable is if it is continuallyupgradedandportedtothelatesthardwareandsoftwarestandards.Ifanarchiveremainsignored,itwillultimatelybecomeasinaccessibleasmyoldeight-inchPDP-8floppydisks.Informationwillcontinuetorequireconstantmaintenanceandsupporttoremain"alive."Whetherdataorwisdom,informationwillsurviveonlyifwewant it to.Byextension,wecanonlyliveforas

longaswecareaboutourselves.Alreadyourknowledgetocontroldiseaseandagingisadvancedtothepointthatyourattitudetowardyourownlongevityisnowthemostimportantinfluenceonyourlong-rangehealth. Our civilization's trove of knowledge does not simply survive by itself.Wemust continuallyrediscover, reinterpret, and reformat the legacy of culture and technology that our forebears havebestowed on us. All of this information will be fleeting if no one cares about it. Translating ourcurrently hardwired thoughts into software will not necessarily provide us with immortality. It willsimply place the means to determine how long we want our lives and thoughts to last in our ownfigurativehands.MOLLY2004:Sowhatyou'resayingisthatI'mjustafile?MOLLY2104:Well,notastaticfile,butadynamicfile.Butwhatdoyoumean"just"?Whatcouldbe

moreimportant?MOLLY2004:Well,Ithrowfilesawayallthetime,evendynamicones.MOLLY2104:Notallfilesarecreatedequal.MOLLY2004:Isupposethat'strue.IwasdevastatedwhenIlostmyonlycopyofmyseniorthesis.I

lostsixmonthsofworkandhadtostartover.MOLLY2104:Ah,yes,thatwasawful.Irememberitwell,eventhoughitwasoveracenturyago.It

wasdevastatingbecauseitwasasmallpartofmyself.Ihadinvestedmythoughtsandcreativityin that file of information. So think how precious all of your—my—accumulated thoughts,experience,skills,andhistoryare.

...onWarfare:TheRemote,Robotic,Robust,Size-Reduced,Virtual-RealityParadigmAs weapons have become more intelligent, there has been a dramatic trend toward more precisemissionswithfewercasualties.Itmaynotseemthatwaywhenviewedalongsidethetendencytowardmore detailed, realistic television-news coverage. The great battles ofWorldWars I and II and theKoreanWar,inwhichtensofthousandsofliveswerelostoverthecourseofafewdays,werevisuallyrecorded only by occasional grainy newsreels. Today, we have a front-row seat for almost everyengagement. Each war has its complexities, but the overall movement toward precision intelligentwarfareisclearbyexaminingthenumberofcasualties.Thistrendissimilartowhatwearebeginningtosee inmedicine,where smartweapons againstdisease are able toperformspecificmissionswith farfewersideeffects.Thetrendissimilarforcollateralcasualties,althoughitmaynotseemthatwayfromcontemporarymediacoverage(recallthataboutfiftymillionciviliansdiedinWorldWarII).

IamoneoffivemembersoftheArmyScienceAdvisoryGroup(ASAG),whichadvisestheU.S.Armyonprioritiesforitsscienceresearch.Althoughourbriefings,deliberations,andrecommendationsareconfidential,Icansharesomeoverall technologicaldirectionsthatarebeingpursuedbythearmyandalloftheU.S.armedforces.Dr.JohnA.Parmentola,directorforresearchandlaboratorymanagementfortheU.S.ArmyandliaisontotheASAG,describestheDepartmentofDefense's"transformation"processasamovetowardan armed force that is "highly responsive, network-centric, capable of swift decision, superior in allechelons,and[abletoprovide]overwhelmingmassedeffectsacrossanybattlespace."46HedescribestheFutureCombatSystem(FCS),nowunderdevelopmentandscheduledtorolloutduringtheseconddecadeofthiscentury,as"smaller,lighter,faster,morelethal,andsmarter." Dramatic changes are planned for futurewar-fighting deployments and technology.Althoughdetails are likely to change, the army envisions deploying Brigade Combat Teams (BCTs) of about2,500soldiers,unmannedroboticsystems,andFCSequipment.AsingleBCTwould representabout3,300 "platforms," each with its own intelligent computational capabilities. The BCTwould have acommonoperatingpicture(COP)ofthebattlefield,whichwouldbeappropriatelytranslatedforit,witheach soldier receiving information through a variety ofmeans, including retinal (and other forms of"headsup")displaysand,inthefuture,directneuralconnection.Thearmy'sgoalistobecapableofdeployingaBCTin96hoursandafulldivisionin120hours.The load for each soldier, which is now about one hundred pounds of equipment, will initially bereducedthroughnewmaterialsanddevicestofortypounds,whiledramaticallyimprovingeffectiveness.Someoftheequipmentwouldbeoffloadedto"roboticmules."AnewuniformmaterialhasbeendevelopedusinganovelformofKevlarwithsilicananoparticlessuspendedinpolyethyleneglycol.Thematerialisflexibleinnormaluse,butwhenstresseditinstantlyforms a nearly impenetrable mass that is stab resistant. The army's Institute for SoldierNanotechnologiesatMITisdevelopingananotechnology-basedmaterialcalled"exomuscle"toenablecombatantstogreatlyincreasetheirphysicalstrengthwhenmanipulatingheavyequipment.47TheAbramstankhasaremarkablesurvivalrecord,withonlythreecombatcasualtiesinitstwenty

years of combat use. This is the result of both advanced armormaterials and of intelligent systemsdesignedtodefeat incomingweapons,suchasmissiles.However, the tankweighsmore thanseventytons,afigurethatwillneedtobesignificantlyreducedtomeetFCSgoalsforsmallersystems.Newlightweightyetultrastrongnanomaterials(suchasplasticscombinedwithnanotubes,whichare fifty times stronger than steel), as well as increased computer intelligence to counteract missileattacks,areexpectedtodramaticallylowertheweightofgroundcombatsystems.Thetrendtowardunmannedaerialvehicles(DAVs),whichstartedwiththearmedPredatorintherecentAfghanistan and Iraq campaigns, will accelerate. Army research includes the development ofmicro-DAYsthesizeofbirdsthatwillbefast,accurate,andcapableofperformingbothreconnaissanceand combat missions. Even smaller DAVs the size of bumblebees are envisioned. The navigationalabilityofanactualbumblebee,whichisbasedonacomplexinteractionbetweenitsleftandrightvisionsystems,hasrecentlybeenreverseengineeredandwillbeappliedtothesetinyflyingmachines.AtthecenteroftheFCSisaself-organizing,highlydistributedcommunicationsnetworkcapableof gathering information from each soldier and each piece of equipment and in turn providing theappropriateinformationdisplaysandfilesbacktoeachhumanandmachineparticipant.Therewillbenocentralizedcommunicationshubsthatcouldbevulnerabletohostileattack.Informationwillrapidlyrouteitselfarounddamagedportionsofthenetwork.Anobvioustoppriorityistodeveloptechnologycapableofmaintainingintegrityofcommunicationandpreventingeithereavesdroppingormanipulationofinformationbyhostileforces.Thesameinformation-securitytechnologywillbeappliedtoinfiltrate,disrupt, confuse, or destroy enemy communications through both electronicmeans and cyberwarfareusingsoftwarepathogens. TheFCSisnotaone-shotprogram; it representsapervasivefocusofmilitarysystems towardremotely guided, autonomous, miniaturized, and robotic systems, combined with robust, self-organizing,distributed,andsecurecommunications.TheU.S.JointForcesCommand'sProjectAlpha(responsibleforacceleratingtransformativeideasthroughout thearmedservices)envisionsa2025fightingforce that"is largelyrobotic," incorporatingtactical autonomous combatants (TACs) that "have some level of autonomy-adjustable autonomy orsupervisedautonomyorfullautonomywithin...missionbounds."48TheTACswillbeavailableinawiderangeofsizes,rangingfromnanobotsandmicrobotsuptolargeUAVsandothervehicles,aswellasautomatedsystemsthatcanwalkthroughcomplexterrains.OneinnovativedesignbeingdevelopedbyNASAwithmilitaryapplicationsenvisionedisintheformofasnake.49Oneoftheprogramscontributingtothe2020sconceptofself-organizingswarmsofsmallrobotsistheAutonomous IntelligentNetwork andSystems (AINS)programof theOffice ofNavalResearch,whichenvisionsadronearmyofunmanned,autonomousrobotsinthewater,ontheground,andintheair. The swarms will have human commanders with decentralized command and control and whatprojectheadAllenMoshfeghcallsan"impregnableInternetinthesky."50

Extensiveresearchisgoingintodesigningswarmintelligence.51Swarmintelligencedescribestheway that complex behaviors can arise from large numbers of individual agents, each followingrelativelysimple rules.52 Swarms of insects are often able to devise intelligent solutions to complexproblems,suchasdesigningthearchitectureofacolony,despitethefactthatnosinglememberoftheswarmpossessestherequisiteskills.DARPAannouncedin2003thatabattalionof120militaryrobots(builtbyI-Robot,acompanycofounded by robotics pioneerRodneyBrooks)was to be fittedwith swarm-intelligence software toenableit tomimictheorganizedbehaviorof insects.53Asroboticsystemsbecomephysicallysmallerand larger in number, the principles of self-organizing swarm intelligence will play an increasinglyimportantrole.Thereisalsorecognitioninthemilitarythatdevelopmenttimesneedtobereduced.Historically,thetypicaltimeperiodformilitaryprojectstogofromresearchtodeploymenthasbeenlongerthanadecade. But with the technology paradigm-shift rate coming down by half every decade, thesedevelopmenttimesneedtokeeppace,asmanyweaponssystemsarealreadyobsoletebythetimethey

reach the field. Oneway to accomplish this is to develop and test newweapons using simulations,which enable weapons systems to be designed, implemented, and tested far more quickly than thetraditionalmeansofbuildingprototypesandtestingthem(oftenbyblowingthemup)inactualuse.Anotherkeytrendistomovepersonnelawayfromcombattoimprovesoldiers'ratesofsurvival.This canbedonebyallowinghumans todriveandpilot systems remotely.Taking thepilotoutof avehicleallowsittotakepartinriskiermissionsandtobedesignedtobefarmoremaneuverable.Italsoallowsthedevicestobecomeverysmallbydispensingwiththeextensiverequirementsforsupportinghuman life. The generals are moving even farther away. Tommy Franks conducted the war inAfghanistanfromhisbunkerinQatar.Smart Dust. DARPA is developing devices even tinier than birds and bumblebees called "smartdust"—complex sensor systems not much bigger than a pinhead. Once fully developed, swarms ofmillionsofthesedevicescouldbedroppedintoenemyterritorytoprovidehighlydetailedsurveillanceand ultimately support offensive warfaremissions (for example, releasing nanoweapons). Power forsmart-dust systems will be provided by nanoengineered fuel cells, as well as by conversion ofmechanicalenergyfromtheirownmovement,wind,andthermalcurrents. Want to find a key enemy?Need to locate hiddenweapons?Massive numbers of essentiallyinvisible spies could monitor every square inch of enemy territory, identify every person (throughthermalandelectromagneticimaging,eventuallyDNAtests,andothermeans)andeveryweaponandevencarryoutmissionstodestroyenemytargets.Nanoweapons. The next step beyond smart dustwill be nanotechnology-basedweapons,whichwillmake obsolete weapons of larger size. The only way for an enemy to counteract such a massivelydistributed force will be with its own nanotechnology. In addition, enhancing nanodevices with theabilitytoself-replicatewillextendtheircapabilitiesbutintroducesgravedangers,asubjectIaddressinchapter8. Nanotechnology is alreadybeingapplied to awide rangeofmilitary functions.These includenanotechcoatingsfor improvedarmor; laboratoriesonachipfor rapidchemicalandbiological-agentdetection and identification; nanoscale catalysts for decontaminating areas; smart materials that canrestructure themselves for different situations; biocidal nanoparticles incorporated into uniforms toreduceinfectionfrominjuries;nanotubescombinedwithplasticstocreateextremelystrongmaterials;and self-healingmaterials.For example, theUniversityof Illinoishasdeveloped self-healingplasticsthat incorporatemicrospheres of liquidmonomers and a catalyst into a plasticmatrix;when a crackappears,themicrospheresbreak,automaticallysealingthecrack.54SmartWeapons.We'vealreadymovedfromdumbmissiles launchedwithhopes theywill find theirtargetstointelligentcruisemissilesthatusepatternrecognitiontomakethousandsoftacticaldecisionson their own.Bullets, however, have remained essentially small dumbmissiles, and providing themwithameasureofintelligenceisanothermilitaryobjective.Asmilitaryweaponsbecomesmallerinsizeandlargerinnumber,itwon'tbedesirableorfeasibletomaintainhumancontrolovereachdevice.Soincreasingthelevelofautonomouscontrolisanotherimportantgoal.Oncemachine intelligencecatchesupwithbiologicalhumanintelligence,manymoresystemswillbefullyautonomous.VR.Virtual-realityenvironmentsarealreadyinusetocontrolremotelyguidedsystemssuchastheU.S.AirForce'sArmedPredatorUAV.55Evenifasoldierisinsideaweaponssystem(suchasanAbramstank),wedon'texpecthimorhertojustlookoutsidethewindowtoseewhatisgoingon.Virtual-realityenvironmentsareneededtoprovideaviewof theactualenvironmentandallowforeffectivecontrol.Human commanders in charge of swarm weapons will also need specialized virtual-realityenvironmentstoenvisionthecomplexinformationthatthesedistributedsystemsarecollecting.

By the late 2030s and 2040s, as we approach human body version 3.0 and the predominance ofnonbiological intelligence, the issue of cyberwarfare will move to center stage.When everything isinformation, the ability to control your own information and disrupt your enemy's communication,command,andcontrolwillbeaprimarydeterminantofmilitarysuccess....onLearning

Scienceisorganizedknowledge.Wisdomisorganizedlife.

—IMMANUELKANT(1724–1804)Mosteducationintheworldtoday,includinginthewealthiercommunities,isnotmuchchangedfromthe model offered by the monastic schools of fourteenth-century Europe. Schools remain highlycentralized institutions built upon the scarce resources of buildings and teachers. The quality ofeducation also varies enormously, depending on the wealth of the local community (the Americantraditionoffundingeducationfrompropertytaxesclearlyexacerbatesthisinequality),thuscontributingtothehave/havenotdivide.Aswithallofourotherinstitutionswewillultimatelymovetowardadecentralizededucationalsysteminwhicheverypersonwillhavereadyaccesstothehighest-qualityknowledgeandinstruction.Wearenowintheearlystagesofthistransformation,butalreadytheadventoftheavailabilityofvastknowledge on theWeb, useful search engines, high-quality openWeb courseware, and increasinglyeffectivecomputer-assistedinstructionareprovidingwidespreadandinexpensiveaccesstoeducation. Mostmajoruniversitiesnowprovideextensivecoursesonline,manyofwhichare free.MIT'sOpenCourseWare (OCW) initiative has been a leader in this effort.MIT offers nine hundred of itscourses—halfofallitscourseofferings—forfreeontheWeb.56Thesehavealreadyhadamajorimpactoneducationaroundtheworld.Forexample,BrigitteBouissouwrites,"AsamathteacherinFrance,Iwant to thankMIT ... for [these] very lucid lectures, which are a great help for preparingmy ownclasses." Sajid Latif, an educator in Pakistan, has integrated the MIT OCW courses into his owncurriculum.HisPakistani students regularlyattendvirtually-MITclassesasa substantialpartof theireducation.57MITintendstohaveeveryoneofitscoursesonlineandopensource(thatis,freeofchargefornoncommercialuse)by2007.TheU.S.ArmyalreadyconductsallofitsnonphysicaltrainingusingWeb-basedinstruction.Theaccessible,inexpensive,andincreasinglyhigh-qualitycoursewareavailableontheWebisalsofuelingatrendtowardhomeschooling. The cost of the infrastructure for high-quality audiovisual Internet-based communication iscontinuingtofallrapidly,atarateofabout50percentperyear,aswediscussedinchapter2.Bytheendof thedecadeitwillbefeasibleforunderdevelopedregionsof theworld toprovidevery inexpensiveaccess to high-quality instruction for all grade levels from preschool to doctoral studies. Access toeducationwillno longerberestrictedby the lackofavailabilityof trained teachers ineach townandvillage. Ascomputer-assistedinstruction(CAl)becomesmoreintelligenttheabilitytoindividualizethelearningexperienceforeachstudentwillgreatlyimprove.Newgenerationsofeducationalsoftwarearecapableofmodelingthestrengthsandweaknessesofeachstudentanddevelopingstrategiestofocusontheproblemareaofeachlearner.AcompanythatIfounded,KurzweilEducationalSystems,providessoftware that is used in tens of thousands of schools by students with reading disabilities to accessordinaryprintedmaterialsandimprovetheirreadingskills.58Becauseofcurrentbandwidthlimitationsandthelackofeffectivethree-dimensionaldisplays,thevirtualenvironmentprovidedtodaythroughroutineWebaccessdoesnotyetfullycompetewith"beingthere,"butthatwillchange.Intheearlypartoftheseconddecadeofthiscenturyvisual-auditoryvirtual-realityenvironmentswillbefullimmersion,veryhighresolution,andveryconvincing.Mostcolleges

willfollowMIT'slead,andstudentswillincreasinglyattendclassesvirtually.Virtualenvironmentswillprovide high-quality virtual laboratories where experiments can be conducted in chemistry, nuclearphysics,oranyotherscientificfield.StudentswillbeabletointeractwithavirtualThomasJeffersonorThomasEdisonoreventobecomeavirtualThomasJefferson.Classeswillbeavailable forallgradelevels in many languages. The devices needed to enter these high-quality, high-resolution virtualclassroomswillbeubiquitousandaffordableeveninthirdworldcountries.Studentsatanyage,fromtoddlerstoadults,willbeabletoaccessthebesteducationintheworldatanytimeandfromanyplace.Thenatureofeducationwillchangeonceagainwhenwemergewithnonbiologicalintelligence.Wewillthenhavetheabilitytodownloadknowledgeandskills,atleastintothenonbiologicalportionofourintelligence.Ourmachinesdothisroutinelytoday.Ifyouwanttogiveyourlaptopstate-of-the-artskillsinspeechorcharacterrecognition,languagetranslation,orInternetsearching,yourcomputerhasonly to quickly download the right patterns (the software). We don't yet have comparablecommunication ports in our biological brains to quickly download the interneuronal connection andneurotransmitterpatterns that representour learning.That isoneofmanyprofound limitationsof thebiologicalparadigmwenowuseforourthinking,alimitationwewillovercomeintheSingularity....onWork

Ifeveryinstrumentcouldaccomplishitsownwork,obeyingoranticipatingthewillofothers,ifthe shuttle could weave, and the pick touch the lyre, without a hand to guide them, chiefworkmenwouldnotneedservants,normastersslaves.

—ARISTOTLEBeforetheinventionofwriting,almosteveryinsightwashappeningforthefirsttime(atleastto the knowledgeof the small groups of humans involved).Whenyou are at the beginning,everything isnew. Inourera,almosteverythingwedo in thearts isdonewithawarenessofwhathasbeendonebeforeandbefore.Intheearlypost-humanera, thingswillbenewagainbecauseanythingthatrequiresgreaterthanhumanabilityhasnotalreadybeendonebyHomerordaVinciorShakespeare.

—VERNORVINGE59Nowpartof[myconsciousness]livesontheInternetandseemstostaythereallthetime....Astudentmayhaveatextbookopen.Thetelevisionisonwiththesoundoff....They'vegotmusicon headphones… there's a homeworkwindow, alongwith e-mail and instantmessaging….Onemulti-tasking student prefers the onlineworld to the face-to-faceworld. "Real life," hesaid,"isjustonemorewindow."

—CHRISTINE BOESE, REPORTING ON FINDINGS BY MITPROFESSORSHERRYTURKLE60

In1651ThomasHobbesdescribed"thelifeofman"as"solitary,poor,nasty,brutish,andshort."61Thiswas a fair assessment of life at the time, but we have largely overcome this harsh characterizationthrough technologicaladvances,at least in thedevelopedworld.Even inunderdevelopednations lifeexpectancy lagsonly slightlybehind.Technology typically startsoutwithunaffordableproducts thatdon'tworkverywell, followedbyexpensiveversions thatworkabitbetter,and thenby inexpensiveproducts thatwork reasonablywell.Finally the technologybecomeshighlyeffective,ubiquitous,andalmostfree.Radioandtelevisionfollowedthispattern,asdidthecellphone.ContemporaryWebaccessisattheinexpensive-and-working-reasonably-wellstage.

Today the delay between early and late adoption is about a decade, but in keepingwith thedoublingoftheparadigm-shiftrateeverydecade,thisdelaywillbeonlyaboutfiveyearsinthemiddleoftheseconddecadeandonlyacoupleofyearsinthemid-2020s.Giventheenormouswealth-creationpotentialofGNRtechnologies,wewillseetheundercIasslargelydisappearoverthenexttwotothreedecades(seethediscussionsofthe2004WorldBankreportinchapters2and9).Thesedevelopmentsarelikelytobemet,however,withincreasingfundamentalistandLudditereactiontotheacceleratingpaceofchange.WiththeadventofMNT-basedmanufacturing,thecostofmakinganyphysicalproductwillbereduced to pennies per pound, plus the cost of the information guiding the process, with the latterrepresentingthetruevalue.Wearealreadynotthatfarfromthisreality;software-basedprocessesguideeverystepofmanufacturing today, fromdesignandmaterialsprocurement toassembly inautomatedfactories.Theportionofamanufacturedproduct'scostattributabletotheinformationprocessesusedinitscreationvaries fromonecategoryofproduct toanotherbut is increasingacross theboard, rapidlyapproaching100percent.Bythelate2020sthevalueofvirtuallyallproducts—clothes,food,energy,andofcourseelectronics—willbealmostentirelyintheirinformation.Asisthecasetoday,proprietaryandopen-sourceversionsofeverytypeofproductandservicewillcoexist.IntellectualProperty.Iftheprimaryvalueofproductsandservicesresidesintheirinformation,thentheprotectionof informationrightswillbecritical tosupportingthebusinessmodels thatprovidethecapitaltofundthecreationofvaluableinformation.Theskirmishestodayintheentertainmentindustryregardingillegaldownloadingofmusicandmoviesareaharbingerofwhatwillbeaprofoundstruggle,once essentially everything of value is composed of information. Clearly, existing or new businessmodelsthatallowforthecreationofvaluableintellectualproperty(IP)needtobeprotected,otherwisethesupplyofIPwillitselfbethreatened.However,thepressurefromtheeaseofcopyinginformationisarealitythatisnotgoingaway,soindustrieswillsufferiftheydonotkeeptheirbusinessmodelsinlinewithpublicexpectations.Inmusic,forexample,ratherthanprovideleadershipwithnewparadigms,therecordingindustrystuckrigidly(untiljustrecently)withtheideaofanexpensiverecordalbum,abusinessmodelthathasremainedunchangedfromthetimemyfatherwasayoung,strugglingmusicianinthe1940s.Thepublicwill avoidwide-scalepiratingof information servicesonly if commercial prices arekept atwhat areperceivedtobereasonablelevels.Themobile-phonesectorisaprimeexampleofanindustrythathasnotinvitedrampantpiracy.Thecostofcell-phonecallshasfallenrapidlywithimprovingtechnology.Ifthemobile-phoneindustryhadkeptcallingratesatthelevelwheretheywerewhenIwasachild(atimewhenpeopledroppedwhatevertheyweredoingattheraretimesthatsomeonecalledlongdistance),wewould be seeing comparable pirating of cell-phone calls,which is technically nomore difficult thanpirating music. But cheating on cell-phone calls is widely regarded as criminal behavior, largelybecauseofthegeneralperceptionthatcell-phonechargesareappropriate.IPbusinessmodelsinvariablyexistontheedgeofchange.Movieshavebeendifficulttodownloadbecauseoftheirlargefilesize,butthatisrapidlybecominglessofanissue.Themovieindustryneedsto lead the charge toward new standards, such as high-definition movies on demand. Musicianstypicallymakemostoftheirmoneywithliveperformances,butthatmodelwillalsocomeunderattackearlyinthenextdecade,whenwewillhavefull-immersionvirtualreality.Eachindustrywillneedtocontinuallyreinventitsbusinessmodels,whichwillrequireasmuchcreativityasthecreationoftheIPitself.Thefirstindustrialrevolutionextendedthereachofourbodies,andthesecondisextendingthereachofourminds.AsImentioned,employmentinfactoriesandfarmshasgonefrom60percentto6percent in theUnitedStates in thepastcentury.Over thenextcoupleofdecades,virtuallyall routinephysicalandmentalworkwillbeautomated.Computationandcommunicationwillnotinvolvediscreteproductssuchashandhelddevicesbutwillbeaseamlesswebofintelligentresourcesthatareallaroundus.Alreadymostcontemporarywork is involved in thecreationandpromotionof IP inone formoranother,aswellasdirectpersonalservicesfromonepersontoanother(health,fitness,education,andsoon). These trends will continue with the creation of IP—including all of our artistic, social, and

scientificcreativity—andwillbegreatlyenhancedbytheexpansionofourintellectthroughthemergerwith nonbiological intelligence. Personal services will largely move to virtual-reality environments,especiallywhenvirtualrealitybeginstoencompassallofthesenses.Decentralization.Thenextseveraldecadeswillseeamajortrendtowarddecentralization.Todaywehavehighlycentralizedandvulnerableenergyplantsanduseshipsandfuel lines to transportenergy.Theadventofnanoengineeredfuelcellsandsolarpowerwillenableenergyresourcestobemassivelydistributedandintegratedintoourinfrastructure.MNTmanufacturingwillbehighlydistributedusinginexpensive nanofabrication minifactories. The ability to do nearly anything with anyone fromanywhere inanyvirtual-realityenvironmentwillmakeobsolete thecentralized technologiesofofficebuildingsandcities.Withversion3.0bodiesabletomorphintodifferentformsatwillandourlargelynonbiologicalbrainsnolongerconstrainedtothelimitedarchitecturethatbiologyhasbestowedonus,thequestionofwhat is human will undergo intensive examination. Each transformation described here does notrepresentasuddenleapbutratherasequenceofmanysmallsteps.Althoughthespeedwithwhichthesesteps are being taken is hastening, mainstream acceptance generally follows rapidly. Consider newreproductive technologies such as in vitro fertilization,whichwere controversial at first but quicklybecamewidelyusedandaccepted.Ontheotherhand,changewillalwaysproducefundamentalistandLuddite counteractions, which will intensify as the pace of change increases. But despite apparentcontroversy,theoverwhelmingbenefitstohumanhealth,wealth,expression,creativity,andknowledgequicklybecomeapparent....onPlay

Technologyisawayoforganizingtheuniversesothatpeopledon'thavetoexperienceit.—MAXFRISCH,HOMOFABER

Lifeiseitheradaringadventureornothing.

—HELENKELLERPlayisjustanotherversionofworkandhasanintegralroleinthehumancreationofknowledgeinallofits forms. A child playing with dolls and blocks is acquiring knowledge essentially by creating itthroughhis or her own experience. People playingwith dancemoves are engaged in a collaborativecreativeprocess(considerthekidsonstreetcomersinthenation'spoorestneighborhoodswhocreatedbreakdancing,whichlaunchedthehip-hopmovement).EinsteinputasidehisworkfortheSwisspatentoffice and engaged inplayfulmind experiments, resulting in the creationof his enduring theories ofspecialandgeneralrelativity.Ifwaristhefatherofinvention,thenplayisitsmother. Already there is no clear distinction between increasingly sophisticated video games andeducational software.TheSims2, a game released inSeptember 2004, usesAI-based characters thathave their own motivations and intentions. With no prepared scripts the characters behave inunpredictableways,withthestorylineemergingoutoftheirinteractions.Althoughconsideredagame,it offers players insights into developing social awareness.Similarly games that simulate sportswithincreasinglyrealisticplayimpartskillsandunderstanding.Bythe2020s,full-immersionvirtualrealitywillbeavastplaygroundofcompellingenvironmentsand experiences. Initially VR will have certain benefits in terms of enabling communications withothersinengagingwaysoverlongdistancesandfeaturingagreatvarietyofenvironmentsfromwhichtochoose.Althoughtheenvironmentswillnotbecompletelyconvincingatfirst,bythelate2020stheywill be indistinguishable from real reality andwill involve all of the senses, aswell as neurologicalcorrelationsofouremotions.Asweenter the2030s therewon'tbecleardistinctionsbetweenhumanandmachine,betweenrealandvirtualreality,orbetweenworkandplay.

. . .ontheIntelligentDestinyoftheCosmos:WhyWeAreProbablyAloneintheUniverse

Theuniverseisnotonlyqueererthanwesuppose,butqueererthanwecansuppose.—J.B.S.HALDANE

Whatistheuniversedoingquestioningitselfviaoneofitssmallestproducts?

—D.E.JENKINS,ANGLICANTHEOLOGIANWhatistheuniversecomputing?Asfaraswecantell,itisnotproducingasingleanswertoasingle question....Instead the universe is computing itself. Powered by Standard Modelsoftware,theuniversecomputesquantumfields,chemicals,bacteria,humanbeings,stars,andgalaxies. As it computes, it maps out its own spacetime geometry to the ultimate precisionallowedbythelawsofphysics.Computationisexistence.

—SETHLLOYDANDY.JACKNG62

Ournaiveviewofthecosmos,datingbacktopre-Copernicandays,wasthattheEarthwasatthecenteroftheuniverseandhumanintelligenceitsgreatestgift(nexttoGod).Themoreinformedrecentviewisthat,evenifthelikelihoodofastar'shavingaplanetwithatechnology-creatingspeciesisverylow(forexample,oneinamillion),therearesomanystars(thatis,billionsoftrillionsofthem),thatthereareboundtobemany(billionsortrillions)withadvancedtechnology. ThisistheviewbehindSETI—theSearchforExtraterrestrialIntelligence—andisthecommoninformedviewtoday.However,therearereasonstodoubtthe"SETIassumption"thatETIisprevalent.First,considerthecommonSETIview.CommoninterpretationsoftheDrakeequation(seebelow)concludethattherearemany(asinbillions)ofETlsintheuniverse,thousandsormillionsinourgalaxy.Wehaveonlyexaminedatinyportionofthehaystack(theuniverse),soourfailuretodatetofindtheneedle(anETIsignal)shouldnotbeconsidereddiscouraging.Ourefforts toexplorethehaystackarescalingup.ThefollowingdiagramfromSky&TelescopeillustratesthescopeoftheSETIprojectbyplottingthe capability of the varied scanning efforts against three major parameters: distance from Earth,frequencyoftransmission,andthefractionofthesky.63

The plot includes two future systems. The Allen Telescope Array, named after MicrosoftcofounderPaulAllen,isbasedonusingmanysmallscanningdishesratherthanoneorasmallnumberoflargedishes,withthirty-twoofthedishesscheduledtobeonlinein2005.Whenallofits350dishesareoperational (projected in2008), itwillbeequivalent toa2½-acredish (10,000squaremeters). Itwillbecapableoflisteningtoupto100millionfrequencychannelssimultaneously,andabletocovertheentiremicrowavespectrum.Oneofitsintendedtaskswillbetoscanmillionsofstarsinourgalaxy.Theproject relieson intelligentcomputation thatcanextracthighlyaccurate signals frommany low-costdishes.64OhioStateUniversityisbuildingtheOmnidirectionalSearchSystem,whichreliesonintelligentcomputation to interpret signals from a large array of simple antennas. Using principles ofinterferometry(thestudyofhowsignalsinterferewitheachother),ahigh-resolutionimageoftheentiresky can be computed from the antenna data.65 Other projects are expanding the range ofelectromagneticfrequency,forexample,toexploretheinfraredandopticalranges.66

Therearesixotherparametersinadditiontothethreeshowninthechartonthepreviouspage—forexample, polarization (the plane of thewavefront in relation to the direction of the electromagneticwaves).Oneoftheconclusionswecandrawfromtheabovegraphisthatonlyverythinslicesofthisnine-dimensional"parameterspace"havebeenexploredbySETI.So,thereasoninggoes,weshouldnotbesurprisedthatwehavenotyetuncoveredevidenceofanETI.However,wearenotjustsearchingforasingleneedle.Basedonthelawofacceleratingreturns,onceanETIreachesprimitivemechanicaltechnologies,itisonlyafewcenturiesbeforeitreachesthevastcapabilitiesI'veprojectedforthetwenty-secondcenturyhereonEarth.RussianastronomerN.S.Kardashev describes a "type II" civilization as one that has harnessed the power of its star forcommunicationusingelectromagneticradiation(about4Î1026watts,basedonoursun).67Accordingtomy projections (see chapter 3), our civilization will reach that level by the twenty-second century.Given that the levelof technologicaldevelopmentof themanycivilizationsprojectedbymanySETItheoristsshouldbespreadoutovervastperiodsoftime,thereshouldbemanygreatlyaheadofus.Sothere should bemany type II civilizations. Indeed, there has been sufficient time for some of thesecivilizations tohavecolonized theirgalaxies andachieveKardashev's type III: a civilization thathasharnessedtheenergyofitsgalaxy(about4Î1037watts,basedonourgalaxy).Evenasingleadvancedcivilization shouldbe emittingbillionsor trillionsof "needles"—that is, transmissions representing avastnumberofpointsintheSETIparameterspaceasartifactsandsideeffectsofitsmyriadinformationprocesses.EvenwiththethinslicesoftheparameterspacescannedbytheSETIprojecttodate,itwouldbehardtomissatypeIIcivilization,letaloneatypeIII.Ifwethenfactorintheexpectationthatthereshouldbeavastnumberoftheseadvancedcivilizations,itisoddthatwehaven'tnoticedthem.That'stheFermiParadox.TheDrakeEquation.TheSETIsearchhasbeenmotivatedinlargepartbyastronomerFrankDrake's1961 equation for estimating the number of intelligent (or, more precisely, radio-transmitting)civilizationsinourgalaxy.68(Presumably,thesameanalysiswouldpertaintoothergalaxies.)ConsidertheSETIassumptionfromtheperspectiveoftheDrakeformula,whichstates:

Thenumberofradio-transmittingcivilizations=NÎfpÎneÎflÎfiÎfcÎfLwhere:

N=thenumberofstarsintheMilkyWaygalaxy.Currentestimatesarearound100billion(1011).fp = the fractionof stars that haveorbitingplanets.Current estimates range fromabout20percentto50percent.ne: For each star with orbiting planets, what is the average number of planets capable ofsustaining life?This factor ishighlycontroversial.Someestimatesareoneorhigher (that is,everystarwithplanetshas,onaverage,atleastoneplanetthatcansustainlife)tomuchlowerfactors,suchasoneinonethousandorevenless.fl:Fortheplanetscapableofsustaininglife,onwhatfractionofthesedoeslifeactuallyevolve?Estimatesarealloverthemap,fromapproximately100percenttoabout0percent.fi:Foreachplanetonwhichlifeevolves,whatisthefractiononwhichintelligentlifeevolves?flandfiare themostcontroversial factors in theDrakeequation.Hereagain,estimatesrangefromnearly100percent(thatis,oncelifegetsafoothold,intelligentlifeissuretofollow)tocloseto0percent(thatis,intelligentlifeisveryrare).

fc: For each planet with intelligent life, what is the fraction that communicates with radiowaves?Theestimatesforfctendtobehigherthanforflandfi,basedonthe(sensible)reasoningthat once you have an intelligent species, the discovery and use of radio communication islikely.fL = the fraction of the universe's life during which an average communicating civilizationcommunicateswith radiowaves.69 Ifwe take our civilization as an example,we have beencommunicatingwithradiotransmissionsforaboutonehundredyearsoutoftheroughlyten-totwenty-billion-year history of the universe, so fL for the Earth is about 10–8 so far. If wecontinue communicating with radio waves for, say, another nine hundred years, the factorwould then be 10–7. This factor is affected by a number of considerations. If a civilizationdestroys itself because it is unable to handle the destructive power of technologies thatmaytend to develop along with radio communication (such as nuclear fusion or self-replicatingnanotechnology), then radio transmissionswould cease.Wehave seen civilizations onEarth(the Mayans, for example) suddenly end their organized societies and scientific pursuits(althoughpreradio).Ontheotherhanditseemsunlikelythateverycivilizationwouldendthisway,sosuddendestructionislikelytobeonlyamodestfactorinreducingthenumberofradio-capablecivilizations.

Amore salient issue is that of civilizations progressing from electromagnetic (that is, radio)transmissions tomore capablemeans of communicating.Here onEarthwe are rapidlymoving fromradiotransmissionstowires,usingcableandfiberopticsforlong-distancecommunication.Sodespiteenormous increases in overall communication bandwidth, the amount of electromagnetic informationsentintospacefromourplanethasneverthelessremainedfairlysteadyforthepastdecade.Ontheotherhand we do have increasing means of wireless communication (for example, cell phones and newwireless Internet protocols, such as the emerging WiMAX standard). Rather than use wires,communication may rely on exotic mediums such as gravity waves. However, even in this case,althoughtheelectromagneticmeansofcommunicationmaynolongerbethecuttingedgeofanETl'scommunicationtechnology,itislikelytocontinuetobeusedforatleastsomeapplications(inanycase,fLdoestakeintoconsiderationthepossibilitythatacivilizationwouldstopsuchtransmissions).ItisclearthattheDrakeequationcontainsmanyimponderables.ManySETIadvocateswhohavestudied it carefully argue that it implies that theremust be significant numbers of radio-transmittingcivilizationsinourgalaxyalone.Forexample,ifweassumethat50percentofthestarshaveplanets(fp=0.5),thateachofthesestarshasanaverageoftwoplanetsabletosustainlife(ne=2),thatonhalfoftheseplanetslifehasactuallyevolved(fl=0.5),thathalfoftheseplanetshasevolvedintelligentlife(fi=0.5),thathalfoftheseareradio-capable(fc=0.5),andthattheaverageradio-capablecivilizationhasbeenbroadcastingforonemillionyears(fL=10–4),theDrakeequationtellsusthatthereare1,250,000radio-capable civilizations in our galaxy. For example, the SETI Institute's senior astronomer, SethShostak, estimates that there are between ten thousand and one million planets in the Milky Waycontainingaradio-broadcastingcivilization.70CarlSaganestimatedaroundamillioninthegalaxy,andDrakeestimatedaroundtenthousand.71Buttheparametersabovearearguablyveryhigh.Ifwemakemoreconservativeassumptionsonthedifficultyofevolvinglife—andintelligentlifeinparticular—wegetaverydifferentoutcome.Ifweassumethat50percentofthestarshaveplanets(fp=0.5),thatonlyonetenthofthesestarshaveplanetsable to sustain life (ne = 0.1 based on the observation that life-supporting conditions are not thatprevalent),thaton1percentoftheseplanetslifehasactuallyevolved(fl=0.01basedonthedifficultyoflifestartingonaplanet),that5percentoftheselife-evolvingplanetshaveevolvedintelligentlife(fi=0.05,basedontheverylongperiodoftimethistookonEarth),thathalfoftheseareradio-capable(fc=

0.5),andthattheaverageradio-capablecivilizationhasbeenbroadcastingfortenthousandyears(fL=10–6),theDrakeequationtellsusthatthereisaboutone(1.25tobeexact)radio-capablecivilizationintheMilkyWay.Andwealreadyknowofone.Intheend,itisdifficulttomakeastrongargumentfororagainstETIbasedonthisequation.IftheDrakeformula tellsusanything, it is theextremeuncertaintyofourestimates.Whatwedoknowfornow,however,isthatthecosmosappearssilent—thatis,we'vedetectednoconvincingevidenceofETItransmissions.TheassumptionbehindSETIisthatlife—andintelligentlife—issoprevalentthattheremustbemillionsifnotbillionsofradio-capablecivilizationsintheuniverse(oratleastwithinourlightsphere,whichreferstoradio-broadcastingcivilizationsthatweresendingoutradiowavesearlyenoughto reachEarthby today).Nota singleoneof them,however,hasmade itselfnoticeable toourSETIefforts thus far. So let's consider the basic SETI assumption regarding the number of radio-capablecivilizations from the perspective of the law of accelerating returns. As we have discussed, anevolutionaryprocessinherentlyaccelerates.Moreover,theevolutionoftechnologyisfarfasterthantherelativelyslowevolutionaryprocessthatgivesrisetoatechnology-creatingspeciesinthefirstplace.Inourowncasewewentfromapre-electricity,computerlesssocietythatusedhorsesasitsfastestland-based transportation to the sophisticated computational and communications technologies we havetodayinonlytwohundredyears.Myprojectionsshow,asnotedabove,thatwithinanothercenturywewill multiply our intelligence by trillions of trillions. So only three hundred years will have beennecessarytotakeusfromtheearlystirringsofprimitivemechanicaltechnologiestoavastexpansionofour intelligence andability to communicate.Thus, once a species creates electronics and sufficientlyadvancedtechnologytobeamradiotransmissions,itisonlyamatterofamodestnumberofcenturiesforittovastlyexpandthepowersofitsintelligence. The three centuries this will have taken on Earth is an extremely brief period of time on acosmologicalscale,giventhattheageoftheuniverseisestimatedatthirteentofourteenbillionyears.72Mymodelimpliesthatoncea348civilizationachievesourownlevelofradiotransmission,ittakesnomore thana century—twoat themost—toachievea type II civilization. Ifweaccept theunderlyingSETI assumption that there aremany thousands if not millions of radio-capable civilizations in ourgalaxy—andthereforebillionswithinourlightsphereintheuniverse—thesecivilizationsmustexistindifferentstagesoverbillionsofyearsofdevelopment.Somewouldbebehindus,andsomewouldbeahead.Itisnotcrediblethateverysingleoneofthecivilizationsthataremoreadvancedthanusisgoingtobeonlyafewdecadesahead.Mostofthosethatareaheadofuswouldbeaheadbymillions,ifnotbillions,ofyears.Yetsinceaperiodofonlyafewcenturiesissufficienttoprogressfrommechanicaltechnologytothe vast explosion of intelligence and communication of theSingularity, under theSETI assumptionthereshouldbebillionsofcivilizationsinourlightsphere(thousandsormillionsinourgalaxy)whosetechnologyisaheadofourstoanunimaginabledegree.InatleastsomediscussionsoftheSETIproject,weseethesamekindoflinearthinkingthatpermeateseveryotherfield,assumptionsthatcivilizationswillreachourleveloftechnology,andthattechnologywillprogressfromthatpointverygraduallyforthousands if not millions of years. Yet the jump from the first stirrings of radio to powers that gobeyondameretypeIIcivilizationtakesonlyafewhundredyears.Sotheskiesshouldbeablazewithintelligenttransmissions.Yettheskiesarequiet.Itisoddandintriguingthatwefindthecosmossosilent.AsEnricoFermiaskedinthesummerof1950,"Whereiseverybody?"73Asufficientlyadvancedcivilizationwouldnotbelikelytorestrictitstransmissionstosubtlesignalsonobscurefrequencies.WhyarealltheETIssoshy?Therehavebeenattemptstorespondtotheso-calledFermiParadox(which,granted,isaparadoxonly if one accepts the optimistic parameters thatmost observers apply to theDrake equation).Onecommon response is that a civilization may obliterate itself once it reaches radio capability. Thisexplanationmightbe acceptable ifwewere talking aboutonly a few suchcivilizations, butwith thecommonSETIassumptionsimplyingbillionsofthem,itisnotcredibletobelievethateveryoneofthemdestroyeditself.

Otherargumentsrunalongthissameline.Perhaps"they"havedecidednottodisturbus(givenhowprimitiveweare)andarejustwatchingusquietly(anethicalguidelinethatwillbefamiliartoStarTrekfans).Again,itishardtobelievethateverysuchcivilizationoutofthebillionsthatshouldexisthas made the same decision. Or, perhaps, they have moved on to more capable communicationparadigms.Idobelievethatmorecapablecommunicationmethodsthanelectromagneticwaves—evenveryhigh-frequencyones—arelikelytobefeasibleandthatanadvancedcivilization(suchaswewillbecomeoverthenextcentury)islikelytodiscoverandexploitthem.Butitisveryunlikelythattherewouldbeabsolutelynoroleleftforelectromagneticwaves,evenasaby-productofothertechnologicalprocesses,inanyofthesemanymillionsofcivilizations.Incidentally,thisisnotanargumentagainstthevalueoftheSETIproject,whichshouldhavehighpriority,becausethenegativefindingisnolessimportantthanapositiveresult.The Limits of Computation Revisited. Let's consider some additional implications of the law ofacceleratingreturnstointelligenceinthecosmos.Inchapter3Idiscussedtheultimatecoldlaptopandestimatedtheoptimalcomputationalcapacityofaone-liter,one-kilogramcomputerataround1042cps,whichissufficienttoperformtheequivalentoftenthousandyearsofthethinkingoftenbillionhumanbrainsintenmicroseconds.Ifweallowmoreintelligentmanagementofenergyandheat,thepotentialinonekilogramofmattertocomputemaybeashighas1050cps.Thetechnicalrequirementstoachievecomputationalcapacitiesinthisrangearedaunting,butasIpointed out, the appropriate mental experiment is to consider the vast engineering ability of acivilization with 1042 cps per kilogram, not the limited engineering ability of humans today. Acivilizationat1042cpsislikelytofigureouthowtogetto1043cpsandthento1044andsoon.(Indeed,wecanmakethesameargumentateachsteptogettothenext.)Oncecivilizationreachestheselevelsitisobviouslynotgoingtorestrictitscomputationtoonekilogramofmatter,anymorethanwedosotoday.Let'sconsiderwhatourcivilizationcanaccomplishwiththemassandenergyinourownvicinity.TheEarthcontainsamassofabout6Î1024kilograms.Jupiterhasamassofabout1.9Î1027kilograms.Ifweignorethehydrogenandhelium,wehaveabout1.7Î1026kilogramsofmatterinthesolarsystem,notincludingthesun(whichultimatelyisalsofairgame).Theoverallsolarsystem,whichisdominatedbythesun,hasamassofabout2Î1030kilograms.Asacrudeupper-boundanalysis,ifweapplythemassinthesolarsystemtoour1050estimateofthelimitofcomputationalcapacityperkilogramofmatter(basedonthelimitsfornanocomputing),wegetalimitof1080cpsforcomputationinour"vicinity."Obviously,therearepracticalconsiderationsthatarelikelytoprovidedifficultyinreachingthiskindofupperlimit.Butevenifwedevotedonetwentiethof1percent(0.0005)ofthematterofthesolarsystem to computational and communication resources, we get capacities of 1069 cps for "cold"computingand1077cpsfor"hot"computing.74Engineeringestimateshavebeenmadeforcomputingatthesescalesthattakeintoconsiderationcomplex design requirements such as energy usage, heat dissipation, internal communication speeds,the composition ofmatter in the solar system, andmany other factors. These designs use reversiblecomputing, but as I pointed out in chapter 3, we still need to consider the energy requirements forcorrecting errors and communicating results. In an analysis by computational neuroscientist AndersSandberg, the computational capacity of an Earth-size computational "object" called Zeus wasreviewed.75 The conceptual design of this "cold" computer, consisting of about 1025 kilograms ofcarbon (about 1.8 times the mass of the Earth) in the form of diamondoid consists of 5 Î 1037computationalnodes,eachofwhichusesextensiveparallelprocessing.Zeusprovidesanestimatedpeakof 1061 cps of computation or, if used for data storage, 1047 bits. A primary limiting factor for thedesignisthenumberofbiterasurespermitted(itallowsupto2.6Î1032biterasurespersecond),whichareprimarilyusedtocorrecterrorsfromcosmicraysandquantumeffects.In1959astrophysicistFreemanDysonproposedaconceptofcurvedshellsaroundastarasaway

toprovidebothenergyandhabitatsforanadvancedcivilization.OneconceptionoftheDysonSphereisquite literallya thinspherearoundastar togatherenergy.76Thecivilization lives in the sphere, andgivesoffheat(infraredenergy)outsidethesphere(awayfromthestar).Another(andmorepractical)versionoftheDysonSphereisaseriesofcurvedshells,eachofwhichblocksonlyaportionofthestar'sradiation.InthiswayDysonShellscanbedesignedtohavenoeffectonexistingplanets,particularlythose,liketheEarth,thatharboranecologythatneedstobeprotected.AlthoughDysonproposedhisconceptasameansofprovidingvastamountsofspaceandenergyforanadvancedbiologicalcivilization, itcanalsobeusedasthebasisforstar-scalecomputers.SuchDysonShells could orbit our sunwithout affecting the sunlight reaching theEarth.Dyson imaginedintelligent biological creatures living in the shells or spheres, but since civilization moves rapidlytowardnonbiologicalintelligenceonceitdiscoverscomputation,therewouldbenoreasontopopulatetheshellswithbiologicalhumans.AnotherrefinementoftheDysonconceptisthattheheatradiatedbyoneshellcouldbecapturedandusedbyaparallelshellthatisplacedatapositionfartherfromthesun.ComputerscientistRobertBradburypointsoutthattherecouldbeanynumberofsuchlayersandproposesacomputeraptlycalleda"Matrioshkabrain,"organizedasa seriesofnestedshellsaround the sunoranother star.Onesuchconceptual design analyzed bySandberg is calledUranos,which is designed to use 1 percent of thenonhydrogen,nonheliummassinthesolarsystem(notincludingthesun),orabout1024kilograms,abitsmaller than Zeus.77 Uranos provides about 1039 computational nodes, an estimated 1051 cps ofcomputation,andabout1052bitsofstorage. Computation is already a widely distributed—rather than centralized—resource, and myexpectationisthatthetrendwillcontinuetowardgreaterdecentralization.However,asourcivilizationapproaches the densities of computation envisioned above, the distribution of the vast number ofprocessors is likely to have characteristics of these conceptual designs. For example, the idea ofMatrioshka shellswould takemaximal advantage of solar power and heat dissipation.Note that thecomputational powers of these solar system-scale computers will be achieved, according to myprojectionsinchapter2,aroundtheendofthiscentury.BiggerorSmaller.Giventhatthecomputationalcapacityofoursolarsystemisintherangeof1070to1080 cps,wewill reach these limits early in the twenty-secondcentury, according tomyprojections.Thehistoryofcomputation tellsus that thepowerofcomputationexpandsboth inwardandoutward.Over the last several decades we have been able to place twice as many computational elements(transistors) on each integrated circuit chip about every two years, which represents inward growth(towardgreaterdensitiesofcomputationperkilogramofmatter).Butwearealsoexpandingoutward,inthat the number of chips is expanding (currently) at a rate of about 8.3 percent per year.78 It isreasonable to expect both types of growth to continue, and for the outward growth rate to increasesignificantlyonceweapproachthelimitsofinwardgrowth(withthree-dimensionalcircuits).Moreover,oncewebumpupagainstthelimitsofmatterandenergyinoursolarsystemtosupporttheexpansionofcomputation,wewillhavenochoicebut toexpandoutwardas theprimaryformofgrowth.Wediscussedearlierthespeculationthatfinerscalesofcomputationmightbefeasible—onthescale of subatomic particles. Such pico- or femtotechnology would permit continued growth ofcomputationbycontinuedshrinkingoffeaturesizes.Evenifthisisfeasible,however,therearelikelytobe major technical challenges in mastering subnanoscale computation, so the pressure to expandoutwardwillremain.ExpandingBeyondtheSolarSystem.Oncewedoexpandourintelligencebeyondthesolarsystem,atwhatratewillthistakeplace?Theexpansionwillnotstartoutatthemaximumspeed;itwillquicklyachieveaspeedwithinavanishinglysmallchangefromthemaximumspeed(speedoflightorgreater).Some critics have objected to this notion, insisting that itwould be very difficult to send people (oradvancedorganismsfromanyotherETIcivilization)andequipmentatnearthespeedoflightwithout

crushing them.Of course,we could avoid this problemby accelerating slowly, but another problemwouldbecollisionswithinterstellarmaterial.Butagain,thisobjectionentirelymissesthepointofthenature of intelligence at this stage of development. Early ideas about the spread of ETI through thegalaxyanduniversewerebasedonthemigrationandcolonizationpatternsfromourhumanhistoryandbasicallyinvolvedsendingsettlementsofhumans(or,inthecaseofotherETIcivilizations,intelligentorganisms) to other star systems. This would allow them to multiply through normal biologicalreproductionandthencontinuetospreadinlikemannerfromthere.Butaswehaveseen,bylateinthiscenturynonbiologicalintelligenceontheEarthwillbemanytrillionsof timesmorepowerful thanbiological intelligence, sosendingbiologicalhumansonsuchamission would not make sense. The samewould be true for any other ETI civilization. This is notsimplyamatterofbiologicalhumanssendingroboticprobes.Humancivilizationbythattimewillbenonbiologicalforallpracticalpurposes.Thesenonbiologicalsentrieswouldnotneedtobeverylargeandinfactwouldprimarilycompriseinformation. It is true, however, that just sending information would not be sufficient, for somematerial-based device that can have a physical impact on other star and planetary systemsmust bepresent. However, it would be sufficient for the probes to be self-replicating nanobots (note that ananobot has nanoscale features but that the overall size of a nanobot ismeasured inmicrons).79Wecould send swarms of many trillions of them, with some of these "seeds" taking root in anotherplanetary system and then replicating by finding the appropriatematerials, such as carbon and otherneededelements,andbuildingcopiesofthemselves.Onceestablished,thenanobotcolonycouldobtaintheadditionalinformationitneedstooptimizeits intelligencefrompure informationtransmissionsthat involveonlyenergy,notmatter,andthataresent at the speed of light. Unlike large organisms such as humans, these nanobots, being extremelysmall, could travel at close to the speed of light. Another scenario would be to dispense with theinformationtransmissionsandembedtheinformationneededinthenanobots'ownmemory.That'sanengineeringdecisionwecanleavetothesefuturesuperengineers.Thesoftwarefilescouldbespreadoutamongbillionsofdevices.Onceoneorafewofthemgeta"foothold"byself-replicatingatadestination,thenowmuchlargersystemcouldgatherupthenanobotstravelinginthevicinitysothatfromthattimeon,thebulkofthenanobotssentinthatdirectiondonotsimply flyby. In thisway, thenowestablished colony cangather up the information, aswell as thedistributedcomputationalresources,itneedstooptimizeitsintelligence.TheSpeedofLightRevisited. In thisway themaximumspeedof expansionof a solar system-sizeintelligence(thatis,atypeIIcivilization)intotherestoftheuniversewouldbeveryclosetothespeedoflight.Wecurrentlyunderstandthemaximumspeedtotransmitinformationandmaterialobjectstobethespeedoflight,butthereareatleastsuggestionsthatthismaynotbeanabsolutelimit. Wehave to regard the possibility of circumventing the speed of light as speculative, andmyprojections of the profound changes that our civilizationwill undergo in this centurymake no suchassumption. However, the potential to engineer around this limit has important implications for thespeedwithwhichwewillbeabletocolonizetherestoftheuniversewithourintelligence.Recentexperimentshavemeasuredtheflighttimeofphotonsatnearlytwicethespeedoflight,aresult of quantum uncertainty on their position.80 However, this result is really not useful for thisanalysis, because it doesnot actually allow information tobe communicated faster than the speedoflight,andwearefundamentallyinterestedincommunicationspeed.Anotherintriguingsuggestionofanactionatadistancethatappearstooccuratspeedsfargreaterthan thespeedof light isquantumdisentanglement.Twoparticlescreated togethermaybe"quantumentangled,"meaning thatwhile a given property (such as the phase of its spin) is not determined ineitherparticle,theresolutionofthisambiguityofthetwoparticleswilloccuratthesamemoment.Inotherwords,iftheundeterminedpropertyismeasuredinoneoftheparticles,itwillalsobedeterminedastheexactsamevalueatthesameinstantintheotherparticle,evenifthetwohavetraveledfarapart.Thereisanappearanceofsomesortofcommunicationlinkbetweentheparticles.

Thisquantumdisentanglementhasbeenmeasuredatmanytimesthespeedoflight,meaningthatresolutionofthestateofoneparticleappearstoresolvethestateoftheotherparticleinanamountoftime that is a small fraction of the time it would take if the informationwere transmitted from oneparticle to theotherat thespeedof light(in theory, the timelapse iszero).Forexample,Dr.NicolasGisin of the University of Geneva sent quantum-entangled photons in opposite directions throughopticalfibersacrossGeneva.Whenthephotonsweresevenmilesapart,theyeachencounteredaglassplate.Each photon had to "decide"whether to pass throughor bounce off the plate (which previousexperimentswithnon-quantum-entangledphotonshaveshowntobearandomchoice).Yetbecausethetwo photons were quantum entangled, they made the same decision at the same moment. Manyrepetitionsprovidedtheidenticalresult.81 Theexperimentshavenotabsolutelyruledouttheexplanationofahiddenvariable—thatis,anunmeasurablestateofeachparticlethatisinphase(settothesamepointinacycle),sothatwhenoneparticleismeasured(forexample,hastodecideitspaththroughoroffaglassplate),theotherhasthesamevalueofthisinternalvariable.Sothe"choice"isgeneratedbyanidenticalsettingofthishiddenvariable,ratherthanbeingtheresultofactualcommunicationbetweenthetwoparticles.However,mostquantumphysicistsrejectthisinterpretation.Yetevenifweaccepttheinterpretationoftheseexperimentsasindicatingaquantumlinkbetweenthe two particles, the apparent communication is transmitting only randomness (profound quantumrandomness)at speeds fargreater than the speedof light,notpredetermined information, suchas thebitsinafile.Thiscommunicationofquantumrandomdecisionstodifferentpointsinspacecouldhavevalue, however, in applications such as providing encryption codes. Two different locations couldreceivethesamerandomsequence,whichcouldthenbeusedbyonelocationtoencryptamessageandbytheothertodecipherit.Itwouldnotbepossibleforanyoneelsetoeavesdropontheencryptioncodewithout destroying the quantum entanglement and thereby being detected. There are alreadycommercialencryptionproductsincorporatingthisprinciple.Thisisafortuitousapplicationofquantummechanics because of the possibility that another application of quantum mechanics—quantumcomputing—mayput an end to the standardmethodof encryptionbasedon factoring large numbers(whichquantumcomputing,withalargenumberofentangledqubits,wouldbegoodat).Yetanotherfaster-than-the-speed-of-lightphenomenonisthespeedwithwhichgalaxiescanrecedefromeachother as a result of the expansionof theuniverse. If thedistancebetween twogalaxies isgreater thanwhat is called theHubbledistance, then thesegalaxiesare receding fromoneanotheratfasterthanthespeedoflight.82ThisdoesnotviolateEinstein'sspecialtheoryofrelativity,becausethisvelocityiscausedbyspaceitselfexpandingratherthanthegalaxiesmovingthroughspace.However,italsodoesn'thelpustransmitinformationatspeedsfasterthanthespeedoflight.Wormholes. There are two exploratory conjectures that suggest ways to circumvent the apparentlimitation of the speed of light. The first is to usewormholes—folds of the universe in dimensionsbeyondthethreevisibleones.Thisdoesnotreallyinvolvetravelingatspeedsfasterthanthespeedoflightbutmerelymeansthatthetopologyoftheuniverseisnotthesimplethree-dimensionalspacethatnaive physics implies.However, ifwormholes or folds in the universe are ubiquitous, perhaps theseshortcutswouldallowustogeteverywherequickly.Orperhapswecanevenengineerthem. In1935EinsteinandphysicistNathanRosenformulated"Einstein-Rosen"bridgesasawayofdescribing electrons andotherparticles in termsof tiny space-time tunnels.83 In 1955physicist JohnWheelerdescribedthesetunnelsas"wormholes,"introducingthetermforthefirsttime.84Hisanalysisofwormholesshowedthemtobefullyconsistentwiththetheoryofgeneralrelativity,whichdescribesspaceasessentiallycurvedinanotherdimension. In 1988 California Institute of Technology physicistsMichaelMorris, Kip Thorne, and UriYurtsever explained in some detail how such wormholes could be engineered.85 Responding to aquestionfromCarlSagantheydescribedtheenergyrequirementstokeepwormholesofvaryingsizesopen.They also pointed out that basedonquantum fluctuation, so-called empty space is continuallygenerating tiny wormholes the size of subatomic particles. By adding energy and following other

requirements of both quantum physics and general relativity (two fields that have been notoriouslydifficulttounify),thesewormholescouldbeexpandedtoallowobjectslargerthansubatomicparticlestotravelthroughthem.Sendinghumansthroughthemwouldnotbeimpossiblebutextremelydifficult.However,asIpointedoutabove,wereallyonlyneedtosendnanobotsplusinformation,whichcouldpassthroughwormholesmeasuredinmicronsratherthanmeters. Thorne and hisPh.D. studentsMorris andYurtsever also described amethod consistentwithgeneral relativity and quantum mechanics that could establish wormholes between the Earth andfarawaylocations.Theirproposedtechniqueinvolvesexpandingaspontaneouslygenerated,subatomic-sizewormholetoalargersizebyaddingenergy,thenstabilizingitusingsuperconductingspheresinthetwoconnected"wormholemouths."Afterthewormholeisexpandedandstabilized,oneofitsmouths(entrances)istransportedtoanotherlocation,whilekeepingitsconnectiontotheotherentrance,whichremainsonEarth.ThorneofferedtheexampleofmovingtheremoteentranceviaasmallrocketshiptothestarVega,whichistwenty-fivelight-yearsaway.Bytravelingatveryclosetothespeedoflight,thejourney,asmeasuredbyclocksontheship,wouldberelativelybrief.Forexample,if theshiptraveledat99.995percentofthespeedoflight,theclocksontheshipwouldmoveaheadbyonlythreemonths.Althoughthe time for the voyage, as measured on Earth, would be around twenty-five years, the stretchedwormholewouldmaintainthedirectlinkbetweenthelocationsaswellasthepointsintimeofthetwolocations. Thus, even as experienced onEarth, itwould take only threemonths to establish the linkbetweenEarthandVega,becausethetwoendsofthewormholewouldmaintaintheirtimerelationship.Suitableengineeringimprovementscouldallowsuchlinkstobeestablishedanywhereintheuniverse.By traveling arbitrarily close to the speed of light, the time required to establish a link—for bothcommunicationsandtransportation—tootherlocationsintheuniverse,eventhosemillionsofbillionsoflightyearsaway,couldberelativelybrief. MattVisser ofWashingtonUniversity in St. Louis has suggested refinements to theMorris-Thorne-Yurtseverconceptthatprovideamorestableenvironment,whichmightevenallowhumanstotravelthroughwormholes.86Inmyview,however,thisisunnecessary.Bythetimeengineeringprojectsof this scale might be feasible, human intelligence will long since have been dominated by itsnonbiologicalcomponent.Sendingmolecular-scaleself-replicatingdevicesalongwithsoftwarewillbesufficientandmucheasier.AndersSandbergestimatesthataone-nanometerwormholecouldtransmitaformidable1069bitspersecond.87PhysicistDavidHochbergandVanderbiltUniversity'sThomasKephartpointoutthatshortlyaftertheBigBang,gravitywasstrongenoughtohaveprovidedtheenergyrequiredtospontaneouslycreatemassivenumbersofself-stabilizingwormholes.88Asignificantportionofthesewormholesislikelytostill be around andmay be pervasive, providing a vast network of corridors that reach far andwidethroughouttheuniverse.Itmightbeeasiertodiscoverandusethesenaturalwormholesthantocreatenewones.ChangingtheSpeedofLight.Thesecondconjectureistochangethespeedoflightitself.Inchapter3,Imentionedthefindingthatappearstoindicatethatthespeedoflighthasdifferedby4.5partsoutof108overthepasttwobillionyears.In2001astronomerJohnWebbdiscoveredthattheso-calledfine-structureconstantvariedwhenheexaminedlightfromsixty-eightquasars(verybrightyounggalaxies).89Thespeedoflightisoneoffourconstantsthatthefine-structureconstantcomprises,sotheresultisanothersuggestionthatvaryingconditionsintheuniversemaycausethespeedoflighttochange.CambridgeUniversityphysicistJohnBarrowandhiscolleaguesareintheprocessofrunningatwo-yeartabletopexperimentthatwilltesttheabilitytoengineerasmallchangeinthespeedoflight.90 Suggestions that the speed of light can vary are consistent with recent theories that it wassignificantlyhigherduringtheinflationaryperiodoftheuniverse(anearlyphaseinitshistory,whenitunderwentvery rapidexpansion).Theseexperiments showingpossiblevariation in the speedof light

clearlyneedcorroborationandareshowingonlysmallchanges.Butifconfirmed,thefindingswouldbeprofound,becauseitistheroleofengineeringtotakeasubtleeffectandgreatlyamplifyit.Again,thementalexperimentweshouldperformnowisnotwhethercontemporaryhumanscientists,suchasweare,canperformtheseengineeringfeatsbutwhetherornotahumancivilizationthathasexpandeditsintelligencebytrillionsoftrillionswillbeabletodoso.Fornowwecansaythatultrahighlevelsofintelligencewillexpandoutwardatthespeedoflight,while recognizing that our contemporary understanding of physics suggests that thismay not be theactual limitof thespeedofexpansionor,even if thespeedof lightproves tobe immutable, that thislimitmaynotrestrictreachingotherlocationsquicklythroughwormholes.TheFermiParadoxRevisited.Recallthatbiologicalevolutionismeasuredinmillionsandbillionsofyears.Soifthereareothercivilizationsoutthere,theywouldbespreadoutintermsofdevelopmentbyhugespansoftime.TheSETIassumptionimpliesthatthereshouldbebillionsofETIs(amongallthegalaxies),sothereshouldbebillionsthatliefaraheadofusintheirtechnologicalprogress.Yetittakesonlyafewcenturiesatmostfromtheadventofcomputationforsuchcivilizationstoexpandoutwardatatleastlightspeed.Giventhis,howcanitbethatwehavenotnoticedthem?TheconclusionIreachisthatitislikely(althoughnotcertain)thattherearenosuchothercivilizations.Inotherwords,weareinthelead.That'sright,ourhumblecivilizationwithitspickuptrucks,fastfood,andpersistentconflicts(andcomputation!)isintheleadintermsofthecreationofcomplexityandorderintheuniverse. Nowhowcanthatbe?Isn't thisextremelyunlikely,giventhesheernumberoflikelyinhabitedplanets?Indeeditisveryunlikely.Butequallyunlikelyistheexistenceofouruniverse,withitssetoflawsofphysicsandrelatedphysicalconstants,soexquisitely,preciselywhatisneededfortheevolutionoflifetobepossible.Butbytheanthropicprinciple,iftheuniversedidn'tallowtheevolutionoflifewewouldn'tbeheretonoticeit.Yethereweare.Sobyasimilaranthropicprinciple,we'rehereintheleadintheuniverse.Again,ifweweren'there,wewouldnotbenoticingit.Let'sconsidersomeargumentsagainstthisperspective.Perhapsthereareextremelyadvancedtechnologicalcivilizationsoutthere,butweareoutsidetheirlightsphereofintelligence.Thatis,theyhaven'tgottenhereyet.Okay,inthiscase,SETIwillstillfailtofindETIsbecausewewon'tbeabletosee(orhear)them,atleastnotunlessanduntilwefindawaytobreakoutofourlightsphere(ortheETldoesso)bymanipulatingthespeedoflightorfindingshortcuts,asIdiscussedabove. Perhaps theyareamongus,buthavedecided toremain invisible tous. If theyhavemade thatdecision, theyare likely to succeed in avoidingbeingnoticed.Again, it ishard tobelieve that everysingleETlhasmadethesamedecision. JohnSmart has suggested inwhat he calls the "transcension" scenario that once civilizationssaturate their local region of spacewith their intelligence, they create a new universe (one thatwillallow continued exponential growth of complexity and intelligence) and essentially leave thisuniverse.91Smartsuggests that thisoptionmaybesoattractive that it is theconsistentandinevitableoutcomeofanETl'shavingreachedanadvancedstageofitsdevelopment,andittherebyexplainstheFermiParadox. Incidentally,Ihavealwaysconsideredthescience-fictionnotionof largespaceshipspilotedbyhuge,squishycreaturessimilartoustobeveryunlikely.SethShostakcommentsthat"thereasonableprobability is that any extraterrestrial intelligence we will detect will be machine intelligence, notbiologicalintelligencelikeus."Inmyviewthisisnotsimplyamatterofbiologicalbeingssendingoutmachines (aswedo today)but rather thatanycivilizationsophisticatedenough tomake the tripherewould have long since passed the point ofmergingwith its technology andwould not need to sendphysicallybulkyorganismsandequipment. If they exist,whywould they come here?Onemissionwould be for observation—to gatherknowledge (just aswe observe other species on Earth today).Anotherwould be to seekmatter andenergy to provide additional substrate for its expanding intelligence.The intelligence and equipmentneeded for such exploration and expansion (by an ETl, or by us when we get to that stage of

development)wouldbeextremelysmall,basicallynanobotsandinformationtransmissions.Itappearsthatoursolarsystemhasnotyetbeenturnedintosomeoneelse'scomputer.Andifthisothercivilizationisonlyobservingusforknowledge'ssakeandhasdecidedtoremainsilent,SETlwillfailtofindit,becauseifanadvancedcivilizationdoesnotwantustonoticeit,itwouldsucceedinthatdesire.Keepinmindthatsuchacivilizationwouldbevastlymoreintelligentthanwearetoday.Perhapsitwill reveal itself to uswhenwe achieve the next level of our evolution, specificallymerging ourbiologicalbrainswithour technology,which is to say, after theSingularity.However,given that theSETlassumptionimpliesthattherearebillionsofsuchhighlydevelopedcivilizations,itseemsunlikelythatallofthemhavemadethesamedecisiontostayoutofourway.TheAnthropic Principle Revisited. We are struck with two possible applications of an anthropicprinciple,onefortheremarkablebiofriendlylawsofouruniverse,andonefortheactualbiologyofourplanet.Let'sfirstconsidertheanthropicprincipleasappliedtotheuniverseinmoredetail.Thequestionconcerning the universe arises becausewe notice that the constants in nature are preciselywhat arerequired for the universe to have grown in complexity. If the cosmological constant, the Planckconstant, and the many other constants of physics were set to just slightly different values, atoms,molecules,stars,planets,organisms,andhumanswouldhavebeenimpossible.Theuniverseappearstohaveexactlytherightrulesandconstants.(ThesituationisreminiscentofStevenWolfram'sobservationthat certain cellular-automata rules [see the sidebar on p. 85] allow for the creation of remarkablycomplex and unpredictable patterns, whereas other rules lead to very uninteresting patterns, such asalternatinglinesorsimpletrianglesinarepeatingorrandomconfiguration.)Howdoweaccountfortheremarkabledesignofthelawsandconstantsofmatterandenergyinour universe that have allowed for the increasing complexity we see in biological and technologyevolution?FreemanDysononcecommentedthat"theuniverseinsomesenseknewwewerecoming."ComplexitytheoristJamesGardnerdescribesthequestioninthisway:

Physicists feel that the task of physics is to predict what happens in the lab, and they areconvincedthatstringtheory,orMtheorycandothis....Buttheyhavenoideawhytheuniverseshould...havethestandardmodel,withthevaluesofits40+parametersthatweobserve.Howcananyonebelievethatsomethingsomessyistheuniquepredictionofstringtheory?Itamazesmethatpeoplecanhavesuchblinkeredvision,thattheycanconcentratejustonthefinalstateoftheuniverse,andnotaskhowandwhyitgotthere.92

The perplexity of how it is that the universe is so "friendly" to biology has led to variousformulationsoftheanthropicprinciple.The"weak"versionoftheanthropicprinciplepointsoutsimplythatifitwerenotthecase,wewouldn'tbeheretowonderaboutit.Soonlyinauniversethatallowedforincreasingcomplexitycouldthequestionevenbeasked.Strongerversionsoftheanthropicprinciplestate that there must bemore to it; advocates of these versions are not satisfied with a mere luckycoincidence.ThishasopenedthedoorforadvocatesofintelligentdesigntoclaimthatthisistheproofofGod'sexistencethatscientistshavebeenaskingfor.The Multiverse. Recently a more Darwinian approach to the strong anthropic principle has beenproposed. Consider that it is possible for mathematical equations to have multiple solutions. Forexample,ifwesolveforxintheequationx2=4,xmaybe2or-2.Someequationsallowforaninfinitenumberofsolutions.Intheequation(a–b)Îx=0,xcantakeonanyoneofaninfinitenumberofvaluesifa=b (since anynumbermultipliedby zero equals zero). It turnsout that the equations for recentstring theories allow in principle for an infinite number of solutions. To bemore precise, since thespatialandtemporalresolutionoftheuniverseislimitedtotheverysmallPlanckconstant,thenumberofsolutionsisnotliterallyinfinitebutmerelyvast.Stringtheoryimplies,therefore,thatmanydifferentsetsofnaturalconstantsarepossible.

Thishasledtotheideaofthemultiverse:thatthereexistavastnumberofuniverses,ofwhichourhumbleuniverseisonlyone.Consistentwithstringtheory,eachoftheseuniversescanhaveadifferentsetofphysicalconstants.EvolvingUniverses.LeonardSusskind,thediscovererofstringtheory,andLeeSmolin,atheoreticalphysicistandexpertonquantumgravity,havesuggestedthatuniversesgiverisetootheruniversesinanatural, evolutionary process that gradually refines the natural constants. In otherwords it is not byaccidentthattherulesandconstantsofouruniverseareidealforevolvingintelligentlifebutratherthattheythemselvesevolvedtobethatway.InSmolin'stheorythemechanismthatgivesrisetonewuniversesisthecreationofblackholes,sothose universes best able to produce black holes are the ones that are most likely to reproduce.According toSmolinauniversebestable tocreate increasingcomplexity—that is,biological life—isalsomostlikelytocreatenewuniverse-generatingblackholes.Asheexplains,"Reproductionthroughblack holes leads to amultiverse in which the conditions for life are common—essentially becausesomeoftheconditionsliferequires,suchasplentifulcarbon,alsoboosttheformationofstarsmassiveenoughtobecomeblackholes."93Susskind'sproposaldiffersindetailfromSmolin'sbutisalsobasedon black holes, aswell as the nature of "inflation," the force that caused the very early universe toexpandrapidly.IntelligenceastheDestinyoftheUniverse.InTheAgeofSpiritualMachines,Iintroducedarelatedidea—namely,thatintelligencewouldultimatelypermeatetheuniverseandwoulddecidethedestinyofthecosmos:

Howrelevantisintelligencetotheuniverse?...Thecommonwisdomisnotvery.Starsarebornanddie; galaxies go through their cycles of creation anddestruction; the universe itselfwasborn in a big bang andwill endwith a crunchor awhimper,we're not yet surewhich.Butintelligence has little to do with it. Intelligence is just a bit of froth, an ebullition of littlecreatures darting in and out of inexorable universal forces. Themindlessmechanism of theuniverse iswindingupor down to a I distant future, and there's nothing intelligence candoaboutit.That'sthecommonwisdom.ButIdon'tagreewithit.Myconjectureisthatintelligencewillultimatelyprovemorepowerfulthanthesebigimpersonalforces....Sowilltheuniverseendinabigcrunch,orinaninfiniteexpansionofdeadstars,orinsome other manner? In my view, the primary issue is not the mass of the universe, or thepossibleexistenceofantigravity,orofEinstein'sso-calledcosmologicalconstant.Rather, thefateoftheuniverseisadecisionyettobemade,onewhichwewillintelligentlyconsiderwhenthetimeisright.94

Complexity theorist JamesGardner combinedmy suggestion on the evolution of intelligencethroughout the universe with Smolin's and Susskind's concepts of evolving universes. Gardnerconjectures that it is specifically the evolution of intelligent life that enables offspring universes.95Gardner builds on British astronomerMartin Rees's observation that "whatwe call the fundamentalconstants—thenumbersthatmattertophysicists—maybesecondaryconsequencesofthefinaltheory,rather than directmanifestations of its deepest andmost fundamental level." To Smolin it ismerelycoincidencethatblackholesandbiologicallifebothneedsimilarconditions(suchaslargeamountsofcarbon),soinhisconceptionthereisnoexplicitroleforintelligence,otherthanthatithappenstobetheby-productofcertainbiofriendlycircumstances.InGardner'sconceptionitisintelligentlifethatcreatesitssuccessors.Gardnerwritesthat"weandotherlivingcreaturesthroughoutthecosmosarepartofavast,stillundiscoveredtransterrestrialcommunityoflivesandintelligencesspreadacrossbillionsofgalaxiesandcountless parsecs who are collectively engaged in a portentousmission of truly cosmic importance.

UndertheBiocosmvision,weshareacommonfatewiththatcommunity—tohelpshapethefutureofthe universe and transform it from a collection of lifeless atoms into a vast, transcendentmind."ToGardnerthelawsofnature,andthepreciselybalancedconstants,"functionasthecosmiccounterpartofDNA:theyfurnishthe'recipe'bywhichtheevolvingcosmosacquiresthecapacitytogeneratelifeandevermorecapableintelligence."MyownviewisconsistentwithGardner'sbeliefinintelligenceasthemostimportantphenomenonintheuniverse.IdohaveadisagreementwithGardneronhissuggestionofa"vast ... transterrestrialcommunityoflivesandintelligencesspreadacrossbillionsofgalaxies."Wedon'tyetseeevidencethatsuchacommunitybeyondEarthexists.Thecommunitythatmattersmaybejustourownunassumingcivilization here. As I pointed out above, although we can fashion all kinds of reasons why eachparticularintelligentcivilizationmayremainhiddenfromus(forexample,theydestroyedthemselves,or theyhavedecided to remain invisibleor stealthy,or they've switchedall of their communicationsaway from electromagnetic transmissions, and so on), it is not credible to believe that every singlecivilizationoutofthebillionsthatshouldbethere(accordingtotheSETIassumption)hassomereasontobeinvisible.TheUltimateUtilityFunction.WecanfashionaconceptualbridgebetweenSusskind'sandSmolin'sideaofblackholesbeingthe"utilityfunction"(thepropertybeingoptimizedinanevolutionaryprocess)ofeachuniverseinthemultiverseandtheconceptionofintelligenceastheutilityfunctionthatIsharewithGardner.AsIdiscussedinchapter3,thecomputationalpowerofacomputerisafunctionofitsmass and its computational efficiency. Recall that a rock has significant mass but extremely lowcomputationalefficiency(thatis,virtuallyallofthetransactionsofitsparticlesareeffectivelyrandom).Mostof theparticle interactions in ahumanare randomalso,butona logarithmic scalehumansareroughlyhalfwaybetweenarockandtheultimatesmallcomputer.Acomputerintherangeoftheultimatecomputerhasaveryhighcomputationalefficiency.Onceweachieveanoptimalcomputationalefficiency,theonlywaytoincreasethecomputationalpowerofacomputerwouldbetoincreaseitsmass.Ifweincreasethemassenough,itsgravitationalforcebecomesstrongenoughtocauseittocollapseintoablackhole.Soablackholecanberegardedastheultimatecomputer.Ofcourse,notanyblackholewilldo.Mostblackholes,likemostrocks,areperforminglotsofrandom transactions but no useful computation. But awell-organized black holewould be themostpowerfulconceivablecomputerintermsofcpsperliter.HawkingRadiation. There has been a long-standing debate about whether or not we can transmitinformation into a black hole, have it usefully transformed, and then retrieve it. StephenHawking'sconceptionoftransmissionsfromablackholeinvolvesparticle-antiparticlepairs thatarecreatedneartheeventhorizon(thepointofnoreturnnearablackhole,beyondwhichmatterandenergyareunableto escape).When this spontaneous creation occurs, as it does everywhere in space, the particle andantiparticletravelinoppositedirections.Ifonememberofthepairtravelsintotheeventhorizon(nevertobeseenagain),theotherwillflyawayfromtheblackhole.SomeoftheseparticleswillhavesufficientenergytoescapeitsgravitationandresultinwhathasbeencalledHawkingradiation.96PriortoHawking'sanalysisitwasthoughtthatblackholeswere,well,black;withhisinsightwerealizedthattheyactuallygiveoffacontinualshowerofenergeticparticles.But according toHawking this radiation is random, since it originates from randomquantum eventsneartheeventboundary.Soablackholemaycontainanultimatecomputer,accordingtoHawking,butaccording tohisoriginal conception,no informationcanescapeablackhole, so this computer couldnevertransmititsresults. In 1997Hawking and fellow physicistKipThorne (thewormhole scientist)made a betwithCaliforniaInstituteofTechnology'sJohnPreskill.HawkingandThornemaintainedthattheinformationthatenteredablackholewaslost,andanycomputationthatmightoccurinsidetheblackhole,usefulorotherwise, could never be transmitted outside of it,whereas Preskillmaintained that the information

could be recovered.97 The loser was to give the winner some useful information in the form of anencyclopedia. In the intervening years the consensus in the physics community steadilymoved away fromHawking, and on July 21, 2004,Hawking admitted defeat and acknowledged that Preskill had beencorrectafterall: that informationsent intoablackhole isnot lost. Itcouldbe transformed inside theblackhole and then transmittedoutside it.According to this understanding,what happens is that theparticle that flies away from the black hole remains quantum entangled with its antiparticle thatdisappearedintotheblackhole.Ifthatantiparticleinsidetheblackholebecomesinvolvedinausefulcomputation,thentheseresultswillbeencodedinthestateofitstangledpartnerparticleoutsideoftheblackhole.AccordinglyHawkingsentPreskillanencyclopediaonthegameofcricket,butPreskillrejectedit,insistingonabaseballencyclopedia,whichHawkinghadflownoverforaceremonialpresentation.AssumingthatHawking'snewpositionisindeedcorrect,theultimatecomputersthatwecancreatewouldbeblackholes.Thereforeauniversethatiswelldesignedtocreateblackholeswouldbeonethatiswelldesignedtooptimizeitsintelligence.SusskindandSmolinarguedmerelythatbiologyandblackholesbothrequirethesamekindofmaterials,soauniversethatwasoptimizedforblackholeswouldalso be optimized for biology.Recognizing that blackholes are the ultimate repository of intelligentcomputation, however,we can conclude that theutility functionof optimizingblack-hole productionandthatofoptimizingintelligenceareoneandthesame.Why Intelligence IsMore Powerful than Physics. There is another reason to apply an anthropicprinciple. It may seem remarkably unlikely that our planet is in the lead in terms of technologicaldevelopment, but as I pointed out above, by aweak anthropic principle, if we had not evolved,wewouldnotbeherediscussingthisissue. As intelligencesaturates thematterandenergyavailable to it, it turnsdumbmatter intosmartmatter. Although smart matter still nominally follows the laws of physics, it is so extraordinarilyintelligentthatitcanharnessthemostsubtleaspectsofthelawstomanipulatematterandenergytoitswill.Soitwouldatleastappearthatintelligenceismorepowerfulthanphysics.WhatIshouldsayisthat intelligence is more powerful than cosmology. That is, once matter evolves into smart matter(matter fullysaturatedwith intelligentprocesses), itcanmanipulateothermatterandenergy todo itsbidding (through suitably powerful engineering). This perspective is not generally considered indiscussionsoffuturecosmology.Itisassumedthatintelligenceisirrelevanttoeventsandprocessesonacosmologicalscale. Onceaplanetyieldsatechnology-creatingspeciesandthatspeciescreatescomputation(ashashappenedhere), it is only amatter of a fewcenturies before its intelligence saturates thematter andenergy in its vicinity, and it begins to expand outward at at least the speed of light (with somesuggestions of circumventing this limit). Such a civilization will then overcome gravity (throughexquisiteandvasttechnology)andothercosmologicalforces—or,tobefullyaccurate,itwillmaneuverandcontroltheseforces—andengineertheuniverseitwants.ThisisthegoaloftheSingularity.AUniverse-ScaleComputer.Howlongwillittakeforourcivilizationtosaturatetheuniversewithourvastlyexpandedintelligence?SethLloydestimatesthereareabout1080particlesintheuniverse,withatheoreticalmaximumcapacityofabout1090 cps. Inotherwords auniverse-scale computerwouldbeabletocomputeat1090cps.98Toarriveatthoseestimates,Lloydtooktheobserveddensityofmatter—about one hydrogen atom per cubic meter—and from this figure computed the total energy in theuniverse.Dividing this energy figure by thePlanck constant, he got about 1090 cps.Theuniverse isabout1017secondsold,soinroundnumberstherehavebeenamaximumofabout10107calculationsinitthusfar.Witheachparticleabletostoreabout1010bitsinallofitsdegreesoffreedom(includingitsposition,trajectory,spin,andsoon),thestateoftheuniverserepresentsabout1090bitsofinformationateachpointintime.

Wedonotneedtocontemplatedevotingallofthemassandenergyoftheuniversetocomputation.If we were to apply 0.01 percent, that would still leave 99.99 percent of the mass and energyunmodified,butwouldstillresultinapotentialofabout1086cps.Basedonourcurrentunderstanding,wecanonlyapproximatetheseordersofmagnitude.Intelligenceatanythingclosetotheselevelswillbesovastthatitwillbeabletoperformtheseengineeringfeatswithenoughcaresoasnottodisruptwhatevernaturalprocessesitconsidersimportanttopreserve.TheHolographicUniverse.Anotherperspectiveonthemaximuminformationstorageandprocessingcapability of the universe comes from a speculative recent theory of the nature of information.According to the "holographic universe" theory the universe is actually a two-dimensional array ofinformationwrittenonitssurface,soitsconventionalthree-dimensionalappearanceisanillusion.99Inessence,theuniverse,accordingtothistheory,isagianthologram.Theinformationiswrittenataveryfinescale,governedbythePlanckconstant.SothemaximumamountofinformationintheuniverseisitssurfaceareadividedbythesquareofthePlanckconstant,whichcomestoabout10120bits.Theredoesnotappeartobeenoughmatterintheuniversetoencodethismuch information, so the limitsof theholographicuniversemaybehigher thanwhat is actuallyfeasible. Inanyevent theorderofmagnitudeof thenumberofordersofmagnitudesof thesevariousestimatesisinthesamerange.Thenumberofbitsthatauniversereorganizedforusefulcomputationwillbeabletostoreis10raisedtoapowersomewherebetween80and120.Again,ourengineering,eventhatofourvastlyevolvedfutureselves;willprobablyfallshortofthesemaximums.Inchapter2Ishowedhowweprogressedfrom10–5to108cpsperthousanddollarsduringthetwentiethcentury.Basedonacontinuationofthesmooth,doublyexponentialgrowththatwesawinthetwentiethcentury,Iprojectedthatwewouldachieveabout1060cpsperthousanddollarsby2100.Ifweestimateamodesttrilliondollarsdevotedtocomputation,that'satotalofabout1069cpsbytheendofthiscentury.Thiscanbeachievedwiththematterandenergyinoursolarsystem. Toget toaround1090cps requiresexpanding through the restof theuniverse.Continuing thedouble-exponential growth curve shows that we can saturate the universewith our intelligencewellbefore the end of the twenty-second century, provided thatwe are not limited by the speed of light.Eveniftheup-to-thirtyadditionalpowersoftensuggestedbytheholographic-universetheoryareborneout,westillreachsaturationbytheendofthetwenty-secondcentury.Again,ifitisatallpossibletocircumventthespeed-of-lightlimitation,thevastintelligencewewill have with solar system-scale intelligence will be able to design and implement the requisiteengineeringtodoso.IfIhadtoplaceabet,Iwouldputmymoneyontheconjecturethatcircumventingthespeedoflightispossibleandthatwewillbeabletodothiswithinthenextcoupleofhundredyears.Butthatisspeculationonmypart,aswedonotyetunderstandtheseissuessufficientlytomakeamoredefinitivestatement.Ifthespeedoflightisanimmutablebarrier,andnoshortcutsthroughwormholesexistthatcanbeexploited,itwilltakebillionsofyears,nothundreds,tosaturatetheuniversewithourintelligence,andwewillbelimitedtoourlightconewithintheuniverse.Ineithereventtheexponentialgrowthofcomputationwillhitawallduringthetwenty-secondcentury.(Butwhatawall!) This largedifference in timespans—hundredsofyearsversusbillionsofyears (to saturate theuniversewithour intelligence)—demonstrateswhy the issueof circumventing the speedof lightwillbecomesoimportant.Itwillbecomeaprimarypreoccupationofthevastintelligenceofourcivilizationinthetwenty-secondcentury.ThatiswhyIbelievethatifwormholesorothercircumventingmeansarefeasible,wewillbehighlymotivatedtofindandexploitthem.Ifitispossibletoengineernewuniversesandestablishcontactwiththem,thiswouldprovideyetfurther means for an intelligent civilization to continue its expansion. Gardner's view is that theinfluenceofan intelligentcivilization increatinganewuniverse lies insetting thephysical lawsandconstantsofthebabyuniverse.Butthevastintelligenceofsuchacivilizationmayfigureoutwaystoexpand itsownintelligence intoanewuniversemoredirectly.The ideaofspreadingour intelligencebeyond this universe is, of course, speculative, as none of the multiverse theories allows for

communicationfromoneuniversetoanother,exceptforpassingonbasiclawsandconstants.Evenifwearelimitedtotheoneuniversewealreadyknowabout,saturatingitsmatterandenergywithintelligenceisourultimatefate.Whatkindofuniversewillthatbe?Well,justwaitandsee.MOLLY2004:SowhentheuniversereachesEpochSix{thestageatwhichthenonbiologicalportion

ofourintelligencespreadsthroughtheuniverse],what'sitgoingtodo?CHARLESDARWIN: I'm not sure we can answer that. As you said, it's like bacteria asking one

anotherwhathumanswilldo.MOLLY2004:SotheseEpochSixentitieswillconsiderusbiologicalhumanstobelikebacteria?GEORGE2048:That'scertainlynothowIthinkofyou.MOLLY2104:George,you'reonlyEpochFive,soIdon'tthinkthatanswersthequestion.CHARLES:Gettingbacktothebacteria,whattheywouldsay,iftheycouldtalk—MOLLY2004:—andthink.CHARLES:Yes, that, too.Theywould say thathumanswilldo the same thingsaswebacteriado—

namely,eat,avoiddanger,andprocreate.MOLLY2104:Oh,butourprocreationissomuchmoreinteresting.MOLLY 2004: Actually, Molly of the future, it's our human pre-Singularity procreation that's

interesting.Yourvirtualprocreationis,actually,alotlikethatofthebacteria.Sexhasnothingtodowithit.

MOLLY2104: It's true we've separated sexuality from reproduction, but that's not exactly new tohumancivilizationin2004.Andbesides,unlikebacteria,wecanchangeourselves.

MOLLY2004:Actually,you'veseparatedchangeandevolutionfromreproductionaswell.MOLLY2104:Thatwasalsoessentiallytruein2004.MOLLY2004:Okay,okay.Butaboutyourlist,Charles,wehumansalsodothingslikecreateartand

music.Thatkindofseparatesusfromotheranimals.GEORGE2048:Indeed,Molly,thatisfundamentallywhattheSingularityisabout.TheSingularityis

thesweetestmusic,thedeepestart,themostbeautifulmathematics....MOLLY2004:Isee,sothemusicandartoftheSingularitywillbetomyera'smusicandartascirca

2004musicandartareto...NEDLUDD:Themusicandartofbacteria.MOLLY2004:Well,I'veseensomeartisticmoldpatterns.NED:Yes,butI'msureyoudidn'treverethem.MOLLY2004:No,actually,Iwipedthemaway.NED:Okay,mypointthen.MOLLY2004:I'mstilltryingtoenvisionwhattheuniversewillbedoinginEpochSix.TIMOTHYLEARY:Theuniversewillbeflyinglikeabird.MOLLY2004:Butwhatisitflyingin?Imeanit'severything.TIMOTHY:That'slikeasking,Whatisthesoundofonehandclapping?MOLLY2004:Hmmm,sotheSingularityiswhattheZenmastershadinmindallalong.

A

CHAPTERSEVEN

IchbineinSingularitarian

Themostcommonofallfolliesistobelievepassionatelyinthepalpablynottrue.

—H.L.MENCKENPhilosophies of life rooted in centuries-old traditions contain much wisdom concerningpersonal, organizational, and social living. Many of us also find shortcomings in thosetraditions. How could they not reach some mistaken conclusions when they arose in pre-scientific times?At the same time, ancient philosophies of life have little or nothing to sayaboutfundamentalissuesconfrontingusasadvancedtechnologiesbegintoenableustochangeouridentityasindividualsandashumansandaseconomic,cultural,andpoliticalforceschangeglobalrelationships.

—MAXMORE,"PRINCIPLESOFEXTROPY"Theworlddoesnotneedanothertotalisticdogma.

—MAXMORE,"PRINCIPLESOFEXTROPY"Yes,wehaveasoul.Butit'smadeoflotsoftinyrobots.

—GIULIOGIORELLISubstrate is morally irrelevant, assuming it doesn't affect functionality or consciousness. Itdoesn'tmatter, from amoral point of view,whether somebody runs on silicon or biologicalneurons(justas itdoesn'tmatterwhetheryouhavedarkorpaleskin).On thesamegrounds,thatwerejectracismandspeciesism,weshouldalsorejectcarbon-chauvinism,orbioism.

—NICK BOSTROM, "ETHICS FOR INTELLIGENT MACHINES: APROPOSAL,2001"

Philosophershavelongnotedthattheirchildrenwerebornintoamorecomplexworldthanthatoftheirancestors.Thisearlyandperhapsevenunconsciousrecognitionofacceleratingchangemayhavebeenthecatalystformuchoftheutopian,apocalyptic,andmillennialistthinkinginourWestern tradition. But the modern difference is that now everyone notices the pace ofprogressonsomelevel,notsimplythevisionaries.

—JOHNSMART

SingularitarianissomeonewhounderstandstheSingularityandhasreflectedonitsmeaningforhisorherownlife. Ihavebeenengagedinsuchreflectionforseveraldecades.Needless tosay, it'snotaprocessthatonecanevercomplete.Istartedponderingtherelationshipofourthinkingtoourcomputational technology as a teenager in the 1960s. In the 1970s I began to study the

accelerationoftechnology,andIwrotemyfirstbookonthesubjectinthelate1980s.SoI'vehad time to contemplate the impact on society—and on myself—of the overlappingtransformationsnowunderway.

GeorgeGilderhasdescribedmyscientificandphilosophicalviewsas"asubstitutevisionforthosewhohavelostfaithinthetraditionalobjectofreligiousbelief."1Gilder'sstatementisunderstandable,asthere are at least apparent similarities between anticipation of theSingularity and anticipation of thetransformationsarticulatedbytraditionalreligions.ButIdidnotcometomyperspectiveasaresultofsearchingforanalternativetocustomaryfaith.Theoriginofmyquesttounderstandtechnologytrendswaspractical:anattempttotimemyinventionsandtomakeoptimaltacticaldecisionsinlaunchingtechnologyenterprises.Overtimethismodelingoftechnologytookonalifeofitsownandledmetoformulateatheoryoftechnologyevolution.Itwasnota huge leap from there to reflect on the impact of these crucial changes on social and culturalinstitutions andonmyown life.So,whilebeing aSingularitarian is not amatterof faithbutoneofunderstanding, pondering the scientific trends I've discussed in this book inescapably engenders newperspectiveson the issues that traditional religionshaveattempted toaddress: thenatureofmortalityandimmortality,thepurposeofourlives,andintelligenceintheuniverse.BeingaSingularitarianhasoftenbeenanalienatingandlonelyexperienceformebecausemostpeopleIencounterdonotsharemyoutlook.Most"bigthinkers"aretotallyunawareofthisbigthought.Inamyriadofstatementsandcommentspeopletypicallyevidencethecommonwisdomthathumanlifeisshort,thatourphysicalandintellectualreachislimited,andthatnothingfundamentalwillchangeinour lifetimes. Iexpect thisnarrowview tochangeas the implicationsofacceleratingchangebecomeincreasinglyapparent,buthavingmorepeoplewithwhomtosharemyoutlookisamajorreasonthatIwrotethisbook.SohowdowecontemplatetheSingularity?Aswiththesun,it'shardtolookatdirectly;it'sbetterto squint at it out of the cornersof our eyes.AsMaxMore states, the last thingweneed is anotherdogma, nor do we need another cult, so Singularitarianism is not a system of beliefs or unifiedviewpoints.Whileitisfundamentallyanunderstandingofbasictechnologytrends,itissimultaneouslyaninsightthatcausesonetorethinkeverything,fromthenatureofhealthandwealthtothenatureofdeathandself. Tome,beingaSingularitarianmeansmanythings,ofwhichthefollowingisasmallsampling.Thesereflectionsarticulatemypersonalphilosophy,notaproposalforanewdoctrine.

·We have themeans right now to live long enough to live forever.2 Existing knowledge can beaggressivelyappliedtodramaticallyslowdownagingprocessessowecanstillbeinvitalhealthwhen themoreradical life-extending therapies frombiotechnologyandnanotechnologybecomeavailable. Butmost baby boomerswon'tmake it because they are unaware of the acceleratingagingprocessesintheirbodiesandtheopportunitytointervene.

·InthisspiritIamaggressivelyreprogrammingmybiochemistry,whichisnowaltogetherdifferentthanitwouldotherwisebe.3Takingsupplementsandmedicationsisnotalastresorttobereservedonlyforwhensomethinggoeswrong.Thereisalreadysomethingwrong.Ourbodiesaregovernedbyobsoletegeneticprogramsthatevolvedinabygoneera,soweneedtoovercomeourgeneticheritage.Wealreadyhavetheknowledgetobegintoaccomplishthis,somethingIamcommittedtodoing.

·Mybodyistemporary.Itsparticlesturnoveralmostcompletelyeverymonth.Onlythepatternofmybodyandbrainhavecontinuity.

·Weshouldstrivetoimprovethesepatternsbyoptimizingthehealthofourbodiesandextendingthereachofourminds.Ultimately,wewillbeabletovastlyexpandourmentalfacultiesbymergingwithourtechnology.

·Weneedabody,butonceweincorporateMNTfabricationintoourselves,wewillbeabletochangeourbodiesatwill.

·Onlytechnologycanprovidethescaletoovercomethechallengeswithwhichhumansocietyhas

struggled for generations. For example, emerging technologies will provide the means ofproviding and storing clean and renewable energy, removing toxins and pathogens from ourbodies and the environment, andproviding the knowledge andwealth to overcomehunger andpoverty.

·Knowledgeispreciousinallitsforms:music,art,science,andtechnology,aswellastheembeddedknowledgeinourbodiesandbrains.Anylossofthisknowledgeistragic.

·Information isnotknowledge.Theworld isawash in information; it is the roleof intelligence tofindandacton thesalientpatterns.Forexample,wehavehundredsofmegabitsof informationflowingthroughoursenseseverysecond,thebulkofwhichisintelligentlydiscarded.Itisonlythekeyrecognitionsandinsights(allformsofknowledge)thatweretain.Thusintelligenceselectivelydestroysinformationtocreateknowledge.

·Death is a tragedy. It is not demeaning to regard a person as a profound pattern (a form ofknowledge),whichislostwhenheorshedies.That,atleast,isthecasetoday,sincewedonotyethave the means to access and back up this knowledge. When people speak of losing part ofthemselveswhena lovedonedies, theyarespeakingquite literally, sincewe lose theability toeffectivelyusetheneuralpatternsinourbrainthathadself-organizedtointeractwiththatperson.

·Aprimaryroleoftraditionalreligionisdeathistrationalization—thatis,rationalizingthetragedyofdeathasagood thing.MalcolmMuggeridgearticulates thecommonview that"if itweren't fordeath,lifewouldbeunbearable."Buttheexplosionofart,science,andotherformsofknowledgethat the Singularity will bring will make life more than bearable; it will make life trulymeaningful.

·In my view the purpose of life—and of our lives—is to create and appreciate ever-greaterknowledge,tomovetowardgreater"order."AsIdiscussedinchapter2,increasingorderusuallymeansincreasingcomplexity,butsometimesaprofoundinsightwillincreaseorderwhilereducingcomplexity.

·AsIseeitthepurposeoftheuniversereflectsthesamepurposeasourlives:tomovetowardgreaterintelligenceandknowledge.Ourhumanintelligenceandourtechnologyformthecuttingedgeofthisexpandingintelligence(giventhatwearenotawareofanyextraterrestrialcompetitors).

·Havingreachedatippingpoint,wewillwithinthiscenturybereadytoinfuseoursolarsystemwithour intelligence throughself-replicatingnonbiological intelligence. Itwill thenspreadout to therestoftheuniverse.

·Ideasaretheembodimentandtheproductofintelligence.Theideasexisttosolvemostanyproblemthatweencounter.Theprimaryproblemswecannotsolveareonesthatwecannotarticulateandaremostlyonesofwhichwearenotyetaware.Fortheproblemsthatwedoencounter, thekeychallengeistoexpressthempreciselyinwords(andsometimesinequations).Havingdonethat,wehavetheabilitytofindtheideastoconfrontandresolveeachsuchproblem.

·We can apply the enormous leverage provided by the acceleration of technology. A notableexample isachieving radical lifeextension through"abridge toabridge toabridge" (applyingtoday's knowledge as a bridge to biotechnology, which in turn will bridge us to the era ofnanotechnology).4Thisoffersawayto live indefinitelynow,eventhoughwedon'tyethavealltheknowledgenecessaryfor radical lifeextension. Inotherwordswedon'thave tosolveeveryproblemtoday.Wecananticipatethecapabilityoftechnologiesthatarecoming—infiveyearsorten years or twenty—andwork these into our plans. That is how I designmy own technologyprojects,andwecandothesamewiththelargeproblemsfacingsocietyandwithourownlives.

ContemporaryphilosopherMaxMoredescribes thegoalofhumanityasa transcendence tobe"achieved through science and technology steered by human values."5 More cites Nietzsche'sobservation "Man is a rope, fastened between animal and overman—a rope over an abyss."We caninterpretNietzsche tobepointingout thatwehave advancedbeyondother animalswhile seeking tobecomesomethingfargreater.WemightregardNietzsche'sreferencetotheabysstoalludetotheperilsinherentintechnology,whichIaddressinthenextchapter.

MorehasatthesametimeexpressedconcernthatanticipatingtheSingularitycouldengenderapassivityinaddressingtoday'sissues."6Becausetheenormouscapabilitytoovercomeage-oldproblemsison thehorizon, theremaybea tendency togrowdetached frommundane,present-dayconcerns. IshareMore's antipathy toward "passiveSingularitarianism,"One reason for aproactive stance is thattechnology is adouble-edged sword and as such alwayshas thepotential of going awry as it surgestoward theSingularity,withprofoundlydisturbingconsequences.Evensmalldelays in implementingemerging technologies can condemn millions of people to continued suffering and death. As oneexample of many, excessive regulatory delays in implementing lifesaving therapies end up costingmanylives.(Welosemillionsofpeopleperyeararoundtheworldfromheartdiseasealone.) More also worries about a cultural rebellion "seduced by religious and cultural urgings for'stability''peace'andagainst'hubris'and'theunknown'"thatmayderailtechnologicalacceleration.7Inmyviewanysignificantderailmentoftheoveralladvancementoftechnologyisunlikely.Evenepochaleventssuchastwoworldwars(inwhichontheorderofonehundredmillionpeopledied),thecoldwar,andnumerouseconomic,cultural,andsocialupheavalshavefailedtomaketheslightestdentinthepaceoftechnologytrends.Butthereflexive,thoughtlessantitechnologysentimentsincreasinglybeingvoicedintheworldtodaydohavethepotentialtoexacerbatealotofsuffering.StillHuman? Some observers refer to the post-Singularity period as "posthuman" and refer to theanticipation of this period as posthumanism. However, to me being human means being part of acivilizationthatseekstoextenditsboundaries.Wearealreadyreachingbeyondourbiologybyrapidlygainingthetoolstoreprogramandaugmentit.Ifweregardahumanmodifiedwithtechnologyasnolonger human,wherewouldwedraw thedefining line? Is a humanwith a bionic heart still human?Howaboutsomeonewithaneurological implant?Whatabouttwoneurological implants?Howaboutsomeone with ten nanobots in his brain? How about 500 million nanobots? Should we establish aboundaryat650millionnanobots:underthat,you'restillhumanandoverthat,you'reposthuman? Ourmergerwithour technologyhasaspectsofaslipperyslope,butone that slidesup towardgreaterpromise,notdownintoNietzsche'sabyss.Someobserversrefertothismergerascreatinganew"species."Butthewholeideaofaspeciesisabiologicalconcept,andwhatwearedoingistranscendingbiology. The transformation underlying the Singularity is not just another in a long line of steps inbiologicalevolution.Weareupendingbiologicalevolutionaltogether.BillGates: I agree with you 99 percent.What I like about your ideas is that they are grounded in

science,butyouroptimismisalmostareligiousfaith.I'moptimisticalso.RAY:Yes,well,weneedanewreligion.Aprincipalroleofreligionhasbeentorationalizedeath,since

upuntiljustnowtherewaslittleelseconstructivewecoulddoaboutit.BILL:Whatwouldtheprinciplesofthenewreligionbe?RAY:We'dwant tokeeptwoprinciples:one fromtraditionalreligionandone fromsecularartsand

sciences—fromtraditionalreligion,therespectforhumanconsciousness.BILL:Ahyes,theGoldenRule.RAY:Right,ourmoralityand legalsystemarebasedonrespect for theconsciousnessofothers. If I

hurt another person, that's considered immoral, and probably illegal, because I have causedsufferingtoanotherconsciousperson.IfIdestroyproperty,it'sgenerallyokayifit'smyproperty,andtheprimaryreasonit'simmoralandillegalifit'ssomeoneelse'spropertyisbecauseIhavecausedsufferingnottothepropertybuttothepersonowningit.

BILL:Andthesecularprinciple?RAY: From the arts and sciences, it is the importance of knowledge. Knowledge goes beyond

information. It's information that has meaning for conscious entities: music, art, literature,science,technology.ThesearethequalitiesthatwillexpandfromthetrendsI'mtalkingabout.

BILL: We need to get away from the ornate and strange stories in contemporary religions andconcentrateonsomesimplemessages.Weneedacharismaticleaderforthisnewreligion.

RAY:Acharismaticleaderispartoftheoldmodel.That'ssomethingwewanttogetawayfrom.

BILL:Okay,acharismaticcomputer,then.RAY:Howaboutacharismaticoperatingsystem?BILL:Ha,we'vealreadygotthat.SoisthereaGodinthisreligion?RAY: Not yet, but there will be. Once we saturate the matter and energy in the universe with

intelligence, itwill"wakeup,"beconscious,andsublimely intelligent.That'saboutasclose toGodasIcanimagine.

BILL:That'sgoingtobesiliconintelligence,notbiologicalintelligence.RAY:Well,yes,we'regoingtotranscendbiologicalintelligence.We'llmergewithitfirst,butultimately

thenonbiologicalportionofour intelligencewillpredominate.By theway, it'snot likely tobesilicon,butsomethinglikecarbonnanotubes.

BILL:Yes,Iunderstand—I'mjustreferringtothatassiliconintelligencesincepeopleunderstandwhatthatmeans.ButIdon'tthinkthat'sgoingtobeconsciousinthehumansense.

RAY:Whynot?Ifweemulateinasdetailedamannerasnecessaryeverythinggoingoninthehumanbrainandbodyandinstantiatetheseprocessesinanothersubstrate,andthenofcourseexpanditgreatly,whywouldn'titbeconscious?

BILL:Oh,itwillbeconscious.Ijustthinkitwillbeadifferenttypeofconsciousness.RAY:Maybethisisthe1percentwedisagreeon.Whywoulditbedifferent?BILL:Becausecomputers canmerge together instantly.Tencomputers—oronemillioncomputers—

canbecomeonefaster,biggercomputer.Ashumans,wecan'tdothat.Weeachhaveadistinctindividualitythatcannotbebridged.

RAY: That's just a limitation of biological intelligence. The unbridgeable distinctness of biologicalintelligenceisnotaplus."Silicon"intelligencecanhaveitbothways.Computersdon'thavetopool their intelligence and resources. They can remain "individuals" if they wish. Siliconintelligencecanevenhaveitbothwaysbymergingandretainingindividuality—atthesametime.Ashumans,wetrytomergewithothersalso,butourabilitytoaccomplishthisisfleeting.

BILL:Everythingofvalueisfleeting.RAY:Yes,butitgetsreplacedbysomethingofevengreatervalue.BILL:True,that'swhyweneedtokeepinnovating.TheVexingQuestionofConsciousness

Ifyoucouldblowthebrainuptothesizeofamillandwalkaboutinside,youwouldnotfindconsciousness.

—G.W.LEIBNIZCanoneeverrememberlove?It'sliketryingtosummonupthesmellofrosesinacellar.Youmightseearose,butnevertheperfume.

—ARTHURMILLER8At one's first and simplest attempts to philosophize, one becomes entangled in questions ofwhether when one knows something, one knows that one knows it, and what, when one isthinkingofoneself,isbeingthoughtabout,andwhatisdoingthethinking.Afteronehasbeenpuzzledandbruisedbythisproblemforalongtime,onelearnsnottopressthesequestions:theconceptofaconsciousbeingis,implicitly,realizedtobedifferentfromthatofanunconsciousobject.Insayingthataconsciousbeingknowssomething,wearesayingnotonlythatheknowsit,butthatheknowsthatheknowsit,andthatheknowsthatheknowsthatheknowsit,andsoon,aslongaswecaretoposethequestion:thereis,werecognize,aninfinityhere,butitisnotan infinite regress in the bad sense, for it is the questions that peter out, as being pointless,ratherthantheanswers.

—J.R.LUCAS,OXFORDPHILOSOPHER,INHIS1961ESSAY"MINDS,MACHINES,ANDGÖDEL"9

Dreamsarerealwhiletheylast;canwesaymoreoflife?

—HAVELOCKWill futuremachines be capable of having emotional and spiritual experiences?We have discussedseveralscenariosforanonbiologicalintelligencetodisplaythefullrangeofemotionallyrichbehaviorexhibited by biological humans today. By the late 2020s we will have completed the reverseengineeringof thehumanbrain,whichwillenableus tocreatenonbiologicalsystems thatmatchandexceedthecomplexityandsubtletyofhumans,includingouremotionalintelligence. Asecondscenario is thatwecouldupload thepatternsofanactualhuman intoasuitablenonbiological, thinking substrate. A third, and the most compelling, scenario involves the gradual butinexorableprogressionofhumansthemselvesfrombiologicaltononbiological.Thathasalreadystartedwiththebenignintroductionofdevicessuchasneuralimplantstoamelioratedisabilitiesanddisease.Itwillprogresswiththeintroductionofnanobotsinthebloodstream,whichwillbedevelopedinitiallyformedicalandantiagingapplications.Latermoresophisticatednanobotswillinterfacewithourbiologicalneuronstoaugmentoursenses,providevirtualandaugmentedrealityfromwithinthenervoussystem,assistourmemories,andprovideotherroutinecognitivetasks.Wewillthenbecyborgs,andfromthatfoothold in our brains, the nonbiological portion of our intelligence will expand its powersexponentially.AsIdiscussedinchapters2and3weseeongoingexponentialgrowthofeveryaspectofinformation technology, including price-performance, capacity, and rate of adoption. Given that themassandenergyrequiredtocomputeandcommunicateeachbitofinformationareextremelysmall(seechapter 3), these trends can continue until our nonbiological intelligence vastly exceeds that of thebiologicalportion.Sinceourbiologicalintelligenceisessentiallyfixedinitscapacity(exceptforsomerelatively modest optimization from biotechnology), the nonbiological portion will ultimatelypredominate.Inthe2040s,whenthenonbiologicalportionwillbebillionsoftimesmorecapable,willwestilllinkourconsciousnesstothebiologicalportionofourintelligence?Clearly,nonbiologicalentitieswillclaimtohaveemotionalandspiritualexperiences,justaswedotoday. They—we—will claim to be human and to have the full range of emotional and spiritualexperiencesthathumansclaimtohave.Andthesewillnotbeidleclaims;theywillevidencethesortofrich,complex,andsubtlebehaviorassociatedwithsuchfeelings. Buthowwill theseclaimsandbehaviors—compellingas theywillbe—relate to thesubjectiveexperience of nonbiological humans? We keep coming back to the very real but ultimatelyunmeasurable(byfullyobjectivemeans)issueofconsciousness.Peopleoftentalkaboutconsciousnessasifitwereaclearpropertyofanentitythatcanreadilybeidentified,detected,andgauged.Ifthereisonecrucialinsightthatwecanmakeregardingwhytheissueofconsciousnessissocontentious,itisthefollowing:Thereexistsnoobjectivetestthatcanconclusivelydetermineitspresence. Science isaboutobjectivemeasurementsand their logical implications,but theverynatureofobjectivity is that you cannotmeasure subjective experience—you can onlymeasure correlates of it,suchasbehavior(andbybehavior,Iincludeinternalbehavior—thatis,theactionsofthecomponentsofanentity,suchasneuronsandtheirmanyparts).Thislimitationhastodowiththeverynatureoftheconcepts of "objectivity" and "subjectivity." Fundamentally we cannot penetrate the subjectiveexperienceofanotherentitywithdirectobjectivemeasurement.Wecancertainlymakeargumentsaboutit,suchas,"Lookinsidethebrainofthisnonbiologicalentity;seehowitsmethodsarejustlikethoseofahumanbrain."Or,"Seehowitsbehavior is just likehumanbehavior."But in theend, theseremainjustarguments.Nomatterhowconvincingthebehaviorofanonbiologicalperson,someobserverswillrefuse to accept the consciousness of such an entity unless it squirts neurotransmitters, is based onDNA-guidedproteinsynthesis,orhassomeotherspecificbiologicallyhumanattribute.

Weassumethatotherhumansareconscious,buteventhatisanassumption.Thereisnoconsensusamong humans about the consciousness of nonhuman entities, such as higher animals. Consider thedebatesregardinganimalrights,whichhaveeverythingtodowithwhetheranimalsareconsciousorjustquasimachinesthatoperateby"instinct."Theissuewillbeevenmorecontentiouswithregardtofuturenonbiologicalentitiesthatexhibitbehaviorandintelligenceevenmorehumanlikethanthoseofanimals.Infactthesefuturemachineswillbeevenmorehumanlikethanhumanstoday.Ifthatseemslikeaparadoxicalstatement,considerthatmuchofhumanthoughttodayispettyandderivative.WemarvelatEinstein'sabilitytoconjureupthetheoryofgeneralrelativityfromathoughtexperimentorBeethoven'sabilitytoimaginesymphoniesthathecouldneverhear.Buttheseinstancesofhumanthoughtatitsbestare rare and fleeting; (Fortunately we have a record of these fleeting moments, reflecting a keycapabilitythathasseparatedhumansfromotheranimals.)Ourfutureprimarilynonbiologicalselveswillbe vastlymore intelligent and sowill exhibit these finer qualities of human thought to a far greaterdegree. So howwillwe come to termswith the consciousness thatwill be claimed by nonbiologicalintelligence?Fromapracticalperspectivesuchclaimswillbeaccepted.Foronething,"they"willbeus,sotherewon'tbeanycleardistinctionsbetweenbiologicalandnonbiologicalintelligence.Furthermore,thesenonbiological entitieswillbeextremely intelligent, so they'llbeable toconvinceotherhumans(biological, nonbiological, or somewhere in between) that they are conscious. They'll have all thedelicateemotionalcues thatconvinceus todaythathumansareconscious.Theywillbeable tomakeother humans laugh and cry. And they'll get mad if others don't accept their claims. But this isfundamentallyapoliticalandpsychologicalprediction,notaphilosophicalargument. Ido take issuewith thosewhomaintain thatsubjectiveexperienceeitherdoesn'texistor isaninessentialqualitythatcansafelybeignored.Theissueofwhoorwhatisconsciousandthenatureofthesubjectiveexperiencesofothersarefundamentaltoourconceptsofethics,morality,andlaw.Ourlegalsystemisbasedlargelyontheconceptofconsciousness,withparticularlyseriousattentionpaidtoactionsthatcausesuffering—anespeciallyacuteformofconsciousexperience—toa(conscious)humanorthatendtheconsciousexperienceofahuman(forexample,murder).Humanambivalenceregardingtheabilityofanimalstosufferisreflectedinlegislationaswell.Wehave laws against animal cruelty, with greater emphasis given to more intelligent animals, such asprimates(althoughweappeartohaveablindspotwithregardtothemassiveanimalsufferinginvolvedinfactoryfarming,butthat'sthesubjectofanothertreatise). My point is that we cannot safely dismiss the question of consciousness as merely a politephilosophicalconcern.Itisatthecoreofsociety'slegalandmoralfoundation.Thedebatewillchangewhen a machine—nonbiological intelligence—can persuasively argue on its own that it/he/she hasfeelings that need to be respected. Once it can do sowith a sense of humor—which is particularlyimportantforconvincingothersofone'shumanness—itislikelythatthedebatewillbewon. I expect that actual change in our legal systemwill come initially from litigation rather thanlegislation, as litigation often precipitates such transformations. In a precursor of what is to come,attorneyMartineRothblatt,apartner inMahon,Patusky,Rothblatt&Fisher, filedamockmotiononSeptember 16, 2003, to prevent a corporation fromdisconnecting a conscious computer.Themotionwas argued in a mock trial in the biocyberethics session at the International Bar Associationconference.10 We canmeasure certain correlates of subjective experience (for example, certain patterns ofobjectively measurable neurological activity with objectively verifiable reports of certain subjectiveexperiences, such as hearing a sound).Butwe cannot penetrate to the core of subjective experiencethrough objective measurement. As I mentioned in chapter 1, we are dealing with the differencebetweenthird-person"objective"experience,whichisthebasisofscience,andfirst-person"subjective"experience,whichisasynonymforconsciousness. Consider that we are unable to truly experience the subjective experiences of others. Theexperience-beaming technology of 2029 will enable the brain of one person to experience only thesensoryexperiences(andpotentiallysomeoftheneurologicalcorrelatesofemotionsandotheraspects

of experience) of another person.But thatwill still not convey the same internal experience as thatundergoneby thepersonbeaming theexperience,becausehisorherbrain isdifferent.Everydaywehearreportsabouttheexperiencesofothers,andwemayevenfeelempathyinresponsetothebehaviorthatresultsfromtheirinternalstates.Butbecausewe'reexposedtoonlythebehaviorofothers,wecanonly imagine their subjective experiences. Because it is possible to construct a perfectly consistent,scientificworldviewthatomitstheexistenceofconsciousness,someobserverscometotheconclusionthatit'sjustanillusion.JaronLanier,thevirtual-realitypioneer,takesissue(inthethirdofhissixobjectionstowhathecalls "cybernetic totalism" in his treatise "One Half a Manifesto") with those who maintain "thatsubjective experience either doesn't exist, or is unimportant because it is some sort of ambient orperipheral effect."11 As I pointed out, there is no device or system we can postulate that coulddefinitively detect subjectivity (conscious experience) associatedwith an entity.Any such purporteddevicewouldhavephilosophicalassumptionsbuilt into it.AlthoughIdisagreewithmuchofLanier'streatise(seethe"CriticismfromSoftware"sectioninchapter9),Iconcurwithhimonthisissueandcaneven imagine (andempathizewith!)his feelingsof frustrationat thedictumsof"cybernetic totalists"such as myself (not that I accept this characterization).12 Like Lanier I even accept the subjectiveexperienceofthosewhomaintainthatthereisnosuchthingassubjectiveexperience. Precisely because we cannot resolve issues of consciousness entirely through objectivemeasurement and analysis (science), a critical role exists for philosophy. Consciousness is themostimportantontologicalquestion.Afterall, ifwe truly imagineaworld inwhich there isno subjectiveexperience(aworldinwhichthereisswirlingstuffbutnoconsciousentitytoexperienceit),thatworldmay as well not exist. In some philosophical traditions, both Eastern (certain schools of Buddhistthought, for example), and Western (specifically, observer-based interpretations of quantummechanics),thatisexactlyhowsuchaworldisregarded.RAY:We can debate what sorts of entities are or can be conscious. We can argue about whether

consciousness is an emergent property or caused by some specific mechanism, biological orotherwise. But there's another mystery associated with consciousness, perhaps the mostimportantone.

MOLLY2004:Okay,I'mallears.RAY:Well, even if we assume that all humans who seem to be conscious in fact are, why is my

consciousnessassociatedwiththisparticularperson,me?WhyamIconsciousofthisparticularpersonwhoreadTomSwiftJr.booksasachild,gotinvolvedwithinventions,writesbooksaboutthe future, and so on? Every morning that I wake up, I have the experiences of this specificperson.Whywasn'tIAlanisMorissetteorsomeoneelse?

SIGMUNDFREUD:Hmm,soyou'dliketobeAlanisMorissette?RAY:That'saninterestingproposition,butthat'snotmypoint.MOLLY2004:Whatisyourpoint?Idon'tunderstand.RAY:WhyamIconsciousoftheexperiencesanddecisionsofthisparticularperson?MOLLY2004:Because,silly,that'swhoyouare.SIGMUND:Itseemsthatthere'ssomethingaboutyourselfthatyoudon'tlike.Tellmemoreaboutthat.MOLLY2004:Earlier,Raydidn'tlikebeinghumanaltogether.RAY:Ididn'tsayIdon'tlikebeinghuman.IsaidIdidn'tlikethelimitations,problems,andhighlevel

ofmaintenanceofmyversion1.0body.But this isallbeside thepoint that I'm trying tomakehere.

CHARLESDARWIN: Youwonderwhy you're you?That's a tautology, there's notmuch towonderabout.

RAY: Like many attempts to express the really "hard" problems of consciousness, this is soundingmeaningless.ButifyouaskmewhatIreallywonderabout,itisthis:whyamIcontinuallyawareofthisparticularperson'sexperiencesandfeelings?Asforotherpeople'sconsciousness,Iacceptit,butIdon'texperienceotherpeople'sexperiences,notdirectlyanyway.

SIGMUND:Okay,I'mgettingaclearerpicturenow.Youdon'texperienceotherpeople'sexperiences?Haveyouevertalkedtosomeoneaboutempathy?

RAY:Look,I'mtalkingaboutconsciousnessnowinaverypersonalway.SIGMUND:That'sgood,keepgoing.RAY:Actually, this isagoodexampleofwhat typicallyhappenswhenpeople try tohaveadialogue

aboutconsciousness.Thediscussioninevitablyveersoffintosomethingelse,likepsychologyorbehaviororintelligenceorneurology.ButthemysteryofwhyIamthisparticularpersoniswhatIreallywonderabout.

CHARLES:Youknowyoudocreatewhoyouare.RAY:Yes,that'strue.Justasourbrainscreateourthoughts,ourthoughtsinturncreateourbrains.CHARLES:Soyou'vemadeyourself,andthat'swhyyouarewhoyouare,sotospeak.MOLLY2104:Weexperiencethatverydirectlyin2104.Beingnonbiological,I'mabletochangewhoI

amquitereadily.Aswediscussedearlier,ifI'minthemood,Icancombinemythoughtpatternswithsomeoneelse'sandcreateamergedidentity.It'saprofoundexperience.

MOLLY2004:Well,MissMollyofthefuture,wedothatbackintheprimitivedaysof2004also.Wecallitfallinginlove.

WhoAmI?WhatAmI?

Whyareyouyou?

—THE IMPLIED QUESTION IN THE ACRONYM YRUU (YOUNGRELIGIOUSUNITARIANUNIVERSALISTS), ANORGANIZATION IWASACTIVEINWHENIWASGROWINGUPINTHEEARLY1960S(ITWASTHENCALLEDLRY,LIBERALRELIGIOUSYOUTH).

Whatyouarelookingforiswhoislooking.

—SAINTFRANCISOFASSISII'mnotawareoftoomanythingsIknowwhatIknowifyouknowwhatImean.Philosophyisthetalkonacerealbox.Religionisthesmileonadog....Philosophyisawalkontheslipperyrocks.Religionisalightinthefog....WhatIamiswhatIam.Areyouwhatyouareorwhat?

—EDIEBRICKELL,"WHATIAM"FreedomofwillistheabilitytodogladlythatwhichImustdo.

—CARLJUNGThechanceofthequantumtheoreticianisnottheethicalfreedomoftheAugustinian.

—NORBERTWEINER13Ishouldprefertoanordinarydeath,beingimmersedwithafewfriendsinacaskofMadeira,until that time, then toberecalled to lifeby thesolarwarmthofmydearcountry!But inall

probability,weliveinacenturytoolittleadvanced,andtooneartheinfancyofscience,toseesuchanartbroughtinourtimetoitsperfection.

—BENJAMINFRANKLIN,1773Arelatedbutdistinctquestionhastodowithourownidentities.Wetalkedearlieraboutthepotentialtoupload the patterns of an individual mind—knowledge, skills, personality, memories—to anothersubstrate.Althoughthenewentitywouldactjustlikeme,thequestionremains:isitreallyme?Someofthescenariosforradicallifeextensioninvolvereengineeringandrebuildingthesystemsandsubsystemsthatourbodiesandbrainscomprise.Intakingpartinthisreconstruction,doIlosemyselfalongtheway?Again,thisissuewilltransformitselffromacenturies-oldphilosophicaldialoguetoapressingpracticalmatterinthenextseveraldecades.SowhoamI?SinceIamconstantlychanging,amIjustapattern?Whatifsomeonecopiesthatpattern?AmItheoriginaland/orthecopy?PerhapsIamthisstuffhere—thatis,thebothorderedandchaoticcollectionofmoleculesthatmakeupmybodyandbrain. But there's a problemwith this position.The specific set of particles thatmybody andbraincompriseare infactcompletelydifferentfromtheatomsandmolecules thatIcomprisedonlyashortwhileago.Weknowthatmostofourcellsareturnedoverinamatterofweeks,andevenourneurons,which persist as distinct cells for a relatively long time, nonetheless change all of their constituentmolecules within a month.14 The half-life of a microtubule (a protein filament that provides thestructureofaneuron) isabout tenminutes.Theactin filaments indendritesare replacedabouteveryfortyseconds.Theproteinsthatpowerthesynapsesarereplacedabouteveryhour.NMDAreceptorsinsynapsesstickaroundforarelativelylongfivedays. So Iamacompletelydifferent setof stuff than Iwasamonthago,andall thatpersists is thepattern of organization of that stuff. The pattern changes also, but slowly and in a continuum. I amrather like thepattern thatwatermakes ina streamas it rushespast the rocks in itspath.Theactualmoleculesofwaterchangeeverymillisecond,butthepatternpersistsforhoursorevenyears.Perhaps,therefore,weshouldsayIamapatternofmatterandenergythatpersistsovertime.Butthereisaproblemwiththisdefinition,aswell,sincewewillultimatelybeabletouploadthispatterntoreplicatemybodyandbraintoasufficientlyhighdegreeofaccuracythatthecopyisindistinguishablefromtheoriginal.(Thatis,thecopycouldpassa"RayKurzweil"Turingtest.)Thecopy,therefore,willsharemypattern.Onemightcounterthatwemaynotgeteverydetailcorrect,butastimegoesonourattempts to create a neural and body replica will increase in resolution and accuracy at the sameexponentialpacethatgovernsallinformation-basedtechnologies.Wewillultimatelybeabletocaptureandre-createmypatternofsalientneuralandphysicaldetailstoanydesireddegreeofaccuracy.Althoughthecopysharesmypattern,itwouldbehardtosaythatthecopyismebecauseIwould—orcould—stillbehere.YoucouldevenscanandcopymewhileIwassleeping.Ifyoucometomeinthe morning and say, "Good news, Ray, we've successfully reinstantiated you into a more durablesubstrate,sowewon'tbeneedingyouroldbodyandbrainanymore,"Imaybegtodiffer. Ifyoudothethoughtexperiment,it'sclearthatthecopymaylookandactjustlikeme,butit'snonethelessnotme.Imaynotevenknowthathewascreated.Althoughhewouldhaveallmymemoriesand recallhavingbeenme, from thepoint in timeofhis creationRay2wouldhavehisownuniqueexperiences,andhisrealitywouldbegintodivergefrommine.Thisisarealissuewithregardtocryonics(theprocessofpreservingbyfreezingapersonwhohasjustdied,withaviewtoward"reanimating"himlaterwhenthetechnologyexiststoreversethedamagefrom the early stages of the dying process, the cryonic-preservation process, and the disease orcondition thatkilledhim in the firstplace).Assuminga "preserved"person isultimately reanimated,manyoftheproposedmethodsimplythatthereanimatedpersonwillessentiallybe"rebuilt"withnewmaterials and even entirely new neuromorphically equivalent systems. The reanimated person will,therefore,effectivelybe"Ray2"(thatis,someoneelse).Nowlet'spursuethistrainofthoughtabitfurther,andyouwillseewherethedilemmaarises.Ifwecopymeandthendestroytheoriginal,that'stheendofme,becauseasweconcludedabovethecopy

is not me. Since the copy will do a convincing job of impersonating me, no one may know thedifference,butit'snonethelesstheendofme.Considerreplacingatinyportionofmybrainwithitsneuromorphicequivalent.Okay,I'mstillhere:theoperationwassuccessful(incidentally,nanobotswilleventuallydothiswithoutsurgery).Weknowpeoplelikethisalready,suchasthosewithcochlearimplants,implantsforParkinson'sdisease, andothers.Now replaceanotherportionofmybrain:okay, I'm still here ... andagain....Attheendoftheprocess,I'mstillmyself.Thereneverwasan"oldRay"anda"newRay,"I'mthesameasIwasbefore.Nooneevermissedme,includingme.ThegradualreplacementofRayresultsinRay,soconsciousnessandidentityappeartohavebeenpreserved.However,inthecaseofgradualreplacementthereisnosimultaneousoldmeandnewme.Attheendoftheprocessyouhavetheequivalentofthenewme(thatis,Ray2)andnooldme(Ray1).Sogradualreplacementalsomeanstheendofme.Wemightthereforewonder:atwhatpointdidmybodyandbrainbecomesomeoneelse? On yet another hand (we're running out of philosophical hands here), as I pointed out at thebeginning of this question, I am in fact being continually replaced as part of a normal biologicalprocess.(And,bytheway,thatprocessisnotparticularlygradualbutratherrapid.)Asweconcluded,all that persists ismy spatial and temporal patternofmatter and energy.But the thought experimentaboveshows thatgradual replacementmeans theendofmeeven ifmypattern ispreserved.SoamIconstantlybeingreplacedbysomeoneelsewhojustseemsalotlikethemeofafewmomentsearlier?So,again,whoamI?It'stheultimateontologicalquestion,andweoftenrefertoitastheissueofconsciousness.Ihaveconsciously(punintended)phrasedtheissueentirelyinthefirstpersonbecausethatisitsnature.Itisnotathird-personquestion.Somyquestionisnot"whoareyou?"althoughyoumaywishtoaskthisquestionyourself. When people speak of consciousness they often slip into considerations of behavioral andneurologicalcorrelatesofconsciousness(forexample,whetherornotanentitycanbeself-reflective).Butthesearethird-person(objective)issuesanddonotrepresentwhatDavidChalmerscallsthe"hardquestion"ofconsciousness:howcanmatter(thebrain) leadtosomethingasapparently immaterialasconsciousness?15 Thequestionofwhetherornotanentity isconscious isapparentonly to itself.Thedifferencebetween neurological correlates of consciousness (such as intelligent behavior) and the ontologicalrealityofconsciousnessisthedifferencebetweenobjectiveandsubjectivereality.That'swhywecan'tproposeanobjectiveconsciousnessdetectorwithoutphilosophicalassumptionsbuiltintoit.Idobelievethatwehumanswillcometoacceptthatnonbiologicalentitiesareconscious,becauseultimatelythenonbiologicalentitieswillhaveallthesubtlecuesthathumanscurrentlypossessandthatweassociatewithemotionalandothersubjectiveexperiences.Still,whilewewillbeabletoverifythesubtlecues,wewillhavenodirectaccesstotheimpliedconsciousness.Iwillacknowledgethatmanyofyoudoseemconscioustome,butIshouldnotbetooquicktoacceptthisimpression.PerhapsIamreallylivinginasimulation,andyouareallpartofit.Or,perhapsit'sonlymymemoriesofyouthatexist,andtheseactualexperiencesnevertookplace.OrmaybeIamonlynowexperiencingthesensationofrecallingapparentmemories,butneithertheexperiencenorthememoriesreallyexist.Well,youseetheproblem.Despitethesedilemmasmypersonalphilosophyremainsbasedonpatternism—Iamprincipallyapatternthatpersistsintime.Iamanevolvingpattern,andIcaninfluencethecourseoftheevolutionofmypattern.Knowledgeisapattern,asdistinguishedfrommereinformation,andlosingknowledgeisaprofoundloss.Thus,losingapersonistheultimateloss.MOLLY2004:As faras I'mconcerned,who Iam ispretty straightforward—it'sbasically thisbrain

andbody,whichatleastthismonthisinprettygoodshape,thankyou.RAY:Areyouincludingthefoodinyourdigestivetract, initsvariousstagesofdecompositionalong

theway?MOLLY2004:Okay,youcanexcludethat.Someofitwillbecomeme,butithasn'tbeenenrolledyetin

the"partofMolly"club.RAY:Well,90percentofthecellsinyourbodydon'thaveyourDNA.MOLLY2004:Isthatso?JustwhoseDNAisit,then?RAY:Biological humans have about ten trillion cellswith their ownDNA, but there are about one

hundredtrillionmicroorganismsinthedigestivetract,basicallybacteria.MOLLY2004:Doesn'tsoundveryappealing.Aretheyentirelynecessary?RAY:They're actually part of the society of cells thatmakesMolly alive and thriving. You couldn't

survivewithouthealthygutbacteria.Assumingyourintestinalfloraareingoodbalance,they'renecessaryforyourwell-being.

MOLLY2004:Okay, but I wouldn't count them asme. There are lots of things thatmywell-beingdependson.Likemyhouseandmycar,butIstilldon'tcountthemaspartofme.

RAY:Verywell,it'sreasonabletoleaveouttheentirecontentsoftheGItract,bacteriaandall.That'sactuallyhowthebodyseesit.Eventhoughit'sphysicallyinsidethebody,thebodyconsidersthetracttobeexternalandcarefullyscreenswhatitabsorbsintothebloodstream.

MOLLY2004:AsIthinkmoreaboutwhoIam,IkindoflikeJaronLanier's"circleofempathy."RAY:Tellmemore.MOLLY2004:Basically,thecircleofrealitythatIconsidertobe"me"isnotclear-cut.It'snotsimply

mybody.Ihavelimitedidentificationwith,say,mytoesand,afterourlastdiscussion,evenlesswiththecontentsofmylargeintestine.

RAY:That's reasonable,andevenwith regard toourbrainsweareawareofonlya tinyportionofwhatgoesoninthere.

MOLLY2004: It's true that there are parts ofmy brain that seem to be somebody else, or at leastsomewhereelse.Often, thoughtsanddreams that intrudeonmyawarenessseemtohavecomefromsomeforeignplace.They'reobviouslycomingfrommybrain,butitdoesn'tseemthatway.

RAY:Conversely, loved oneswhoare physically separatemay be so close as to seem to be part ofourselves.

MOLLY2004:Theboundaryofmyselfisseeminglessandlessclear.RAY:Well,justwaituntilwe'repredominantlynonbiological.Thenwe'llbeabletomergeourthoughts

andthinkingatwill,sofindingboundarieswillbeevenmoredifficult.MOLLY 2004: That actually sounds kind of appealing. You know, some Buddhist philosophies

emphasizetheextenttowhichthereisinherentlynoboundaryatallbetweenus.RAY:Soundslikethey'retalkingabouttheSingularity.TheSingularityasTranscendence

Modernityseeshumanityashavingascendedfromwhatis inferiortoit—lifebeginsinslimeandendsinintelligence—whereastraditionalculturesseeitasdescendedfromitssuperiors.AstheanthropologistMarshallSahlinsputsthematter:"Wearetheonlypeoplewhoassumethatwehaveascendedfromapes.Everybodyelsetakesitforgrantedthattheyaredescendedfromgods."

—HUSTONSMITH16

Somephilosophersholdthatphilosophyiswhatyoudotoaproblemuntilit'sclearenoughtosolve itbydoingscience.Othershold that ifaphilosophicalproblemsuccumbs toempiricalmethods,thatshowsitwasn'treallyphilosophicaltobeginwith.

—JERRYA.FODOR17TheSingularitydenotesaneventthatwill takeplaceinthematerialworld,theinevitablenextstepin

theevolutionaryprocessthatstartedwithbiologicalevolutionandhasextendedthroughhuman-directedtechnologicalevolution.However, it ispreciselyintheworldofmatterandenergythatweencountertranscendence,aprincipalconnotationofwhatpeoplerefertoasspirituality.Let'sconsiderthenatureofspiritualityinthephysicalworld. Where shall I start?How aboutwithwater? It's simple enough, but consider the diverse andbeautifulways itmanifests itself: theendlesslyvaryingpatternsas itcascadespast rocks inastream,then surges chaotically down a waterfall (all viewable from my office window, incidentally); thebillowingpatternsofcloudsinthesky;thearrangementofsnowonamountain;thesatisfyingdesignofasinglesnowflake.OrconsiderEinstein'sdescriptionoftheentangledorderanddisorderinaglassofwater(thatis,histhesisonBrownianmotion).Orelsewhereinthebiologicalworld,considertheintricatedanceofspiralsofDNAduringmitosis.Howaboutthelovelinessofatreeasitbendsinthewindanditsleaveschurninatangleddance?Orthebustlingworldweseeinamicroscope?There'stranscendenceeverywhere.Acommentontheword"transcendence"isinorderhere."Totranscend"means"togobeyond,"but thisneednotcompelus toadoptanornatedualistviewthatregards transcendent levelsofreality(suchasthespirituallevel)tobenotofthisworld.Wecan"gobeyond"the"ordinary"powersofthematerialworldthroughthepowerofpatterns.AlthoughIhavebeencalledamaterialist,Iregardmyselfasa"patternist,"It'sthroughtheemergentpowersofthepatternthatwetranscend.Sincethematerialstuffofwhichwearemadeturnsoverquickly,itisthetranscendentpowerofourpatternsthatpersists.Thepowerofpatternstoenduregoesbeyondexplicitlyself-replicatingsystems,suchasorganismsand self-replicating technology. It is the persistence and power of patterns that support life andintelligence.Thepatternisfarmoreimportantthanthematerialstuffthatconstitutesit.Randomstrokesonacanvasarejustpaint.Butwhenarrangedinjusttherightway,theytranscendthematerialstuffandbecomeart.Randomnotesarejustsounds.Sequencedinan"inspired"way,wehavemusic.Apileofcomponentsisjustaninventory.Orderedinaninnovativemanner,andperhapswith the addition of some software (another pattern), we have the "magic" (transcendence) oftechnology. Although some regardwhat is referred to as "spiritual" as the truemeaningof transcendence,transcendencerefers toall levelsofreality: thecreationsof thenaturalworld, includingourselves,aswellasourowncreationsintheformofart,culture,technology,andemotionalandspiritualexpression.Evolution concerns patterns, and it is specifically the depth and order of patterns that grow in anevolutionaryprocess.Asaconsummationoftheevolutioninourmidst,theSingularitywilldeepenallofthesemanifestationsoftranscendence.Anotherconnotationoftheword"spiritual"is"containingspirit,"whichistosaybeingconscious.Consciousness—the seat of "personalness"—is regarded as what is real in many philosophical andreligioustraditions.AcommonBuddhistontologyconsiderssubjective—conscious—experienceastheultimatereality,ratherthanphysicalorobjectivephenomena,whichareconsideredmaya(illusion).TheargumentsImakeinthisbookwithregardtoconsciousnessareforthepurposeofillustratingthis vexing and paradoxical (and, therefore, profound) nature of consciousness: how one set ofassumptions(thatis,thatacopyofmymindfileeithersharesordoesnotsharemyconsciousness)leadsultimatelytoanoppositeview,andviceversa.Wedoassumethathumansareconscious,atleastwhentheyappeartobe.Attheotherendofthespectrumweassumethatsimplemachinesarenot.Inthecosmologicalsensethecontemporaryuniverseactsmorelikeasimplemachinethanaconsciousbeing.Butaswediscussedinthepreviouschapter,thematter and energy in our vicinity will become infused with the intelligence, knowledge, creativity,beauty,andemotionalintelligence(theabilitytolove,forexample)ofourhuman-machinecivilization.Ourcivilizationwill thenexpandoutward, turningall thedumbmatterandenergyweencounter intosublimelyintelligent—transcendent—matterandenergy.Soinasense,wecansaythattheSingularitywillultimatelyinfusetheuniversewithspirit. Evolution moves toward greater complexity, greater elegance, greater knowledge, greaterintelligence, greater beauty, greater creativity, and greater levels of subtle attributes such as love. Inevery monotheistic tradition God is likewise described as all of these qualities, only without any

limitation:infiniteknowledge,infiniteintelligence,infinitebeauty,infinitecreativity,infinitelove,andsoon.Ofcourse,even theacceleratinggrowthofevolutionneverachievesan infinite level,butas itexplodes exponentially it certainly moves rapidly in that direction. So evolution moves inexorablytowardthisconceptionofGod,althoughneverquitereachingthisideal.Wecanregard,therefore,thefreeingofour thinkingfromthesevere limitationsof itsbiological formtobeanessentiallyspiritualundertaking.MOLLY2004:So,doyoubelieveinGod?RAY:Well,it'sathree-letterword—andapowerfulmeme.MOLLY2004:Irealizethewordandtheideaexist.Butdoesitrefertoanythingthatyoubelievein?RAY:Peoplemeanlotsofthingsbyit.MOLLY2004:Doyoubelieveinthosethings?RAY:It'snotpossibletobelieveallthesethings:Godisanall-powerfulconsciouspersonlookingover

us,makingdeals,andgettingangryquiteabit.OrHe—It—isapervasivelifeforceunderlyingallbeautyandcreativity.OrGodcreatedeverythingandthensteppedback....

MOLLY2004:Iunderstand,butdoyoubelieveinanyofthem?RAY:Ibelievethattheuniverseexists.MOLLY2004:Nowwaitaminute,that'snotabelief,that'sascientificfact.RAY:Actually,Idon'tknowforsure'thatanythingexistsotherthanmyownthoughts.MOLLY2004:Okay, I understand that this is the philosophy chapter, but you can read scientific

papers—thousandsof them—that corroborate the existenceof starsandgalaxies. So,all thosegalaxies—wecallthattheuniverse.

RAY:Yes,I'veheardofthat,andIdorecallreadingsomeofthesepapers,butIdon'tknowthatthosepapersreallyexist,orthatthethingstheyrefertoreallyexist,otherthaninmythoughts.

MOLLY2004:Soyoudon'tacknowledgetheexistenceoftheuniverse?RAY:No, I just said that I do believe that it exists, but I'm pointing out that it's a belief That'smy

personalleapoffaith.MOLLY2004:Allright,butIaskedwhetheryoubelievedinGod.RAY:Again,"God"isawordbywhichpeoplemeandifferentthings.Forthesakeofyourquestion,we

canconsiderGodtobetheuniverse,andIsaidthatIbelieveintheexistenceoftheuniverse.MOLLY2004:Godisjusttheuniverse?RAY:Just?It'saprettybigthingtoapplytheword"just"to.Ifwearetobelievewhatsciencetellsus—

andIsaidthatIdo—it'saboutasbigaphenomenonaswecouldimagine.MOLLY2004:Actually,manyphysicistsnowconsiderouruniversetobejustonebubbleamongavast

numberofotheruniverses.But Imeant thatpeopleusuallymeansomethingmoreby theword"God"than"just"thematerialworld.SomepeopledoassociateGodwitheverythingthatexists,buttheystillconsiderGodtobeconscious.SoyoubelieveinaGodthat'snotconscious?

RAY:Theuniverseisnotconscious—yet.Butitwillbe.Strictlyspeaking,weshouldsaythatverylittleof it is conscious today.But thatwill changeand soon. I expect that theuniversewill becomesublimelyintelligentandwillwakeupinEpochSix.TheonlybeliefIampositinghereisthattheuniverseexists.Ifwemakethatleapoffaith,theexpectationthatitwillwakeupisnotsomuchabeliefasaninformedunderstanding,basedonthesamesciencethatsaysthereisauniverse.

MOLLY 2004: Interesting. You know, that's essentially the opposite of the view that there was aconsciouscreatorwhogoteverythingstartedandthenkindofbowedout.You'rebasicallysayingthataconsciousuniversewill"bowin"duringEpochSix.

RAY:Yes,that'stheessenceofEpochSix.

CHAPTEREIGHT

TheDeeplyIntertwinedPromiseandPerilofGNR

Wearebeingpropelledintothisnewcenturywithnoplan,nocontrol,nobrakes....Theonly realistic alternative I see is relinquishment: to limit development of thetechnologies that are too dangerous, by limiting our pursuit of certain kinds ofknowledge.

—BILLJOY,"WHYTHEFUTUREDOESN'TNEEDUS"Environmentalistsmustnowgrapplesquarelywiththeideaofaworldthathasenoughwealthandenoughtechnologicalcapability,andshouldnotpursuemore.

—BILL MCKIBBEN, ENVIRONMENTALIST WHO FIRSTWROTEABOUTGLOBALWARMING1

Progressmighthavebeenallrightonce,butithasgoneontoolong.

—OGDENNASH(1902–1971)In the late 1960s Iwas transformed into a radical environmental activist.A rag-taggroupofactivistsandIsailedaleakyoldhalibutboatacrosstheNorthPacifictoblockthe last hydrogen bomb tests under President Nixon. In the process I co-foundedGreenpeace....Environmentalists were often able to produce arguments that soundedreasonable,whiledoinggooddeedslikesavingwhalesandmakingtheairandwatercleaner. But now the chickens have come home to roost. The environmentalists'campaignagainstbiotechnologyingeneral,andgeneticengineeringinparticular,hasclearlyexposedtheirintellectualandmoralbankruptcy.Byadoptingazerotolerancepolicy toward a technologywith somany potential benefits for humankind and theenvironment, they ... have alienated themselves from scientists, intellectuals, andinternationalists.Itseemsinevitablethatthemediaandthepublicwill,intime,seetheinsanityoftheirposition.

—PATRICKMOOREIthinkthat...flightfromandhatredoftechnologyisself-defeating.TheBuddharestsquite as comfortably in the circuits of a digital computer and the gears of a cycletransmissionashedoesatthetopofamountainorinthepetalsofaflower.TothinkotherwiseistodemeantheBuddha—whichistodemeanoneself.

—ROBERTM. PIRSIG,ZENAND THE ARTOFMOTORCYCLEMAINTENANCE

onsiderthesearticleswe'drathernotseeavailableontheWeb:

CImpress Your Enemies: How to Build Your Own Atomic Bomb from Readily AvailableMaterials2HowtoModifytheInfluenzaVirusinYourCollegeLaboratorytoReleaseSnakeVenomTenEasyModificationstotheE.coliVirusHowtoModifySmallpoxtoCounteracttheSmallpoxVaccineBuildYourOwnChemicalWeaponsfromMaterialsAvailableontheInternetHowtoBuildaPilotless,Self-Guiding,Low-FlyingAirplaneUsingaLow-CostAircraft,GPS,andaNotebookComputer

Or,howaboutthefollowing:

TheGenomesofTenLeadingPathogens

TheFloorPlansofLeadingSkyscrapers

TheLayoutofU.S.NuclearReactors

TheHundredTopVulnerabilitiesofModernSociety

TheTopTenVulnerabilitiesoftheInternetPersonalHealthInformationonOneHundredMillionAmericans

TheCustomerListsofTopPornographySites

AnyonepostingthefirstitemaboveisalmostcertaintogetaquickvisitfromtheFBI,asdidNateCiccolo,afifteen-year-oldhighschoolstudent,inMarch2000.Foraschoolscienceprojecthebuiltapapier-mâchémodelofanatomicbombthatturnedouttobedisturbinglyaccurate.IntheensuingmediastormCiccolotoldABCNews,"Someonejustsortofmentioned,youknow,youcangoontheInternetnowandgetinformation.AndI,sortof,wasn'texactlyuptodateonthings.Tryit.IwentonthereandacoupleofclicksandIwasrightthere.3OfcourseCiccolodidn'tpossessthekeyingredient,plutonium,nordidhehaveanyintentionofacquiringit,butthereportcreatedshockwavesinthemedia,nottomentionamongtheauthoritieswhoworryaboutnuclearproliferation.Ciccolohadreportedfinding563Webpagesonatomic-bombdesigns,andthepublicityresultedinanurgentefforttoremovethem.Unfortunately,tryingtogetridofinformationontheInternetisakintotryingtosweepbacktheoceanwithabroom.Someofthesitescontinuetobeeasilyaccessibletoday.Iwon'tprovideanyURLsinthisbook,buttheyarenothardtofind.Althoughthearticletitlesabovearefictitious,onecanfindextensiveinformationontheInternetaboutallofthesetopics.4TheWebisanextraordinaryresearchtool.Inmyownexperience,researchthatusedtorequireahalfdayatthelibrarycannowbeaccomplishedtypicallyinacoupleofminutesorless. Thishasenormousandobviousbenefits foradvancingbeneficial technologies,but it canalsoempower those whose values are inimical to the mainstream of society. So are we in danger? Theanswerisclearlyyes.Howmuchdanger,andwhattodoaboutit,arethesubjectsofthischapter.Myurgentconcernwiththisissuedatesbackatleastacoupleofdecades.WhenIwroteTheAgeofIntelligentMachinesinthemid-1980s,Iwasdeeplyconcernedwiththeabilityofthen-emerginggenetic

engineering toenable those skilled in theart andwithaccess to fairlywidelyavailableequipment tomodifybacterialandviralpathogens tocreatenewdiseases.5 Indestructiveormerely carelesshandstheseengineeredpathogenscouldpotentiallycombineahighdegreeofcommunicability,stealthiness,anddestructiveness.Sucheffortswerenoteasytocarryoutinthe1980sbutwerenonethelessfeasible.Wenowknowthatbioweaponsprograms in theSovietUnionandelsewhereweredoingexactly this.6At the time Imadeaconsciousdecisiontonottalkaboutthisspecterinmybook,feelingthatIdidnotwanttogivethewrong people any destructive ideas. I didn'twant to turn on the radio one day and hear about adisaster,withtheperpetratorssayingthattheygottheideafromRayKurzweil.PartlyasaresultofthisdecisionIfacedsomereasonablecriticismthatthebookemphasizedthebenefitsoffuturetechnologywhileignoringitspitfalls.WhenIwroteTheAgeofSpiritualMachinesin1997–1998, therefore, I attempted to account for both promise and peril.7 There had been sufficientpublic attention by that time (for example, the 1995movieOutbreak, which portrays the terror andpanic from the release of a newviral pathogen) that I felt comfortable to begin to address the issuepublicly. InSeptember1998,havingjustcompletedthemanuscript,IranintoBillJoy,anesteemedandlongtimecolleagueinthehigh-technologyworld,inabarinLakeTahoe.AlthoughIhadlongadmiredJoy forhiswork inpioneering the leadingsoftware language for interactiveWebsystems (Java)andhavingcofoundedSunMicrosystems,myfocusatthisbriefget-togetherwasnotonJoybutratheronthe third person sitting in our small booth, John Searle. Searle, the eminent philosopher from theUniversity of California at Berkeley, had built a career of defending the deep mysteries of humanconsciousnessfromapparentattackbymaterialistssuchasRayKurzweil(acharacterizationIrejectinthenextchapter).SearleandIhadjustfinisheddebatingtheissueofwhetheramachinecouldbeconsciousduringtheclosingsessionofGeorgeGilder'sTelecosmconference.Thesessionwasentitled"SpiritualMachines"andwasdevotedtoadiscussionofthephilosophicalimplicationsofmyupcomingbook.IhadgivenJoyapreliminarymanuscriptandtriedtobringhimuptospeedonthedebateaboutconsciousnessthatSearleandIwerehaving. As it turnedout, Joywas interested inacompletelydifferent issue, specifically the impendingdangerstohumancivilizationfromthreeemergingtechnologiesIhadpresentedinthebook:genetics,nanotechnology, and robotics (GNR, as discussed earlier).Mydiscussion of the downsides of futuretechnologyalarmedJoy,ashewould later relate inhisnow-famouscoverstory forWired, "Why theFutureDoesn'tNeedUs."8 In the article Joydescribeshowheaskedhis friends in the scientific andtechnology community whether the projections I was making were credible and was dismayed todiscoverhowclosethesecapabilitiesweretorealization.Joy'sarticlefocusedentirelyonthedownsidescenariosandcreatedafirestorm.Herewasoneofthe technology world's leading figures addressing new and dire emerging dangers from futuretechnology. It was reminiscent of the attention that George Soros, the currency arbitrageur andarchcapitalist, receivedwhen hemade vaguely critical comments about the excesses of unrestrainedcapitalism,althoughtheJoycontroversybecamefarmoreintense.TheNewYorkTimesreportedtherewereabouttenthousandarticlescommentingonanddiscussingJoy'sarticle,morethananyotherinthehistoryofcommentaryontechnologyissues.MyattempttorelaxinaLakeTahoeloungethusendedupfosteringtwolong-termdebates,asmydialoguewithJohnSearlehasalsocontinuedtothisday. Despitemy being the origin of Joy's concern,my reputation as a "technology optimist" hasremainedintact,andJoyandIhavebeeninvitedtoavarietyofforumstodebatetheperilandpromise,respectively,offuturetechnologies.AlthoughIamexpectedtotakeupthe"promise"sideofthedebate,Ioftenendupspendingmostofmytimedefendinghispositiononthefeasibilityofthesedangers. Manypeoplehave interpreted Joy's article as anadvocacyofbroad relinquishment,notof alltechnologicaldevelopments,butofthe"dangerousones"likenanotechnology.Joy,whoisnowworkingas a venture capitalistwith the legendarySiliconValley firmofKleiner,Perkins,Caufield&Byers,investing in technologies such as nanotechnology applied to renewable energy and other natural

resources,saysthatbroadrelinquishmentisamisinterpretationofhispositionandwasneverhisintent.In a recent private e-mail communication, he says the emphasis should be on his call to "limitdevelopment of the technologies that are too dangerous" (see the epigraph at the beginning of thischapter),notoncompleteprohibition.Hesuggests, forexample,aprohibitionagainst self-replicatingnanotechnology, which is similar to the guidelines advocated by the Foresight Institute, founded bynanotechnology pioneer EricDrexler and Christine Peterson.Overall, this is a reasonable guideline,althoughIbelievetherewillneedtobetwoexceptions,whichIdiscussbelow(seep.411).Asanotherexample,JoyadvocatesnotpublishingthegenesequencesofpathogensontheInternet,whichIalsoagreewith.Hewouldliketoseescientistsadoptregulationsalongtheselinesvoluntarilyand internationally, andhepoints out that "ifwewait until after a catastrophe,wemayendupwithmoresevereanddamagingregulations."Hesayshehopesthat"wewilldosuchregulationlightly,sothatwecangetmostofthebenefits." Others, suchasBillMcKibben, theenvironmentalistwhowasoneof the first towarnagainstglobal warming, have advocated relinquishment of broad areas such as biotechnology andnanotechnology, or even of all technology. As I discuss in greater detail below (see p. 410),relinquishingbroadfieldswouldbeimpossibletoachievewithoutessentiallyrelinquishingalltechnicaldevelopment.ThatinturnwouldrequireaBraveNewWorldstyleoftotalitariangovernment,banningalltechnologydevelopment.Notonlywouldsuchasolutionbeinconsistentwithourdemocraticvalues,butitwouldactuallymakethedangersworsebydrivingthetechnologyunderground,whereonlytheleastresponsiblepractitioners(forexample,roguestates)wouldhavemostoftheexpertise.IntertwinedBenefits...

Itwasthebestoftimes,itwastheworstoftimes,itwastheageofwisdom,itwastheageoffoolishness, itwas the epoch of belief, itwas the epoch of incredulity, itwas the season ofLight,itwastheseasonofDarkness,itwasthespringofhope,itwasthewinterofdespair,wehadeverythingbeforeus,wehadnothingbeforeus,wewereallgoingdirect toHeaven,wewereallgoingdirecttheotherway.

—CHARLESDICKENS,ATALEOFTWOCITIESIt'slikearguinginfavoroftheplough.Youknowsomepeoplearegoingtoargueagainstit,butyoualsoknowit'sgoingtoexist.

—JAMES HUGHES, SECRETARY OF THE TRANSHUMANISTASSOCIATION AND SOCIOLOGIST AT TRINITY COLLEGE, IN ADEBATE, "SHOULD HUMANS WELCOME OR RESIST BECOMINGPOSTHUMAN?"

Technology has always been a mixed blessing, bringing us benefits such as longer and healthierlifespans,freedomfromphysicalandmentaldrudgery,andmanynovelcreativepossibilitiesontheonehand,whileintroducingnewdangers.Technologyempowersbothourcreativeanddestructivenatures. Substantialportionsofourspecieshavealreadyexperiencedalleviationofthepoverty,disease,hard labor,andmisfortunethathavecharacterizedmuchofhumanhistory.Manyofusnowhavetheopportunitytogainsatisfactionandmeaningfromourwork,ratherthanmerelytoilingtosurvive.Wehave ever more powerful tools to express ourselves. With the Web now reaching deeply into lessdevelopedregionsoftheworld,wewillseemajorstridesintheavailabilityofhigh-qualityeducationand medical knowledge. We can share culture, art, and humankind's exponentially expandingknowledgebaseworldwide.ImentionedtheWorldBank'sreportontheworldwidereductioninpovertyinchapter2anddiscussthatfurtherinthenextchapter. We'vegonefromabout twentydemocracies in theworldafterWorldWarII tomore thanone

hundred today largely through the influence of decentralized electronic communication. The biggestwave of democratization, including the fall of the Iron Curtain, occurred during the 1990swith thegrowthoftheInternetandrelatedtechnologies.Thereis,ofcourse,agreatdealmoretoaccomplishineachoftheseareas.Bioengineeringisintheearlystagesofmakingenormousstridesinreversingdiseaseandagingprocesses. Ubiquitous N and R are two to three decades away and will continue an exponentialexpansion of these benefits. As I reviewed in earlier chapters, these technologies will createextraordinarywealth, thereby overcoming poverty and enabling us to provide for all of ourmaterialneedsbytransforminginexpensiverawmaterialsandinformationintoanytypeofproduct.Wewillspendincreasingportionsofourtimeinvirtualenvironmentsandwillbeabletohaveanytypeofdesiredexperiencewithanyone,realorsimulated,invirtualreality.Nanotechnologywillbringasimilar ability to morph the physical world to our needs and desires. Lingering problems from ourwaning industrial age will be overcome. We will be able to reverse remaining environmentaldestruction. Nanoengineered fuel cells and solar cells will provide clean energy. Nanobots in ourphysical bodieswill destroy pathogens, remove debris such asmisformed proteins and proto fibrils,repairDNA,andreverseaging.Wewillbeabletoredesignallofthesystemsinourbodiesandbrainstobefarmorecapableanddurable. Most significant will be the merger of biological and nonbiological intelligence, althoughnonbiological intelligence will quickly come to predominate. There will be a vast expansion of theconceptofwhatitmeanstobehuman.Wewillgreatlyenhanceourabilitytocreateandappreciateallformsofknowledgefromsciencetothearts,whileextendingourabilitytorelatetoourenvironmentandoneanother.Ontheotherhand......andDangers

"Plants"with"leaves"nomoreefficientthantoday'ssolarcellscouldoutcompeterealplants,crowding the biosphere with an inedible foliage. Tough omnivorous "bacteria" could out-competerealbacteria:Theycouldspreadlikeblowingpollen,replicatedswiftly,andreducethebiospheretodustinamatterofdays.Dangerousreplicatorscouldeasilybetootough,small,and rapidly spreading to stop—at least ifwemake no preparation.We have trouble enoughcontrollingvirusesandfruitflies.

—ERICDREXLERAswellasitsmanyremarkableaccomplishments,thetwentiethcenturysawtechnology'sawesomeability to amplify our destructive nature, from Stalin's tanks to Hitler's trains. The tragic event ofSeptember11,2001,isanotherexampleoftechnologies(jetsandbuildings)takenoverbypeoplewithagendasofdestruction.Westilllivetodaywithasufficientnumberofnuclearweapons(notallofwhichareaccountedfor)toendallmammalianlifeontheplanet.Sincethe1980sthemeansandknowledgehaveexistedinaroutinecollegebioengineeringlabtocreate unfriendly pathogens potentially more dangerous than nuclear weapons." In a war-gamesimulation conducted at Johns Hopkins University called "Dark Winter," it was estimated that anintentionalintroductionofconventionalsmallpoxinthreeU.S.citiescouldresultinonemilliondeaths.Iftheviruswerebioengineeredtodefeattheexistingsmallpoxvaccine,theresultscouldbefarworse.'?Therealityof thisspecterwasmadeclearbya2001experiment inAustralia inwhich themousepoxviruswasinadvertentlymodifiedwithgenesthatalteredtheimmune-systemresponse.Themousepoxvaccinewaspowerlesstostopthisalteredvirus.11Thesedangersresonate inourhistoricalmemories.BubonicplaguekilledonethirdoftheEuropeanpopulation.Morerecentlythe1918flukilledtwentymillionpeopleworldwide.12

Willsuchthreatsprevent theongoingaccelerationof thepower,efficiency,andintelligenceofcomplexsystems(suchashumansandourtechnology)?Thepastrecordofcomplexityincreaseonthisplanet has shown a smooth acceleration, even through a longhistory of catastrophes, both internallygeneratedandexternallyimposed.Thisistrueofbothbiologicalevolution(whichfacedcalamitiessuchasencounterswithlargeasteroidsandmeteors)andhumanhistory(whichhasbeenpunctuatedbyanongoingseriesofmajorwars). However, I believe we can take some encouragement from the effectiveness of the world'sresponse to the SARS(severe acute respiratory syndrome) virus.Although the possibility of an evenmore virulent return of SARS remains uncertain as of the writing of this book, it appears thatcontainment measures have been relatively successful and have prevented this tragic outbreak frombecominga truecatastrophe.Partof the response involvedancient, low-tech toolssuchasquarantineandfacemasks. However,thisapproachwouldnothaveworkedwithoutadvancedtoolsthathaveonlyrecentlybecome available.Researcherswere able to sequence theDNAof the SARS viruswithin thirty-onedaysof theoutbreak—compared to fifteenyears forHIV.That enabled the rapiddevelopment of aneffective test so that carriers could quickly be identified. Moreover, instantaneous globalcommunicationfacilitatedacoordinatedresponseworldwide,afeatnotpossiblewhenvirusesravagedtheworldinancienttimes. AstechnologyacceleratestowardthefullrealizationofGNR,wewillseethesameintertwinedpotentials:a feastofcreativity resultingfromhuman intelligenceexpandedmanyfold,combinedwithmany grave new dangers. A quintessential concern that has received considerable attention isunrestrained nanobot replication.Nanobot technology requires trillions of such intelligently designeddevices to be useful.To scale up to such levels itwill be necessary to enable them to self-replicate,essentiallythesameapproachusedinthebiologicalworld(that'showonefertilizedeggcellbecomesthetrillionsofcellsinahuman).Andinthesamewaythatbiologicalself-replicationgoneawry(thatis,cancer)resultsinbiologicaldestruction,adefectinthemechanismcurtailingnanobotself-replication—theso-calledgray-gooscenario—wouldendangerallphysicalentities,biologicalorotherwise.Livingcreatures—includinghumans—wouldbetheprimaryvictimsofanexponentiallyspreadingnanobotattack.Theprincipaldesignsfornanobotconstructionusecarbonasaprimarybuildingblock.Becauseofcarbon'suniqueabilitytoformfour-waybonds,itisanidealbuildingblockformolecularassemblies. Carbonmolecules can form straight chains, zigzags, rings, nanotubes (hexagonal arraysformed in tubes), sheets, buckyballs (arrays of hexagons and pentagons formed into spheres), and avarietyofothershapes.Becausebiologyhasmadethesameuseofcarbon,pathologicalnanobotswouldfindtheEarth'sbiomassanidealsourceofthisprimaryingredient.Biologicalentitiescanalsoprovidestoredenergy in the formofglucoseandATP.13Useful trace elements such as oxygen, sulfur, iron,calcium,andothersarealsoavailableinthebiomass.Howlongwouldittakeanout-of-controlreplicatingnanobottodestroytheEarth'sbiomass?Thebiomasshasontheorderof1045carbonatoms.14Areasonableestimateofthenumberofcarbonatomsinasinglereplicatingnanobotisabout106.(Notethatthisanalysisisnotverysensitivetotheaccuracyofthesefigures,onlytotheapproximateorderofmagnitude.)Thismalevolentnanobotwouldneedtocreateontheorderof1039copiesof itself toreplacethebiomass,whichcouldbeaccomplishedwith130 replications (each of which would potentially double the destroyed biomass). Rob Freitas hasestimatedaminimumreplicationtimeofapproximatelyonehundredseconds,so130replicationcycleswouldrequireaboutthreeandahalfhours.15However,theactualrateofdestructionwouldbeslowerbecausebiomassisnot"efficiently"laidout.Thelimitingfactorwouldbetheactualmovementofthefrontofdestruction.Nanobotscannottravelveryquicklybecauseoftheirsmallsize.It'slikelytotakeweeksforsuchadestructiveprocesstocircletheglobe.Basedonthisobservationwecanenvisionamoreinsidiouspossibility.Inatwo-phasedattack,thenanobotstakeseveralweekstospreadthroughoutthebiomassbutuseupaninsignificantportionofthecarbonatoms,sayoneoutofeverythousandtrillion(1015).Atthisextremelylowlevelofconcentration

the nanobotswould be as stealthy as possible. Then, at an "optimal" point, the second phasewouldbeginwiththeseednanobotsexpandingrapidlyinplacetodestroythebiomass.Foreachseednanobottomultiply itself a thousand trillionfoldwould require only about fifty binary replications, or aboutninety minutes. With the nanobots having already spread out in position throughout the biomass,movementofthedestructivewavefrontwouldnolongerbealimitingfactor. Thepoint is thatwithoutdefenses, theavailablebiomasscouldbedestroyedbygraygooveryrapidly.AsIdiscussbelow(seep.417),wewillclearlyneedananotechnologyimmunesysteminplacebefore these scenarios become a possibility. This immune system would have to be capable ofcontendingnot justwithobviousdestructionbutwithanypotentiallydangerous(stealthy)replication,evenatverylowconcentrations. MikeTrederandChrisPhoenix—executivedirectoranddirectorof researchof theCenter forResponsible Nanotechnology, respectively—Eric Drexler, Robert Freitas, Ralph Merkle, and othershave pointed out that futureMNTmanufacturing devices can be createdwith safeguards thatwouldprevent the creation of self-replicating nanodevices.16 I discuss some of these strategies below.However, thisobservation,although important,doesnoteliminate the specterofgraygoo.Thereareother reasons (beyond manufacturing) that self-replicating nanobots will need to be created. Thenanotechnologyimmunesystemmentionedabove,forexample,willultimatelyrequireself-replication;otherwiseitwouldbeunabletodefendus.Self-replicationwillalsobenecessaryfornanobotstorapidlyexpand intelligence beyond the Earth, as I discussed in chapter 6. It is also likely to find extensivemilitary applications. Moreover, safeguards against unwanted self-replication, such as the broadcastarchitecturedescribedbelow(seep.412),canbedefeatedbyadeterminedadversaryorterrorist.Freitashasidentifiedanumberofotherdisastrousnanobotscenarios.17Inwhathecallsthe"grayplankton"scenario,maliciousnanobotswoulduseunderwatercarbonstoredasCH4(methane)aswellasCO2dissolvedinseawater.Theseocean-basedsourcescanprovideabouttentimesasmuchcarbonas Earth's biomass. In his "gray dust" scenario, replicating nanobots use basic elements available inairborne dust and sunlight for power. The "gray lichens" scenario involves using carbon and otherelementsonrocks.APanoplyofExistentialRisks

Ifalittleknowledgeisdangerous,whereisapersonwhohassomuchastobeoutofdanger?

—THOMASHENRYIdiscussbelow(seethesection"AProgramforGNRDefense,"p.422)stepswecantaketoaddressthesegraverisks,butwecannothavecompleteassuranceinanystrategythatwedevisetoday.TheserisksarewhatNickBostromcalls"existentialrisks,"whichhedefinesasthedangersintheupper-rightquadrantofthefollowingtable:18

BiologicallifeonEarthencounteredahuman-madeexistentialriskforthefirsttimeinthemiddleofthetwentiethcenturywiththeadventofthehydrogenbombandthesubsequentcold-warbuildupofthermonuclearforces.PresidentKennedyreportedlyestimatedthatthelikelihoodofanall-outnuclearwar during the Cuban missile crisis was between 33 and 50 percent.19 The legendary informationtheoristJohnvonNeumann,whobecamethechairmanoftheAirForceStrategicMissilesEvaluationCommittee and a government adviser on nuclear strategies, estimated the likelihood of nuclearArmageddon(priortotheCubanmissilecrisis)atcloseto100percent.20Giventheperspectiveofthe1960swhat informed observer of those timeswould have predicted that theworldwould have gonethroughthenextfortyyearswithoutanothernontestnuclearexplosion?Despitetheapparentchaosofinternationalaffairswecanbegratefulforthesuccessfulavoidancethusfaroftheemploymentofnuclearweaponsinwar.Butweclearlycannotresteasily,sinceenoughhydrogenbombsstillexist todestroyallhumanlifemanytimesover.21Althoughattractingrelativelylittlepublicdiscussion,themassiveopposingICBMarsenalsoftheUnitedStatesandRussiaremaininplace,despitetheapparentthawingofrelations.Nuclearproliferationandthewidespreadavailabilityofnuclearmaterialsandknow-howisanothergrave concern, although not an existential one for our civilization. (That is, only an all-outthermonuclear war involving the ICBM arsenals poses a risk to survival of all humans.) Nuclearproliferationandnuclearterrorismbelongtothe"profound-local"categoryofrisk,alongwithgenocide.However,theconcerniscertainlyseverebecausethelogicofmutualassureddestructiondoesnotworkinthecontextofsuicideterrorists. Debatablywe'venowaddedanotherexistentialrisk,whichisthepossibilityofabioengineeredvirusthatspreadseasily,hasalongincubationperiod,anddeliversanultimatelydeadlypayload.Somevirusesareeasilycommunicable,suchasthefluandcommoncold.Othersaredeadly,suchasHIV.Itisrare for a virus to combine both attributes. Humans living today are descendants of those whodevelopednaturalimmunitiestomostofthehighlycommunicableviruses.Theabilityofthespeciestosurviveviraloutbreaksisoneadvantageofsexualreproduction,whichtendstoensuregeneticdiversityinthepopulation,sothattheresponsetospecificviralagentsishighlyvariable.Althoughcatastrophic,bubonicplaguedidnotkill everyone inEurope.Otherviruses, suchas smallpox,havebothnegativecharacteristics—theyareeasilycontagiousanddeadly—buthavebeenaround longenough that therehas been time for society to create a technological protection in the form of a vaccine. Geneengineering, however, has the potential to bypass these evolutionary protections by suddenlyintroducingnewpathogensforwhichwehavenoprotection,naturalortechnological.Theprospectofaddinggenesfordeadlytoxinstoeasilytransmitted,commonvirusessuchasthecommoncoldandfluintroducedanotherpossibleexistential-riskscenario.ItwasthisprospectthatledtotheAsilomarconferencetoconsiderhowtodealwithsuchathreatandthesubsequentdraftingofa

set of safety and ethics guidelines. Although these guidelines have worked thus far, the underlyingtechnologiesforgeneticmanipulationaregrowingrapidlyinsophistication.In2003theworldstruggled,successfully,withtheSARSvirus.TheemergenceofSARSresultedfromacombinationofanancientpractice(thevirusissuspectedofhavingjumpedfromexoticanimals,possibly civet cats, tohumans living in closeproximity) andamodernpractice (the infection spreadrapidly across theworld by air travel). SARS provided uswith a dry run of a virus new to humancivilizationthatcombinedeasytransmission,theabilitytosurviveforextendedperiodsoftimeoutsidethehumanbody,andahighdegreeofmortality,withdeathratesestimatedat14to20percent.Again,theresponsecombinedancientandmoderntechniques. Our experiencewithSARS shows thatmost viruses, even if relatively easily transmitted andreasonably deadly, represent grave but not necessarily existential risks. SARS, however, does notappeartohavebeenengineered.SARSspreadseasilythroughexternallytransmittedbodilyfluidsbutisnoteasilyspreadthroughairborneparticles.Itsincubationperiodisestimatedtorangefromonedaytotwoweeks, whereas a longer incubation periodwould allow a THE virus to spread through severalexponentiallygrowinggenerationsbeforecarriersareidentified.22SARSisdeadly,butthemajorityofitsvictimsdosurvive.Itcontinuestobefeasibleforavirustobemalevolentlyengineeredso it spreadsmoreeasily thanSARS,hasanextended incubationperiod,andisdeadlytoessentiallyallvictims.Smallpoxisclosetohavingthesecharacteristics.Althoughwehave a vaccine (albeit a crude one), the vaccinewould not be effective against geneticallymodifiedversionsofthevirus.AsIdescribebelow,thewindowofmaliciousopportunityforbioengineeredviruses,existentialorotherwise, will close in the 2020s when we have fully effective antiviral technologies based onnanobots.23However, because nanotechnologywill be thousands of times stronger, faster, andmoreintelligentthanbiologicalentities,self-replicatingnanobotswillpresentagreaterriskandyetanotherexistential risk. The window for malevolent nanobots will ultimately be closed by strong artificialintelligence, but, not surprisingly, "unfriendly" AI will itself present an even more compellingexistentialrisk,whichIdiscussbelow(seep.420).ThePrecautionaryPrinciple.AsBostrom,Freitas,andotherobserversincludingmyselfhavepointedout,we cannot rely on trial-and-error approaches to dealwith existential risks. There are competinginterpretationsofwhathasbecomeknownasthe"precautionaryprinciple."(Iftheconsequencesofanaction are unknown but judged by some scientists to have even a small risk of being profoundlynegative, it'sbetter tonotcarryout theactionthanrisknegativeconsequences.)But it'sclear thatweneedtoachievethehighestpossiblelevelofconfidenceinourstrategiestocombatsuchrisks.Thisisone reason we're hearing increasingly strident voices demanding that we shut down the advance oftechnology,asaprimarystrategytoeliminatenewexistentialrisksbeforetheyoccur.Relinquishment,however, is not the appropriate response andwill only interferewith the profound benefits of theseemerging technologies while actually increasing the likelihood of a disastrous outcome. MaxMorearticulates the limitationsof theprecautionaryprincipleandadvocates replacing itwithwhathecallsthe"proactionaryprinciple,"whichinvolvesbalancingtherisksofactionandinaction.24Beforediscussinghowtorespondtothenewchallengeofexistentialrisks,it'sworthreviewingafewmorethathavebeenpostulatedbyBostromandothers.TheSmallertheInteraction,theLargertheExplosivePotential.Therehasbeenrecentcontroversyover the potential for future very high-energy particle accelerators to create a chain reaction oftransformedenergystatesatasubatomiclevel.Theresultcouldbeanexponentiallyspreadingareaofdestruction, breaking apart all atoms in our galactic vicinity. A variety of such scenarios has beenproposed,includingthepossibilityofcreatingablackholethatwoulddrawinoursolarsystem.Analysesofthesescenariosshowthemtobeveryunlikely,althoughnotallphysicistsaresanguineaboutthedanger.25Themathematicsoftheseanalysesappearstobesound,butwedonotyethaveaconsensusontheformulasthatdescribethislevelofphysicalreality.Ifsuchdangerssoundfar-fetched,

consider the possibility thatwe have indeed detected increasingly powerful explosive phenomena atdiminishingscalesofmatter. AlfredNobeldiscovereddynamitebyprobingchemical interactionsofmolecules.Theatomicbomb, which is tens of thousands of times more powerful than dynamite, is based on nuclearinteractionsinvolvinglargeatoms,whicharemuchsmallerscalesofmatterthanlargemolecules.Thehydrogen bomb, which is thousands of times more powerful than an atomic bomb, is based oninteractions involving an even smaller scale: small atoms.Although this insight does not necessarilyimply the existence of yet more powerful destructive chain reactions by manipulating subatomicparticles,itdoesmaketheconjectureplausible. My own assessment of this danger is that we are unlikely simply to stumble across such adestructiveevent.Considerhowunlikelyitwouldbetoaccidentallyproduceanatomicbomb.Suchadevicerequiresapreciseconfigurationofmaterialsandactions,andtheoriginalrequiredanextensiveandpreciseengineeringprojecttodevelop.Inadvertentlycreatingahydrogenbombwouldbeevenlessplausible. One would have to create the precise conditions of an atomic bomb in a particulararrangementwithahydrogencoreandotherelements.Stumblingacrosstheexactconditionstocreateanew class of catastrophic chain reaction at a subatomic level appears to be even less likely. Theconsequencesaresufficientlydevastating,however, that theprecautionaryprinciple should leadus totakethesepossibilitiesseriously.Thispotentialshouldbecarefullyanalyzedpriortocarryingoutnewclasses of accelerator experiments. However, this risk is not high onmy list of twenty-first-centuryconcerns.OurSimulationIsTurnedOff.AnotherexistentialriskthatBostromandothershaveidentifiedisthatwe'reactuallylivinginasimulationandthesimulationwillbeshutdown.Itmightappearthatthere'snotalotwecoulddotoinfluencethis.However,sincewe'rethesubjectofthesimulation,wedohavetheopportunitytoshapewhathappensinsideofit.Thebestwaywecouldavoidbeingshutdownwouldbe to be interesting to the observers of the simulation. Assuming that someone is actually payingattentiontothesimulation,it'safairassumptionthatit'slesslikelytobeturnedoffwhenit'scompellingthanotherwise.Wecouldspendalotoftimeconsideringwhatitmeansforasimulationtobeinteresting,butthecreationofnewknowledgewouldbeacriticalpartofthisassessment.Althoughitmaybedifficultforustoconjecturewhatwouldbeinterestingtoourhypothesizedsimulationobserver,itwouldseemthatthe Singularity is likely to be about as absorbing as any developmentwe could imagine andwouldcreatenewknowledgeatanextraordinaryrate.Indeed,achievingaSingularityofexplodingknowledgemaybetheverypurposeofthesimulation.Thus,assuringa"constructive"Singularity(onethatavoidsdegenerateoutcomessuchasexistentialdestructionbygraygooordominancebyamaliciousAI)couldbethebestcoursetopreventthesimulationfrombeingterminated.Ofcourse,wehaveeverymotivationtoachieveaconstructiveSingularityformanyotherreasons. If theworldwe're living in is a simulation on someone's computer, it's a very good one—sodetailed,infact,thatwemayaswellacceptitasourreality.Inanyevent,itistheonlyrealitytowhichwehaveaccess.Ourworldappearstohavealongandrichhistory.Thismeansthateitherourworldisnot,infact,asimulation or, if it is, the simulation has been going a very long time and thus is not likely to stopanytime soon. Of course it is also possible that the simulation includes evidence of a long historywithoutthehistory'shavingactuallyoccurred.AsIdiscussedinchapter6,thereareconjecturesthatanadvancedcivilizationmaycreateanewuniverse to perform computation (or, to put it another way, to continue the expansion of its owncomputation). Our living in such a universe (created by another civilization) can be considered asimulationscenario.Perhapsthisothercivilizationisrunninganevolutionaryalgorithmonouruniverse(that is, the evolution we're witnessing) to create an explosion of knowledge from a technologySingularity.Ifthatistrue,thenthecivilizationwatchingouruniversemightshutdownthesimulationifitappearedthataknowledgeSingularityhadgoneawryanditdidnotlooklikeitwasgoingtooccur. Thisscenarioisalsonothighonmyworrylist,particularlysincetheonlystrategythatwecan

followtoavoidanegativeoutcomeistheoneweneedtofollowanyway.Crashing the Party. Another oft-cited concern is that of a large-scale asteroid or comet collision,whichhasoccurredrepeatedlyintheEarth'shistory,anddidrepresentexistentialoutcomesforspeciesatthesetimes.Thisisnotaperiloftechnology,ofcourse.Rather,technologywillprotectusfromthisrisk (certainlywithin one to a couple of decades).Although small impacts are a regular occurrence,largeanddestructivevisitorsfromspacearerare.Wedon'tseeoneonthehorizon,andit isvirtuallycertainthatbythetimesuchadangeroccurs,ourcivilizationwillreadilydestroytheintruderbeforeitdestroysus.Anotheritemontheexistentialdangerlistisdestructionbyanalienintelligence(notonethatwe'vecreated).Idiscussedthispossibilityinchapter6andIdon'tseethisaslikely,either.GNR:TheProperFocusofPromiseVersusPeril.ThisleavestheGNRtechnologiesastheprimaryconcerns.However, Ido thinkwealsoneed to takeseriously themisguidedand increasinglystridentLudditevoices that advocate relianceonbroad relinquishmentof technological progress to avoid thegenuinedangersofGNR.ForreasonsIdiscussbelow(seep.410),relinquishmentisnottheanswer,butrationalfearcouldleadtoirrationalsolutions.Delaysinovercominghumansufferingarestillofgreatconsequence—forexample,theworseningoffamineinAfricaduetooppositiontoaidfromfoodusingGMOs(geneticallymodifiedorganisms). Broadrelinquishmentwouldrequireatotalitariansystemtoimplement,andatotalitarianbravenew world is unlikely because of the democratizing impact of increasingly powerful decentralizedelectronic and photonic communication. The advent of worldwide, decentralized communicationepitomizedbytheInternetandcellphoneshasbeenapervasivedemocratizingforce.ItwasnotBorisYeltsin standing on a tank that overturned the 1991 coup againstMikhailGorbachev, but rather theclandestinenetworkoffaxmachines,photocopiers,videorecorders,andpersonalcomputersthatbrokedecadesoftotalitariancontrolofinformation.26Themovement towarddemocracyandcapitalismandtheattendanteconomicgrowththatcharacterizedthe1990swereallfueledbytheacceleratingforceoftheseperson-to-personcommunicationtechnologies. Thereareotherquestionsthatarenonexistentialbutnonethelessserious.Theyinclude"Whoiscontrolling the nanobots?" and "Whom are the nanobots talking to?" Future organizations (whethergovernmentsorextremistgroups)orjustacleverindividualcouldputtrillionsofundetectablenanobotsin the water or food supply of an individual or of an entire population. These spybots could thenmonitor, influence, and even control thoughts and actions. In addition existing nanobots could beinfluenced through software viruses and hacking techniques.When there is software running in ourbodies and brains (aswe discussed, a thresholdwe have already passed for some people), issues ofprivacyandsecuritywill takeonanewurgency,andcountersurveillancemethodsofcombatingsuchintrusionswillbedevised.TheInevitabilityofaTransformedFuture.ThediverseGNRtechnologiesareprogressingonmanyfronts.The full realizationofGNRwill result fromhundredsof small steps forward, eachbenign initself.ForGwehavealreadypassed the thresholdofhaving themeans tocreatedesignerpathogens.Advancesinbiotechnologywillcontinuetoaccelerate,fueledbythecompellingethicalandeconomicbenefitsthatwillresultfrommasteringtheinformationprocessesunderlyingbiology. Nanotechnologyis theinevitableendresultof theongoingminiaturizationoftechnologyofallkinds.Thekeyfeaturesforawiderangeofapplications,includingelectronics,mechanics,energy,andmedicine,areshrinkingattherateofafactorofaboutfourperlineardimensionperdecade.Moreover,thereisexponentialgrowthinresearchseekingtounderstandnanotechnologyanditsapplications.(Seethegraphsonnanotechnologyresearchstudiesandpatentsonpp.83and84.) Similarly,ourefforts toreverseengineer thehumanbrainaremotivatedbydiverseanticipatedbenefits, includingunderstanding and reversing cognitivediseases anddecline.The tools for peeringintothebrainareshowingexponentialgainsinspatialandtemporalresolution,andwe'vedemonstrated

theabilitytotranslatedatafrombrainscansandstudiesintoworkingmodelsandsimulations.Insightsfromthebrainreverse-engineeringeffort,overallresearchindevelopingAIalgorithms,andongoingexponentialgainsincomputingplatformsmakestrongAI(AIathumanlevelsandbeyond)inevitable.OnceAIachieveshumanlevels, itwillnecessarilysoarpastitbecauseitwillcombinethestrengths of human intelligence with the speed, memory capacity, and knowledge sharing thatnonbiological intelligence already exhibits. Unlike biological intelligence, nonbiological intelligencewillalsobenefitfromongoingexponentialgainsinscale,capacity,andprice-performance.TotalitarianRelinquishment.Theonlyconceivablewaythattheacceleratingpaceofadvancementonallofthesefrontscouldbestoppedwouldbethroughaworldwidetotalitariansystemthatrelinquishesthe very idea of progress. Even this specterwould be likely to fail in averting the dangers ofGNRbecausetheresultingundergroundactivitywouldtendtofavorthemoredestructiveapplications.Thisisbecausetheresponsiblepractitionersthatwerelyontoquicklydevelopdefensivetechnologieswouldnothaveeasyaccesstotheneededtools.Fortunately,suchatotalitarianoutcomeisunlikelybecausetheincreasingdecentralizationofknowledgeisinherentlyademocratizingforce.PreparingtheDefensesMy own expectation is that the creative and constructive applications of these technologies willdominate,asIbelievetheydotoday.However,weneedtovastlyincreaseourinvestmentindevelopingspecificdefensivetechnologies.AsIdiscussed,weareatthecriticalstagetodayforbiotechnology,andwewillreachthestagewhereweneedtodirectlyimplementdefensivetechnologiesfornanotechnologyduringthelateteenyearsofthiscentury. We don't have to look past today to see the intertwined promise and peril of technologicaladvancement. Imagine describing the dangers (atomic and hydrogen bombs for one thing) that existtodaytopeoplewholivedacoupleofhundredyearsago.Theywouldthinkitmadtotakesuchrisks.Buthowmanypeoplein2005wouldreallywanttogobacktotheshort,brutish,disease-filled,poverty-stricken,disaster-pronelivesthat99percentofthehumanracestruggledthroughacoupleofcenturiesago?27Wemayromanticizethepast,butupuntilfairlyrecentlymostofhumanitylivedextremelyfragilelives in which one all-too-common misfortune could spell disaster. Two hundred years ago lifeexpectancyforfemalesintherecord-holdingcountry(Sweden)wasroughlythirty-fiveyears,verybriefcompared to the longest life expectancy today—almost eighty-five years, for Japanese women. Lifeexpectancyformaleswasroughlythirty-threeyears,comparedtothecurrentseventy-nineyearsintherecord-holding countries.28 It took half the day to prepare the evening meal, and hard laborcharacterizedmosthumanactivity.Therewerenosocialsafetynets.Substantialportionsofourspeciesstillliveinthisprecariousway,whichisatleastonereasontocontinuetechnologicalprogressandtheeconomic enhancement that accompanies it. Only technology, with its ability to provide orders ofmagnitudeof improvement incapabilityandaffordability,has thescale toconfrontproblemssuchaspoverty,disease,pollution,andtheotheroverridingconcernsofsocietytoday. Peopleoftengo through three stages in considering the impactof future technology: aweandwondermentatitspotentialtoovercomeage-oldproblems;thenasenseofdreadatanewsetofgravedangersthataccompanythesenoveltechnologies;followedfinallybytherealizationthattheonlyviableandresponsiblepathistosetacarefulcoursethatcanrealizethebenefitswhilemanagingthedangers. Needless tosay,wehavealreadyexperienced technology'sdownside—forexample,deathanddestructionfromwar.Thecrudetechnologiesofthefirstindustrialrevolutionhavecrowdedoutmanyofthespeciesthatexistedonourplanetacenturyago.Ourcentralizedtechnologies(suchasbuildings,cities,airplanes,andpowerplants)aredemonstrablyinsecure.The"NBC"(nuclear,biological,andchemical)technologiesofwarfarehaveallbeenusedorbeenthreatenedtobeusedinourrecentpast.29ThefarmorepowerfulGNRtechnologies threatenuswith

new, profound local and existential risks. If we manage to get past the concerns about geneticallyaltered designer pathogens, followed by self-replicating entities created through nanotechnology, wewill encounter robotswhose intelligencewill rival and ultimately exceed our own. Such robotsmaymakegreatassistants,butwho's tosaythatwecancountonthemtoremainreliablyfriendlytomerebiologicalhumans?StrongAI.StrongAIpromisestocontinuetheexponentialgainsofhumancivilization.(AsIdiscussedearlier,Iincludethenonbiologicalintelligencederivedfromourhumancivilizationasstillhuman.)Butthe dangers it presents are also profound precisely because of its amplification of intelligence.Intelligence is inherently impossible to control, so the various strategies that have been devised tocontrol nanotechnology (for example, the "broadcast architecture" described below) won't work forstrongAI.There have beendiscussions andproposals to guideAI development towardwhatEliezerYudkowsky calls "friendlyAI"30 (see the section "Protection from 'Unfriendly' StrongAI," p. 420).Theseareusefulfordiscussion,butitisinfeasibletodaytodevisestrategiesthatwillabsolutelyensurethatfutureAIembodieshumanethicsandvalues.Returning to thePast? In his essay and presentations Bill Joy eloquently describes the plagues ofcenturiespastandhownewself-replicatingtechnologies,suchasmutantbioengineeredpathogensandnanobots run amok, may bring back long-forgotten pestilence. Joy acknowledges that technologicaladvances, such as antibiotics and improved sanitation, have freed us from the prevalence of suchplagues, and such constructive applications, therefore, need to continue. Suffering in the worldcontinues and demands our steadfast attention. Should we tell the millions of people afflicted withcancer and other devastating conditions that we are canceling the development of all bioengineeredtreatmentsbecause there is a risk that these same technologiesmay somedaybeused formalevolentpurposes?Havingposedthisrhetoricalquestion,Irealizethatthereisamovementtodoexactlythat,butmostpeoplewouldagreethatsuchbroad-basedrelinquishmentisnottheanswer. The continued opportunity to alleviate human distress is one key motivation for continuingtechnologicaladvancement.Alsocompellingarethealreadyapparenteconomicgainsthatwillcontinuetohasten in thedecadesahead.Theongoingaccelerationofmany intertwined technologiesproducesroads paved with gold. (I use the plural here because technology is clearly not a single path.) In acompetitive environment it is an economic imperative to go down these roads. Relinquishingtechnologicaladvancementwouldbeeconomicsuicideforindividuals,companies,andnations.TheIdeaofRelinquishment

Themajoradvancesincivilizationallbutwreckthecivilizationsinwhichtheyoccur.

—ALFREDNORTHWHITEHEADThis brings us to the issue of relinquishment, which is the most controversial recommendation byrelinquishmentadvocatessuchasBillMcKibben.Idofeelthatrelinquishmentattherightlevelispartof a responsible and constructive response to the genuine perils thatwewill face in the future. Theissue,however,isexactlythis:atwhatlevelarewetorelinquishtechnology?TedKaczynski,whobecameknowntotheworldastheUnabomber,wouldhaveusrenounceallofit.31Thisisneitherdesirablenorfeasible,andthefutilityofsuchapositionisonlyunderscoredbythesenselessnessofKaczynski'sdeplorabletactics. Othervoices, less reckless thanKaczynski's, arenonetheless likewisearguing forbroad-basedrelinquishmentoftechnology.McKibbentakesthepositionthatwealreadyhavesufficienttechnologyandthatfurtherprogressshouldend.Inhislatestbook,Enough:StayingHumaninanEngineeredAge,he metaphorically compares technology to beer: "One beer is good, two beers may be better; eight

beers, you're almost certainly going to regret."32 That metaphor misses the point and ignores theextensivesufferingthatremains in thehumanworldthatwecanalleviate throughsustainedscientificadvance.Althoughnewtechnologies,likeanythingelse,maybeusedtoexcessattimes,theirpromiseisnotjust amatter of adding a fourth cell phone or doubling the number of unwanted e-mails. Rather, itmeansperfectingthetechnologiestoconquercancerandotherdevastatingdiseases,creatingubiquitouswealth to overcome poverty, cleaning up the environment from the effects of the first industrialrevolution(anobjectivearticulatedbyMcKibben),andovercomingmanyotherage-oldproblems.Broad Relinquishment. Another level of relinquishment would be to forgo only certain fields—nanotechnology,forexample—thatmightberegardedastoodangerous.Butsuchsweepingstrokesofrelinquishmentareequallyuntenable.AsIpointedoutabove,nanotechnologyissimplytheinevitableendresultofthepersistenttrendtowardminiaturizationthatpervadesalloftechnology.Itisfarfromasinglecentralizedeffortbutisbeingpursuedbyamyriadofprojectswithmanydiversegoals.Oneobserverwrote:

Afurtherreasonwhyindustrialsocietycannotbereformed ... is thatmodern technologyisaunifiedsysteminwhichallpartsaredependentononeanother.Youcan'tgetridofthe"bad"parts of technology and retain only the "good" parts. Take modern medicine, for example.Progress in medical science depends on progress in chemistry, physics, biology, computerscienceandotherfields.Advancedmedicaltreatmentsrequireexpensive,high-techequipmentthat canbemade available onlyby a technologically progressive, economically rich society.Clearlyyoucan'thavemuchprogressinmedicinewithoutthewholetechnologicalsystemandeverythingthatgoeswithit.

Theobserver I amquotinghere is, again,TedKaczynski.33Althoughonewill properly resistKaczynski as an authority, I believehe is correct on the deeply entanglednature of the benefits andrisks.However,KaczynskiandIclearlypartcompanyonouroverallassessmentoftherelativebalancebetweenthetwo.BillJoyandIhavehadanongoingdialogueonthisissuebothpubliclyandprivately,andwebothbelievethattechnologywillandshouldprogressandthatweneedtobeactivelyconcernedwithitsdarkside.Themostchallengingissuetoresolveisthegranularityofrelinquishmentthatisbothfeasibleanddesirable.Fine-GrainedRelinquishment.Idothinkthatrelinquishmentattherightlevelneedstobepartofourethicalresponsetothedangersoftwenty-first-centurytechnologies.Oneconstructiveexampleofthisisthe ethical guideline proposed by the Foresight Institute: namely, that nanotechnologists agree torelinquishthedevelopmentofphysicalentitiesthatcanself-replicateinanaturalenvironment.34Inmyview, there are two exceptions to this guideline. First, we will ultimately need to provide ananotechnology-based planetary immune system (nanobots embedded in the natural environment toprotectagainstrogueself-replicatingnanobots).RobertFreitasandIhavediscussedwhetherornotsuchan immune system would itself need to be self-replicating. Freitas writes: "A comprehensivesurveillance system coupled with prepositioned resources—resources including high-capacitynonreplicatingnanofactoriesabletochurnoutlargenumbersofnonreplicatingdefendersinresponsetospecific threats—shouldsuffice."35 I agreewithFreitas that aprepositioned immune systemwith theability toaugment thedefenderswillbesufficient inearlystages.ButoncestrongAIismergedwithnanotechnology,and theecologyofnanoengineeredentitiesbecomeshighlyvariedandcomplex,myownexpectationisthatwewillfindthatthedefendingnanorobotsneedtheabilitytoreplicateinplacequickly.Theotherexceptionistheneedforself-replicatingnanobot-basedprobestoexploreplanetarysystemsoutsideofoursolarsystem. Anothergoodexampleofausefulethicalguidelineisabanonself-replicatingphysicalentitiesthat contain their own codes for self-replication. In what nanotechnologist Ralph Merkle calls the

"broadcastarchitecture,"suchentitieswouldhavetoobtainsuchcodesfromacentralizedsecureserver,whichwould guard against undesirable replication.36 The broadcast architecture is impossible in thebiological world, so there's at least one way in which nanotechnology can be made safer thanbiotechnology. In other ways, nanotech is potentially more dangerous because nanobots can bephysicallystrongerthanprotein-basedentitiesandmoreintelligent. As I described in chapter 5, we can apply a nanotechnology-based broadcast architecture tobiology.A nanocomputerwould augment or replace the nucleus in every cell and provide theDNAcodes. A nanobot that incorporated molecular machinery similar to ribosomes (the molecules thatinterpretthebasepairsinthemRNAoutsidethenucleus)wouldtakethecodesandproducethestringsofaminoacids.Sincewecouldcontrolthenanocomputerthroughwirelessmessages,wewouldbeableto shut off unwanted replication, thereby eliminating cancer. We could produce special proteins asneededtocombatdisease.AndwecouldcorrecttheDNAerrorsandupgradetheDNAcode.Icommentfurtheronthestrengthsandweaknessesofthebroadcastarchitecturebelow.DealingwithAbuse.Broadrelinquishment iscontrary toeconomicprogressandethicallyunjustifiedgiven the opportunity to alleviate disease, overcome poverty, and clean up the environment. Asmentioned above, it would exacerbate the dangers. Regulations on safety—essentially fine-grainedrelinquishment—willremainappropriate.However,wealsoneedtostreamlinetheregulatoryprocess.RightnowintheUnitedStates,wehaveafive-toten-yeardelayonnewhealthtechnologiesforFDAapproval(withcomparabledelaysinothernations).Theharmcausedbyholdinguppotentiallifesavingtreatments(forexample,onemillionlives lost in theUnitedStates foreachyearwedelay treatments forheartdisease) isgivenvery littleweightagainstthepossiblerisksofnewtherapies. Other protections will need to include oversight by regulatory bodies, the development oftechnology-specific "immune" responses, and computer-assisted surveillance by law-enforcementorganizations. Many people are not aware that our intelligence agencies already use advancedtechnologies such as automated keyword spotting tomonitor a substantial flow of telephone, cable,satellite,andInternetconversations.Aswegoforward,balancingourcherishedrightsofprivacywithourneedtobeprotectedfromthemalicioususeofpowerful twenty-first-centurytechnologieswillbeoneofmanyprofoundchallenges.Thisisonereasonsuchissuesasanencryption"trapdoor"(inwhichlaw-enforcement authorities would have access to otherwise secure information) and the FBI'sCarnivoree-mail-snoopingsystemhavebeencontroversial.37Asatestcasewecantakeasmallmeasureofcomfortfromhowwehavedealtwithonerecenttechnologicalchallenge.Thereexiststodayanewfullynonbiologicalself-replicatingentitythatdidn'texistjustafewdecadesago:thecomputervirus.Whenthisformofdestructiveintruderfirstappeared,strong concerns were voiced that as they became more sophisticated, software pathogens had thepotential todestroy thecomputer-networkmediuminwhich they live.Yet the"immunesystem" thathas evolved in response to this challenge has been largely effective. Although destructive self-replicatingsoftwareentitiesdocausedamagefromtimetotime,theinjuryisbutasmallfractionofthebenefitwereceivefromthecomputersandcommunicationlinksthatharborthem.Onemightcounterthatcomputervirusesdonothavethelethalpotentialofbiologicalvirusesorofdestructivenanotechnology.This isnot always thecase;we relyon software tooperateour911callcenters,monitorpatientsincritical-careunits,flyandlandairplanes,guideintelligentweaponsinourmilitarycampaigns,handleour financial transactions,operateourmunicipalutilities, andmanyothermission-criticaltasks.Totheextentthatsoftwarevirusesdonotyetposealethaldanger,however,thisobservation only strengthensmy argument. The fact that computer viruses are not usually deadly tohumans only means that more people are willing to create and release them. The vast majority ofsoftware-virusauthorswouldnotreleasevirusesif theythoughttheywouldkillpeople.Italsomeansthatourresponsetothedangeristhatmuchlessintense.Conversely,whenitcomestoself-replicatingentitiesthatatepotentiallylethalonalargescale,ourresponseonalllevelswillbevastlymoreserious.Althoughsoftwarepathogensremainaconcern,thedangerexiststodaymostlyatanuisancelevel.

Keepinmindthatoursuccess incombatingthemhas takenplace inan industry inwhichthere isnoregulationandminimalcertificationforpractitioners.Thelargelyunregulatedcomputerindustryisalsoenormouslyproductive.Onecouldarguethatithascontributedmoretoourtechnologicalandeconomicprogressthananyotherenterpriseinhumanhistory.Butthebattleconcerningsoftwarevirusesandthepanoplyofsoftwarepathogenswillneverend.Wearebecomingincreasinglyreliantonmission-criticalsoftwaresystems,andthesophisticationandpotentialdestructivenessofself-replicatingsoftwareweaponswillcontinuetoescalate.Whenwehavesoftware running in our brains and bodies and controlling the world's nanobot immune system, thestakeswillbeimmeasurablygreater.TheThreat fromFundamentalism. The world is struggling with an especially pernicious form ofreligious fundamentalism in the formof radical Islamic terrorism.Although itmay appear that theseterrorists have no program other than destruction, they do have an agenda that goes beyond literalinterpretations of ancient scriptures: essentially, to turn the clock back on such modern ideas asdemocracy,women'srights,andeducation.Butreligiousextremismisnottheonlyformoffundamentalismthatrepresentsareactionaryforce.At the beginning of this chapter I quoted Patrick Moore, cofounder of Greenpeace, on hisdisillusionment with themovement he helped found. The issue that underminedMoore's support ofGreenpeacewasitstotaloppositiontoGoldenRice,astrainofricegeneticallymodifiedtocontainhighlevelsofbeta-carotene,theprecursortovitaminA.38HundredsofmillionsofpeopleinAfricaandAsialacksufficientvitaminA,withhalfamillionchildrengoingblindeachyearfromthedeficiency,andmillions more contracting other related diseases. About seven ounces a day of Golden Rice wouldprovide100percentofachild'svitaminArequirement.Extensivestudieshaveshownthatthisgrain,aswellasmanyothergeneticallymodifiedorganisms(GMOs),issafe.Forexample,in2001theEuropeanCommissionreleasedeighty-onestudiesthatconcludedthatGMOshave"notshownanynewriskstohuman health or the environment, beyond the usual uncertainties of conventional plant breeding.Indeed,theuseofmoreprecisetechnologyandthegreaterregulatoryscrutinyprobablymakethemevensaferthanconventionalplantsandfoods."39ItisnotmypositionthatallGMOsareinherentlysafe;obviouslysafetytestingofeachproductisneeded. But the anti-GMO movement takes the position that every GMO is by its very naturehazardous,aviewthathasnoscientificbasis. TheavailabilityofGoldenRicehasbeendelayedbyat leastfiveyearsthroughthepressureofGreenpeace and other anti-GMO activists. Moore, noting that this delay will cause millions ofadditionalchildrentogoblind,quotesthegrain'sopponentsasthreatening"toriptheG.M.riceoutofthe fields if farmersdare toplant it."Similarly,Africannationshavebeenpressured to refuseGMOfood aid and genetically modified seeds, thereby worsening conditions of famine.40 Ultimately thedemonstratedabilityoftechnologiessuchasGMOtosolveoverwhelmingproblemswillprevail,butthetemporarydelayscausedbyirrationaloppositionwillnonethelessresultinunnecessarysuffering. Certain segments of the environmental movement have become fundamentalist Luddites—"fundamentalist" because of their misguided attempt to preserve things as they are (or were);"Luddite" because of the reflexive stance against technological solutions to outstanding problems.IronicallyitisGMOplants—manyofwhicharedesignedtoresistinsectsandotherformsofblightandthereby require greatly reduced levels of chemicals, if any—that offer the best hope for reversingenvironmentalassaultfromchemicalssuchaspesticides.Actuallymycharacterizationofthesegroupsas"fundamentalistLuddites"isredundant,becauseLudditism is inherently fundamentalist. It reflects the idea that humanity will be better off withoutchange, without progress. This brings us back to the idea of relinquishment, as the enthusiasm forrelinquishing technology on a broad scale is coming from the same intellectual sources and activistgroupsthatmakeuptheLudditesegmentoftheenvironmentalmovement.FundamentalistHumanism.With G and N technologies now beginning to modify our bodies and

brains,anotherformofoppositiontoprogresshasemergedintheformof"fundamentalisthumanism":oppositiontoanychangeinthenatureofwhatitmeanstobehuman(forexample,changingourgenesand taking other steps toward radical life extension). This effort, too, will ultimately fail, however,becausethedemandfortherapiesthatcanovercomethesuffering,disease,andshortlifespansinherentinourversion1.0bodieswillultimatelyproveirresistible. In the end, it is only technology—especially GNR—that will offer the leverage needed toovercomeproblemsthathumancivilizationhasstruggledwithformanygenerations.DevelopmentofDefensiveTechnologiesandtheImpactofRegulationOneofthereasonsthatcallsforbroadrelinquishmenthaveappealisthattheypaintapictureoffuturedangersassumingtheywillbereleasedinthecontextoftoday'sunpreparedworld.Therealityisthatthesophistication and power of our defensive knowledge and technologies will grow along with thedangers.Aphenomenonlikegraygoo(unrestrainednanobotreplication)willbecounteredwith"bluegoo"("police"nanobotsthatcombatthe"bad"nanobots).Obviouslywecannotsaywithassurancethatwewillsuccessfullyavertallmisuse.Butthesurestwaytopreventdevelopmentofeffectivedefensivetechnologieswouldbetorelinquishthepursuitofknowledgeinanumberofbroadareas.Wehavebeenable to largely control harmful software-virus replication because the requisite knowledge is widelyavailabletoresponsiblepractitioners.Attemptstorestrictsuchknowledgewouldhavegivenrisetoafarlessstablesituation.Responsestonewchallengeswouldhavebeenfarslower,anditislikelythatthebalance would have shifted toward more destructive applications (such as self-modifying softwareviruses).Ifwecomparethesuccesswehavehadincontrollingengineeredsoftwarevirusestothecomingchallengeofcontrollingengineeredbiologicalviruses,wearestruckwithonesalientdifference.AsInotedabove,thesoftwareindustryisalmostcompletelyunregulated.Thesameisobviouslynottrueforbiotechnology.While a bioterrorist does not need to put his "inventions" through the FDA, we dorequire the scientists developing defensive technologies to follow existing regulations, which slowdowntheinnovationprocessateverystep.Moreover,underexistingregulationsandethicalstandards,itisimpossibletotestdefensesagainstbioterroristagents.Extensivediscussionisalreadyunderwaytomodifytheseregulationstoallowforanimalmodelsandsimulationstoreplaceunfeasiblehumantrials.This will be necessary, but I believe we will need to go beyond these steps to accelerate thedevelopmentofvitallyneededdefensivetechnologies.Intermsofpublicpolicythetaskathandistorapidlydevelopthedefensivestepsneeded,whichincludeethical standards, legal standards, anddefensive technologies themselves. It isquite clearlyarace.AsInoted,inthesoftwarefielddefensivetechnologieshaverespondedquicklytoinnovationsintheoffensiveones.Inthemedicalfield,incontrast,extensiveregulationslowsdowninnovation,sowecannothavethesameconfidencewithregardtotheabuseofbiotechnology.Inthecurrentenvironment,whenonepersondiesingene-therapytrials,researchcanbeseverelyrestricted.41Thereisalegitimateneedtomakebiomedicalresearchassafeaspossible,butourbalancingofrisksiscompletelyskewed.Millions of people desperately need the advances promised by gene therapy and other breakthroughbiotechnology advances, but they appear to carry little political weight against a handful of well-publicizedcasualtiesfromtheinevitablerisksofprogress.Thisrisk-balancingequationwillbecomeevenmorestarkwhenweconsidertheemergingdangersofbioengineeredpathogens.What is needed is a change inpublic attitude in tolerance for necessaryrisk. Hastening defensive technologies is absolutely vital to our security. We need to streamlineregulatory procedures to achieve this. At the same time we must greatly increase our investmentexplicitly in defensive technologies. In the biotechnology field thismeans the rapid development ofantiviral medications. We will not have time to formulate specific countermeasures for each newchallengethatcomesalong.Weareclosetodevelopingmoregeneralizedantiviraltechnologies,suchasRNAinterference,andtheseneedtobeaccelerated.

We'readdressingbiotechnologyherebecausethatistheimmediatethresholdandchallengethatwenowface.Asthethresholdforself-organizingnanotechnologyapproaches,wewillthenneedtoinvestspecifically in the development of defensive technologies in that area, including the creation of atechnological immune system. Consider how our biological immune system works.When the bodydetects a pathogen the T cells and other immune-system cells self-replicate rapidly to combat theinvader.Ananotechnology immunesystemwouldworksimilarlyboth in thehumanbodyand in theenvironment and would include nanobot sentinels that could detect rogue self-replicating nanobots.Whena threatwasdetected,defensivenanobotscapableofdestroying the intruderswouldrapidlybecreated(eventuallywithself-replication)toprovideaneffectivedefensiveforce.BillJoyandotherobservershavepointedoutthatsuchanimmunesystemwoulditselfbeadangerbecauseofthepotentialof"autoimmune"reactions(thatis,theimmune-systemnanobotsattackingtheworldtheyaresupposedtodefend).42Howeverthispossibilityisnotacompellingreasontoavoidthecreationofanimmunesystem.Noonewouldarguethathumanswouldbebetteroffwithoutanimmunesystembecauseofthepotentialofdevelopingautoimmunediseases.Althoughtheimmunesystemcanitselfpresentadanger,humanswouldnotlastmorethanafewweeks(barringextraordinaryeffortsatisolation) without one. And even so, the development of a technological immune system fornanotechnologywillhappenevenwithoutexpliciteffortstocreateone.Thishaseffectivelyhappenedwithregardtosoftwareviruses,creatinganimmunesystemnotthroughaformalgrand-designprojectbutratherthroughincrementalresponsestoeachnewchallengeandbydevelopingheuristicalgorithmsforearlydetection.Wecanexpectthesamethingwillhappenaschallengesfromnanotechnology-baseddangers emerge. The point for public policy will be to invest specifically in these defensivetechnologies.Itisprematuretodaytodevelopspecificdefensivenanotechnologies,sincewecannowhaveonlyageneral ideaofwhatweare trying todefendagainst.However, fruitfuldialogueanddiscussiononanticipatingthisissuearealreadytakingplace,andsignificantlyexpandedinvestmentintheseeffortsistobeencouraged.AsImentionedabove,theForesightInstitute,asoneexample,hasdevisedasetofethical standards and strategies for assuring the development of safe nanotechnology, based onguidelinesforbiotechnology.43Whengene-splicingbeganin1975twobiologists,MaxineSingerandPaul Berg, suggested a moratorium on the technology until safety concerns could be addressed. Itseemed apparent that therewas substantial risk if genes for poisonswere introduced into pathogens,suchasthecommoncold,thatspreadeasily.Afteraten-monthmoratoriumguidelineswereagreedtoattheAsilomarconference,which includedprovisions forphysicalandbiologicalcontainment,bansonparticular types of experiments, and other stipulations. These biotechnology guidelines have beenstrictlyfollowed,andtherehavenotbeenreportedaccidentsinthethirty-yearhistoryofthefield. More recently, the organization representing the world's organ transplantation surgeons hasadoptedamoratoriumonthetransplantationofvascularizedanimalorgansintohumans.Thiswasdoneoutoffearofthespreadoflong-dormantHIV-typexenovirusesfromanimalssuchaspigsorbaboonsinto thehumanpopulation.Unfortunately, such amoratoriumcan also slowdown the availabilityoflifesaving xenografts (genetically modified animal organs that are accepted by the human immunesystem)tothemillionsofpeoplewhodieeachyearfromheart,kidney,andliverdisease.GeoethicistMartine Rothblatt has proposed replacing this moratorium with a new set of ethical guidelines andregulations.44 In the case of nanotechnology, the ethics debate has started a couple of decades prior to theavailabilityof theparticularlydangerousapplications.Themost importantprovisionsof theForesightInstituteguidelinesinclude:

·"Artificialreplicatorsmustnotbecapableofreplicationinanatural,uncontrolledenvironment."·"Evolutionwithinthecontextofaself-replicatingmanufacturingsystemisdiscouraged."· "MNT device designs should specifically limit proliferation and provide traceability of anyreplicatingsystems."

·"Distributionofmolecularmanufacturingdevelopmentcapabilityshouldberestrictedwhenever

possible, to responsibleactors thathaveagreed touse theGuidelines.Nosuch restrictionneedapplytoendproductsofthedevelopmentprocess."

OtherstrategiesthattheForesightInstitutehasproposedinclude:

·Replicationshouldrequirematerialsnotfoundinthenaturalenvironment.· Manufacturing (replication) should be separated from the functionality of end products.Manufacturingdevicescancreateendproductsbutcannotreplicatethemselves,andendproductsshouldhavenoreplicationcapabilities.

·Replicationshouldrequirereplicationcodesthatareencryptedandtimelimited.Thebroadcastarchitecturementionedearlierisanexampleofthisrecommendation.

These guidelines and strategies are likely to be effective for preventing accidental release ofdangerousself-replicatingnanotechnologyentities.Butdealingwiththeintentionaldesignandreleaseofsuchentitiesisamorecomplexandchallengingproblem.Asufficientlydeterminedanddestructiveopponent could possibly defeat each of these layers of protections.Take, for example, the broadcastarchitecture. When properly designed, each entity is unable to replicate without first obtainingreplication codes, which are not repeated from one replication generation to the next. However, amodification to such adesign couldbypass thedestructionof the replication codes and therebypassthemontothenextgeneration.Tocounteractthatpossibilityithasbeenrecommendedthatthememoryforthereplicationcodesbelimitedtoonlyasubsetofthefullcode.However,thisguidelinecouldbedefeatedbyexpandingthesizeofthememory.Anotherprotectionthathasbeensuggestedistoencryptthecodesandbuildinprotectionsinthedecryptionsystems,suchastime-expirationlimitations.However,wecanseehoweasyithasbeentodefeatprotections againstunauthorized replicationsof intellectualproperty suchasmusic files.Oncereplication codes and protective layers are stripped away, the information can be replicatedwithouttheserestrictions.Thisdoesn'tmeanthatprotectionisimpossible.Rather,eachlevelofprotectionwillworkonlytoacertainlevelofsophistication.Themeta-lessonhere is thatwewillneedtoplacetwenty-first-centurysociety's highest priority on the continuing advance of defensive technologies, keeping them one ormorestepsaheadofthedestructivetechnologies(oratleastnomorethanaquickstepbehind).Protection from "Unfriendly" Strong AI. Even as effective a mechanism as the broadcastarchitecture,however,won'tserveasprotectionagainstabusesofstrongAI.Thebarriersprovidedbythe broadcast architecture rely on the lack of intelligence in nanoengineered entities. By definition,however,intelligententitieshavetheclevernesstoeasilyovercomesuchbarriers.EliezerYudkowskyhasextensivelyanalyzedparadigms,architectures,andethicalrulesthatmayhelpassure thatonce strongAIhas themeansof accessingandmodifying itsowndesign it remainsfriendlytobiologicalhumanityandsupportiveofitsvalues.Giventhatself-improvingstrongAIcannotberecalled,Yudkowskypointsoutthatweneedto"getitrightthefirsttime,"andthatitsinitialdesignmusthave"zerononrecoverableerrors."45InherentlytherewillbenoabsoluteprotectionagainststrongAI.AlthoughtheargumentissubtleIbelieve that maintaining an open free-market system for incremental scientific and technologicalprogress, in which each step is subject to market acceptance, will provide the most constructiveenvironmentfortechnologytoembodywidespreadhumanvalues.AsIhavepointedout,strongAIisemergingfrommanydiverseeffortsandwillbedeeplyintegratedintoourcivilization'sinfrastructure.Indeed, it will be intimately embedded in our bodies and brains. As such, it will reflect our valuesbecauseitwillbeus.Attemptstocontrolthesetechnologiesviasecretivegovernmentprograms,alongwith inevitable underground development, would only foster an unstable environment in which thedangerousapplicationswouldbelikelytobecomedominant.

DecentraIization.Oneprofoundtrendalreadywellunderwaythatwillprovidegreaterstabilityisthemovementfromcentralizedtechnologiestodistributedonesandfromtherealworldtothevirtualworlddiscussed above. Centralized technologies involve an aggregation of resources such as people (forexample, cities, buildings), energy (such as nuclear-power plants, liquid-natural-gas and oil tankers,energy pipelines), transportation (airplanes, trains), and other items. Centralized technologies aresubject to disruption and disaster. They also tend to be inefficient, wasteful, and harmful to theenvironment.Distributedtechnologies,ontheotherhand,tendtobeflexible,efficient,andrelativelybenignintheirenvironmentaleffects.Thequintessentialdistributed technology is the Internet.The Internethasnot been substantially disrupted to date, and as it continues to grow, its robustness and resiliencecontinuetostrengthen.Ifanyhuborchanneldoesgodown,informationsimplyroutesaroundit.DistributedEnergy.Inenergy,weneedtomoveawayfromtheextremelyconcentratedandcentralizedinstallations on which we now depend. For example, one company is pioneering fuel cells that aremicroscopic,usingMEMStechnology.46Theyaremanufactured likeelectronicchipsbutareactuallyenergy-storage devices with an energy-to-size ratio significantly exceeding that of conventionaltechnology.AsIdiscussedearlier,nanoengineeredsolarpanelswillbeabletomeetourenergyneedsina distributed, renewable, and clean fashion. Ultimately technology along these lines could powereverything from our cell phones to our cars and homes. These types of decentralized energytechnologieswouldnotbesubjecttodisasterordisruption.Asthesetechnologiesdevelop,ourneedforaggregatingpeopleinlargebuildingsandcitieswilldiminish,andpeoplewillspreadout,livingwheretheywantandgatheringtogetherinvirtualreality.CivilLibertiesinanAgeofAsymmetricWarfare.Thenatureofterroristattacksandthephilosophiesof the organizations behind them highlight how civil liberties can be at odds with legitimate stateinterests insurveillanceandcontrol.Our law-enforcementsystem—andindeed,muchofour thinkingaboutsecurity—isbasedon theassumption thatpeoplearemotivated topreserve theirown livesandwell-being.Thatlogicunderliesallourstrategies,fromprotectionat thelocalleveltomutualassureddestructionontheworldstage.Butafoethatvaluesthedestructionofbothitsenemyanditselfisnotamenabletothislineofreasoning. The implications of dealing with an enemy that does not value its own survival are deeplytroublesomeandhaveledtocontroversythatwillonlyintensifyasthestakescontinuetoescalate.Forexample,whentheFBIidentifiesalikelyterroristcell,itwillarresttheparticipants,eventhoughtheremaybeinsufficientevidencetoconvictthemofacrimeandtheymaynotyetevenhavecommittedacrime.Undertherulesofengagementinourwaronterrorism,thegovernmentcontinuestoholdtheseindividuals.Inaleadeditorial,theNewYorkTimesobjectedtothispolicy,whichitdescribedasa"troublingprovision."47Thepaperargued that thegovernmentshould release thesedetaineesbecause theyhavenotyetcommittedacrimeandshouldrearrestthemonlyaftertheyhavedoneso.Ofcoursebythattimesuspected terrorists might well be dead along with a large number of their victims. How can theauthoritiespossiblybreakupavastnetworkofdecentralizedcellsofsuicide terrorists if theyhavetowaitforeachonetocommitacrime? On theotherhand thisvery logichasbeen routinelyusedby tyrannical regimes to justify thewaivingofthejudicialprotectionswehavecometocherish.Itislikewisefairtoarguethatcurtailingcivillibertiesinthiswayisexactlytheaimoftheterrorists,whodespiseournotionsoffreedomsandpluralism.However,Idonotseetheprospectofanytechnology"magicbullet"thatwouldessentiallychangethisdilemma.Theencryptiontrapdoormaybeconsideredatechnicalinnovationthatthegovernmenthasbeenproposinginanattempttobalancelegitimateindividualneedsforprivacywiththegovernment'sneedforsurveillance.Alongwith this typeof technologywealsoneedtherequisitepolitical innovation toprovideforeffectiveoversight,byboththejudicialandlegislativebranches,oftheexecutivebranch's

useofthesetrapdoors,toavoidthepotentialforabuseofpower.Thesecretivenatureofouropponentsand their lack of respect for human life including their ownwill deeply test the foundations of ourdemocratictraditions.AProgramforGNRDefense

Wecomefromgoldfish,essentially,butthat[doesn't]meanweturnedaroundandkilledallthegoldfish.Maybe [theAIs]will feedusonceaweek....Ifyouhadamachinewitha10 to the18thpowerIQoverhumans,wouldn'tyouwantittogovern,oratleastcontrolyoureconomy?

—SETHSHOSTAKHowcanwesecure theprofoundbenefitsofGNRwhileameliorating itsperils?Here'sareviewofasuggestedprogramforcontainingtheGNRrisks:Themosturgentrecommendationistogreatlyincreaseourinvestmentindefensivetechnologies.SincewearealreadyintheGera,thebulkofthisinvestmenttodayshouldbein(biological)antiviralmedicationsandtreatments.Wehavenewtoolsthatarewellsuitedtothistask.RNAinterference,forexample,canbeusedtoblockgeneexpression.Virtuallyallinfections(aswellascancer)relyongeneexpressionatsomepointduringtheirlifecycles. Efforts toanticipate thedefensive technologiesneededtosafelyguideNandRshouldalsobesupported,and theseshouldbesubstantially increasedaswegetcloser to thefeasibilityofmolecularmanufacturingandstrongAI, respectively.Asignificantsidebenefitwouldbe toaccelerateeffectivetreatmentsforinfectiousdiseaseandcancer.I'vetestifiedbeforeCongressonthisissue,advocatingtheinvestmentoftensofbillionsofdollarsperyear(lessthan1percentoftheGDP)toaddressthisnewandunder-recognizedexistentialthreattohumanity."

·Weneedtostreamlinetheregulatoryprocessforgeneticandmedicaltechnologies.Theregulationsdonotimpedethemalevolentuseoftechnologybutsignificantlydelaytheneededdefenses.Asmentioned,weneedtobetterbalancetherisksofnewtechnology(forexample,newmedications)againsttheknownharmofdelay.

·Aglobal programof confidential, random serummonitoring for unknownor evolvingbiologicalpathogensshouldbefunded.Diagnostictoolsexist torapidlyidentifytheexistenceofunknownprotein or nucleic acid sequences. Intelligence is key to defense, and such a program couldprovideinvaluableearlywarningofanimpendingepidemic.Sucha"pathogensentinel"programhasbeenproposed formanyyearsbypublic health authorities but hasnever received adequatefunding.

·Well-defined and targeted temporarymoratoriums, such as the one that occurred in the geneticsfield in 1975, may be needed from time to time. But such moratoriums are unlikely to benecessarywith nanotechnology. Broad efforts at relinquishingmajor areas of technology serveonlytocontinuevasthumansufferingbydelayingthebeneficialaspectsofnewtechnologies,andactuallymakethedangersworse.

·Effortstodefinesafetyandethicalguidelinesfornanotechnologyshouldcontinue.Suchguidelineswillinevitablybecomemoredetailedandrefinedaswegetclosertomolecularmanufacturing.

·Tocreate thepolitical support to fund theefforts suggestedabove, it isnecessary toraise publicawarenessof thesedangers.Because,of course, there exists thedownsideof raisingalarmandgeneratinguninformedbackingforbroadantitechnologymandates,wealsoneedtocreateapublicunderstandingoftheprofoundbenefitsofcontinuingadvancesintechnology.

·These risks cut across international boundaries—which is, of course, nothing new; biologicalviruses,softwareviruses,andmissilesalreadycrosssuchboundarieswithimpunity.Internationalcooperation was vital to containing the SARS virus and will become increasingly vital inconfrontingfuturechallenges.Worldwideorganizationssuchas theWorldHealthOrganization,

whichhelpedcoordinatetheSARSresponse,needtobestrengthened.·Acontentiouscontemporarypoliticalissueistheneedforpreemptiveactiontocombatthreats,suchas terrorists with access to weapons of mass destruction or rogue nations that support suchterrorists.Suchmeasureswillalwaysbecontroversial,butthepotentialneedforthemisclear.Anuclearexplosioncandestroyacityinseconds.Aself-replicatingpathogen,whetherbiologicalornanotechnology based, could destroy our civilization in amatter of days orweeks.We cannotalwaysaffordtowaitforthemassingofarmiesorotherovertindicationsofillintentbeforetakingprotectiveaction.

·Intelligenceagenciesandpolicingauthoritieswillhaveavitalroleinforestallingthevastmajorityofpotentiallydangerous incidents.Theireffortsneedto involve themostpowerful technologiesavailable.Forexample,beforethisdecadeisover,devicesthesizeofdustparticleswillbeabletocarryout reconnaissancemissions.Whenwe reach the2020sandhave software running inourbodiesandbrains,governmentauthoritieswillhavealegitimateneedonoccasiontomonitorthesesoftwarestreams.Thepotential forabuseofsuchpowers isobvious.Wewillneedtoachieveamiddleroadofpreventingcatastrophiceventswhilepreservingourprivacyandliberty.

·TheaboveapproacheswillbeinadequatetodealwiththedangerfrompathologicalR(strongAI).Ourprimary strategy in this area shouldbe tooptimize the likelihood that futurenonbiologicalintelligencewillreflectourvaluesofliberty,tolerance,andrespectforknowledgeanddiversity.Thebestwaytoaccomplishthisistofosterthosevaluesinoursocietytodayandgoingforward.Ifthis sounds vague, it is. But there is no purely technical strategy that isworkable in this area,becausegreaterintelligencewillalwaysfindawaytocircumventmeasuresthataretheproductofalesserintelligence.Thenonbiologicalintelligencewearecreatingisandwillbeembeddedinoursocieties and will reflect our values. The transbiological phase will involve nonbiologicalintelligencedeeplyintegratedwithbiologicalintelligence.Thiswillamplifyourabilities,andourapplicationofthesegreaterintellectualpowerswillbegovernedbythevaluesofitscreators.Thetransbiologicalerawillultimatelygivewaytothepostbiologicalera,butitistobehopedthatourvalueswillremaininfluential.Thisstrategyiscertainlynotfoolproof,butitistheprimarymeanswehavetodaytoinfluencethefuturecourseofstrongAI.

Technology will remain a double-edged sword. It represents vast power to be used for allhumankind'spurposes.GNRwillprovidethemeanstoovercomeage-oldproblemssuchasillnessandpoverty, but it will also empower destructive ideologies. We have no choice but to strengthen ourdefenses while we apply these quickening technologies to advance our human values, despite anapparentlackofconsensusonwhatthosevaluesshouldbe.MOLLY2004:Okay,nowrun thatstealthyscenariobymeagain—youknow, theonewhere thebad

nanobots spreadquietly through thebiomass toget themselves intopositionbutdon't actuallyexpandtonoticeablydestroyanythinguntilthey'respreadaroundtheglobe.

RAY:Well,thenanobotswouldspreadatverylowconcentrations,sayonecarbonatomper1015inthebiomass,sotheywouldbeseededthroughoutthebiomass.Thus,thespeedofphysicalspreadofthedestructivenanobotswouldnotbealimitingfactorwhentheysubsequentlyreplicateinplace.If they skipped the stealth phase and expanded instead from a single point, the spreadingnanodiseasewouldbenoticed,andthespreadaroundtheworldwouldberelativelyslow.

MOLLY2004:Sohowarewegoingtoprotectourselvesfromthat?Bythetimetheystartphasetwo,we'vegotonlyaboutninetyminutes,ormuchlessifyouwanttoavoidenormousdamage.

RAY:Becauseof thenatureofexponentialgrowth, thebulkof thedamagegetsdone in the last fewminutes, but your point iswell taken.Under any scenario,wewon't have a chancewithout ananotechnologyimmunesystem.Obviously,wecan'twaituntilthebeginningofaninety-minutecycle of destruction to begin thinking about creating one. Such a system would be verycomparabletoourhumanimmunesystem.Howlongwouldabiologicalhumancirca2004lastwithoutone?

MOLLY2004:Notlong,Isuppose.Howdoesthisnano-immunesystempickupthesebadnanobotsifthey'reonlyoneinathousandtrillion?

RAY:Wehavethesameissuewithourbiological immunesystem.Detectionofevenasingleforeignproteintriggersrapidactionbybiologicalantibodyfactories,sotheimmunesystemis thereinforcebythetimeapathogenachievesanearcriticallevel.We'llneedasimilarcapabilityforthenanoimmunesystem.

CHARLESDARWIN:Nowtellme,dotheimmune-systemnanobotshavetheabilitytoreplicate?RAY:Theywouldneedtobeabletodothis;otherwisetheywouldnotbeabletokeeppacewiththe

replicatingpathogenicnanobots.Therehavebeenproposalstoseedthebiomasswithprotectiveimmune-system nanobots at a particular concentration, but as soon as the bad nanobotssignificantly exceeded this fixed concentration the immune system would lose. Robert Freitasproposes nonreplicating nanofactories able to turn out additional protective nanorobots whenneeded.Ithinkthisislikelytodealwiththreatsforawhile,butultimatelythedefensivesystemwill need the ability to replicate its immune capabilities in place to keep pace with emergingthreats.

CHARLES: So aren't the immune-system nanobots entirely equivalent to the phase one malevolentnanobots?Imeanseedingthebiomassisthefirstphaseofthestealthscenario.

RAY:Buttheimmune-systemnanobotsareprogrammedtoprotectus,notdestroyus.CHARLES:Iunderstandthatsoftwarecanbemodified.RAY:Hacked,youmean?CHARLES:Yes,exactly.Soiftheimmune-systemsoftwareismodifiedbyahackertosimplyturnonits

self-replicationabilitywithoutend—RAY:—yes,well,we'llhavetobecarefulaboutthat,won'twe?MOLLY2004:I'llsay.RAY: We have the same problem with our biological immune system. Our immune system is

comparablypowerful,andifitturnsonusthat'sanautoimmunedisease,whichcanbeinsidious.Butthere'sstillnoalternativetohavinganimmunesystem.

MOLLY2004:Soasoftwareviruscouldturnthenanobotimmunesystemintoastealthdestroyer?RAY:That'spossible. It's fair toconclude thatsoftwaresecurity isgoing tobe thedecisive issue for

many levels of the human-machine civilization. With everything becoming information,maintainingthesoftwareintegrityofourdefensivetechnologieswillbecritical tooursurvival.Even on an economic level, maintaining the business model that creates information will becriticaltoourwell-being.

MOLLY2004: This makes me feel rather helpless. I mean, with all these good and bad nanobotsbattlingitout,I'lljustbeahaplessbystander.

RAY:That'shardlyanewphenomenon.Howmuchinfluencedoyouhavein2004onthedispositionofthetensofthousandsofnuclearweaponsintheworld?

MOLLY2004:AtleastIhaveavoiceandavoteinelectionsthataffectforeign-policyissues.RAY:There'sno reason for that to change.Providing fora reliablenanotechnology immune system

willbeoneofthegreatpoliticalissuesofthe2020sand2030s.MOLLY2004:ThenwhataboutstrongAI?RAY:Thegoodnewsisthatitwillprotectusfrommalevolentnanotechnologybecauseitwillbesmart

enoughtoassistusinkeepingourdefensivetechnologiesaheadofthedestructiveones.NEDLUDD:Assumingit'sonourside.RAY:Indeed.

I

CHAPTERNINE

ResponsetoCritics

Thehumanmindlikesastrangeideaaslittleasthebodylikesastrangeproteinandresistsitwithasimilarenergy.

—W.I.BEVERIDGEIfa...scientistsaysthatsomethingispossibleheisalmostcertainlyright,butifhesaysthatitisimpossibleheisveryprobablywrong.

—ARTHURC.CLARKEAPanoplyofCriticisms

nTheAgeofSpiritualMachines, Ibegan toexamine someof theaccelerating trends that Ihavesoughttoexploreingreaterdepthinthisbook.ASMinspiredabroadvarietyofreactions,includingextensivediscussionsoftheprofound,imminentchangesitconsidered(forexample,thepromise-versus-peril debatepromptedbyBill Joy'sWired story, "Why theFutureDoesn'tNeedUs," as Ireviewedinthepreviouschapter).Theresponsealsoincludedattemptstoargueonmanylevelswhy

such transformative changeswould not, could not, or should not happen.Here is a summary of thecritiquesIwillberespondingtointhischapter:

·The "criticism fromMalthus": It's amistake to extrapolate exponential trends indefinitely, sincetheyinevitablyrunoutofresourcestomaintaintheexponentialgrowth.Moreover,wewon'thaveenoughenergytopowertheextraordinarilydensecomputationalplatformsforecast,andevenifwedidtheywouldbeashotasthesun.Exponentialtrendsdoreachanasymptote,butthematterandenergyresourcesneededforcomputationandcommunicationaresosmallpercomputeandperbit that thesetrendscancontinuetothepointwherenonbiological intelligenceis trillionsoftrillions of timesmore powerful than biological intelligence. Reversible computing can reduceenergyrequirements,aswellasheatdissipation,bymanyordersofmagnitude.Evenrestrictingcomputation to "cold" computers will achieve nonbiological computing platforms that vastlyoutperformbiologicalintelligence.

·The"criticismfromsoftware":We'remakingexponentialgainsinhardware,butsoftwareisstuckin the mud. Although the doubling time for progress in software is longer than that forcomputationalhardware,softwareisalsoacceleratingineffectiveness,efficiency,andcomplexity.Manysoftwareapplications, rangingfromsearchengines togames, routinelyuseAI techniquesthatwereonlyresearchprojectsadecadeago.Substantialgainshavealsobeenmadeintheoverallcomplexityofsoftware,insoftwareproductivity,andintheefficiencyofsoftwareinsolvingkeyalgorithmicproblems.Moreover,wehave an effective gameplan to achieve the capabilities ofhuman intelligence in a machine: reverse engineering the brain to capture its principles ofoperation and then implementing those principles in brain-capable computing platforms. Everyaspect of brain reverse engineering is accelerating: the spatial and temporal resolution of brain

scanning,knowledgeabouteverylevelofthebrain'soperation,andeffortstorealisticallymodelandsimulateneuronsandbrainregions.

·The"criticism fromanalogprocessing":Digital computation is too rigidbecausedigitalbitsareeitheronoroff.Biologicalintelligenceismostlyanalog,sosubtlegradationscanbeconsidered.It's true that the humanbrain uses digital-controlled analogmethods, butwe can also use suchmethodsinourmachines.Moreover,digitalcomputationcansimulateanalogtransactionstoanydesiredlevelofaccuracy,whereastheconversestatementisnottrue.

·The "criticism from the complexity of neural processing": The information processes in theinterneuronalconnections(axons,dendrites,synapses)arefarmorecomplexthanthesimplisticmodelsusedinneuralnets.True,butbrain-regionsimulationsdon'tusesuchsimplifiedmodels.We have achieved realistic mathematical models and computer simulations of neurons andinterneuronal connections that do capture the nonlinearities and intricacies of their biologicalcounterparts.Moreover,wehave found that thecomplexityofprocessingbrain regions isoftensimpler than the neurons they comprise.We already have effectivemodels and simulations forseveraldozenregionsofthehumanbrain.Thegenomecontainsonlyaboutthirtytoonehundredmillionbytesofdesigninformationwhenredundancyisconsidered,sothedesigninformationforthebrainisofamanageablelevel.

·The "criticism from microtubules and quantum computing": The microtubules in neurons arecapableofquantumcomputing,andsuchquantumcomputingisaprerequisiteforconsciousness.To "upload" a personality, one would have to capture its precise quantum state.No evidenceexists to support either of these statements. Even if true, there is nothing that bars quantumcomputingfrombeingcarriedoutinnonbiologicalsystems.Weroutinelyusequantumeffectsinsemiconductors(tunnelingintransistors,forexample),andmachine-basedquantumcomputingisalsoprogressing.Asforcapturingaprecisequantumstate,I'minaverydifferentquantumstatethanIwasbeforewritingthissentence.SoamIalreadyadifferentperson?PerhapsIam,butifonecapturedmystateaminuteago,anuploadbasedonthatinformationwouldstillsuccessfullypassa"RayKurzweil"Turingtest.

·The "criticism from the Church-Turing thesis":We can show that there are broad classes ofproblemsthatcannotbesolvedbyanyTuringmachine.ItcanalsobeshownthatTuringmachinescan emulate any possible computer (that is, there exists a Turing machine that can solve anyproblemthatanycomputercansolve),sothisdemonstratesaclearlimitationontheproblemsthatacomputercansolve.Yethumansarecapableofsolvingtheseproblems,somachineswillneveremulate human intelligence. Humans are no more capable of universally solving such"unsolvable"problemsthanmachines.Humanscanmakeeducatedguessestosolutionsincertaininstances,butmachinescandothesamethingandcanoftendosomorequickly.

·The "criticism from failure rates":Computer systems are showing alarming rates of catastrophicfailureastheircomplexityincreases.ThomasRaywritesthatweare"pushingthelimitsofwhatwe can effectively design and build through conventional approaches." We have developedincreasinglycomplexsystemstomanageabroadvarietyofmission-criticaltasks,andfailureratesin these systems are very low. However, imperfection is an inherent feature of any complexprocess,andthatcertainlyincludeshumanintelligence.

·The "criticism from 'lock-in' ": The pervasive and complex support systems (and the hugeinvestments in thesesystems)requiredbysuch fieldsasenergyand transportationareblockinginnovation, so this will prevent the kind of rapid change envisioned for the technologiesunderlyingtheSingularity.Itisspecificallyinformationprocessesthataregrowingexponentiallyincapabilityandprice-performance.Wehavealreadyseenrapidparadigmshiftsineveryaspectof information technology, unimpededby any lock-in phenomenon (despite large infrastructureinvestments in such areas as the Internet and telecommunications). Even the energy andtransportation sectors will witness revolutionary changes from new nanotechnology-basedinnovations.

·The "criticism from ontology": John Searle describes several versions of his Chinese Roomanalogy.InoneformulationamanfollowsawrittenprogramtoanswerquestionsinChinese.The

manappearstobeansweringquestionscompetentlyinChinese,butsinceheis"justmechanicallyfollowingawrittenprogram,hehasnorealunderstandingofChineseandnorealawarenessofwhathe isdoing.The"man"in theroomdoesn'tunderstandanything,because,afterall,"he isjust a computer," according to Searle. So clearly, computers cannot understandwhat they aredoing, since they are just following rules. Searle'sChineseRoomarguments are fundamentallytautological, as they just assume his conclusion that computers cannot possibly have any realunderstanding.PartofthephilosophicalsleightofhandinSearle'ssimpleanalogiesisamatterofscale.Hepurports todescribeasimplesystemand thenasks the reader toconsiderhowsuchasystemcouldpossiblyhaveanyrealunderstanding.Butthecharacterizationitself ismisleading.TobeconsistentwithSearle'sownassumptions theChineseRoomsystemthatSearledescribeswouldhavetobeascomplexasahumanbrainandwould,therefore,haveasmuchunderstandingasahumanbrain.Themanintheanalogywouldbeactingasthecentral-processingunit,onlyasmallpartofthesystem.Whilethemanmaynotseeit,theunderstandingisdistributedacrosstheentirepatternoftheprogramitselfandthebillionsofnoteshewouldhavetomaketofollowtheprogram.Consider that I understandEnglish, but noneofmyneuronsdo.Myunderstanding isrepresented in vast patterns of neurotransmitter strengths, synaptic clefts, and interneuronalconnections.

·The"criticismfromtherich-poordivide":It's likely that through these technologies therichmayobtaincertainopportunitiesthattherestofhumankinddoesnothaveaccessto.This,ofcourse,wouldbenothingnew,butIwouldpointoutthatbecauseoftheongoingexponentialgrowthofprice-performance,allofthesetechnologiesquicklybecomesoinexpensiveastobecomealmostfree.

·The"criticismfromthe likelihoodofgovernmentregulation":Governmental regulationwill slowdownandstoptheaccelerationoftechnology.Althoughtheobstructivepotentialofregulationisanimportantconcern,ithashadasofyetlittlemeasurableeffectonthetrendsdiscussedinthisbook.Absentaworldwide totalitarianstate, theeconomicandother forcesunderlying technicalprogress will only grow with ongoing advances. Even controversial issues such as stem-cellresearchendupbeinglikestonesinastream,theflowofprogressrushingaroundthem.

·The"criticismfromtheism":AccordingtoWilliamA.Dembski,"contemporarymaterialistssuchasRayKurzweil ...see themotionsandmodificationsofmatterassufficient toaccount forhumanmentality."Butmaterialismispredictable,whereasrealityisnot.Predictability[is]materialism'smainvirtue...andhollowness[is]itsmainfault."Complexsystemsofmatterandenergyarenotpredictable,sincetheyarebasedonavastnumberofunpredictablequantumevents.Evenifweaccepta"hiddenvariables"interpretationofquantummechanics(whichsaysthatquantumeventsonlyappeartobeunpredictablebutarebasedonundetectablehiddenvariables),thebehaviorofacomplexsystemwouldstillbeunpredictableinpractice.Allofthetrendsshowthatweareclearlyheadedfornonbiologicalsystemsthatareascomplexastheirbiologicalcounterparts.Suchfuturesystemswillbenomore"hollow" thanhumansand inmanycaseswillbebasedon thereverseengineering of human intelligence.We don't need to go beyond the capabilities of patterns ofmatterandenergytoaccountforthecapabilitiesofhumanintelligence.

·The "criticism from holism":To quoteMichael Denton, organisms are "self-organizing, ... self-referential,...self-replicating,...reciprocal,...self-formative,and...holistic."Suchorganicformscan be created only through biological processes, and such forms are "immutable, ...impenetrable,and...fundamentalrealitiesofexistence."1It'struethatbiologicaldesignrepresentsa profound set of principles. However, machines can use—and already are using—these sameprinciples,andthereisnothingthatrestrictsnonbiologicalsystemsfromharnessingtheemergentpropertiesofthepatternsfoundinthebiologicalworld.

I'veengagedincountlessdebatesanddialoguesrespondingtothesechallengesinadiversevarietyofforums.OneofmygoalsforthisbookistoprovideacomprehensiveresponsetothemostimportantcriticismsIhaveencountered.Mostofmyrejoinders to thesecritiquesonfeasibilityand inevitability

havebeendiscussedthroughoutthisbook,butinthischapterIwanttoofferadetailedreplytoseveralofthemoreinterestingones.TheCriticismfromIncredulityPerhaps themost candid criticism of the future I have envisioned here is simple disbelief that suchprofoundchangescouldpossiblyoccur.ChemistRichardSmalley,forexample,dismisses the ideaofnanobotsbeingcapableofperformingmissionsinthehumanbloodstreamasjust"silly."Butscientists'ethics call for caution in assessing the prospects for current work, and such reasonable prudenceunfortunatelyoftenleadsscientists toshyawayfromconsideringthepowerofgenerationsofscienceandtechnologyfarbeyondtoday'sfrontier.Withtherateofparadigmshiftoccurringevermorequickly,thisingrainedpessimismdoesnotservesociety'sneedsinassessingscientificcapabilitiesinthedecadesahead.Considerhowincredibletoday'stechnologywouldseemtopeopleevenacenturyago.Arelatedcriticismisbasedonthenotionthatitisdifficulttopredictthefuture,andanynumberofbad predictions from other futurists in earlier eras can be cited to support this. Predicting whichcompanyorproductwillsucceedisindeedverydifficult,ifnotimpossible.Thesamedifficultyoccursin predicting which technical design or standard will prevail. (For example, how will the wireless-communicationprotocolsWiMAX,CDMA,and3Gfareoverthenextseveralyears?)However,asthisbook has extensively argued, we find remarkably precise and predictable exponential trends whenassessingtheoveralleffectiveness(asmeasuredbyprice-performance,bandwidth,andothermeasuresof capability)of information technologies.For example, the smoothexponentialgrowthof theprice-performanceofcomputingdatesbackoveracentury.Given that theminimumamountofmatterandenergyrequiredtocomputeortransmitabitofinformationisknowntobevanishinglysmall,wecanconfidently predict the continuation of these information-technology trends at least through this nextcentury.Moreover,wecanreliablypredictthecapabilitiesofthesetechnologiesatfuturepointsintime. Consider that predicting the path of a singlemolecule in a gas is essentially impossible, butpredicting certain properties of the entire gas (composed of a great many chaotically interactingmolecules) can reliably be predicted through the laws of thermodynamics. Analogously, it is notpossible toreliablypredict theresultsofaspecificprojectorcompany,but theoverallcapabilitiesofinformation technology (comprised of many chaotic activities) can nonetheless be dependablyanticipatedthroughthelawofacceleratingreturns. Many of the furious attempts to argue why machines—nonbiological systems—cannot everpossibly compare tohumans appear tobe fueledby thisbasic reactionof incredulity.Thehistoryofhumanthoughtismarkedbymanyattemptstorefusetoacceptideasthatseemtothreatentheacceptedviewthatourspeciesisspecial.Copernicus'sinsightthattheEarthwasnotatthecenteroftheuniversewasresisted,aswasDarwin'sthatwewereonlyslightlyevolvedfromotherprimates.Thenotionthatmachines couldmatch and even exceed human intelligence appears to challenge human status onceagain.Inmyviewthereissomethingessentiallyspecial,afterall,abouthumanbeings.WewerethefirstspeciesonEarthtocombineacognitivefunctionandaneffectiveopposableappendage(thethumb),sowewereabletocreatetechnologythatwouldextendourownhorizons.NootherspeciesonEarthhasaccomplished this. (To be precise,we're the only surviving species in this ecological niche—others,suchastheNeanderthals,didnotsurvive.)AndasIdiscussedinchapter6,wehaveyettodiscoveranyothersuchcivilizationintheuniverse.TheCriticismfromMalthusExponentialTrendsDon'tLastForever. The classicalmetaphorical example of exponential trendshittingawallisknownas"rabbitsinAustralia."Aspecieshappeninguponahospitablenewhabitatwill

expand itsnumbersexponentiallyuntil itsgrowthhits the limitsof theabilityof thatenvironment tosupport it. Approaching this limit to exponential growth may even cause an overall reduction innumbers—forexample,humansnoticinga spreadingpestmay seek toeradicate it.Another commonexampleisamicrobethatmaygrowexponentiallyinananimalbodyuntilalimitisreached:theabilityofthatbodytosupportit,theresponseofitsimmunesystem,orthedeathofthehost.Eventhehumanpopulationisnowapproachingalimit.Familiesinthemoredevelopednationshavemasteredmeansofbirthcontrolandhavesetrelativelyhighstandardsfortheresourcestheywishtoprovidetheirchildren.Asaresultpopulationexpansioninthedevelopedworldhaslargelystopped.Meanwhilepeopleinsome(butnotall)underdevelopedcountrieshavecontinuedtoseeklargefamiliesasameansofsocialsecurity,hopingthatatleastonechildwillsurvivelongenoughtosupporttheminoldage.However,withthelawofacceleratingreturnsprovidingmorewidespreadeconomicgains,theoverallgrowthinhumanpopulationisslowing.Soisn'tthereacomparablelimittotheexponentialtrendsthatwearewitnessingforinformationtechnologies?Theanswerisyes,butnotbeforetheprofoundtransformationsdescribedthroughoutthisbooktakeplace.AsIdiscussedinchapter3,theamountofmatterandenergyrequiredtocomputeortransmitonebitisvanishinglysmall.Byusingreversiblelogicgates,theinputofenergyisrequiredonlytotransmitresults and to correct errors. Otherwise, the heat released from each computation is immediatelyrecycledtofuelthenextcomputation. As I discussed in chapter 5, nanotechnology-based designs for virtually all applications—computation,communication,manufacturing,andtransportation—willrequiresubstantiallylessenergythan they do today. Nanotechnologywill also facilitate capturing renewable energy sources such assunlight.We couldmeet all of our projected energy needs of thirty trillionwatts in 2030with solarpowerifwecapturedonly0.03percent(three ten-thousandths)of thesun'senergyas ithit theEarth.Thiswillbefeasiblewithextremelyinexpensive,lightweight,andefficientnanoengineeredsolarpanelstogetherwithnano-fuelcellstostoreanddistributethecapturedenergy.AVirtuallyUnlimitedLimit.AsIdiscussedinchapter3anoptimallyorganized2.2-poundcomputerusing reversible logic gates has about 1025 atoms and can store about 1027 bits. Just consideringelectromagnetic interactions between the particles, there are at least 1015 state changes per bit persecond that can be harnessed for computation, resulting in about 1042 calculations per second in theultimate"cold"2.2-poundcomputer.Thisisabout1016timesmorepowerfulthanallbiologicalbrainstoday.Ifweallowourultimatecomputertogethot,wecanincreasethisfurtherbyasmuchas108-fold.And we obviously won't restrict our computational resources to one kilogram of matter but willultimatelydeployasignificantfractionofthematterandenergyontheEarthandinthesolarsystemandthenspreadoutfromthere.Specificparadigmsdoreachlimits.WeexpectthatMoore'sLaw(concerningtheshrinkingofthesizeoftransistorsonaflatintegratedcircuit)willhitalimitoverthenexttwodecades.ThedateforthedemiseofMoore'sLawkeepsgettingpushedback.The first estimatespredicted2002,butnowIntelsays itwon't takeplaceuntil 2022.But as I discussed in chapter 2, every time a specific computingparadigmwas seen to approach its limit, research interest and pressure increased to create the nextparadigm.Thishasalreadyhappened four times in thecentury-longhistoryofexponentialgrowth incomputation (from electromagnetic calculators to relay-based computers to vacuum tubes to discretetransistorstointegratedcircuits).Wehavealreadyachievedmanyimportantmilestonestowardthenext(sixth)paradigmofcomputing:three-dimensionalself-organizingcircuitsatthemolecularlevel.Sotheimpendingendofagivenparadigmdoesnotrepresentatruelimit.Therearelimitstothepowerofinformationtechnology,buttheselimitsarevast.Iestimatedthecapacityofthematterandenergyinoursolarsystemtosupportcomputationtobeatleast1070cps(seechapter6).Giventhatthereareatleast1020starsintheuniverse,wegetabout1090cpsfor it,whichmatchesSethLloyd'sindependentanalysis.Soyes,therearelimits,butthey'renotverylimiting.

TheCriticismfromSoftwareA common challenge to the feasibility of strong AI, and therefore the Singularity, begins bydistinguishing between quantitative and qualitative trends. This argument acknowledges, in essence,that certain brute-force capabilities such as memory capacity, processor speed, and communicationsbandwidths are expanding exponentially but maintains that the software (that is, the methods andalgorithms)arenot.Thisisthehardware-versus-softwarechallenge,anditisasignificantone.Virtual-realitypioneerJaronLanier,forexample,characterizesmypositionandthatofotherso-calledcybernetictotalistsas,we'lljustfigureoutthesoftwareinsomeunspecifiedway—apositionhereferstoasasoftware"deusexmachina,"2Thisignores,however,thespecificanddetailedscenariothatI'vedescribedbywhichthesoftwareofintelligencewillbeachieved.Thereverseengineeringofthehumanbrain,anundertakingthatismuchfurtheralongthanLanierandmanyotherobserversrealize,willexpandourAItoolkittoincludetheself-organizingmethodsunderlyinghumanintelligence.I'llreturntothistopicinamoment,butfirstlet'saddresssomeotherbasicmisconceptionsabouttheso-calledlackofprogressinsoftware.SoftwareStability.Laniercallssoftwareinherently"unwieldy"and"brittle"andhasdescribedatgreatlengthavarietyoffrustrationsthathehasencounteredinusingit.Hewritesthat"gettingcomputerstoperform specific tasks of significant complexity in a reliable butmodifiableway,without crashes orsecuritybreaches,isessentiallyimpossible."3 It isnotmyintentiontodefendallsoftware,but it'snottruethatcomplexsoftwareisnecessarilybrittleandpronetocatastrophicbreakdown.Manyexamplesof complex mission-critical software operate with very few, if any, breakdowns: for example, thesophisticated software programs that control an increasing percentage of airplane landings, monitorpatientsincritical-carefacilities,guideintelligentweapons,controltheinvestmentofbillionsofdollarsinautomatedpatternrecognition-basedhedgefunds,andservemanyotherfunctions.4Iamnotawareofany airplane crashes that have been caused by failures of automated landing software; the same,however,cannotbesaidforhumanreliability.Software Responsiveness. Lanier complains that "computer user interfaces tend to respond moreslowly touser interfaceevents, suchas akeypress, than theydid fifteenyears earlier....What'sgonewrong?"5 I would invite Lanier to attempt using an old computer today. Even if we put aside thedifficulty of setting one up (which is a different issue), he has forgotten just how unresponsive,unwieldy, and limited theywere.Trygetting some realworkdone to today's standardswith twenty-year-oldpersonal-computersoftware.It'ssimplynottruetosaythattheoldsoftwarewasbetterinanyqualitativeorquantitativesense.Althoughit'salwayspossibletofindpoor-qualitydesign,responsedelays,whentheyoccur,aregenerally the resultofnewfeaturesandfunctions. Ifuserswerewilling to freeze the functionalityoftheir software, the ongoing exponential growth of computing speed and memory would quicklyeliminate software-response delays.But themarket demands ever-expanded capability.Twenty yearsagotherewerenosearchenginesoranyotherintegrationwiththeWorldWideWeb(indeed,therewasnoWeb),onlyprimitivelanguage,formatting,andmultimediatools,andsoon.Sofunctionalityalwaysstaysontheedgeofwhat'sfeasible.Thisromancingofsoftwarefromyearsordecadesagoiscomparabletopeople'sidyllicviewoflife hundreds of years ago, when people were "unencumbered" by the frustrations of working withmachines.Lifewasunfettered,perhaps,butitwasalsoshort,labor-intensive,povertyfilled,anddiseaseanddisasterprone.SoftwarePrice-Performance.With regard to the price-performance of software, the comparisons in

every area are dramatic. Consider the table on p. 103 on speech-recognition software. In 1985 fivethousand dollars bought you a software package that provided a thousand-word vocabulary, did notoffer continuous-speechcapability, required threehoursof trainingonyourvoice, andhad relativelypooraccuracy.In2000foronlyfiftydollars,youcouldpurchaseasoftwarepackagewithahundred-thousand-word vocabulary that provided continuous-speech capability, required only fiveminutes oftraining on your voice, had dramatically improved accuracy, offered natural-language understanding(foreditingcommandsandotherpurposes),andincludedmanyotherfeatures.6SoftwareDevelopmentProductivity.How about software development itself? I've been developingsoftwaremyselfforfortyyears,soIhavesomeperspectiveonthetopic.Iestimatethedoublingtimeofsoftware development productivity to be approximately six years,which is slower than the doublingtime for processor price-performance, which is approximately one year today. However, softwareproductivityisnonethelessgrowingexponentially.Thedevelopment tools,class libraries,andsupportsystems available today are dramatically more effective than those of decades ago. In my currentprojects teamsof just threeor fourpeopleachieve inafewmonthsobjectives thatarecomparable towhattwenty-fiveyearsagorequiredateamofadozenormorepeopleworkingforayearormore.SoftwareComplexity.Twentyyearsagosoftwareprogramstypicallyconsistedofthousandstotensofthousands of lines. Today, mainstream programs (for example, supply-channel control, factoryautomation, reservation systems, biochemical simulation) aremeasured inmillions of lines ormore.SoftwareformajordefensesystemssuchastheJointStrikeFightercontainstensofmillionsoflines. Software to control software is itself rapidly increasing in complexity. IBM is pioneering theconcept of autonomic computing, inwhich routine information-technology support functionswill beautomated.7Thesesystemswillbeprogrammedwithmodelsoftheirownbehaviorandwillbecapable,according to IBM, of being "self-configuring, self-healing, self-optimizing, and self-protecting." Thesoftware tosupportautonomiccomputingwillbemeasured in tensofmillionsof linesofcode (witheach line containing tens of bytes of information). So in terms of information complexity, softwarealready exceeds the tens of millions of bytes of usable information in the human genome and itssupportingmolecules. The amount of information contained in a program, however, is not the best measure ofcomplexity.Asoftwareprogrammaybelongbutmaybebloatedwithuselessinformation.Ofcourse,thesamecanbesaidforthegenome,whichappearstobeveryinefficientlycoded.Attemptshavebeenmade to formulatemeasuresofsoftwarecomplexity-forexample, theCyclomaticComplexityMetric,developed by computer scientists Arthur Watson and Thomas McCabe at the National Institute ofStandards and Technology.8 This metric measures the complexity of program logic and takes intoaccountthestructureofbranchinganddecisionpoints.Theanecdotalevidencestronglysuggestsrapidlyincreasingcomplexityifmeasuredbytheseindexes,althoughthereisinsufficientdatatotrackdoublingtimes.However,thekeypointisthatthemostcomplexsoftwaresystemsinuseinindustrytodayhavehigher levels of complexity than software programs that are performing neuromorphic-basedsimulationsofbrainregions,aswellasbiochemicalsimulationsofindividualneurons.Wecanalreadyhandle levels of software complexity that exceedwhat is needed tomodel and simulate the parallel,self-organizing,fractalalgorithmsthatwearediscoveringinthehumanbrain.AcceleratingAlgorithms. Dramatic improvements have taken place in the speed and efficiency ofsoftware algorithms (on constant hardware). Thus the price-performance of implementing a broadvarietyofmethodstosolvethebasicmathematicalfunctionsthatunderlieprogramslikethoseusedinsignalprocessing,patternrecognition,andartificialintelligencehasbenefitedfromtheaccelerationofbothhardwareandsoftware.Theseimprovementsvarydependingontheproblem,butarenonethelesspervasive. For example, consider the processing of signals, which is awidespread and computationallyintensivetaskforcomputersaswellasforthehumanbrain.GeorgiaInstituteofTechnology'sMarkA.

Richards andMIT'sGaryA. Shaw have documented a broad trend toward greater signal-processingalgorithmefficiency.9 For example, to findpatterns in signals it is oftennecessary to solvewhat arecalledpartialdifferentialequations.AlgorithmsexpertJonBentleyhasshownacontinualreductioninthenumberofcomputingoperationsrequiredtosolvethisclassofproblem.10Forexample,from1945to 1985, for a representative application (finding an elliptic partial differential solution for a three-dimensional grid with sixty-four elements on each side), the number of operation counts has beenreducedbyafactorofthreehundredthousand.Thisisa38percentincreaseinefficiencyeachyear(notincludinghardwareimprovements). Another example is the ability to send information on unconditioned phone lines,which hasimprovedfrom300bitspersecondto56,000bpsintwelveyears,a55percentannualincrease.11Someofthisimprovementwastheresultofimprovementsinhardwaredesign,butmostofitisafunctionofalgorithmicinnovation. OneofthekeyprocessingproblemsisconvertingasignalintoitsfrequencycomponentsusingFouriertransforms,whichexpresssignalsassumsofsinewaves.Thismethodisusedinthefrontendofcomputerizedspeechrecognitionandinmanyotherapplications.Humanauditoryperceptionalsostartsbybreaking the speech signal into frequency components in the cochlea.The1965 "radix-2Cooley-Tukeyalgorithm"fora"fastFouriertransform"reducedthenumberofoperationsrequiredfora1,024-point Fourier transform by about two hundred.12An improved "radix-a"method further boosted theimprovementtoeighthundred.Recently"wavelet"transformshavebeenintroduced,whichareabletoexpressarbitrarysignalsassumsofwaveformsmorecomplexthansinewaves.Thesemethodsprovidefurtherdramaticincreasesintheefficiencyofbreakingdownasignalintoitskeycomponents. Theexamplesabovearenotanomalies;mostcomputationallyintensive"core"algorithmshaveundergonesignificantreductionsinthenumberofoperationsrequired.Otherexamplesincludesorting,searching, autocorrelation (and other statistical methods), and information compression anddecompression. Progress has also been made in parallelizing algorithms—that is, breaking a singlemethod intomultiplemethods that can be performed simultaneously.As I discussed earlier, parallelprocessing inherently runs at a lower temperature.Thebrainusesmassiveparallelprocessingasonestrategytoachievemorecomplexfunctionsandfasterreactiontimes,andwewillneedtoutilize thisapproachinourmachinestoachieveoptimalcomputationaldensities. There is an inherentdifferencebetween the improvements inhardwareprice-performanceandimprovements in softwareefficiencies.Hardware improvementshavebeen remarkablyconsistentandpredictable.Aswemastereachnewlevelofspeedandefficiencyinhardwarewegainpowerfultoolstocontinuetothenextlevelofexponentialimprovement.Softwareimprovements,ontheotherhand,arelesspredictable.RichardsandShawcallthem"worm-holesindevelopmenttime,"becausewecanoftenachieve theequivalentofyearsofhardware improvement througha singlealgorithmic improvement.Notethatwedonotrelyonongoingprogressinsoftwareefficiency,sincewecancountontheongoingacceleration of hardware. Nonetheless, the benefits from algorithmic breakthroughs contributesignificantlytoachievingtheoverallcomputationalpowertoemulatehumanintelligence,andtheyarelikelytocontinuetoaccrue.The Ultimate Source of Intelligent Algorithms. The most important point here is that there is aspecificgameplanforachievinghuman-level intelligenceinamachine:reverseengineertheparallel,chaotic, self-organizing, and fractal methods used in the human brain and apply these methods tomodern computational hardware. Having tracked the exponentially increasing knowledge about thehuman brain and itsmethods (see chapter 4), we can expect that within twenty years wewill havedetailedmodelsandsimulationsoftheseveralhundredinformation-processingorganswecollectivelycallthehumanbrain. Understanding the principles of operationof human intelligencewill add to our toolkit ofAIalgorithms.Manyofthesemethodsusedextensivelyinourmachinepattern-recognitionsystemsexhibitsubtleandcomplexbehaviorsthatarenotpredictablebythedesigner.Self-organizingmethodsarenotaneasyshortcuttothecreationofcomplexandintelligentbehavior,buttheyareoneimportantwaythe

complexity of a system can be increased without incurring the brittleness of explicitly programmedlogicalsystems. AsIdiscussedearlier, thehumanbrain itself iscreatedfromagenomewithonly thirty toonehundred million bytes of useful, compressed information. How is it, then, that an organ with onehundred trillion connections can result from a genome that is so small? (I estimate that just theinterconnectiondataaloneneededtocharacterizethehumanbrainisonemilliontimesgreaterthantheinformationinthegenome.)13Theansweristhatthegenomespecifiesasetofprocesses,eachofwhichutilizeschaoticmethods (that is, initial randomness, thenself-organization) to increase theamountofinformationrepresented.Itisknown,forexample,thatthewiringoftheinterconnectionsfollowsaplanthatincludesagreatdealofrandomness.Asanindividualencountershisenvironmenttheconnectionsandtheneurotransmitter-levelpatternsself-organizetobetterrepresenttheworld,buttheinitialdesignisspecifiedbyaprogramthatisnotextremeinitscomplexity.Itisnotmypositionthatwewillprogramhumanintelligencelinkbylinkinamassiverule-basedexpert system.Nordoweexpect thebroadsetofskills representedbyhuman intelligence toemergefromamassivegeneticalgorithm.Lanierworriescorrectlythatanysuchapproachwouldinevitablygetstuckinsomelocalminima(adesignthatisbetterthandesignsthatareverysimilartoitbutthatisnotactually optimal). Lanier also interestingly points out, as does Richard Dawkins, that biologicalevolution "missed thewheel" (in thatnoorganismevolved tohaveone).Actually, that's not entirelyaccurate-there are smallwheel-like structures at theprotein level, for example the ionicmotor in thebacterialflagellum,whichisusedfortransportationinathree-dimensionalenvironment.14Withlargerorganisms,wheelsarenotveryuseful,ofcourse,withoutroads,whichiswhytherearenobiologicallyevolvedwheelsfortwo-dimensionalsurfacetransportation.15However,evolutiondidgenerateaspeciesthatcreatedbothwheelsandroads,soitdidsucceedincreatingalotofwheels,albeitindirectly.Thereisnothingwrongwithindirectmethods;weusetheminengineeringallthetime.Indeed,indirectionishowevolutionworks(thatis,theproductsofeachstagecreatethenextstage). Brain reverse engineering is not limited to replicating eachneuron. In chapter 5we sawhowsubstantialbrainregionscontainingmillionsorbillionsofneuronscouldbemodeledbyimplementingparallel algorithms that are functionally equivalent. The feasibility of such neuromorphic approacheshas been demonstratedwithmodels and simulations of a couple dozen regions.As I discussed, thisoften results in substantially reduced computational requirements, as shown by LloydWatts, CarverMead,andothers.Lanierwritesthat"ifthereeverwasacomplex,chaoticphenomenon,weareit."Iagreewiththatbutdon't see this asanobstacle.Myownareaof interest is chaoticcomputing,which ishowwedopattern recognition,which in turn is the heart of human intelligence.Chaos is part of the process ofpatternrecognition—itdrivestheprocess—andthereisnoreasonthatwecannotharnessthesemethodsinourmachinesjustastheyareutilizedinourbrains.Lanierwritesthat"evolutionhasevolved,introducingsex,forinstance,butevolutionhasneverfound away to be any speed but very slow."ButLanier's comment is only applicable to biologicalevolution,nottechnologicalevolution.That'spreciselywhywe'vemovedbeyondbiologicalevolution.Lanier is ignoring the essential nature of an evolutionary process: it accelerates because each stageintroducesmorepowerfulmethodsforcreatingthenextstage.We'vegonefrombillionsofyearsforthefirststepsofbiologicalevolution(RNA)tothefastpaceoftechnologicalevolutiontoday.TheWorldWideWeb emerged in only a few years, distinctly faster than, say, the Cambrian explosion. Thesephenomenaareallpartofthesameevolutionaryprocess,whichstartedoutslow,isnowgoingrelativelyquickly,andwithinafewdecadeswillgoastonishinglyfast. Lanier writes that "the whole enterprise of Artificial Intelligence is based on an intellectualmistake."Untilsuchtimethatcomputersatleastmatchhumanintelligenceineverydimension,itwillalwaysremainpossibleforskepticstosaytheglassishalfempty.EverynewachievementofAIcanbedismissedbypointingoutothergoalsthathavenotyetbeenaccomplished.Indeed,thisisthefrustrationoftheAIpractitioner:onceanAIgoalisachieved,itisnolongerconsideredasfallingwithintherealmof AI and becomes instead just a useful general technique. AI is thus often regarded as the set of

problemsthathavenotyetbeensolved.Butmachinesareindeedgrowinginintelligence,andtherangeoftasksthattheycanaccomplish—tasks that previously required intelligent human attention—is rapidly increasing.Aswe discussed inchapters5and6therearehundredsofexamplesofoperationalnarrowAItoday. Asoneexampleofmany, Ipointedout in thesidebar"DeepFritzDraws"onpp.274–78 thatcomputerchesssoftwarenolongerreliesjustoncomputationalbruteforce.In2002DeepFritz,runningon just eight personal computers, performed as well as IBM's Deep Blue in 1997 based onimprovementsinitspattern-recognitionalgorithms.Weseemanyexamplesofthiskindofqualitativeimprovement in software intelligence. However, until such time as the entire range of humanintellectualcapabilityisemulated,itwillalwaysbepossibletominimizewhatmachinesarecapableofdoing. Oncewehave achieved completemodels of human intelligence,machineswill be capable ofcombining the flexible, subtle human levels of pattern recognition with the natural advantages ofmachine intelligence, in speed, memory capacity, and, most important, the ability to quickly shareknowledgeandskills.TheCriticismfromAnalogProcessingManycritics,suchasthezoologistandevolutionary-algorithmscientistThomasRay,chargetheoristslikemewhopostulateintelligentcomputerswithanalleged"failuretoconsidertheuniquenatureofthedigitalmedium."16Firstofall,mythesisincludestheideaofcombininganaloganddigitalmethodsinthesamewaythat thehumanbraindoes.Forexample,moreadvancedneuralnetsarealreadyusinghighlydetailedmodels of human neurons, including detailed nonlinear, analog activation functions. There's asignificantefficiencyadvantagetoemulatingthebrain'sanalogmethods.Analogmethodsarealsonotthe exclusive province of biological systems.We used to refer to "digital computers" to distinguishthemfromthemoreubiquitousanalogcomputerswidelyusedduringWorldWarII.TheworkofCarverMeadhas shown theabilityof siliconcircuits to implementdigital-controlledanalogcircuitsentirelyanalogous to,and indeedderivedfrom,mammalianneuronalcircuits.Analogmethodsare readilyre-created by conventional transistors, which are essentially analog devices. It is only by adding themechanismofcomparingthetransistor'soutputtoathresholdthatitismadeintoadigitaldevice. More important, there isnothing that analogmethodscanaccomplish thatdigitalmethodsareunable to accomplish just aswell.Analogprocesses canbe emulatedwithdigitalmethods (byusingfloatingpointrepresentations),whereasthereverseisnotnecessarilythecase.TheCriticismfromtheComplexityofNeuralProcessingAnothercommoncriticismisthatthefinedetailofthebrain'sbiologicaldesignissimplytoocomplextobemodeledandsimulatedusingnonbiologicaltechnology.Forexample,ThomasRaywrites:

Thestructureandfunctionofthebrainoritscomponentscannotbeseparated.Thecirculatorysystemprovideslifesupportforthebrain,butitalsodelivershormonesthatareanintegralpartofthechemicalinformationprocessingfunctionofthebrain.Themembraneofaneuronisastructuralfeaturedefiningthelimitsandintegrityofaneuron,but it isalsothesurfacealongwhich depolarization propagates signals. The structural and life-support functions cannot beseparatedfromthehandlingofinformation.17

Raygoesontodescribeseveralofthe"broadspectrumofchemicalcommunicationmechanisms"thatthebrainexhibits.

Infact,allofthesefeaturescanreadilybemodeled,andagreatdealofprogresshasalreadybeenmade in this endeavor. The intermediate language is mathematics, and translating the mathematicalmodelsintoequivalentnonbiologicalmechanisms(examplesincludecomputersimulationsandcircuitsusing transistors in their native analogmode) is a relatively straightforward process.Thedelivery ofhormonesbythecirculatorysystem,forexample,isanextremelylow-bandwidthphenomenon,whichisnot difficult to model and replicate. The blood levels of specific hormones and other chemicalsinfluenceparameterlevelsthataffectagreatmanysynapsessimultaneously.ThomasRayconcludesthat"ametalliccomputationsystemoperatesonfundamentallydifferentdynamic properties and could never precisely and exactly 'copy' the function of a brain." Followingclosely the progress in the related fields of neurobiology, brain scanning, neuron and neural-regionmodeling,neuron-electroniccommunication,neural implants, and relatedendeavors,we find thatourability to replicate thesalient functionalityofbiological informationprocessingcanmeetanydesiredlevelofprecision. Inotherwords thecopiedfunctionalitycanbe"closeenough"foranyconceivablepurpose or goal, including satisfying a Turing-test judge. Moreover, we find that efficientimplementationsofthemathematicalmodelsrequiresubstantiallylesscomputationalcapacitythanthetheoreticalpotentialofthebiologicalneuronclustersbeingmodeled.Inchapter4,Ireviewedanumberofbrain-regionmodels(Watts'sauditoryregions,thecerebellum,andothers)thatdemonstratethis.BrainComplexity.ThomasRayalsomakesthepointthatwemighthavedifficultycreatingasystemequivalent to "billionsof linesof code,"which is the level of complexityhe attributes to thehumanbrain.Thisfigure,however,ishighlyinflated,foraswehaveseenourbrainsarecreatedfromagenomeofonlyabout thirty toonehundredmillionbytesofunique information (eighthundredmillionbyteswithout compression, but compression is clearly feasible given the massive redundancy), of whichperhapstwothirdsdescribetheprinciplesofoperationofthebrain.Itisself-organizingprocessesthatincorporate significantelementsof randomness (aswellasexposure to the realworld) thatenable sorelativelysmallanamountofdesigninformationtobeexpandedtothethousandsoftrillionsofbytesofinformation represented in a mature human brain. Similarly, the task of creating human-levelintelligence in a nonbiological entity will involve creating not a massive expert system comprisingbillions of rules or lines of code but rather a learning, chaotic, self-organizing system, one that isultimatelybiologicallyinspired. Ray goes on to write, "The engineers among usmight propose nanomolecular devices withfullerene switches, or evenDNA-like computers. But I am sure theywould never think of neurons.Neuronsareastronomicallylargestructurescomparedtothemoleculeswearestartingwith."Thisisexactlymyownpoint.Thepurposeofreverseengineeringthehumanbrainisnottocopythe digestive or other unwieldy processes of biological neurons but rather to understand their keyinformation-processingmethods.Thefeasibilityofdoingthishasalreadybeendemonstratedindozensofcontemporaryprojects.Thecomplexityoftheneuronclustersbeingemulatedisscalingupbyordersofmagnitude,alongwithallofourothertechnologicalcapabilities.AComputer's Inherent Dualism. Neuroscientist Anthony Bell of RedwoodNeuroscience Institutearticulatestwochallengestoourabilitytomodelandsimulatethebrainwithcomputation.Inthefirsthemaintainsthat

acomputerisanintrinsicallydualisticentity,withitsphysicalset-updesignednottointerferewithitslogicalset-up,whichexecutesthecomputation.Inempiricalinvestigation,wefindthatthebrainisnotadualisticentity.Computerandprogrammaybetwo,butmindandbrainareone.Thebrainisthusnotamachine,meaningitisnotafinitemodel(orcomputer)instantiatedphysicallyinsuchawaythatthephysicalinstantiationdoesnotinterferewiththeexecutionofthemodel(orprogram).18

Thisargumentiseasilydispensedwith.Theabilitytoseparateinacomputertheprogramfromthe

physicalinstantiationthatperformsthecomputationisanadvantage,notalimitation.Firstofall,wedohaveelectronicdeviceswithdedicatedcircuitryinwhichthe"computerandprogram"arenottwo,butone.Suchdevicesarenotprogrammablebutarehardwiredforonespecificsetofalgorithms.NotethatIamnotjustreferringtocomputerswithsoftware(called"firmware")inread-onlymemory,asmaybefoundinacellphoneorpocketcomputer.Insuchasystem,theelectronicsandthesoftwaremaystillbeconsidereddualisticeveniftheprogramcannoteasilybemodified.Iamreferringinsteadtosystemswithdedicatedlogicthatcannotbeprogrammedatall—suchasapplication-specificintegratedcircuits(used,forexample,forimageandsignalprocessing).Thereisacost efficiency in implementing algorithms in thisway, andmany electronic consumer products usesuchcircuitry.Programmablecomputerscostmorebutprovidetheflexibilityofallowingthesoftwaretobechangedandupgraded.Programmablecomputerscanemulatethefunctionalityofanydedicatedsystem, including the algorithms thatwe are discovering (through the efforts to reverse engineer thebrain)forneuralcomponents,neurons,andbrainregions. There isnovalidity tocallinga system inwhich the logicalalgorithm is inherently tied to itsphysical design "not a machine." If its principles of operation can be understood, modeled inmathematical terms,and then instantiatedonanother system(whether thatother system isamachinewithunchangeablededicatedlogicorsoftwareonaprogrammablecomputer),thenwecanconsiderittobeamachineandcertainlyanentitywhosecapabilitiescanbere-createdinamachine.AsIdiscussedextensivelyinchapter4,therearenobarrierstoourdiscoveringthebrain'sprinciplesofoperationandsuccessfullymodelingandsimulatingthem,fromitsmolecularinteractionsupward.Bellreferstoacomputer's"physicalset-up[thatis]designednottointerferewithitslogicalset-up," implying that the brain does not have this "limitation."He is correct that our thoughts do helpcreateourbrains,andasIpointedoutearlierwecanobservethisphenomenonindynamicbrainscans.Butwecanreadilymodelandsimulateboththephysicalandlogicalaspectsofthebrain'splasticityinsoftware. The fact that software in a computer is separate from its physical instantiation is anarchitecturaladvantage in that itallows thesamesoftware tobeapplied toever-improvinghardware.Computersoftware,likethebrain'schangingcircuits,canalsomodifyitself,aswellasbeupgraded. Computerhardwarecan likewisebeupgradedwithout requiringachange insoftware. It is thebrain's relatively fixed architecture that is severely limited.Although the brain is able to create newconnectionsandneurotransmitterpatterns, it is restrictedtochemicalsignalingmore thanonemilliontimesslowerthanelectronics,tothelimitednumberofinterneuronalconnectionsthatcanfitinsideourskulls, and to having no ability to be upgraded, other than through the merger with nonbiologicalintelligencethatI'vebeendiscussing.LevelsandLoops.Bellalsocommentsontheapparentcomplexityofthebrain:

Molecularandbiophysicalprocessescontrolthesensitivityofneuronstoincomingspikes(bothsynaptic efficiency and post-synaptic responsivity), the excitability of the neuron to producespikes, the patterns of spikes it can produce and the likelihood of new synapses forming(dynamicrewiring),tolistonlyfourofthemostobviousinterferencesfromthesubneurallevel.Furthermore, transneural volume effects such as local electric fields and the transmembranediffusionofnitricoxidehavebeenseentoinfluence,respectively,coherentneuralfiring,andthedeliveryofenergy(bloodflow)tocells,thelatterofwhichdirectlycorrelateswithneuralactivity. Thelistcouldgoon.Ibelievethatanyonewhoseriouslystudiesneuromodulators, ionchannels or synaptic mechanism and is honest, would have to reject the neuron level as aseparatecomputinglevel,evenwhilefindingittobeausefuldescriptivelevel.19

AlthoughBellmakesthepointherethattheneuronisnottheappropriatelevelatwhichtosimulatethe brain, his primary argument here is similar to that of Thomas Ray above: the brain is morecomplicatedthansimplelogicgates.

Hemakesthisexplicit:

Toarguethatonepieceofstructuredwateroronequantumcoherenceisanecessarydetailinthe functional description of the brain would clearly be ludicrous. But if, in every cell,moleculesderivesystematicfunctionalityfromthesesubmolecularprocesses,iftheseprocessesare used all the time, allover the brain, to reflect, record and propagate spatiotemporalcorrelationsofmolecularfluctuations,toenhanceordiminishtheprobabilitiesandspecificitiesofreactions,thenwehaveasituationqualitativelydifferentfromthelogicgate.

Atonelevelheisdisputingthesimplisticmodelsofneuronsandinterneuronalconnectionsusedinmany neural-net projects. Brain-region simulations don't use these simplified models, however, butratherapplyrealisticmathematicalmodelsbasedontheresultsfrombrainreverseengineering.TherealpointthatBellismakingisthatthebrainisimmenselycomplicated,withtheconsequentimplicationthat itwill thereforebeverydifficult tounderstand,model,andsimulate itsfunctionality.TheprimaryproblemwithBell'sperspectiveisthathefailstoaccountfortheself-organizing,chaotic,and fractal nature of the brain's design. It's certainly true that the brain is complex, but a lot of thecomplication ismoreapparent than real. Inotherwords, theprinciplesof thedesignof thebrainaresimplerthantheyappear. To understand this, let's first consider the fractal nature of the brain's organization, which Idiscussedinchapter2.Afractalisarulethatisiterativelyappliedtocreateapatternordesign.Theruleisoftenquite simple,butbecauseof the iteration the resultingdesigncanbe remarkablycomplex.Afamousexampleof this is theMandelbrotsetdevisedbymathematicianBenoitMandelbrot.20Visualimages of the Mandelbrot set are remarkably complex, with endlessly complicated designs withindesigns.Aswelookatfinerandfinerdetail inanimageof theMandelbrotset, thecomplexitynevergoes away, and we continue to see ever finer complication. Yet the formula underlying all of thiscomplexityisamazinglysimple:theMandelbrotsetischaracterizedbyasingleformulaZ=Z2+C,inwhich Z is a "complex" (meaning two-dimensional) number and C is a constant. The formula isiterativelyapplied,andtheresultingtwo-dimensionalpointsaregraphedtocreatethepattern.

The point here is that a simple design rule can create a lot of apparent complexity. StephenWolframmakes a similar point using simple rules on cellular automata (see chapter 2).This insightholds true for the brain's design. As I've discussed, the compressed genome is a relatively compactdesign, smaller than some contemporary software programs. As Bell points out, the actualimplementationofthebrainappearsfarmorecomplexthanthis.JustaswiththeMandelbrotset,aswelookatfinerandfinerfeaturesofthebrain,wecontinuetoseeapparentcomplexityateachlevel.Atamacrolevelthepatternofconnectionslookscomplicated,andatamicrolevelsodoesthedesignofasingleportionof aneuron suchasadendrite. I'vementioned that itwould takeat least thousandsoftrillionsofbytes tocharacterize thestateofahumanbrain,but thedesign isonly tensofmillionsofbytes. So the ratio of the apparent complexity of the brain to the design information is at least onehundredmillion to one. The brain's information starts out as largely random information, but as thebraininteractswithacomplexenvironment(thatis,asthepersonlearnsandmatures),thatinformationbecomesmeaningful.Theactualdesigncomplexityisgovernedbythecompressedinformationinthedesign(thatis,thegenomeandsupportingmolecules),notbythepatternscreatedthroughtheiterativeapplicationofthedesignrules. Iwouldagree that the roughly thirty toonehundredmillionbytesof information in thegenome do not represent a simple design (certainly farmore complex than the six characters in thedefinitionoftheMandelbrotset),butit isa levelofcomplexitythatwecanalreadymanagewithourtechnology. Many observers are confused by the apparent complexity in the brain's physicalinstantiation, failing to recognize that the fractal nature of the design means that the actual designinformationisfarsimplerthanwhatweseeinthebrain.Ialsomentionedinchapter2thatthedesigninformationinthegenomeisaprobabilisticfractal,meaning that the rules are appliedwith a certain amount of randomness each time a rule is iterated.There is, for example, very little information in the genome describing the wiring pattern for thecerebellum,whichcomprisesmorethanhalftheneuronsinthebrain.Asmallnumberofgenesdescribethebasicpatternofthefourcell typesinthecerebellumandthensayinessence,"Repeatthispatternseveral billion times with some random variation in each repetition." The result may look verycomplicated,butthedesigninformationisrelativelycompact. Bell is correct that trying to compare thebrain's design to a conventional computerwouldbefrustrating. The brain does not follow a typical top-down (modular) design. It uses its probabilisticfractaltypeoforganizationtocreateprocessesthatarechaotic—thatis,notfullypredictable.Thereisawell-developedbodyofmathematicsdevoted tomodelingand simulatingchaotic systems,whichareusedtounderstandphenomenasuchasweatherpatternsandfinancialmarkets,thatisalsoapplicabletothebrain.Bellmakesnomentionofthisapproach.Heargueswhythebrainisdramaticallydifferentfromconventionallogicgatesandconventionalsoftwaredesign,whichleadstohisunwarrantedconclusionthatthebrainisnotamachineandcannotbemodeledbyamachine.Whileheiscorrectthatstandardlogicgatesandtheorganizationofconventionalmodularsoftwarearenottheappropriatewaytothinkaboutthebrain,thatdoesnotmeanthatweareunabletosimulatethebrainonacomputer.Becausewecan describe the brain's principles of operation inmathematical terms, and sincewe canmodel anymathematicalprocess(includingchaoticones)onacomputer,weareabletoimplementthesetypesofsimulations.Indeed,we'remakingsolidandacceleratingprogressindoingso.DespitehisskepticismBellexpressescautiousconfidencethatwewillunderstandourbiologyandbrainswellenoughtoimproveonthem.Hewrites:"Willtherebeatranshumanage?Forthisthereisastrongbiologicalprecedent in thetwomajorstepsinbiologicalevolution.Thefirst, theincorporationintoeukaryoticbacteriaofprokaryoticsymbiotes,andthesecond,theemergenceofmulticellularlife-formsfromcoloniesofeukaryotes....Ibelievethatsomethinglike[atranshumanistage]mayhappen."TheCriticismfromMicrotubulesandQuantumComputing

Quantummechanicsismysterious,andconsciousnessismysterious.Q.E.D.:Quantummechanicsandconsciousnessmustberelated.

—CHRISTOF KOCH, MOCKING ROGER PENROSE'S THEORY OFQUANTUMCOMPUTING INNEURONTUBULESASTHESOURCEOFHUMANCONSCIOUSNESS21

Over the past decade Roger Penrose, a noted physicist and philosopher, in conjunction with StuartHameroff, an anesthesiologist, has suggested that fine structures in the neurons called microtubulesperform an exotic form of computation called "quantum computing." As I discussed, quantumcomputing is computing using what are called qubits, which take on all possible combinations ofsolutionssimultaneously.Themethodcanbeconsideredtobeanextremeformofparallelprocessing(becauseeverycombinationofvaluesofthequbitsistestedsimultaneously).Penrosesuggeststhatthemicrotubulesand theirquantum-computingcapabilitiescomplicate theconceptof re-creatingneuronsandreinstantiatingmindfiles.22Healsohypothesizesthatthebrain'squantumcomputingisresponsiblefor consciousness and that systems, biological or otherwise, cannot be conscious without quantumcomputing.Althoughsomescientistshaveclaimedtodetectquantumwavecollapse(resolutionofambiguousquantumpropertiessuchasposition,spin,andvelocity)inthebrain,noonehassuggestedthathumancapabilitiesactuallyrequireacapacityforquantumcomputing.PhysicistSethLloydsaid:

Ithinkthatitisincorrectthatmicrotubulesperformcomputingtasksinthebrain,inthewaythat [Penrose] and Hameroff have proposed. The brain is a hot, wet place. It is not a veryfavorable environment for exploiting quantum coherence. The kinds of superpositions andassembly/disassemblyofmicrotubulesforwhichtheysearchdonotseemtoexhibitquantumentanglement....Thebrainclearlyisn'taclassical,digitalcomputerbyanymeans.Butmyguessisthatitperformsmostofitstasksina"classical"manner.Ifyouweretotakealargeenoughcomputer, and model all of the neurons, dendrites, synapses, and such, [then] you couldprobablygetthethingtodomostofthetasksthatbrainsperform.Idon'tthinkthatthebrainisexploitinganyquantumdynamicstoperformtasks.23

Anthony Bell also remarks that "there is no evidence that large-scalemacroscopic quantumcoherences,suchasthoseinsuperfluidsandsuperconductors,occurinthebrain."24 However,even if thebraindoesdoquantumcomputing, thisdoesnotsignificantlychange theoutlookforhuman-levelcomputing(andbeyond),nordoesitsuggestthatbrainuploadingisinfeasible.Firstofall,ifthebraindoesdoquantumcomputingthiswouldonlyverifythatquantumcomputingisfeasible.Therewouldbenothinginsuchafindingtosuggest thatquantumcomputingisrestrictedtobiologicalmechanisms.Biologicalquantum-computingmechanisms,iftheyexist,couldbereplicated.Indeed, recent experiments with small-scale quantum computers appear to be successful. Even theconventionaltransistorreliesonthequantumeffectofelectrontunneling.Penrose'spositionhasbeeninterpretedtoimplythatitisimpossibletoperfectlyreplicateasetofquantum states, so therefore perfect downloading is impossible.Well, how perfect does a downloadhavetobe?Ifwedevelopdownloadingtechnologytothepointwherethe"copies"areasclosetotheoriginalastheoriginalpersonistohim-orherselfoverthecourseofoneminute,thatwouldbegoodenoughforanyconceivablepurposeyetwouldnotrequirecopyingquantumstates.Asthetechnologyimproves,theaccuracyofthecopycouldbecomeascloseastheoriginaltowithineverbrieferperiodsoftime(onesecond,onemillisecond,onemicrosecond).WhenitwaspointedouttoPenrosethatneurons(andevenneuralconnections)weretoobigforquantumcomputing,hecameupwith the tubule theoryasapossiblemechanismforneuralquantumcomputing.Ifoneissearchingforbarrierstoreplicatingbrainfunctionitisaningenioustheory,butit

fails to introduceanygenuinebarriers.However, there is littleevidence tosuggest thatmicrotubules,which provide structural integrity to the neural cells, perform quantum computing and that thiscapabilitycontributestothethinkingprocess.Evengenerousmodelsofhumanknowledgeandpotentialaremorethanaccountedforbycurrentestimatesofbrainsize,basedoncontemporarymodelsofneuronfunctioning that do not includemicrotubule-based quantum computing.Recent experiments showingthathybridbiological!nonbiologicalnetworksperformsimilarly toall-biologicalnetworks,whilenotdefinitive,arestronglysuggestivethatourmicrotubulelessmodelsofneuronfunctioningareadequate.LloydWatts'ssoftwaresimulationofhis intricatemodelofhumanauditoryprocessingusesordersofmagnitude less computation than the networks of neurons he is simulating, and again there is nosuggestionthatquantumcomputingisneeded.Ireviewedotherongoingeffortstomodelandsimulatebrain regions in chapter 4, while in chapter 3 I discussed estimates of the amount of computationnecessarytosimulateallregionsofthebrainbasedonfunctionallyequivalentsimulationsofdifferentregions.Noneoftheseanalysesdemonstratesthenecessityforquantumcomputinginordertoachievehuman-levelperformance.Somedetailedmodelsofneurons(inparticularthosebyPenroseandHameroff)doassignaroletothe microtubules in the functioning and growth of dendrites and axons. However, successfulneuromorphicmodelsofneuralregionsdonotappeartorequiremicrotubulecomponents.Forneuronmodelsthatdoconsidermicrotubules,resultsappeartobesatisfactorybymodelingtheiroverallchaoticbehavior without modeling each microtubule filament individually. However, even if the Penrose-Hameroff tubules are an important factor, accounting for themdoesn't change the projections I havediscussed above to any significant degree. According to my model of computational growth, if thetubulesmultiplied neuron complexity by even a factor of one thousand (and keep inmind that ourcurrent tubuleless neuron models are already complex, including on the order of one thousandconnectionsperneuron,multiplenonlinearities,andotherdetails),thiswoulddelayourreachingbraincapacitybyonly aboutnineyears. Ifwe'reoff by a factorof onemillion, that's still a delayofonlyseventeenyears.Afactorofabillionisaroundtwenty-fouryears(recallthatcomputationisgrowingbyadoubleexponential).25TheCriticismfromtheChurch-TuringThesisEarlyinthetwentiethcenturymathematiciansAlfredNorthWhiteheadandBertrandRussellpublishedtheirseminalwork,PrincipiaMathematica,whichsoughttodetermineaxiomsthatcouldserveasthebasis for all of mathematics.26 However, they were unable to prove conclusively that an axiomaticsystemthatcangeneratethenaturalnumbers(thepositiveintegersorcountingnumbers)wouldnotgiverisetocontradictions.Itwasassumedthatsuchaproofwouldbefoundsoonerorlater,butinthe1930sayoungCzechmathematician,KurtGödel,stunnedthemathematicalworldbyprovingthatwithinsuchasystemthereinevitablyexistpropositionsthatcanbeneitherprovednordisproved.Itwaslatershownthat suchunprovablepropositionsareascommonasprovableones.Gödel's incompleteness theorem,whichisfundamentallyaproofdemonstratingthattherearedefinitelimitstowhatlogic,mathematics,andbyextensioncomputationcando,hasbeencalled themost important inallmathematics, and itsimplicationsarestillbeingdebated.27 AsimilarconclusionwasreachedbyAlanTuringinthecontextofunderstandingthenatureofcomputation. When in 1936 Turing presented the Turing machine (described in chapter 2) as atheoreticalmodel of a computer, which continues today to form the basis ofmodern computationaltheory,hereportedanunexpecteddiscoverysimilartoGödel's.28Inhispaperthatyearhedescribedtheconceptofunsolvableproblems—thatis,problemsthatarewelldefined,withuniqueanswersthatcanbeshowntoexist,butthatwecanalsoshowcanneverbecomputedbyaTuringmachine.ThefactthatthereareproblemsthatcannotbesolvedbythisparticulartheoreticalmachinemaynotseemparticularlystartlinguntilyouconsidertheotherconclusionofTuring'spaper:thattheTuringmachine can model any computational process. Turing showed that there are as many unsolvable

problemsassolvableones,thenumberofeachbeingthelowestorderofinfinity,theso-calledcountableinfinity (that is, counting the number of integers). Turing also demonstrated that the problem ofdetermining the truth or falsity of any logical proposition in an arbitrary system of logic powerfulenoughtorepresentthenaturalnumberswasoneexampleofanunsolvedproblem,aresultsimilartoGödel's. (In other words, there is no procedure guaranteed to answer this question for all suchpropositions.)AroundthesametimeAlonzoChurch,anAmericanmathematicianandphilosopher,publishedatheoremthatexaminedasimilarquestioninthecontextofarithmetic.ChurchindependentlycametothesameconclusionasTuring.29Taken together, theworksofTuring,Church, andGödelwere the firstformalproofsthattherearedefinitelimitstowhatlogic,mathematics,andcomputationcando.Inaddition,ChurchandTuringalsoadvanced,independently,anassertionthathasbecomeknownas the Church-Turing thesis. This thesis has both weak and strong interpretations. The weakinterpretationisthatifaproblemthatcanbepresentedtoaTuringmachineisnotsolvablebyone,thenitisnotsolvablebyanymachine.ThisconclusionfollowsfromTuring'sdemonstrationthattheTuringmachinecouldmodelanyalgorithmicprocess.Itisonlyasmallstepfromtheretodescribethebehaviorofamachineasfollowinganalgorithm. Thestronginterpretation is thatproblemsthatarenotsolvableonaTuringmachinecannotbesolved by human thought, either. The basis of this thesis is that human thought is performed by thehumanbrain(withsomeinfluencebythebody),thatthehumanbrain(andbody)comprisesmatterandenergy, that matter and energy follow natural laws, that these laws are describable inmathematicalterms,andthatmathematicscanbesimulatedtoanydegreeofprecisionbyalgorithms.Thereforethereexist algorithms that can simulate human thought. The strong version of the Church-Turing thesispostulatesanessentialequivalencebetweenwhatahumancanthinkorknowandwhatiscomputable. It is important to note that although the existence of Turing's unsolvable problems is amathematicalcertainty,theChurch-Turingthesisisnotamathematicalpropositionatall.Itis,rather,aconjecture that, in various disguises, is at the heart of some of our most profound debates in thephilosophyofmind.30ThecriticismofstrongAIbasedontheChurch-Turingthesisarguesthefollowing:sincethereareclearlimitationstothetypesofproblemsthatacomputercansolve,yethumansarecapableofsolvingthese problems, machines will never emulate the full range of human intelligence. This conclusion,however, is not warranted. Humans are no more capable of universally solving such "unsolvable"problemsthanmachinesare.Wecanmakeeducatedguesses tosolutions incertain instancesandcanapplyheuristicmethods(proceduresthatattempttosolveproblemsbutthatarenotguaranteedtowork)that succeedonoccasion.But both these approaches are also algorithmically basedprocesses,whichmeans thatmachinesarealsocapableofdoingthem.Indeed,machinescanoftensearchforsolutionswithfargreaterspeedandthoroughnessthanhumanscan.ThestrongformulationoftheChurch-Turingthesisimpliesthatbiologicalbrainsandmachinesareequallysubjecttothelawsofphysics,andthereforemathematicscanmodelandsimulatethemequally.We've alreadydemonstrated the ability tomodel and simulate the functionofneurons, sowhynot asystemof ahundredbillionneurons?Sucha systemwoulddisplay the samecomplexity and lackofpredictability as human intelligence. Indeed, we already have computer algorithms (for example,genetic algorithms) with results that are complex and unpredictable and that provide intelligentsolutions to problems. If anything, the Church-Turing thesis implies that brains and machines areessentiallyequivalent. To seemachines' ability to useheuristicmethods, consider oneof themost interestingof theunsolvable problems, the "busy beaver" problem, formulated byTiborRado in 1962.31 EachTuringmachine has a certain number of states that its internal program can be in,which correspond to thenumberofstepsinitsinternalprogram.Thereareanumberofdifferent4-stateTuringmachinesthatarepossible, a certain number of 5-state machines, and so on. In the "busy beaver" problem, given apositive integer n, we construct all the Turing machines that have n states. The number of suchmachineswillalwaysbefinite.Nextweeliminatethosen-statemachinesthatgetintoaninfiniteloop

(thatis,neverhalt).Finally,weselectthemachine(onethatdoeshalt)thatwritesthelargestnumberof1sonitstape.Thenumberof1sthatthisTuringmachinewritesiscalledthebusybeaverofn. Radoshowedthatthereisnoalgorithm—thatis,noTuringmachine—thatcancomputethisfunctionforallns. The crux of the problem is sorting out those n-statemachines that get into infinite loops. If weprogramaTuringmachinetogenerateandsimulateallpossiblen-stateTuringmachines,thissimulatoritselfgetsintoaninfiniteloopwhenitattemptstosimulateoneofthen-statemachinesthatgetsintoaninfiniteloop.Despiteitsstatusasanunsolvableproblem(andoneofthemostfamous),wecandeterminethebusy-beaverfunctionforsomens.(Interestingly, it isalsoanunsolvableproblemtoseparate thosensforwhichwecandetermine thebusybeaverofn fromthoseforwhichwecannot.)Forexample, thebusybeaverof6iseasilydeterminedtobe35.Withsevenstates,aTuringmachinecanmultiply,sothebusybeaverof7ismuchbigger:22,961.Witheightstates,aTuringmachinecancomputeexponentials,so the busy beaver of 8 is even bigger: approximately 1043.We can see that this is an "intelligent"function,inthatitrequiresgreaterintelligencetosolveforlargerns.Bythetimewegetto10,aTuringmachinecanperformtypesofcalculationsthatareimpossibleforahumantofollow(withouthelpfromacomputer).Sowewereabletodeterminethebusybeaverof10onlywithacomputer'sassistance.Theanswerrequiresanexoticnotationtowritedown,inwhichwehaveastackofexponents,theheightofwhichisdeterminedbyanotherstackofexponents,theheightofwhichisdeterminedbyanotherstackofexponents,andsoon.Becauseacomputercankeeptrackofsuch complex numbers,whereas the human brain cannot, it appears that computerswill provemorecapableofsolvingunsolvableproblemsthanhumanswill.TheCriticismfromFailureRatesJaronLanier,ThomasRay,andotherobserversallcitehighfailureratesoftechnologyasabarriertoitscontinuedexponentialgrowth.Forexample,Raywrites:

Themostcomplexofourcreationsareshowingalarmingfailurerates.Orbitingsatellitesandtelescopes,spaceshuttles,interplanetaryprobes,thePentiumchip,computeroperatingsystems,all seem to be pushing the limits of what we can effectively design and build throughconventional approaches....Our most complex software (operating systems andtelecommunications control systems) already contains tens of millions of lines of code. Atpresentitseemsunlikelythatwecanproduceandmanagesoftwarewithhundredsofmillionsorbillionsoflinesofcode.32

First, we might ask what alarming failure rates Ray is referring to. As mentioned earlier,computerized systems of significant sophistication routinely fly and land our airplanes automaticallyandmonitorintensivecareunitsinhospitals,yetalmostnevermalfunction.Ifalarmingfailureratesareof concern, they're more often attributable to human error. Ray alludes to problems with Intelmicroprocessor chips, but these problems have been extremely subtle, have caused almost norepercussions,andhavequicklybeenrectified. The complexity of computerized systems has indeed been scaling up, as we have seen, andmoreoverthecuttingedgeofoureffortstoemulatehumanintelligencewillutilizetheself-organizingparadigms thatwe find in the human brain.Aswe continue our progress in reverse engineering thehumanbrain,wewilladdnewself-organizingmethodstoourpatternrecognitionandAItoolkit.AsIhave discussed, self-organizing methods help to alleviate the need for unmanageable levels ofcomplexity.AsIpointedoutearlier,wewillnotneedsystemswith"billionsoflinesofcode"toemulatehumanintelligence.Itisalsoimportanttopointoutthatimperfectionisaninherentfeatureofanycomplexprocess,andthatcertainlyincludeshumanintelligence.

TheCriticismfrom"Lock-In" JaronLanier andother critics have cited the prospect of a "lock-in," a situation inwhich oldtechnologiesresistdisplacementbecauseofthelargeinvestmentintheinfrastructuresupportingthem.They argue that pervasive and complex support systems have blocked innovation in such fields astransportation,whichhavenotseentherapiddevelopmentthatwe'veseenincornputation.33Theconceptoflock-inisnottheprimaryobstacletoadvancingtransportation.Iftheexistenceofacomplexsupportsystemnecessarilycausedlock-in,thenwhydon'tweseethisphenomenonaffectingtheexpansionofeveryaspectoftheInternet?Afterall,theInternetcertainlyrequiresanenormousandcomplex infrastructure.Because it is specifically theprocessingandmovementof information that isgrowingexponentially,however,onereason thatanareasuchas transportationhas reachedaplateau(thatis,restingatthetopofanS-curve)isthatmanyifnotmostofitspurposeshavebeensatisfiedbyexponentiallygrowingcommunicationtechnologies.Myownorganization,forexample,hascolleaguesindifferentpartsofthecountry,andmostofourneedsthatintimespastwouldhaverequiredapersonor a package to be transported can be met through the increasingly viable virtual meetings (andelectronic distribution of documents and other intellectual creations)made possible by a panoply ofcommunicationtechnologies,someofwhichLanierhimselfisworkingtoadvance.Moreimportant,wewill see advances in transportation facilitated by the nanotechnology-based energy technologies Idiscussed in chapter 5.However,with increasingly realistic, high-resolution full-immersion forms ofvirtual reality continuing to emerge, our needs to be together will increasingly be met throughcomputationandcommunication. AsIdiscussedinchapter5, thefulladventofMNT-basedmanufacturingwillbringthelawofaccelerating returns to such areas as energy and transportation. Once we can create virtually anyphysicalproductfrominformationandveryinexpensiverawmaterials,thesetraditionallyslow-movingindustrieswillseethesamekindofannualdoublingofprice-performanceandcapacitythatweseeininformationtechnologies.Energyandtransportationwilleffectivelybecomeinformationtechnologies.Wewillseetheadventofnanotechnology-basedsolarpanelsthatareefficient,lightweight,andinexpensive, aswell as comparablypowerful fuel cells andother technologies to store anddistributethat energy. Inexpensive energy will in turn transform transportation. Energy obtained fromnanoengineered solar cells and other renewable technologies and stored in nanoengineered fuel cellswillprovidecleanandinexpensiveenergyforeverytypeoftransportation.Inaddition,wewillbeabletomanufacturedevices—includingflyingmachinesofvaryingsizes—foralmostnocost,otherthanthecost of the design (which needs to be amortized only once). It will be feasible, therefore, to buildinexpensivesmallflyingdevicesthatcantransportapackagedirectlytoyourdestinationinamatterofhourswithout going through intermediaries such as shipping companies.Larger but still inexpensivevehicleswillbeabletoflypeoplefromplacetoplacewithnanoengineeredmicrowings.Informationtechnologiesarealreadydeeplyinfluentialineveryindustry.Withthefullrealizationof the GNR revolutions in a few decades, every area of human endeavor will essentially compriseinformationtechnologiesandthuswilldirectlybenefitfromthelawofacceleratingreturns.TheCriticismfromOntology:CanaComputerBeConscious?

Becausewedonotunderstandthebrainverywellweareconstantlytemptedtousethelatesttechnologyasamodelfortryingtounderstandit.Inmychildhoodwewerealwaysassuredthatthe brainwas a telephone switchboard. ("What else could it be?") Iwas amused to see thatSherrington, the great British neuroscientist, thought that the brain worked like a telegraphsystem. Freud often compared the brain to hydraulic and electromagnetic systems. Leibnizcomparedittoamill,andIamtoldsomeoftheancientGreeksthoughtthebrainfunctionslike

acatapult.Atpresent,obviously,themetaphoristhedigitalcomputer.—JOHNR.SEARLE,“MINDS,BRAINS,ANDSCIENCE”

Can a computer—a nonbiological intelligence—be conscious?We have first, of course, to agree onwhatthequestionmeans.AsIdiscussedearlier,thereareconflictingperspectivesonwhatmayatfirstappeartobeastraightforwardissue.Regardlessofhowweattempttodefinetheconcept,however,wemustacknowledgethatconsciousnessiswidelyregardedasacrucial,ifnotessential,attributeofbeinghuman.34 JohnSearle, distinguished philosopher at theUniversity ofCalifornia atBerkeley, is popularamong his followers for what they believe is a staunch defense of the deep mystery of humanconsciousnessagainsttrivializationbystrong-AI"reductionists"likeRayKurzweil.AndeventhoughIhave always foundSearle's logic in his celebratedChineseRoomargument to be tautological, I hadexpectedanelevating treatiseon theparadoxesofconsciousness.Thus it iswithsomesurprise that IfindSearlewritingstatementssuchas,

"human brains cause consciousness by a series of specific neurobiological processes in thebrain";"The essential thing is to recognize that consciousness is a biological process like digestion,lactation,photosynthesis,ormitosis";"Thebrain isamachine,abiologicalmachine tobesure,butamachineall thesame.So thefirststepistofigureouthowthebraindoesitandthenbuildanartificialmachinethathasanequallyeffectivemechanismforcausingconsciousness";and"Weknowthatbrainscauseconsciousnesswithspecificbiologicalmechanisms."35

So who is being the reductionist here? Searle apparently expects that we can measure thesubjectivityofanotherentityasreadilyaswemeasuretheoxygenoutputofphotosynthesis.SearlewritesthatI"frequentlyciteIBM'sDeepBlueasevidenceofsuperiorintelligenceinthecomputer."Ofcourse, theopposite is thecase: IciteDeepBluenot tobelabor the issueofchessbutrather to examine the dear contrast it illustrates between the human and contemporary machineapproaches to the game. As I pointed out earlier, however, the pattern-recognition ability of chessprograms is increasing, so chess machines are beginning to combine the analytical strength oftraditionalmachineintelligencewithmorehumanlikepatternrecognition.Thehumanparadigm(ofself-organizingchaoticprocesses)offersprofoundadvantages:wecanrecognizeandrespondtoextremelysubtlepatterns.Butwecanbuildmachineswiththesameabilities.That,indeed,hasbeenmyownareaoftechnicalinterest.SearleisbestknownforhisChineseRoomanalogyandhaspresentedvariousformulationsofitover twenty years. One of the more complete descriptions of it appears in his 1992 book, TheRediscoveryoftheMind:

Ibelieve thebest-knownargumentagainststrongAIwasmyChinese roomargument ... thatshowed thata systemcould instantiateaprogramsoas togiveaperfect simulationof somehumancognitivecapacity,suchasthecapacitytounderstandChinese,eventhoughthatsystemhadnounderstandingofChinesewhatever.SimplyimaginethatsomeonewhounderstandsnoChinese is locked in a room with a lot of Chinese symbols and a computer program foranswering questions inChinese.The input to the system consists inChinese symbols in theform of questions; the output of the system consists in Chinese symbols in answer to thequestions.WemightsupposethattheprogramissogoodthattheanswerstothequestionsareindistinguishablefromthoseofanativeChinesespeaker.Butallthesame,neitherthepersoninside nor any other part of the system literally understands Chinese; and because the

programmedcomputerhasnothingthatthissystemdoesnothave,theprogrammedcomputer,qua computer, does not understandChinese either.Because the program is purely formal orsyntacticalandbecausemindshavementalorsemanticcontents,anyattempttoproduceamindpurelywithcomputerprogramsleavesouttheessentialfeaturesofthemind.36

Searle'sdescriptions illustratea failure toevaluate theessenceofeitherbrainprocessesor thenonbiologicalprocessesthatcouldreplicatethem.Hestartswiththeassumptionthatthe"man"intheroomdoesn't understandanythingbecause, after all, "he is just a computer," thereby illuminatinghisownbias.NotsurprisinglySearlethenconcludesthatthecomputer(asimplementedbytheman)doesn'tunderstand.Searlecombinesthistautologywithabasiccontradiction:thecomputerdoesn'tunderstandChinese, yet (according to Searle) can convincingly answer questions inChinese.But if an entity—biologicalorotherwise—reallydoesn'tunderstandhumanlanguage, itwillquicklybeunmaskedbyacompetent interlocutor. In addition, for the program to respond convincingly, itwould have to be ascomplex as a human brain. The observers would long be dead while the man in the room spendsmillionsofyearsfollowingaprogrammanymillionsofpageslong. Most important, themanisactingonlyasthecentralprocessingunit,asmallpartofasystem.Whilethemanmaynotseeit,theunderstandingisdistributedacrosstheentirepatternoftheprogramitselfandthebillionsofnoteshewouldhavetomaketofollowtheprogram.IunderstandEnglish,butnoneofmyneuronsdo.Myunderstandingisrepresentedinvastpatternsofneurotransmitterstrengths,synapticclefts,andinterneuronalconnections.Searlefailstoaccountforthesignificanceofdistributedpatternsofinformationandtheiremergentproperties. A failure to see that computing processes are capable of being—just like the human brain—chaotic,unpredictable,messy,tentative,andemergentisbehindmuchofthecriticismoftheprospectofintelligentmachinesthatwehearfromSearleandotheressentiallymaterialistphilosophers.InevitablySearlecomesbacktoacriticismof"symbolic"computing:thatorderlysequentialsymbolicprocessescannot re-create true thinking. I think that's correct (depending, of course, on what level we aremodelinganintelligentprocess),butthemanipulationofsymbols(inthesensethatSearleimplies)isnottheonlywaytobuildmachines,orcomputers.So-calledcomputers(andpartoftheproblemistheword"computer,"becausemachinescandomore than"compute")arenot limited tosymbolicprocessing.Nonbiologicalentitiescanalsouse theemergent self-organizing paradigm,which is a trendwell underway and one thatwill become evenmoreimportantoverthenextseveraldecades.Computersdonothavetouseonly0and1,nordotheyhavetobealldigital.Evenifacomputerisalldigital,digitalalgorithmscansimulateanalogprocessestoanydegreeofprecision(orlackofprecision).Machinescanbemassivelyparallel.Andmachinescanusechaoticemergenttechniquesjustasthebraindoes. Theprimarycomputingtechniquesthatwehaveusedinpattern-recognitionsystemsdonotusesymbolmanipulationbut rather self-organizingmethods such as those described in chapter 5 (neuralnets, Markov models, genetic algorithms, and more complex paradigms based on brain reverseengineering). Amachine that could really do what Searle describes in the Chinese Room argumentwouldnotmerelybemanipulatinglanguagesymbols,becausethatapproachdoesn'twork.ThisisattheheartofthephilosophicalsleightofhandunderlyingtheChineseRoom.Thenatureofcomputingisnotlimitedtomanipulatinglogicalsymbols.Somethingisgoingoninthehumanbrain,andthereisnothingthatpreventsthesebiologicalprocessesfrombeingreverseengineeredandreplicatedinnonbiologicalentities.AdherentsappeartobelievethatSearle'sChineseRoomargumentdemonstratesthatmachines(thatis,nonbiologicalentities)cannevertrulyunderstandanythingofsignificance,suchasChinese.First,itisimportanttorecognizethatforthissystem—thepersonandthecomputer—to,asSearleputsit,"giveaperfect simulationof somehumancognitive capacity, suchas the capacity tounderstandChinese,"andtoconvincinglyanswerquestionsinChinese,itmustessentiallypassaChineseTuringtest.Keepinmind thatwearenot talkingaboutansweringquestions froma fixed listof stockquestions (becausethat's a trivial task) but answering any unanticipated question or sequence of questions from a

knowledgeablehumaninterrogator.Now,thehumanintheChineseRoomhaslittleornosignificance.Heisjustfeedingthingsintothecomputerandmechanicallytransmittingitsoutput(or,alternatively,justfollowingtherulesintheprogram). And neither the computer nor the human needs to be in a room. Interpreting Searle'sdescriptiontoimplythatthemanhimselfisimplementingtheprogramdoesnotchangeanythingotherthantomakethesystemfarslowerthanrealtimeandextremelyerrorprone.Boththehumanandtheroomareirrelevant.Theonlythingthatissignificantisthecomputer(eitheranelectroniccomputerorthecomputercomprisingthemanfollowingtheprogram).Forthecomputertoreallyperformthis"perfectsimulation,"itwouldindeedhavetounderstandChinese.Accordingtotheverypremiseithas"thecapacitytounderstandChinese,"soitisthenentirelycontradictorytosaythat"theprogrammedcomputer...doesnotunderstandChinese." Acomputerandcomputerprogramasweknowthemtodaycouldnotsuccessfullyperformthedescribed task.So ifweare tounderstand thecomputer tobe like today'scomputers, then it cannotfulfillthepremise.Theonlywaythatitcoulddosowouldbeifithadthedepthandcomplexityofahuman. Turing's brilliant insight in proposing his testwas that convincingly answering any possiblesequenceofquestions froman intelligenthumanquestioner inahuman language reallyprobesallofhumanintelligence.Acomputerthatiscapableofaccomplishingthis—acomputerthatwillexistafewdecadesfromnow—willneedtobeofhumancomplexityorgreaterandwillindeedunderstandChineseinadeepway,becauseotherwiseitwouldneverbeconvincinginitsclaimtodoso. Merely stating, then, that thecomputer "doesnot literallyunderstandChinese"doesnotmakesense,foritcontradictstheentirepremiseoftheargument.Toclaimthatthecomputerisnotconsciousisnotacompellingcontention,either.TobeconsistentwithsomeofSearle'sotherstatements,wehaveto conclude that we really don't know if it is conscious or not. With regard to relatively simplemachines, including to day's computers, while we can't state for certain that these entities are notconscious,theirbehavior,includingtheirinnerworkings,doesn'tgiveusthatimpression.ButthatwillnotbetrueforacomputerthatcanreallydowhatisneededintheChineseRoom.Suchamachinewillatleastseemconscious,evenifwecannotsaydefinitivelywhetheritisornot.Butjustdeclaringthatitisobviousthatthecomputer(ortheentiresystemofthecomputer,person,androom)isnotconsciousisfarfromacompellingargument.InthequoteaboveSearlestatesthat"theprogramispurelyformalorsyntactical,"ButasIpointedoutearlier,thatisabadassumption,basedonSearle'sfailuretoaccountfortherequirementsofsuchatechnology.ThisassumptionisbehindmuchofSearle'scriticismofAI.Aprogramthatispurelyformalorsyntacticalwillnotbeable tounderstandChinese,and itwon't"giveaperfectsimulationofsomehumancognitivecapacity."Butagain,wedon'thavetobuildourmachinesthatway.Wecanbuildtheminthesamefashionthat nature built the human brain: using chaotic emergent methods that are massively parallel.Furthermore,thereisnothinginherentintheconceptofamachinethatrestrictsitsexpertisetothelevelof syntaxaloneandprevents it frommasteringsemantics. Indeed, if themachine inherent inSearle'sconception of the Chinese Room had not mastered semantics, it would not be able to convincinglyanswerquestionsinChineseandthuswouldcontradictSearle'sownpremise.Inchapter4Idiscussedtheongoingefforttoreverseengineerthehumanbrainandtoapplythesemethods tocomputingplatformsof sufficientpower.So, likeahumanbrain, ifwe teachacomputerChinese,itwillunderstandChinese.Thismayseemtobeanobviousstatement,butitisonewithwhichSearletakesissue.Tousehisownterminology,Iamnottalkingaboutasimulationpersebutratheraduplicationofthecausalpowersofthemassiveneuronclusterthatconstitutesthebrain,atleastthosecausalpowerssalientandrelevanttothinking. Will such a copy be conscious? I don't think theChineseRoom tells us anything about thisquestion.ItisalsoimportanttopointoutthatSearle'sChineseRoomargumentcanbeappliedtothehumanbrainitself.Althoughitisclearlynothisintent,hislineofreasoningimpliesthatthehumanbrainhasnounderstanding.Hewrites:"Thecomputer...succeedsbymanipulatingformalsymbols.Thesymbolsthemselvesarequitemeaningless:theyhaveonlythemeaningwehaveattachedtothem.Thecomputer

knowsnothingof this, it just shuffles the symbols."Searle acknowledges thatbiologicalneurons aremachines, so ifwe simply substitute the phrase "humanbrain" for "computer" and "neurotransmitterconcentrationsandrelatedmechanisms"for"formalsymbols,"weget:

The [human brain] ... succeeds bymanipulating [neurotransmitter concentrations and relatedmechanisms]. The [neurotransmitter concentrations and related mechanisms] themselves arequitemeaningless:theyhaveonlythemeaningwehaveattachedtothem.The[humanbrain]knows nothing of this, it just shuffles the [neurotransmitter concentrations and relatedmechanisms].

Ofcourse,neurotransmitterconcentrationsandotherneuraldetails (forexample, interneuronalconnection and neurotransmitter patterns) have nomeaning in and of themselves. Themeaning andunderstanding that emerge in the humanbrain are exactly that: anemergent property of its complexpatterns of activity. The same is true for machines. Although "shuffling symbols" does not havemeaninginandofitself,theemergentpatternshavethesamepotentialroleinnonbiologicalsystemsasthey do in biological systems such as the brain. Hans Moravec has written, "Searle is looking forunderstandinginthewrongplaces....[He]seeminglycannotacceptthatrealmeaningcanexistinmerepatterns.37Let'saddressasecondversionoftheChineseRoom.InthisconceptiontheroomdoesnotincludeacomputeroramansimulatingacomputerbuthasaroomfullofpeoplemanipulatingslipsofpaperwithChinese symbols on them—essentially, a lot of people simulating a computer. This system wouldconvincinglyanswerquestionsinChinese,butnoneoftheparticipantswouldknowChinese,norcouldwe say that the whole system really knows Chinese—at least not in a conscious way. Searle thenessentiallyridiculestheideathatthis"system"couldbeconscious.Whatarewetoconsiderconscious,heasks:theslipsofpaper?Theroom? Oneof theproblemswith thisversionof theChineseRoomargument is that itdoesnotcomeremotelyclose to really solving the specificproblemofansweringquestions inChinese. Instead it isreallyadescriptionofamachinelikeprocess thatuses theequivalentofa table lookup,withperhapssomestraightforwardlogicalmanipulations,toanswerquestions.Itwouldbeabletoansweralimitednumberofcannedquestions,but if itwere toansweranyarbitraryquestion that itmightbeasked, itwouldreallyhavetounderstandChineseinthesamewaythataChinese-speakingpersondoes.Again,itisessentiallybeingaskedtopassaChineseTuringtest,andassuch,wouldhavetobeasclever,andaboutascomplex,asahumanbrain.Straightforwardtablelookupalgorithmsaresimplynotgoingtoachievethat. Ifwewant to re-create a brain that understandsChineseusingpeople as little cogs in the re-creation,wewouldreallyneedbillionsofpeoplesimulatingtheprocessesinahumanbrain(essentiallythepeoplewouldbe simulatinga computer,whichwouldbe simulatinghumanbrainmethods).Thiswould require a rather large room, indeed.And even if extremely efficiently organized, this systemwouldrunmanythousandsoftimesslowerthantheChinese-speakingbrainitisattemptingtore-create.Now,it'struethatnoneofthesebillionsofpeoplewouldneedtoknowanythingaboutChinese,andnoneofthemwouldnecessarilyknowwhatisgoingoninthiselaboratesystem.Butthat'sequallytrueof theneuralconnections ina realhumanbrain.Noneof thehundred trillionconnections inmybrainknowsanythingaboutthisbookIamwriting,nordoanyofthemknowEnglish,noranyoftheother things that I know. None of them is conscious of this chapter, nor of any of the things I amconsciousof.Probablynoneof them is conscious at all.But the entire systemof them—that is,RayKurzweil—isconscious.AtleastI'mclaimingthatI'mconscious(andsofar,theseclaimshavenotbeenchallenged).SoifwescaleupSearle'sChineseRoomtobetherathermassive"room"itneedstobe,who'stosaythattheentiresystemofbillionsofpeoplesimulatingabrainthatknowsChineseisn'tconscious?CertainlyitwouldbecorrecttosaythatsuchasystemknowsChinese.Andwecan'tsaythatitisnotconsciousanymorethanwecansaythataboutanyotherbrainprocess.Wecan'tknowthesubjective

experienceofanotherentity(andinatleastsomeofSearle'sotherwritings,heappearstoacknowledgethislimitation).Andthismassivemultibillion-person"room"isanentity.Andperhapsitisconscious.Searleisjustdeclaringipsofactothatitisn'tconsciousandthatthisconclusionisobvious.Itmayseemthat way when you call it a room and talk about a limited number of peoplemanipulating a smallnumberofslipsofpaper.ButasIsaid,suchasystemdoesn'tremotelywork.AnotherkeytothephilosophicalconfusionimplicitintheChineseRoomargumentisspecificallyrelated to the complexity and scale of the system. Searle says thatwhereas he cannot prove that histypewriter or tape recorder is not conscious, he feels it is obvious that they are not.Why is this soobvious?Atleastonereasonisbecauseatypewriterandataperecorderarerelativelysimpleentities.Buttheexistenceorabsenceofconsciousnessisnotsoobviousinasystemthatisascomplexasthehumanbrain—indeed,onethatmaybeadirectcopyoftheorganizationand"causalpowers"ofarealhumanbrain.Ifsucha"system"actshumanandknowsChineseinahumanway,isitconscious?Nowtheanswerisnolongersoobvious.WhatSearleissayingintheChineseRoomargumentisthatwetakeasimple"machine"andthenconsiderhowabsurditistoconsidersuchasimplemachinetobeconscious.Thefallacyhaseverythingtodowith thescaleandcomplexityof thesystem.Complexityalonedoesnotnecessarilygiveusconsciousness,buttheChineseRoomtellsusnothingaboutwhetherornotsuchasystemisconscious.Kurzweil'sChineseRoom. Ihavemyownconceptionof theChineseRoom—call itRayKurzweil'sChineseRoom.Inmythoughtexperimentthereisahumaninaroom.TheroomhasdecorationsfromtheMingdynasty,includingapedestalonwhichsitsamechanicaltypewriter.Thetypewriterhasbeenmodifiedso that its keys are marked with Chinese symbols instead of English letters. And the mechanicallinkages have been cleverly altered so that when the human types in a question in Chinese, thetypewriter does not type the question but instead types the answer to the question.Now, the personreceivesquestions inChinesecharacters anddutifullypresses theappropriatekeyson the typewriter.The typewriter types out not the question, but the appropriate answer. The human then passes theansweroutsidetheroom. SoherewehavearoomwithahumaninitwhoappearsfromtheoutsidetoknowChineseyetclearly does not. And clearly the typewriter does not know Chinese, either. It is just an ordinarytypewriter with its mechanical linkagesmodified. So despite the fact that theman in the room cananswerquestionsinChinese,whoorwhatcanwesaytrulyknowsChinese?Thedecorations?Now,youmighthavesomeobjectionstomyChineseRoom.Youmightpointoutthatthedecorationsdon'tseemtohaveanysignificance.Yes,that'strue.Neitherdoesthepedestal.Thesamecanbesaidforthehumanandfortheroom.Youmightalsopointoutthatthepremiseisabsurd.JustchangingthemechanicallinkagesinamechanicaltypewritercouldnotpossiblyenableittoconvincinglyanswerquestionsinChinese(nottomention the fact that we can't fit the thousands of Chinese-character symbols on the keys of atypewriter).Yes,that'savalidobjection,aswell.TheonlydifferencebetweenmyChineseRoomconceptionand the several proposed by Searle is that it is patently obvious in my conception that it couldn'tpossiblyworkandisbyitsverynatureabsurd.ThatmaynotbequiteasapparenttomanyreadersorlistenerswithregardtotheSearleChineseRooms.However,itisequallythecase.Andyetwecanmakemyconceptionwork,justaswecanmakeSearle'sconceptionswork.Allyouhavetodoistomakethetypewriterlinkagesascomplexasahumanbrain.Andthat'stheoretically(ifnotpractically)possible.But thephrase"typewriter linkages"doesnot suggest suchvastcomplexity.ThesameistrueofSearle'sdescriptionofapersonmanipulatingslipsofpaperorfollowingabookofrulesoracomputerprogram.Theseareallequallymisleadingconceptions. Searlewrites:"Actualhumanbrainscauseconsciousnessbyaseriesofspecificneurobiologicalprocessesinthebrain."However,hehasyettoprovideanybasisforsuchastartlingview.ToilluminateSearle'sperspective,Iquotefromaletterhesentme:

It may turn out that rather simple organisms like termites or snails are conscious....Theessential thing is to recognize that consciousness is a biological process like digestion,lactation,photosynthesis,ormitosis,andyoushouldlookforitsspecificbiologyasyoulookforthespecificbiologyoftheseotherprocesses.38

Ireplied:

Yes,itistruethatconsciousnessemergesfromthebiologicalprocess(es)ofthebrainandbody,butthereisatleastonedifference.IfIaskthequestion,"doesaparticularentityemitcarbondioxide,"Icananswerthatquestionthroughclearobjectivemeasurement.IfIaskthequestion,"isthisentityconscious,"Imaybeabletoprovideinferentialarguments—possiblystrongandconvincingones—butnotclearobjectivemeasurement.

Withregardtothesnail,Iwrote:

Nowwhenyousaythatasnailmaybeconscious,Ithinkwhatyouaresayingisthefollowing:that we may discover a certain neurophysiological basis for consciousness (call it "x") inhumans such thatwhen thisbasiswaspresenthumanswereconscious,andwhen itwasnotpresenthumanswerenotconscious.Sowewouldpresumablyhaveanobjectivelymeasurablebasis for consciousness.And then ifwe found that ina snail,we could conclude that itwasconscious. But this inferential conclusion is just a strong suggestion, it is not a proof ofsubjective experience on the snail's part. Itmaybe that humans are conscious because theyhave"x"aswellassomeotherqualitythatessentiallyallhumansshare,callthis"y."The"y"mayhavetodowithahuman'slevelofcomplexityorsomethinghavingtodowiththewayweareorganized,orwiththequantumpropertiesofourmicrotubules(althoughthismaybepartof"x"),orsomethingelseentirely.Thesnailhas"x"butdoesn'thave"y"andsoitmaynotbeconscious.

Howwouldonesettle suchanargument?Youobviouslycan'task thesnail.Even ifwecouldimagineawaytoposethequestion,anditansweredyes,thatstillwouldn'tprovethatitwasconscious.Youcan'ttellfromitsfairlysimpleandmore-or-lesspredictablebehavior.Pointingoutthatithas"x"maybeagoodargument,andmanypeoplemaybeconvincedby it.But it's justanargument—notadirect measurement of the snail's subjective experience. Once again, objective measurement isincompatiblewiththeveryconceptofsubjectiveexperience.Manysuchargumentsaretakingplacetoday—thoughnotsomuchaboutsnailsasabouthigher-level animals. It is apparent to me that dogs and cats are conscious (and Searle has said that heacknowledges this as well). But not all humans accept this. I can imagine scientific ways ofstrengtheningtheargumentbypointingoutmanysimilaritiesbetweentheseanimalsandhumans,butagainthesearejustarguments,notscientificproof. Searleexpects to findsomeclearbiological"cause"ofconsciousness,andheseemsunable toacknowledgethateitherunderstandingorconsciousnessmayemergefromanoverallpatternofactivity.Other philosophers, such as Daniel Dennett, have articulated such "pattern emergent" theories ofconsciousness.Butwhether it is "caused"bya specificbiologicalprocessorbyapatternofactivity,Searle provides no foundation for how we would measure or detect consciousness. Finding aneurological correlate of consciousness in humans does not prove that consciousness is necessarilypresentinotherentitieswiththesamecorrelate,nordoesitprovethattheabsenceofsuchacorrelateindicates the absence of consciousness. Such inferential arguments necessarily stop short of directmeasurement. In this way, consciousness differs from objectively measurable processes such aslactationandphotosynthesis.AsIdiscussedinchapter4,wehavediscoveredabiologicalfeatureuniquetohumansandafewotherprimates: thespindlecells.Andthesecellswiththeirdeepbranchingstructuresdoappeartobe

heavilyinvolvedwithourconsciousresponses,especiallyemotionalones.Isthespindlecellstructurethe neurophysiological basis "x" for human consciousness?What sort of experiment could possiblyprove that? Cats and dogs don't have spindle cells. Does that prove that they have no consciousexperience? Searlewrites:"It isoutof thequestion, forpurelyneurobiological reasons, tosuppose that thechairorthecomputerisconscious."Iagreethatchairsdon'tseemtobeconscious,butasforcomputersofthefuturethathavethesamecomplexity,depth,subtlety,andcapabilitiesashumans,Idon'tthinkwecanruleoutthispossibility.Searlejustassumesthattheyarenot,andthatitis"outofthequestion"tosupposeotherwise.ThereisreallynothingmoreofasubstantivenaturetoSearle's"arguments"thanthistautology.Now,partoftheappealofSearle'sstanceagainstthepossibilityofacomputer'sbeingconsciousisthat the computers we know today just don't seem to be conscious. Their behavior is brittle andformulaic,eveniftheyareoccasionallyunpredictable.ButasIpointedoutabove,computerstodayareontheorderofonemilliontimessimplerthanthehumanbrain,whichisatleastonereasontheydon'tshareallof theendearingqualitiesofhuman thought.But thatdisparity is rapidlyshrinkingandwillultimately reverse itself inacoupleofdecades.Theearly twenty-first-centurymachines I am talkingaboutinthisbookwillappearandactverydifferentlythantherelativelysimplecomputersoftoday. Searlearticulates theview thatnonbiological entities arecapableofonlymanipulating logicalsymbolsandheappearstobeunawareofotherparadigms.Itistruethatmanipulatingsymbolsislargelyhowrule-basedexpertsystemsandgame-playingprogramswork.Butthecurrenttrendisinadifferentdirection,towardself-organizingchaoticsystemsthatemploybiologicallyinspiredmethods,includingprocessesderiveddirectlyfromthereverseengineeringofthehundredsofneuronclusterswecallthehumanbrain. Searle acknowledges that biological neurons aremachines—indeed, that the entire brain is amachine. As I discussed in chapter 4, we have already recreated in an extremely detailed way the"causal powers" of individual neurons as well as those of substantial neuron clusters. There is noconceptualbarriertoscalingtheseeffortsuptotheentirehumanbrain.TheCriticismfromtheRich-PoorDivideAnotherconcernexpressedbyJaronLanierandothersisthe"terrifying"possibilitythatthroughthesetechnologies the richmay gain certain advantages and opportunities towhich the rest of humankinddoesnothaveaccess.39Suchinequality,ofcourse,wouldbenothingnew,butwithregardtothisissuethe law of accelerating returns has an important and beneficial impact. Because of the ongoingexponentialgrowthofprice-performance,allofthesetechnologiesquicklybecomesoinexpensiveastobecomealmostfree. Lookat theextraordinaryamountofhigh-quality informationavailableatnocoston theWebtodaythatdidnotexistatalljustafewyearsago.AndifonewantstopointoutthatonlyafractionoftheworldtodayhasWebaccess,keepinmindthattheexplosionoftheWebisstillinitsinfancy,andaccess is growing exponentially. Even in the poorest countries of Africa,Web access is expandingrapidly.Eachexampleofinformationtechnologystartsoutwithearly-adoptionversionsthatdonotworkverywellandthatareunaffordableexceptbytheelite.Subsequentlythetechnologyworksabitbetterandbecomesmerely expensive.Then itworksquitewell andbecomes inexpensive.Finally itworksextremelywellandisalmostfree.Thecellphone,forexample, issomewherebetweenthese last twostages.Considerthatadecadeagoifacharacterinamovietookoutaportabletelephone,thiswasanindicationthatthispersonmustbeverywealthy,powerful,orboth.Yettherearesocietiesaroundtheworldinwhichthemajorityofthepopulationwerefarmingwiththeirhandstwodecadesagoandnowhave thriving information-based economieswithwidespread use of cell phones (for example, Asiansocieties, including rural areas of China). This lag from very expensive early adopters to very

inexpensive,ubiquitous adoptionnow takes about adecade.But inkeepingwith thedoublingof theparadigm-shiftrateeachdecade,thislagwillbeonlyfiveyearsadecadefromnow.Intwentyyears,thelagwillbeonlytwotothreeyears(seechapter2).Therich-poordivideremainsacriticalissue,andateachpointintimethereismorethatcanandshouldbedone.Itistragic,forexample,thatthedevelopednationswerenotmoreproactiveinsharingAIDSdrugswithpoorcountriesinAfricaandelsewhere,withmillionsofliveslostasaresult.Buttheexponentialimprovementintheprice-performanceofinformationtechnologiesisrapidlymitigatingthisdivide. Drugs are essentially an information technology, and we see the same doubling of price-performance each year as we do with other forms of information technology such as computers,communications, andDNAbase-pair sequencing.AIDSdrugs startedoutnotworkingverywell andcostingtensofthousandsofdollarsperpatientperyear.TodaythesedrugsworkreasonablywellandareapproachingonehundreddollarsperpatientperyearinpoorcountriessuchasthoseinAfrica.Inchapter2IcitedtheWorldBankreportfor2004ofhighereconomicgrowthinthedevelopingworld(over6percent)comparedtotheworldaverage(of4percent),andanoverallreductioninpoverty(forexample,areductionof43percentinextremepovertyintheEastAsianandPacificregionsince1990).Moreover,economistXavierSala-i-Martinexaminedeightmeasuresofglobalinequalityamongindividuals,andfoundthatallweredecliningoverthepastquartercentury.40TheCriticismfromtheLikelihoodofGovernmentRegulation

Theseguystalkinghereactasthoughthegovernmentisnotpartoftheirlives.Theymaywishitweren't,butitis.Asweapproachtheissuestheydebatedheretoday,theyhadbetterbelievethat those issueswill be debated by thewhole country. Themajority ofAmericanswill notsimplysitstillwhilesomeelitestripsofftheirpersonalitiesanduploadsthemselvesintotheircyberspace paradise. They will have something to say about that. There will be vehementdebateaboutthatinthiscountry.

—LEON FUERTH, FORMER NATIONAL SECURITY ADVISER TOVICEPRESIDENTALGORE,ATTHE2002FORESIGHTCONFERENCE

Human lifewithoutdeathwouldbesomethingother thanhuman;consciousnessofmortalitygivesrisetoourdeepestlongingsandgreatestaccomplishments.

—LEON KASS, CHAIR OF THE PRESIDENTIAL COMMISSION ONBIOETHICS,2003

The criticism concerning governmental control is that regulation will slow down and stop theaccelerationoftechnology.Althoughregulationisavitalissue,ithasactuallyhadnomeasurableeffectonthetrendsdiscussedinthisbook,whichhaveoccurredwithextensiveregulationinplace.Shortofaworldwidetotalitarianstate,theeconomicandotherforcesunderlyingtechnicalprogresswillonlygrowwithongoingadvances.Considertheissueofstem-cellresearch,whichhasbeenespeciallycontroversial,andforwhichtheU.S.governmentisrestrictingitsfunding.Stem-cellresearchisonlyoneofnumerousideasconcernedwithcontrollingandinfluencingtheinformationprocessesunderlyingbiologythatarebeingpursuedaspart of the biotechnology revolution. Even within the field of cell therapies the controversy overembryonicstem-cellresearchhasservedonlytoaccelerateotherwaysofaccomplishingthesamegoal.For example, transdifferentiation (converting one type of cell such as a skin cell into other types ofcells)hasmovedaheadquickly.AsIreportedinchapter5,scientistshaverecentlydemonstratedtheabilitytoreprogramskincellsintoseveralothercelltypes.Thisapproachrepresentstheholygrailofcelltherapiesinthatitpromisesanunlimitedsupplyofdifferentiatedcellswiththepatient'sownDNA.ItalsoallowscellstobeselectedwithoutDNAerrorsandwillultimatelybeabletoprovideextendedtelomerestrings(tomakethecellsmoreyouthful).Evenembryonicstem-cellresearchitselfhasmovedahead,forexample,withprojects

likeHarvard'smajornewresearchcenterandCalifornia'ssuccessfulthree-billion-dollarbondinitiativetosupportsuchwork.Althoughtherestrictionsonstem-cellresearchareunfortunate,itishardtosaythatcell-therapyresearch,letalonethebroadfieldofbiotechnology,hasbeenaffectedtoasignificantdegree. Some governmental restrictions reflect the perspective of fundamentalist humanism,which Iaddressedinthepreviouschapter.Forexample,theCouncilofEuropeproclaimedthat"humanrightsimply theright to inheritageneticpattern thathasnotbeenartificiallychanged."41Perhaps themostinterestingaspectofthecouncil'sedictisitsposingarestrictionasaright.Inthesamespirit,Iassumethecouncilwouldadvocatethehumanrightnottobecuredfromnaturaldiseasebyunnaturalmeans,just as activists "protected" starvingAfrican nations from the indignity of consuming bioengineeredcrops.42Ultimatelythebenefitsoftechnicalprogressoverwhelmsuchreflexiveantitechnologysentiments.Themajority of crops in theUnited States are alreadyGMOs,whileAsian nations are aggressivelyadoptingthetechnologytofeedtheirlargepopulations,andevenEuropeisnowbeginningtoapproveGMOfoods.Theissueisimportantbecauseunnecessaryrestrictions,althoughtemporary,canresultinexacerbatedsufferingofmillionsofpeople.Buttechnicalprogressisadvancingonthousandsoffronts,fueledbyirresistibleeconomicgainsandprofoundimprovementsinhumanhealthandwell-being. LeonFuerth's observation quoted above reveals an inherentmisconception about informationtechnologies. Information technologies are not available only to an elite. As discussed, desirableinformation technologies rapidlybecomeubiquitousandalmost free. It isonlywhen theydon'tworkverywell(thatis,inanearlystageofdevelopment)thattheyareexpensiveandrestrictedtoanelite.Earlyintheseconddecadeofthiscentury,theWebwillprovidefullimmersionvisual-auditoryvirtualrealitywithimageswrittendirectlytoourretinasfromoureyeglassesandlensesandveryhigh-bandwidthwirelessInternetaccesswoveninourclothing.Thesecapabilitieswillnotberestrictedjusttotheprivileged.Justlikecellphones,bythetimetheyworkwelltheywillbeeverywhere. In the 2020s we will routinely have nanobots in our bloodstream keeping us healthy andaugmentingourmentalcapabilities.Bythetimetheseworkwell theywillbeinexpensiveandwidelyused. As I discussed above, reducing the lag between early and late adoption of informationtechnologieswillitselfacceleratefromthecurrentten-yearperiodtoonlyacoupleofyearstwodecadesfrom now. Once nonbiological intelligence gets a foothold in our brains, it will at least double incapability each year, as is the nature of information technology. Thus it will not take long for thenonbiologicalportionofourintelligencetopredominate.Thiswillnotbealuxuryreservedfortherich,any more than search engines are today. And to the extent that there will be a debate about thedesirability of such augmentation, it's easy to predict who will win, since those with enhancedintelligencewillbefarbetterdebaters.The Unbearable Slowness of Social Institutions. MIT senior research scientist Joel Cutcher-Gershenfeldwrites:"Justlookingbackoverthecourseofthepastcenturyandahalf,therehavebeenasuccession of political regimeswhere eachwas the solution to an earlier dilemma, but created newdilemmasinthesubsequentera.Forexample,TammanyHallandthepoliticalpatronmodelwereavastimprovementoverthedominantsystembasedonlandedgentry—manymorepeoplewereincludedinthepoliticalprocess.Yet,problemsemergedwithpatronage,which led to thecivil servicemodel—astrongsolutiontotheprecedingproblembyintroducingthemeritocracy.Then,ofcourse,civilservicebecamethebarriertoinnovationandwemovetoreinventinggovernment.Andthestorycontinues."43Gershenfeldispointingoutthatsocialinstitutionsevenwheninnovativeintheirdaybecome"adragoninnovation."FirstIwouldpointoutthattheconservatismofsocialinstitutionsisnotanewphenomenon.Itispartoftheevolutionaryprocessofinnovation,andthelawofacceleratingreturnshasalwaysoperatedinthiscontext.Second,innovationhasawayofworkingaroundthelimitsimposedbyinstitutions.Theadventofdecentralizedtechnologyempowerstheindividualtobypassallkindsofrestrictions,anddoesrepresentaprimarymeansforsocialchangetoaccelerate.Asoneofmanyexamples,theentirethicket

of communications regulations is in the process of being bypassed by emerging point-to-pointtechniquessuchasvoiceoverInternetprotocol(VOIP).Virtualrealitywillrepresentanothermeansofhasteningsocialchange.Peoplewillultimatelybeable to have relationships and engage in activities in immersive and highly realistic virtual-realityenvironmentsthattheywouldnotbeableorwillingtodoinrealreality.Astechnologybecomesmoresophisticateditincreasinglytakesontraditionalhumancapabilitiesandrequireslessadaptation.Youhadtobetechnicallyadepttouseearlypersonalcomputers,whereasusing computerized systems today, such as cell phones, music players, andWeb browsers, requiresmuchlesstechnicalability.Intheseconddecadeofthiscentury,wewillroutinelybeinteractingwithvirtual humans that, although not yet Turing-test capable, will have sufficient natural languageunderstandingtoactasourpersonalassistantsforawiderangeoftasks.Therehasalwaysbeenamixofearlyandlateadoptersofnewparadigms.Westillhavepeopletodaywhowanttoliveaswedidintheseventhcentury.Thisdoesnotrestraintheearlyadoptersfromestablishing new attitudes and social conventions, for example newWeb-based communities.A fewhundredyears ago,onlyahandfulofpeople suchasLeonardodaVinci andNewtonwereexploringnewwaysofunderstandingandrelatingtotheworld.Today,theworldwidecommunitythatparticipatesinandcontributestothesocialinnovationofadoptingandadaptingtonewtechnologicalinnovationisasubstantialportionofthepopulation,anotherreflectionofthelawofacceleratingreturns.TheCriticismfromTheismAnothercommonobjectionexplicitlygoesbeyondsciencetomaintainthatthereisaspirituallevelthataccountsforhumancapabilitiesandthatisnotpenetrablebyobjectivemeans.WilliamA.Dembski,adistinguishedphilosopherandmathematician,decriestheoutlookofsuchthinkersasMarvinMinsky,Daniel Dennett, Patricia Churchland, and RayKurzweil, whom he calls "contemporarymaterialists"who"seethemotionsandmodificationsofmatterassufficienttoaccountforhumanmentality."44Dembskiascribes"predictability[as]materialism'smainvirtue"andcites"hollowness[as]itsmainfault."He goes on to say that "humans have aspirations.We long for freedom, immortality, and thebeatificvision.WearerestlessuntilwefindourrestinGod.Theproblemforthematerialist,however,isthattheseaspirationscannotberedeemedinthecoinofmatter."Heconcludesthathumanscannotbemeremachinesbecauseof"thestrictabsenceofextra-materialfactorsfromsuchsystems."IwouldpreferthatwecallDembski'sconceptofmaterialism"capabilitymaterialism,"orbetteryet"capabilitypatternism,"Capabilitymaterialism/patternismisbasedontheobservationthatbiologicalneurons and their interconnections aremade up of sustainable patterns ofmatter and energy. It alsoholdsthattheirmethodscanbedescribed,understood,andmodeledwitheitherreplicasorfunctionallyequivalentre-creations.Iusetheword"capability"becauseitencompassesallof therich,subtle,anddiverseways inwhich humans interactwith theworld, not just those narrower skills that onemightlabelasintellectual.Indeed,ourabilitytounderstandandrespondtoemotionsisatleastascomplexanddiverseasourabilitytoprocessintellectualissues.JohnSearle,forexample,acknowledgesthathumanneuronsarebiologicalmachines.Fewseriousobservers have postulated capabilities or reactions of human neurons that require Dembski's "extra-materialfactors."Relyingonthepatternsofmatterandenergyinthehumanbodyandbraintoexplainitsbehaviorandproficienciesneednotdiminishourwondermentatitsremarkablequalities.Dembskihasanoutdatedunderstandingoftheconceptof"machine." Dembskialsowrites that"unlikebrains,computersareneatandprecise....[C]omputersoperatedeterministically."Thisstatementandothersrevealaviewofmachines,orentitiesmadeupofpatternsofmatterandenergy("material"entities), that is limitedto the literallysimplemindedmechanismsofnineteenth-centuryautomatons.Thesedevices,with theirhundredsandeven thousandsofparts,werequitepredictableandcertainlynotcapableoflongingsforfreedomandothersuchendearingqualitiesofthehumanentity.Thesameobservationslargelyholdtruefor today'smachines,withtheirbillionsof

parts. But the same cannot necessarily be said for machines withmillions of billions of interacting"parts,"entitieswiththecomplexityofthehumanbrainandbody. Moreoveritisincorrecttosaythatmaterialismispredictable.Eventoday'scomputerprogramsroutinelyuse simulated randomness. Ifoneneeds truly randomevents inaprocess, therearedevicesthatcanprovidethisaswell.Fundamentally,everythingweperceiveinthematerialworldistheresultof many trillions of quantum events, each of which displays a profound and irreducible quantumrandomness at the core of physical reality (or so it seems—the scientific jury is still out on the truenatureoftheapparentrandomnessunderlyingquantumevents}.Thematerialworld—atboththemacroandmicrolevels—isanythingbutpredictable. Althoughmany computer programs do operate thewayDembski describes, the predominanttechniquesinmyownfieldofpatternrecognitionusebiology-inspiredchaotic-computingmethods.Inthesesystemstheunpredictableinteractionofmillionsofprocesses,manyofwhichcontainrandomandunpredictableelements,provideunexpectedyetappropriateanswerstosubtlequestionsofrecognition.Thebulkofhumanintelligenceconsistsofjustthesesortsofpattern-recognitionprocesses. As for our responses to emotions and our highest aspirations, these are properly regarded asemergentproperties—profoundones tobesurebutnonethelessemergentpatterns thatresult fromtheinteraction of the human brain with its complex environment. The complexity and capacity ofnonbiological entities is increasing exponentially and will match biological systems including thehumanbrain(alongwith therestof thenervoussystemandtheendocrinesystem)withinacoupleofdecades. Indeed, many of the designs of future machines will be biologically inspired—that is,derivativeofbiologicaldesigns. (This isalready trueofmanycontemporarysystems.) It ismy thesisthatbysharingthecomplexityaswellastheactualpatternsofhumanbrains,thesefuturenonbiologicalentitieswilldisplaytheintelligenceandemotionallyrichreactions(suchas"aspirations")ofhumans.Willsuchanonbiologicalentitybeconscious?Searleclaimsthatwecan(atleastintheory)readilyresolvethisquestionbyascertainingifithasthecorrect"specificneurobiologicalprocesses."Itismyviewthatmanyhumans,ultimatelythevastmajorityofhumans,willcometobelievethatsuchhuman-derived but nonetheless nonbiological intelligent entities are conscious, but that's a political andpsychological prediction, not a scientific or philosophical judgment. My bottom line: I agree withDembski that this is not a scientific question, because it cannot be resolved through objectiveobservation.Someobserverssaythatifit'snotascientificquestion,it'snotanimportantorevenarealquestion.Myview(andI'msureDembskiagrees)isthatpreciselybecausethequestionisnotscientific,itisaphilosophicalone—indeed,thefundamentalphilosophicalquestion. Dembski writes: "We need to transcend ourselves to find ourselves. Now the motions andmodificationsofmatteroffernoopportunityfortranscendingourselves....Freud...Marx...Nietzsche,...eachregardedthehopefortranscendenceasadelusion."Thisviewoftranscendenceasanultimategoalisreasonablystated.ButIdisagreethatthematerialworldoffersno"opportunityfortranscending."Thematerial world inherently evolves, and each stage transcends the stage before it. As I discussed inchapter 7, evolution moves toward greater complexity, greater elegance, greater knowledge, greaterintelligence,greaterbeauty,greatercreativity,greater love.AndGodhasbeencalledall these things,onlywithoutanylimitation:infiniteknowledge,infiniteintelligence,infinitebeauty,infinitecreativity,and infinite love. Evolution does not achieve an infinite level, but as it explodes exponentially itcertainlymovesinthatdirection.SoevolutionmovesinexorablytowardourconceptionofGod,albeitneverreachingthisideal.Dembskicontinues:

A machine is fully determined by the constitution, dynamics, and interrelationships of itsphysical parts "[M]achines" stresses the strict absence of extra-material factors Thereplacementprinciple is relevant to thisdiscussionbecause it implies thatmachineshavenosubstantive history....But a machine, properly speaking, has no history. Its history is asuperfluous rider—an addendum that could easily have been different without altering themachine....Foramachine,allthatisiswhatitisatthismoment....Machinesaccessorfailtoaccess items in storage....Mutatis mutandis, items that represent counterfactual occurrences

(i.e., things that never happened) but which are accessible can be, as far as themachine isconcerned,justasthoughtheydidhappen.

It need hardly be stressed that thewhole point of this book is thatmany of our dearly heldassumptions about the nature of machines and indeed of our own human nature will be called intoquestion in the next several decades.Dembski's conception of "history" is just another aspect of ourhumanitythatnecessarilyderivesfromtherichness,depth,andcomplexityofbeinghuman.Conversely,nothavingahistoryintheDembskisenseisjustanotherattributeofthesimplicityofthemachinesthatwehaveknownuptothistime.Itispreciselymythesisthatmachinesofthe2030sandbeyondwillbeof such great complexity and richness of organization that their behavior will evidence emotionalreactions,aspirations,and,yes,history.SoDembskiismerelydescribingtoday'slimitedmachinesandjustassumingthat theselimitationsareinherent,a lineofargumentequivalent tostatingthat"today'smachinesarenotascapableashumans,thereforemachineswillneverreachthislevelofperformance."Dembskiisjustassuminghisconclusion. Dembski's viewof the ability ofmachines to understand their own history is limited to their"accessing" items in storage. Futuremachines, however,will possess not only a record of their ownhistorybut an ability tounderstand thathistory and to reflect insightfullyupon it.As for "items thatrepresentcounterfactualoccurrences,"surelythesamecanbesaidforourhumanmemories.Dembski'slengthydiscussionofspiritualityissummedupthus:

ButhowcanamachinebeawareofGod'spresence?Recallthatmachinesareentirelydefinedbytheconstitution,dynamics,andinterrelationshipsamongtheirphysicalparts.ItfollowsthatGodcannotmakehispresenceknowntoamachinebyactinguponitandtherebychangingitsstate.Indeed,themomentGodactsuponamachinetochangeitsstate,itnolongerproperlyisamachine,foranaspectofthemachinenowtranscendsitsphysicalconstituents.ItfollowsthatawarenessofGod'spresencebyamachinemustbeindependentofanyactionbyGodtochangethestateof themachine.How thendoes themachinecome toawarenessofGod'spresence?Theawarenessmustbeself-induced.Machinespiritualityisthespiritualityofself-realization,not the spirituality of an activeGodwho freely gives himself in self-revelation and therebytransformsthebeingswithwhichheisincommunion.ForKurzweiltomodify"machine"withtheadjective"spiritual"thereforeentailsanimpoverishedviewofspirituality.

Dembskistatesthatanentity(forexample,aperson)cannotbeawareofGod'spresencewithoutGod'sactinguponher,yetGodcannotactuponamachine,sothereforeamachinecannotbeawareofGod'spresence.Such reasoning is entirely tautological andhuman-centric.Godcommunesonlywithhumans, andonlybiologicalones at that. I havenoproblemwithDembski's subscribing to this as apersonalbelief,buthefailstomakethe"strongcase"thathepromises,that"humansarenotmachines—period."AswithSearle,Dembskijustassumeshisconclusion. LikeSearle,Dembskicannotseemtograsptheconceptof theemergentpropertiesofcomplexdistributedpatterns.Hewrites:

Anger presumably is correlated with certain localized brain excitations. But localized brainexcitations hardly explain anger any better than overt behaviors associated with anger, likeshouting obscenities. Localized brain excitations may be reliably correlated with anger, butwhataccountsforonepersoninterpretingacommentasaninsultandexperiencinganger,andanother person interpreting that same comment as a joke and experiencing laughter? A fullmaterialist accountofmindneeds tounderstand localizedbrain excitations in termsofotherlocalizedbrainexcitations.Insteadwefindlocalizedbrainexcitations(representing,say,anger)having to be explained in terms of semantic contents (representing, say, insults). But thismixtureofbrainexcitationsandsemanticcontentshardlyconstitutesamaterialistaccountofmindorintelligentagency.

Dembskiassumesthatangeriscorrelatedwitha"localizedbrainexcitation,"butangerisalmostcertainly the reflection of complex distributed patterns of activity in the brain. Even if there is alocalizedneuralcorrelateassociatedwithanger,itnonethelessresultsfrommultifacetedandinteractingpatterns. Dembski's question as towhy different people react differently to similar situations hardlyrequires us to resort to his extramaterial factors for an explanation. The brains and experiences ofdifferentpeopleareclearlynotthesame,andthesedifferencesarewellexplainedbydifferencesintheirphysicalbrainsresultingfromvaryinggenesandexperiences.Dembski'sresolutionoftheontologicalproblemisthattheultimatebasisofwhatexistsiswhathecalls the "real world of things" that are not reducible to material stuff. Dembski does not list what"things"wemightconsiderasfundamental,butpresumablyhumanmindswouldbeonthelist,asmightbeotherthings,suchasmoneyandchairs.Theremaybeasmallcongruenceofourviewsinthisregard.I regard Dembski's "things" as patterns. Money, for example, is a vast and persisting pattern ofagreements,understandings,andexpectations."RayKurzweil"isperhapsnotsovastapatternbutthusfarisalsopersisting.Dembskiapparentlyregardspatternsasephemeralandnotsubstantial,butIhaveaprofoundrespectforthepowerandenduranceofpatterns.Itisnotunreasonabletoregardpatternsasafundamentalontological reality.Weareunable to really touchmatter andenergydirectly,butwedodirectlyexperiencethepatternsunderlyingDembski's"things."Fundamentaltothisthesisisthatasweapply our intelligence, and the extension of our intelligence called technology, to understanding thepowerful patterns in our world (for example, human intelligence), we can re-create—and extend!—thesepatternsinothersubstrates.Thepatternsaremoreimportantthanthematerialsthatembodythem.Finally,ifDembski'sintelligence-enhancingextramaterialstuffreallyexists,thenI'dliketoknowwhereIcangetsome.TheCriticismfromHolismAnothercommoncriticismsaysthefollowing:machinesareorganizedasrigidlystructuredhierarchiesofmodules,whereasbiologyisbasedonholistically-organizedelementsinwhicheveryelementaffectseveryother.Theuniquecapabilitiesofbiology(suchashumanintelligence)canresultonlyfromthistypeofholisticdesign.Furthermore,onlybiologicalsystemscanusethisdesignprinciple.MichaelDenton,abiologistattheUniversityofOtagoinNewZealand,pointsouttheapparentdifferencesbetweenthedesignprinciplesofbiologicalentitiesandthoseofthemachineshehasknown.Denton eloquently describes organisms as "self-organizing, self-referential, ... self-replicating, ...reciprocal,...self-formative,andholistic."45Hethenmakestheunsupportedleap—aleapoffaith,onemight say—that such organic forms can be created only through biological processes and that suchformsare"immutable,...impenetrable,and...fundamental"realitiesofexistence.IdoshareDenton's"awestruck"senseof"wonderment"atthebeauty,intricacy,strangeness,andinterrelatedness of organic systems, ranging from the "eerie other-worldly ... impression" left byasymmetricproteinshapes to theextraordinarycomplexityofhigher-orderorganssuchas thehumanbrain. Further, I agree with Denton that biological design represents a profound set of principles.However, it is precisely my thesis, which neither Denton nor other critics from the holistic schoolacknowledge or respond to, thatmachines (that is, entities derivative of human-directed design) canaccess—andalreadyareusing—thesesameprinciples.Thishasbeen the thrustofmyownworkandrepresentsthewaveofthefuture.Emulatingtheideasofnatureisthemosteffectivewaytoharnesstheenormouspowersthatfuturetechnologywillmakeavailable. Biologicalsystemsarenotcompletelyholistic,andcontemporarymachinesarenotcompletelymodular;bothexistonacontinuum.Wecanidentifyunitsoffunctionalityinnaturalsystemsevenatthemolecular level, and discerniblemechanisms of action are evenmore evident at the higher level oforgans and brain regions. The process of understanding the functionality and informationtransformationsperformedinspecificbrainregionsiswellunderway,aswediscussedinchapter4.Itismisleadingtosuggestthateveryaspectofthehumanbraininteractswitheveryotheraspect

and that it is therefore impossible tounderstand itsmethods.Researchershavealready identifiedandmodeled the transformations of information in several dozen of its regions. Conversely there arenumerous examples of contemporarymachines thatwere not designed in amodular fashion, and inwhichmanyofthedesignaspectsaredeeplyinterconnected,suchastheexamplesofgeneticalgorithmsdescribedinchapter5.Dentonwrites:

Todayalmost allprofessionalbiologistshaveadopted themechanistic/ reductionist approachand assume that the basic parts of an organism (like the cogs of a watch) are the primaryessentialthings,thatalivingorganism(likeawatch)isnomorethanthesumofitsparts,andthatitisthepartsthatdeterminethepropertiesofthewholeandthat(likeawatch)acompletedescription of all the properties of an organism may be had by characterizing its parts inisolation.

Denton,too,isignoringheretheabilityofcomplexprocessestoexhibitemergentpropertiesthatgobeyond"itspartsinisolation."Heappearstorecognizethispotentialinnaturewhenhewrites:"Inaveryrealsenseorganicforms...representgenuinelyemergentrealities."However,itishardlynecessarytoresort toDenton's"vitalisticmodel" toexplainemergent realities.Emergentpropertiesderivefromthepowerofpatterns,andnothingrestrictspatternsandtheiremergentpropertiestonaturalsystems.Dentonappearstoacknowledgethefeasibilityofemulatingthewaysofnaturewhenhewrites:

Successinengineeringneworganicformsfromproteinsuptoorganismswillthereforerequirea completelynovel approach, a sort of designing from"the topdown."Because theparts oforganicwholesonlyexistinthewhole,organicwholescannotbespecifiedbitbybitandbuiltupfromasetofrelativelyindependentmodules;consequentlytheentireundividedunitymustbespecifiedtogetherintoto.

HereDentonprovidessoundadviceanddescribesanapproach toengineering that Iandotherresearchers use routinely in the areas of pattern recognition, complexity (chaos) theory, and self-organizing systems. Denton appears to be unaware of these methodologies, however, and afterdescribingexamplesofbottom-up,component-drivenengineeringandtheirlimitationsconcludeswithnojustificationthatthereisanunbridgeablechasmbetweenthetwodesignphilosophies.Thebridgeis,infact,alreadyunderconstruction. AsIdiscussed inchapter5,wecancreateourown"eerieother-worldly"buteffectivedesignsthroughappliedevolution. Idescribedhow toapply theprinciplesofevolution tocreating intelligentdesigns through genetic algorithms. In my own experience with this approach, the results are wellrepresentedbyDenton'sdescriptionoforganicmoleculesinthe"apparentillogicofthedesignandthelackofanyobviousmodularityorregularity, ... thesheerchaosofthearrangement, ... [andthe]non-mechanicalimpression."Geneticalgorithmsandotherbottom-upself-organizingdesignmethodologies(suchasneuralnets,Markovmodels,andothersthatwediscussedinchapter5)incorporateanunpredictableelement,sothattheresultsofsuchsystemsaredifferenteverytimetheprocessisrun.Despitethecommonwisdomthatmachines aredeterministic and thereforepredictable, there arenumerous readily available sourcesofrandomnessavailabletomachines.Contemporarytheoriesofquantummechanicspostulateaprofoundrandomnessatthecoreofexistence.Accordingtocertaintheoriesofquantummechanics,whatappearsto be the deterministic behavior of systems at a macro level is simply the result of overwhelmingstatistical preponderances based on enormous numbers of fundamentally unpredictable events.Moreover, thework of StephenWolfram and others has demonstrated that even a system that is intheory fully deterministic can nonetheless produce effectively random and, most important, entirelyunpredictableresults.Geneticalgorithmsandsimilarself-organizingapproachesgiverisetodesignsthatcouldnothavebeenarrivedatthroughamodularcomponent-drivenapproach.The"strangeness,...[the]chaos,...thedynamic interaction" of parts to the whole that Denton attributes exclusively to organic structures

describeverywellthequalitiesoftheresultsofthesehuman-initiatedchaoticprocesses.InmyownworkwithgeneticalgorithmsIhaveexaminedtheprocessbywhichsuchanalgorithmgraduallyimprovesadesign.Ageneticalgorithmdoesnotaccomplishitsdesignachievementsthroughdesigningindividualsubsystemsoneatatimebuteffectsanincremental"allatonce"approach,makingmany small distributed changes throughout the design that progressively improve the overall fit or"power" of the solution. The solution itself emerges gradually and unfolds from simplicity tocomplexity.Whilethesolutionsitproducesareoftenasymmetricandungainlybuteffective,justasinnature,theycanalsoappearelegantandevenbeautiful. Denton iscorrect inobserving thatmostcontemporarymachines, suchas today'sconventionalcomputers, are designed using the modular approach. There are certain significant engineeringadvantages to this traditional technique.Forexample,computershavemuchmoreaccuratememoriesthan humans and can perform logical transformations far more effectively than unaided humanintelligence.Most important, computers can share theirmemories andpatterns instantly.The chaoticnonmodularapproachofnaturealsohasclearadvantagesthatDentonwellarticulates,asevidencedbythedeeppowersofhumanpatternrecognition.Butitisawhollyunjustifiedleaptosaythatbecauseofthe current (and diminishing!) limitations of human-directed technology that biological systems areinherently,evenontologically,aworldapart.Theexquisitedesignsofnature(theeye,forexample)havebenefitedfromaprofoundevolutionaryprocess.Ourmostcomplexgeneticalgorithmstodayincorporategeneticcodes'oftensofthousandsofbits,whereasbiologicalentitiessuchashumansarecharacterizedbygeneticcodesofbillionsofbits(onlytensofmillionsofbyteswithcompression). However, as is the case with all information-based technology, the complexity of geneticalgorithms and other nature-inspired methods is increasing exponentially. If we examine the rate atwhichthiscomplexityisincreasing,wefindthattheywillmatchthecomplexityofhumanintelligencewithinabouttwodecades,whichisconsistentwithmyestimatesdrawnfromdirecttrendsinhardwareandsoftware.Dentonpointsoutwehavenotyetsucceededinfoldingproteinsinthreedimensions,"evenoneconsistingofonly100components."However,itisonlyintherecentfewyearsthatwehavehadthetoolseventovisualizethesethree-dimensionalpatterns.Moreover,modelingtheinteratomicforceswillrequireontheorderofonehundredthousandbillion(1014)calculationspersecond.Inlate2004IBMintroducedaversionof itsBlueGene/L supercomputerwith a capabilityof seventy teraflops (nearly1014cps),which,asthenamesuggests,isexpectedtoprovidetheabilitytosimulateproteinfolding. Wehavealready succeeded incutting, splicing, and rearranginggenetic codesandharnessingnature'sownbiochemicalfactories toproduceenzymesandothercomplexbiologicalsubstances.It istrue that most contemporary work of this type is done in two dimensions, but the requisitecomputationalresourcestovisualizeandmodelthefarmorecomplexthree-dimensionalpatternsfoundinnaturearenotfarfromrealization.IndiscussionsoftheproteinissuewithDentonhimself,heacknowledgedthattheproblemwouldeventuallybesolved,estimatingthatitwasperhapsadecadeaway.Thefactthatacertaintechnicalfeathasnotyetbeenaccomplishedisnotastrongargumentthatitneverwillbe.Dentonwrites:

Fromknowledgeof thegenesofanorganismit is impossible topredict theencodedorganicforms.Neitherthepropertiesnorstructureofindividualproteinsnorthoseofanyhigherorderforms—such as ribosomes andwhole cells—can be inferred even from themost exhaustiveanalysisofthegenesandtheirprimaryproducts,linearsequencesofaminoacids.

AlthoughDenton'sobservationaboveisessentiallycorrect,itbasicallypointsoutthatthegenomeisonlypartoftheoverallsystem.TheDNAcodeisnotthewholestory,andtherestofthemolecularsupportsystemisrequiredforthesystemtoworkandforittobeunderstood.Wealsoneedthedesignof the ribosomeandothermolecules thatmake theDNAmachineryfunction.However,adding thesedesignsdoesnotsignificantlychangetheamountofdesigninformationinbiology.

Butre-creatingthemassivelyparallel,digitallycontrolledanalog,hologramlike,self-organizing,andchaoticprocessesofthehumanbraindoesnotrequireustofoldproteins.Asdiscussedinchapter4there are dozens of contemporary projects that have succeeded in creating detailed re-creations ofneurological systems. These include neural implants that successfully function inside people's brainswithout folding any proteins. However, while I understand Denton's argument about proteins to beevidenceregardingtheholisticwaysofnature,asIhavepointedouttherearenoessentialbarrierstoouremulatingthesewaysinourtechnology,andwearealreadywelldownthispath.Insummary,Dentonisfartooquicktoconcludethatcomplexsystemsofmatterandenergyinthephysicalworldare incapableofexhibiting the"emergent ...vitalcharacteristicsoforganismssuchasself-replication,'morphing,'self-regeneration,self-assemblyandtheholisticorderofbiologicaldesign"and that, therefore, "organisms andmachines belong to different categories of being."Dembski andDentonsharethesamelimitedviewofmachinesasentitiesthatcanbedesignedandconstructedonlyinamodularway.Wecanbuildandalreadyarebuilding"machines"thathavepowersfargreaterthanthesum of their parts by combining the self-organizing design principles of the natural world with theacceleratingpowersofourhuman-initiatedtechnology.Itwillbeaformidablecombination.

H

Epilogue

IdonotknowwhatImayappeartotheworld,buttomyselfIseemtohavebeenonlylikeaboyplayingontheseashore,anddivertingmyselfinnowandthenfindingasmootherpebbleoraprettiershellthanordinary,whilstthegreatoceanoftruthlayundiscoveredbeforeme.

—ISAACNEWTON1

Themeaning of life is creative love. Not love as an inner feeling, as a private sentimentalemotion,butloveasadynamicpowermovingoutintotheworldanddoingsomethingoriginal.

—TOMMORRIS,IFARISTOTLERANNGENERALMOTORSNoexponentialisforever...butwecandelay"forever."

—GORDONE.MOORE,2004

owSingular?How singular is theSingularity?Will it happen in an instant?Let's consideragain thederivationof theword. Inmathematicsa singularity isavalue that isbeyondanylimit-inessence,infinity.(Formallythevalueofafunctionthatcontainssuchasingularityissaidtobeundefinedatthesingularitypoint,butwecanshowthatthevalueofthefunctionatnearbypointsexceedsanyspecificfinitevalue).2

The Singularity, as we have discussed it in this book, does not achieve infinite levels ofcomputation,memory,oranyothermeasurableattribute.Butitcertainlyachievesvastlevelsofallofthesequalities,includingintelligence.Withthereverseengineeringofthehumanbrainwewillbeabletoapplytheparallel,self-organizing,chaoticalgorithmsofhumanintelligencetoenormouslypowerfulcomputational substrates.This intelligencewill thenbe inaposition to improve itsowndesign,bothhardwareandsoftware,inarapidlyacceleratingiterativeprocess.Buttherestillappearstobealimit.Thecapacityoftheuniversetosupportintelligenceappearstobeonlyabout1090calculationspersecond,asIdiscussedinchapter6.Therearetheoriessuchastheholographicuniversethatsuggestthepossibilityofhighernumbers(suchas10120),buttheselevelsarealldecidedlyfinite.Ofcourse,thecapabilitiesofsuchanintelligencemayappearinfiniteforallpracticalpurposestoourcurrentlevelofintelligence.Auniversesaturatedwithintelligenceat1090cpswouldbeonetrilliontrillion trillion trillion trillion timesmorepowerful thanall biologicalhumanbrainsonEarth today.3

Evenaone-kilogram"cold"computerhasapeakpotentialof1042cps,asIreviewedinchapter3,whichistenthousandtrillion(1016)timesmorepowerfulthanallbiologicalhumanbrains.4Giventhepowerofexponentialnotation,wecaneasilyconjureupbiggernumbers,evenifwelackthe imagination tocontemplateallof their implications.Wecan imagine thepossibilityofour futureintelligence spreading into other universes. Such a scenario is conceivable given our currentunderstandingofcosmology,althoughspeculative.Thiscouldpotentiallyallowourfutureintelligencetogobeyondanylimits.Ifwegainedtheabilitytocreateandcolonizeotheruniverses(andifthereisa

way to do this, the vast intelligence of our future civilization is likely to be able to harness it), ourintelligencewouldultimatelybecapableofexceedinganyspecificfinitelevel.That'sexactlywhatwecansayforsingularitiesinmathematicalfunctions.Howdoesouruseof"singularity"inhumanhistorycomparetoitsuseinphysics?Thewordwasborrowedfrommathematicsbyphysics,whichhasalwaysshownapenchantforanthropomorphicterms(suchas"charm"and"strange"fornamesofquarks).Inphysics"singularity"theoreticallyrefers toapointofzerosizewithinfinitedensityofmassandthereforeinfinitegravity.But because of quantumuncertainty there is no actual point of infinite density, and indeed quantummechanicsdisallowsinfinitevalues. Just like the Singularity as I have discussed it in this book, a singularity in physics denotesunimaginablylargevalues.Andtheareaofinterestinphysicsisnotactuallyzeroinsizebutratherisanevent horizon around the theoretical singularity point inside a black hole (which is not even black).Insidetheeventhorizonparticlesandenergy,suchaslight,cannotescapebecausegravityistoostrong.Thusfromoutsidetheeventhorizon,wecannotseeeasilyinsidetheeventhorizonwithcertainty.However,theredoesappeartobeawaytoseeinsideablackhole,becauseblackholesgiveoffashowerofparticles.Particle-antiparticlepairsarecreatedneartheeventhorizon(ashappenseverywherein space), and for some of these pairs, one of the pair is pulled into the black holewhile the othermanages to escape. These escaping particles form a glow calledHawking radiation, named after itsdiscoverer,StephenHawking.Thecurrentthinkingisthatthisradiationdoesreflect(inacodedfashion,andasaresultofaformofquantumentanglementwiththeparticlesinside)whatishappeninginsidetheblackhole.Hawkinginitiallyresistedthisexplanationbutnowappearstoagree. So,we find our use of the term "Singularity" in this book to be no less appropriate than thedeployment of this term by the physics community. Just aswe find it hard to see beyond the eventhorizon of a black hole, we also find it difficult to see beyond the event horizon of the historicalSingularity.How canwe,with our brains each limited to 1016 to1019 cps, imaginewhat our futurecivilizationin2099withits1060cpswillbecapableofthinkinganddoing? Nevertheless, just as we can draw conclusions about the nature of black holes through ourconceptual thinking, despite never having actually been inside one, our thinking today is powerfulenoughtohavemeaningfulinsightsintotheimplicationsoftheSingularity.That'swhatI'vetriedtodointhisbook.HumanCentrality.Acommonviewisthatsciencehasconsistentlybeencorrectingouroverlyinflatedviewofourownsignificance.StephenJayGouldsaid, "Themost important scientific revolutionsallinclude,as theironlycommonfeature, thedethronementofhumanarrogancefromonepedestalafteranotherofpreviousconvictionsaboutourcentralityinthecosmos."5 Butitturnsoutthatwearecentral,afterall.Ourabilitytocreatemodels-virtualrealities-inourbrains, combined with our modest-looking thumbs, has been sufficient to usher in another form ofevolution: technology.Thatdevelopment enabled thepersistenceof the acceleratingpace that startedwithbiologicalevolution.Itwillcontinueuntiltheentireuniverseisatourfingertips.

ResourcesandContactInformationSingularity.comNewdevelopmentsinthediversefieldsdiscussedinthisbookareaccumulatingatanacceleratingpace.Tohelpyoukeeppace,IinviteyoutovisitSingularity.com,whereyouwillfind

·Recentnewsstories·Acompilationofthousandsofrelevantnewsstoriesgoingbackto2001fromKurzweilAI.net(seebelow)

·HundredsofarticlesonrelatedtopicsfromKurzweilAI.net·Researchlinks·Dataandcitationforallgraphs·Materialaboutthisbook·Excerptsfromthisbook·Onlineendnotes

KurzweilAI.netYou are also invited to visit our award-winningWeb site, KurzweilAI.net, which includes over sixhundredarticlesbyoveronehundred"bigthinkers"(manyofwhomarecitedinthisbook),thousandsofnewsarticles,listingsofevents,andotherfeatures.Overthepastsixmonths,wehavehadmorethanonemillionreaders.MemesonKurzweilAI.netinclude:

·TheSingularity·WillMachinesBecomeConscious?·LivingForeveroHowtoBuildaBrain·VirtualRealities·Nanotechnology·DangerousFutures·VisionsoftheFuture·Point/Counterpoint

Youcansignupforourfree(dailyorweekly)e-newsletterbyputtingyoure-mailaddressinthesimple one-line form on theKurzweilAI.net home page.We do not share your e-mail address withanyone.Fantastic-Voyage.netandRayandTerry.comFor those of youwhowould like to optimize your health today, and tomaximize your prospects ofliving longenough to actuallywitness andexperience theSingularity, visitFantastic-Voyage.net andRayandTerry.com. I developed these sites with Terry Grossman, M.D., my health collaborator and

coauthor of Fantastic Voyage: Live Long Enough to Live Forever. These sites contain extensiveinformationaboutimprovingyourhealthwithtoday'sknowledgesothatyoucanbeingoodhealthandspiritswhenthebiotechnologyandnanotechnologyrevolutionsarefullymature.ContactingtheAuthorRayKurzweilcanbereachedatray@singularity.com.

APPENDIX

TheLawofAcceleratingReturnsRevisitedThefollowinganalysisprovidesthebasisofunderstandingevolutionarychangeasadoublyexponentialphenomenon(thatis,exponentialgrowthinwhichtherateofexponentialgrowth-theexponent-isitselfgrowingexponentially).Iwilldescribeherethegrowthofcomputationalpower,althoughtheformulasare similar for other aspects of evolution, especially information-based processes and technologies,including our knowledge of human intelligence, which is a primary source of the software ofintelligence.Weareconcernedwiththreevariables:V:Velocity(thatis,power)ofcomputation(measuredincalculationspersecondperunitcost)W:Worldknowledgeasitpertainstodesigningandbuildingcomputationaldevicest:TimeAsafirst-orderanalysis,weobservethatcomputerpowerisalinearfunctionofW.Wealsonotethat W is cumulative. This is based on the observation that relevant technology algorithms areaccumulated in an incremental way. In the case of the human brain, for example, evolutionarypsychologistsarguethatthebrainisamassivelymodularintelligencesystem,evolvedovertimeinanincremental manner. Also, in this simple model, the instantaneous increment to knowledge isproportional to computational power. These observations lead to the conclusion that computationalpowergrowsexponentiallyovertime.Inotherwords,computerpowerisalinearfunctionoftheknowledgeofhowtobuildcomputers.Thisisactuallyaconservativeassumption.Ingeneral,innovationsimproveVbyamultiple,notinanadditiveway. Independent innovations (each representing a linear increment to knowledge)multiplyone another's effects. For example, a circuit advance such as CMOS (complementary metal oxidesemiconductor),amoreefficientICwiringmethodology,aprocessorinnovationsuchaspipelining,oranalgorithmicimprovementsuchasthefastFouriertransform,allincreaseVbyindependentmultiples.Asnoted,ourinitialobservationsare:Thevelocityofcomputationisproportionaltoworldknowledge:(1)

Therateofchangeofworldknowledgeisproportionaltothevelocityofcomputation:(2)

Substituting(1)into(2)gives:(3)

Thesolutiontothisis:(4)

andWgrowsexponentiallywithtime(eisthebaseofthenaturallogarithms). ThedatathatI'vegatheredshowsthat thereisexponentialgrowthintherateof(exponentfor)exponentialgrowth(wedoubledcomputerpowereverythreeyearsearly in thetwentiethcenturyandeverytwoyearsinthemiddleofthecentury,andaredoublingiteveryoneyearnow).Theexponentiallygrowingpoweroftechnologyresultsinexponentialgrowthoftheeconomy.Thiscanbeobservedgoingbackatleastacentury.Interestingly,recessions,includingtheGreatDepression,canbemodeledasafairlyweakcycleontopoftheunderlyingexponentialgrowth.Ineachcase,theeconomy"snapsback"towhereitwouldhavebeenhadtherecession/depressionneverexistedinthefirstplace.Wecanseeeven more rapid exponential growth in specific industries tied to the exponentially growingtechnologies,suchasthecomputerindustry.Ifwefactorintheexponentiallygrowingresourcesforcomputation,wecanseethesourceforthesecondlevelofexponentialgrowth.Onceagainwehave:(5)

But now we include the fact that the resources deployed for computation, N, are also growingexponentially:(6)

The rate of change of world knowledge is now proportional to the product of the velocity ofcomputationandthedeployedresources:(7)

Substituting(5)and(6)into(7)weget:(8)

Thesolutiontothisis:(9)

andworldknowledgeaccumulatesatadoubleexponentialrate.Nowlet'sconsidersomereal-worlddata.Inchapter3,Iestimatedthecomputationalcapacityofthe human brain, based on the requirements for functional simulation of all brain regions, to beapproximately 1016 cps. Simulating the salient nonlinearities in every neuron and interneuronalconnectionwouldrequireahigherlevelofcomputing:1011neuronstimesanaverage103connectionsperneuron(withthecalculationstakingplaceprimarilyintheconnections)times102transactionspersecondtimes103calculationsper transaction—atotalofabout1019cps.Theanalysisbelowassumesthelevelforfunctionalsimulation(1016cps).

AnalysisThree

Consideringthedataforactualcalculatingdevicesandcomputersduringthetwentiethcentury:LetS=cps/$1K:calculationspersecondfor$1,000.Twentieth-centurycomputingdatamatches:

Wecandeterminethegrowthrate,G,overaperiodoftime:

whereSciscps/$1Kforcurrentyear,Spiscps/$1Kofpreviousyear,Yciscurrentyear,andYpispreviousyear.Humanbrain=1016calculationspersecond.Humanrace=10billion(1010)humanbrains=1026calculationspersecond.Weachieveonehumanbraincapability(1016cps)for$1,000aroundtheyear2023.Weachieveonehumanbraincapability(1016cps)foronecentaroundtheyear2037.Weachieveonehumanracecapability(1026cps)for$1,000aroundtheyear2049.

Ifwe factor in the exponentially growing economy, particularlywith regard to the resourcesavailable for computation (already about one trilliondollarsper year),we can see that nonbiologicalintelligencewillbebillionsoftimesmorepowerfulthanbiologicalintelligencebeforethemiddleofthecentury.Wecanderivethedoubleexponentialgrowthinanotherway.Inotedabovethattherateofaddingknowledge (dW/dt)was at least proportional to the knowledge at each point in time. This is clearlyconservativegiventhatmanyinnovations(incrementstoknowledge)haveamultiplicativeratherthanadditiveimpactontheongoingrate.However,ifwehaveanexponentialgrowthrateoftheform:(10)

whereC>1,thishasthesolution:(11)

whichhasaslowlogarithmicgrowthwhile t<1/lnCbut thenexplodesclosetothesingularityat t =1/lnC.EventhemodestdW/dt=W2resultsinasingularity.Indeedanyformulawithapowerlawgrowthrateoftheform:(12)

wherea>1,leadstoasolutionwithasingularity:(12)

atthetimeT.Thehigherthevalueofa,thecloserthesingularity.Myviewisthatitishardtoimagineinfiniteknowledge,givenapparentlyfiniteresourcesofmatterand energy, and the trends to datematch a double exponential process. The additional term (toW)appearstobeoftheformWÎlog(W).Thistermdescribesanetworkeffect.IfwehaveanetworksuchastheInternet,itseffectorvaluecanreasonablybeshowntobeproportionaltonÎlog(n)wherenisthenumberofnodes.Eachnode(eachuser)benefits,sothisaccountsforthenmultiplier.Thevaluetoeachuser(toeachnode)=log(n).BobMetcalfe(inventorofEthernet)haspostulatedthevalueofanetworkofnnodes=cÎn2,butthisisoverstated.IftheInternetdoublesinsize,itsvaluetomedoesincreasebutitdoesnotdouble.Itcanbeshownthatareasonableestimateisthatanetwork'svaluetoeachuserisproportionaltothelogofthesizeofthenetwork.Thus,itsoverallvalueisproportionaltonÎlog(n).Ifthegrowthrateinsteadincludesalogarithmicnetworkeffect,wegetanequationfortherateofchangethatisgivenby:(14)

Thesolutiontothisisadoubleexponential,whichwehaveseenbeforeinthedata:(15)

Notes

Prologue:ThePowerofIdeas1.Mymotherisatalentedartistspecializinginwatercolorpaintings.Myfatherwasanoted

musician,conductoroftheBellSymphony,founderandformerchairmanoftheQueensboroughCollegeMusicDepartment.

2.TheTomSwiftJr.series,whichwaslaunchedin1954byGrossetandDunlapandwrittenbyaseriesofauthorsunderthepseudonymVictorAppleton,continueduntil1971.TheteenageTomSwift,alongwithhispalBudBarclay,racedaroundtheuniverseexploringstrangeplaces,conqueringbadguys,andusingexoticgadgetssuchashouse-sizedspacecraft,aspacestation,aflyinglab,acycloplane,anelectrichydrolung,adivingseacopter,andarepellatron(whichrepelledthings;underwater,forexample,itwouldrepelwater,thusformingabubbleinwhichtheboyscouldlive).ThefirstninebooksintheseriesareTomSwiftandHisFlyingLab(1954),TomSwiftandHisJetmarine(1954),TomSwiftandHisRocketShip(1954),TomSwiftandHisGiantRobot(1954),TomSwiftandHisAtomicEarthBlaster(1954),TomSwiftandHisOutpostinSpace(1955),TomSwiftandHisDivingSeacopter(1956),TomSwiftintheCavesofNuclearFire(1956),andTomSwiftonthePhantomSatellite(1956).

3.TheprogramwascalledSelect.Studentsfilledoutathree-hundred-itemquestionnaire.Thecomputersoftware,whichcontainedadatabaseofabouttwomillionpiecesofinformationonthreethousandcolleges,selectedsixtofifteenschoolsthatmatchedthestudent'sinterests,background,andacademicstanding.WeprocessedabouttenthousandstudentsonourownandthensoldtheprogramtothepublishingcompanyHarcourt,Brace,andWorld.

4.TheAgeofIntelligentMachines,publishedin1990byMITPress,wasnamedBestComputerScienceBookbytheAssociationofAmericanPublishers.Thebookexploresthedevelopmentofartificialintelligenceandpredictsarangeofphilosophic,social,andeconomicimpactsofintelligentmachines.Thenarrativeiscomplementedbytwenty-threearticlesonAIfromthinkerssuchasSherryTurkle,DouglasHofstadter,MarvinMinsky,SeymourPapert,andGeorgeGilder.Fortheentiretextofthebook,seehttp://www.KurzweilAI.net/aim.

5.Keymeasuresofcapability(suchasprice-performance,bandwidth,andcapacity)increasebymultiples(thatis,themeasuresaremultipliedbyafactorforeachincrementoftime)ratherthanbeingaddedtolinearly.

6.DouglasR.Hofstadter,Gödel;Escher,Bach:AnEternalGoldenBraid(NewYork:BasicBooks,1979).

ChapterOne:TheSixEpochs1.AccordingtotheTranstopiasite(http://transtopia.org/faq.html#1.11),"Singularitarian"was

"originallydefinedbyMarkPlus('91)tomean'onewhobelievestheconceptofaSingularity.'"Anotherdefinitionofthistermis"'Singularityactivist'or'friendoftheSingularity';thatis,onewhoactssoastobringaboutaSingularity[MarkPlus,1991;SingularitarianPrinciples,EliezerYudkowsky,2000]."Thereisnotuniversalagreementonthisdefinition,andmany

TranshumanistsarestillSingularitariansintheoriginalsense—thatis,"believersintheSingularityconcept"ratherthan"activists"or"friends."EliezerS.Yudkowsky,inTheSingularitarianPrinciples,version1.0.2(January1,2000),http://yudkowsky.net/sing/principles.ext.html,proposedanalternatedefinition:"ASingularitarianissomeonewhobelievesthattechnologicallycreatingagreater-than-humanintelligenceisdesirable,andwhoworkstothatend.ASingularitarianisfriend,advocate,defender,andagentofthefutureknownastheSingularity."Myview:onecanadvancetheSingularityandinparticularmakeitmorelikelytorepresentaconstructiveadvanceofknowledgeinmanywaysandinmanyspheresofhumandiscourse—forexample,advancingdemocracy,combatingtotalitarianandfundamentalistbeliefsystemsandideologies,andcreatingknowledgeinallofitsdiverseforms:music,art,literature,science,andtechnology.IregardaSingularitarianassomeonewhounderstandsthetransformationsthatarecominginthiscenturyandwhohasreflectedontheirimplicationsforhisorherownlife.

2.Wewillexaminethedoublingratesofcomputationinthenextchapter.Althoughthenumberoftransistorsperunitcosthasdoubledeverytwoyears,transistorshavebeengettingprogressivelyfaster,andtherehavebeenmanyotherlevelsofinnovationandimprovement.Theoverallpowerofcomputationperunitcosthasrecentlybeendoublingeveryyear.Inparticular,theamountofcomputation(incomputationspersecond)thatcanbebroughttobeartoacomputerchessmachinedoubledeveryyearduringthe1990s.

3.JohnvonNeumann,paraphrasedbyStanislawUlam,"TributetoJohnvonNeumann,"BulletinoftheAmericanMathematicalSociety64.3,pt.2(May1958):1–49.VonNeumann(1903–1957)wasborninBudapestintoaJewishbankingfamilyandcametoPrincetonUniversitytoteachmathematicsin1930.In1933hebecameoneofthesixoriginalprofessorsinthenewInstituteforAdvancedStudyinPrinceton,wherehestayeduntiltheendofhislife.Hisinterestswerefarranging:hewastheprimaryforceindefiningthenewfieldofquantummechanics;alongwithcoauthorOskarMorgenstern,hewroteTheoryofGamesandEconomicBehavior,atextthattransformedthestudyofeconomics;andhemadesignificantcontributionstothelogicaldesignofearlycomputers,includingbuildingMANIAC(MathematicalAnalyzer,NumeralIntegrator,andComputer)inthelate1930s.HereishowOskarMorgensterndescribedvonNeumannintheobituary"JohnvonNeumann,1903–1957,"intheEconomicJournal(March1958:174):"VonNeumannexercisedanunusuallylargeinfluenceuponthethoughtofothermeninhispersonalrelations....Hisstupendousknowledge,theimmediateresponse,theunparalleledintuitionheldvisitorsinawe.Hewouldoftensolvetheirproblemsbeforetheyhadfinishedstatingthem.Hismindwassouniquethatsomepeoplehaveaskedthemselves—theytooeminentscientists—whetherhedidnotrepresentanewstageinhumanmentaldevelopment."

4.Seenotes20and21inchapter2.5.TheconferencewasheldFebruary19–21,2003,inMonterey,California.Amongthetopics

coveredwerestem-cellresearch,biotechnology,nanotechnology,cloning,andgeneticallymodifiedfood.Foralistofbooksrecommendedbyconferencespeakers,seehttp://www.thefutureoflife.com/books.htm.

6.TheInternet,asmeasuredbythenumberofnodes(servers),wasdoublingeveryyearduringthe1980sbutwasonlytensofthousandsofnodesin1985.Thisgrewtotensofmillionsofnodesby1995.ByJanuary2003,theInternetSoftwareConsortium(http://www.isc.org/ds/host-count-history.html)counted172millionWebhosts,whicharetheservershostingWebsites.Thatnumberrepresentsonlyasubsetofthetotalnumberofnodes.

7.Atthebroadestlevel,theanthropicprinciplestatesthatthefundamentalconstantsofphysicsmustbecompatiblewithourexistence;iftheywerenot,wewouldnotbeheretoobservethem.Oneofthecatalystsforthedevelopmentoftheprincipleisthestudyofconstants,suchasthegravitationalconstantandtheelectromagnetic-couplingconstant.Ifthevaluesoftheseconstantsweretostraybeyondaverynarrowrange,intelligentlifewouldnotbepossibleinouruniverse.Forexample,iftheelectromagnetic-couplingconstantwerestronger,therewouldbenobonding

betweenelectronsandotheratoms.Ifitwereweaker,electronscouldnotbeheldinorbit.Inotherwords,ifthissingleconstantstrayedoutsideanextremelynarrowrange,moleculeswouldnotform.Ouruniverse,then,appearstoproponentsoftheanthropicprincipletobefine-tunedfortheevolutionofintelligentlife.(DetractorssuchasVictorStengerclaimthefine-tuningisnotsofineafterall;therearecompensatorymechanismsthatwouldsupportawiderwindowforlifetoformunderdifferentconditions.)Theanthropicprinciplecomesupagaininthecontextofcontemporarycosmologytheoriesthatpositmultipleuniverses(seenotes8and9,below),eachwithitsownsetoflaws.Onlyinauniverseinwhichthelawsallowedthinkingbeingstoexistcouldwebehereaskingthesequestions.OneoftheseminaltextsinthediscussionisJohnBarrowandFrankTipler,TheAnthropicCosmologicalPrinciple(NewYork:OxfordUniversityPress,1988).SeealsoStevenWeinberg,"ADesignerUniverse?"athttp://www.physlink.comlEducation/essay_weinberg.cfm.

8.Accordingtosomecosmologicaltheories,thereweremultiplebigbangs,notone,leadingtomultipleuniverses(parallelmultiversesor"bubbles").Differentphysicalconstantsandforcesapplyinthedifferentbubbles;conditionsinsome(oratleastone)ofthesebubblessupportcarbon-basedlife.SeeMaxTegmark,"ParallelUniverses,"ScientificAmerican(May2003):41–53;MartinRees,"ExploringOurUniverseandOthers,"ScientificAmerican(December1999):78–83;AndreiLinde,"TheSelf-ReproducingInflationaryUniverse,"ScientificAmerican(November1994):48–55.

9.The"manyworlds"ormultiversetheoryasaninterpretationofquantummechanicswasdevelopedtosolveaproblempresentedbyquantummechanicsandthenhasbeencombinedwiththeanthropicprinciple.AssummarizedbyQuentinSmith:

AseriousdifficultyassociatedwiththeconventionalorCopenhageninterpretationofquantummechanicsisthatitcannotbeappliedtothegeneralrelativityspace-timegeometryofacloseduniverse.Aquantumstateofsuchauniverseisdescribableasawavefunctionwithvaryingspatial-temporalamplitude;theprobabilityofthestateoftheuniversebeingfoundatanygivenpointisthesquareoftheamplitudeofthewavefunctionatthatpoint.Inorderfortheuniversetomakethetransitionfromthesuperpositionofmanypointsofvaryingprobabilitiestooneofthesepoints—theoneinwhichitactuallyis—ameasuringapparatusmustbeintroducedthatcollapsesthewavefunctionanddeterminestheuniversetobeatthatpoint.Butthisisimpossible,forthereisnothingoutsidetheuniverse,noexternalmeasuringapparatus,thatcancollapsethewavefunction.ApossiblesolutionistodevelopaninterpretationofquantummechanicsthatdoesnotrelyonthenotionofexternalobservationormeasurementthatiscentraltotheCopenhageninterpretation.Aquantummechanicscanbeformulatedthatisinternaltoaclosedsystem.ItissuchaninterpretationthatHughEverettdevelopedinhis1957paper,"RelativeStateFormulationofQuantumMechanics."Eachpointinthesuperpositionrepresentedbythewavefunctionisregardedasactuallycontainingonestateoftheobserver(ormeasuringapparatus)andonestateofthesystembeingobserved.Thus"witheachsucceedingobservation(orinteraction),theobserverstate'branches'intoanumberofdifferentstates.Eachbranchrepresentsadifferentoutcomeofthemeasurementandthecorrespondingeigenstatefortheobject-systemstate.Allbranchesexistsimultaneouslyinthesuperpositionafteranygivensequenceofobservations."Eachbranchiscausallyindependentofeachotherbranch,andconsequentlynoobserverwilleverbeawareofany"splitting"process.Theworldwillseemtoeachobserverasitdoesinfactseem.Appliedtotheuniverseasawhole,thismeansthattheuniverseisregularlydividingintonumerousdifferentandcausallyindependentbranches,consequentuponthemeasurement-likeinteractionsamongitsvariousparts.Eachbranchcanberegardedasaseparateworld,witheachworldconstantlysplittingintofurtherworlds.

Giventhatthesebranches—thesetofuniverses—willincludeonesbothsuitableandunsuitableforlife,Smithcontinues,"Atthispointitcanbestatedhowthestronganthropicprincipleincombinationwiththemany-worldsinterpretationofquantummechanicscanbeusedinanattempttoresolvetheapparentproblemmentionedatthebeginningofthisessay.Theseeminglyproblematicfactthataworldwithintelligentlifeisactual,ratherthanoneofthemanylifelessworlds,isfoundnottobeafactatall.Ifworldswithlifeandwithoutlifearebothactual,thenitisnotsurprisingthatthisworldisactualbutissomethingtobeexpected."QuentinSmith,"TheAnthropicPrincipleandMany-WorldsCosmologies,"AustralasianJournalofPhilosophy63.3(September1985),availableathttp://www.qsmithwmu.com/the_anthropic_principle_and_many-worlds_cosmologies.htm.

10.Seechapter4foracompletediscussionofthebrain'sself-organizingprinciplesandtherelationshipofthisprincipleofoperationtopatternrecognition.

11.Witha"linear"plot(whereallgraphdivisionsareequal),itwouldbeimpossibletovisualizeallofthedata(suchasbillionsofyears)inalimitedspace(suchasapageofthisbook).Alogarithmic("log")plotsolvesthatbyplottingtheorderofmagnitudeofthevaluesratherthantheactualvalues,allowingyoutoseeawiderrangeofdata.

12.TheodoreModis,professoratDUXX,GraduateSchoolinBusinessLeadershipinMonterrey,Mexico,attemptedtodevelopa"precisemathematicallawthatgovernstheevolutionofchangeandcomplexityintheUniverse."Toresearchthepatternandhistoryofthesechanges,herequiredananalyticdatasetofsignificanteventswheretheeventsequatetomajorchange.Hedidnotwanttorelysolelyonhisownlist,becauseofselectionbias.Instead,hecompiledthirteenmultipleindependentlistsofmajoreventsinthehistoryofbiologyandtechnologyfromthesesources:

CarlSagan,TheDragonsofEden:SpeculationsontheEvolutionofHumanIntelligence(New

York:BallantineBooks,1989).ExactdatesprovidedbyModis.AmericanMuseumofNaturalHistory.ExactdatesprovidedbyModis.Thedataset"importanteventsinthehistoryoflife"intheEncyclopaediaBritannica.EducationalResourcesinAstronomyandPlanetaryScience(ERAPS),UniversityofArizona,

http://ethel.as.arizona.edu/~collins/astro/subiects/evolve-26.html.PaulD.Boyer,biochemist,winnerofthe1997NobelPrize,privatecommunication.ExactdatesprovidedbyModis.J.D.BarrowandJ.Silk,"TheStructureoftheEarlyUniverse,"ScientificAmerican242.4(April

1980):118–28.J.Heidmann,CosmicOdyssey:ObservatoirdeParis,trans.SimonMitton(Cambridge,U.K.:

CambridgeUniversityPress,1989).J.W.Schopf,ed.,MajorEventsintheHistoryofLife,symposiumconvenedbytheIGPPCenter

fortheStudyofEvolutionandtheOriginofLife,1991(Boston:JonesandBartlett,1991).

PhillipTobias,"MajorEventsintheHistoryofMankind,"chap.6inSchopf,MajorEventsintheHistoryofLife.

DavidNelson,"LectureonMolecularEvolutionI,"http://drnelson.utmem.edu/evolution.html,and"LectureNotesforEvolutionII,"http://drnelson.utmem.edu/evolution2.html.

G.Burenhult,ed.,TheFirstHumans:HumanOriginsandHistoryto10,000BC(SanFrancisco:HarperSanFrancisco,1993).

D.JohansonandB.Edgar,FromLucytoLanguage(NewYork:Simon&Schuster,1996).R.Coren,TheEvolutionaryTrajectory:TheGrowthofInformationintheHistoryandFutureof

Earth,WorldFuturesGeneralEvolutionStudies(Amsterdam:GordonandBreach,1998).

Theselistsdatefromthe1980sand1990s,withmostcoveringtheknownhistoryofthe

universe,whilethreefocusonthenarrowerperiodofhominoidevolution.Thedatesusedbysomeoftheolderlistsareimprecise,butitistheeventsthemselves,andtherelativelocationsoftheseeventsinhistory,thatareofprimaryinterest.Modisthencombinedtheseliststofindclustersofmajorevents,his"canonicalmilestones."Thisresultedin28canonicalmilestonesfromthe203milestoneeventsinthelists.ModisalsousedanotherindependentlistbyCorenasachecktoseeifitcorroboratedhismethods.SeeT.Modis,"ForecastingtheGrowthofComplexityandChange,"TechnologicalForecastingandSocialChange69.4(2002);http://ourworld.compuserve.com/homepages/tmodis/TedWEB.htm.

13.Modisnotesthaterrorscanarisefromvariationsinthesizeoflistsandfromvariationsindatesassignedtoevents(seeT.Modis,"TheLimitsofComplexityandChange,"TheFuturist[May–June2003],http://ourworld.compuserve.com/homepages/tmodis/Futurist.pdf).Soheusedclustersofdatestodefinehiscanonicalmilestones.Amilestonerepresentsanaverage,withknownerrorsassumedtobethestandarddeviation.Foreventswithoutmultiplesources,he"arbitrarilyassign[ed]theaverageerroraserror."Modisalsopointsoutothersourcesoferror—caseswhereprecisedatesareunknownorwherethereisthepossibilityofinappropriateassumptionofequalimportanceforeachdatapoint—whicharenotcaughtinthestandarddeviation.NotethatModis'sdateof54.6millionyearsagoforthedinosaurextinctionisnotfarenoughback.

14.Typicalinterneuronalresettimesareontheorderoffivemilliseconds,whichallowsfortwohundreddigital-controlledanalogtransactionspersecond.Evenaccountingformultiplenonlinearitiesinneuronalinformationprocessing,thisisontheorderofamilliontimesslowerthancontemporaryelectroniccircuits,whichcanswitchinlessthanonenanosecond(seetheanalysisofcomputationalcapacityinchapter2).

15.AnewanalysisbyLosAlamosNationalLabresearchersoftherelativeconcentrationsofradioactiveisotopesintheworld'sonlyknownnaturalnuclearreactor(atOkloinGabon,WestAfrica)hasfoundadecreaseinthefine-structureconstant,oralpha(thespeedoflightisinverselyproportionaltoalpha),overtwobillionyears.Thattranslatesintoasmallincreaseinthespeedoflight,althoughthisfindingclearlyneedstobeconfirmed.See"SpeedofLightMayHaveChangedRecently,"NewScientist,June30,2004,http://www.newscientist.comlnews/news.jsp?id=ns99996092.Seealsohttp://www.sciencedaily.com/releases/2005/05/050512120842.htm.

16.StephenHawkingdeclaredatascientificconferenceinDublinonJuly21,2004,thathehadbeenwronginacontroversialassertionhemadethirtyyearsagoaboutblackholes.Hehadsaidinformationaboutwhathadbeenswallowedbyablackholecouldneverberetrievedfromit.Thiswouldhavebeenaviolationofquantumtheory,whichsaysthatinformationispreserved."I'msorrytodisappointsciencefictionfans,butifinformationispreservedthereisnopossibilityofusingblackholestotraveltootheruniverses,"hesaid."Ifyoujumpintoablackhole,yourmassenergywillbereturnedtoouruniverse,butinamangledform,whichcontainstheinformationaboutwhatyouwerelike,butinanunrecognizablestate."SeeDennisOverbye,"AboutThoseFearsomeBlackHoles?NeverMind,"NewYorkTimes,July22,2004.

17.Aneventhorizonistheouterboundary,orperimeter,ofasphericalregionsurroundingthesingularity(theblackhole'scenter,characterizedbyinfinitedensityandpressure).Insidetheeventhorizon,theeffectsofgravityaresostrongthatnotevenlightcanescape,althoughthereisradiationemergingfromthesurfaceowingtoquantumeffectsthatcauseparticle-antiparticlepairstoform,withoneofthepairbeingpulledintotheblackholeandtheotherbeingemittedasradiation(so-calledHawkingradiation).Thisisthereasonwhytheseregionsarecalled"blackholes,"aterminventedbyProfessorJohnWheeler.AlthoughblackholeswereoriginallypredictedbyGermanastrophysicistKurtSchwarzschildin1916basedonEinstein'stheoryofgeneralrelativity,theirexistenceatthecentersofgalaxieshasonlyrecentlybeenexperimentallydemonstrated.Forfurtherreading,seeKimberlyWeaver,"TheGalacticOddCouple,"

http://www.scientificamerican.com.June10,2003;Jean-PierreLasota,"UnmaskingBlackHoles,"ScientificAmerican(May1999):41–47;StephenHawking,ABriefHistoryofTime:FromtheBigBangtoBlackHoles(NewYork:Bantam,1988).

18.JoelSmollerandBlakeTemple,"Shock-WaveCosmologyInsideaBlackHole,"ProceedingsoftheNationalAcademyofSciences100.20(September30,2003):11216–18.

19.VernorVinge,"FirstWord,"Omni(January1983):10.20.RayKurzweil,TheAgeofIntelligentMachines(Cambridge,Mass.:MITPress,1989).21.HansMoravec,MindChildren:TheFutureofRobotandHumanIntelligence(Cambridge,

Mass.:HarvardUniversityPress,1988).22.VernorVinge,"TheComingTechnologicalSingularity:HowtoSurviveinthePost-Human

Era,"VISION-21Symposium,sponsoredbytheNASALewisResearchCenterandtheOhioAerospaceInstitute,March1993.Thetextisavailableathttp://www.KurzweiW.net/vingesing.

23.RayKurzweil,TheAgeofSpiritualMachines:WhenComputersExceedHumanIntelligence(NewYork:Viking,1999).

24.HansMoravec,Robot:MereMachinetoTranscendentMind(NewYork:OxfordUniversityPress,1999).

25.DamienBroderick,twoworks:TheSpike:AcceleratingintotheUnimaginableFuture(Sydney,Australia:ReedBooks,1997)andTheSpike:HowOurLivesAreBeingTransformedbyRapidlyAdvancingTechnologies,rev.ed.(NewYork:Tor/Forge,2001).

26.OneofJohnSmart'soverviews,"WhatIstheSingularity,"canbefoundathttp://www.KurzweilAI.net/meme/frame.html?main=/articles/art0133.html;foracollectionofJohnSmart'swritingsontechnologyacceleration,theSingularity,andrelatedissues,seehttp://www.singularitywatch.comandhttp://www.Accelerating.org.JohnSmartrunsthe"AcceleratingChange"conference,whichcoversissuesrelatedto"artificialintelligenceandintelligenceamplification."Seehttp://www.accelerating.org/ac2005/index.html.

27.Anemulationofthehumanbrainrunningonanelectronicsystemwouldrunmuchfasterthanourbiologicalbrains.Althoughhumanbrainsbenefitfrommassiveparallelism(ontheorderofonehundredtrillioninterneuronalconnections,allpotentiallyoperatingsimultaneously),theresettimeoftheconnectionsisextremelyslowcomparedtocontemporaryelectronics.

28.Seenotes20and21inchapter2.29.Seetheappendix,"TheLawofAcceleratingReturnsRevisited,"foramathematicalanalysis

oftheexponentialgrowthofinformationtechnologyasitappliestotheprice-performanceofcomputation.

30.Ina1950paperpublishedinMind:AQuarterlyReviewofPsychologyandPhilosophy,thecomputertheoreticianAlanTuringposedthefamousquestions"Canamachinethink?Ifacomputercouldthink,howcouldwetell?"TheanswertothesecondquestionistheTuringtest.Asthetestiscurrentlydefined,anexpertcommitteeinterrogatesaremotecorrespondentonawiderangeoftopicssuchaslove,currentevents,mathematics,philosophy,andthecorrespondent'spersonalhistorytodeterminewhetherthecorrespondentisacomputerorahuman.TheTuringtestisintendedasameasureofhumanintelligence;failuretopassthetestdoesnotimplyalackofintelligence.Turing'soriginalarticlecanbefound.athttp://www.abelard.org/turpap/turpap.htm;seealsotheStanfordEncyclopediaofPhilosophy,http://plato.stanford.edu/entries/turing-test,foradiscussionofthetest.ThereisnosetoftricksoralgorithmsthatwouldallowamachinetopassaproperlydesignedTuringtestwithoutactuallypossessingintelligenceatafullyhumanlevel.AlsoseeRayKurzweil,"AWagerontheTuringTest:WhyIThinkIWillWin,"http://www.KurzweilAI.net/turingwin.

31.SeeJohnH.Byrne,"PropagationoftheActionPotential,"NeuroscienceOnline,https://oac22.hsc.uth.tmc.edu/courses/nba/s1/i3-1.html:"Thepropagationvelocityoftheactionpotentialsinnervescanvaryfrom100meterspersecond(580milesperhour)tolessthanatenthofameterpersecond(0.6milesperhour)."

AlsoseeKennethR.Koehler,"TheActionPotential,"http://www.rwc.uc.edu/koehler/biophys/4d.html:"Thespeedofpropagationformammalianmotorneuronsis10–120m/s,whilefornonmyelinatedsensoryneuronsit'sabout5–25m/s(nonmyelinatedneuronsfireinacontinuousfashion,withoutthejumps;ionleakageallowseffectivelycompletecircuitsbutslowstherateofpropagation)."

32.A2002studypublishedinSciencehighlightedtheroleofthebeta-cateninproteininthehorizontalexpansionofthecerebralcortexinhumans.Thisproteinplaysakeyroleinthefoldingandgroovingofthesurfaceofthecerebralcortex;itisthisfolding,infact,thatincreasesthesurfaceareaofthispartofthebrainandmakesroomformoreneurons.Micethatoverproducedtheproteindevelopedwrinkled,foldedcerebralcortexeswithsubstantiallymoresurfaceareathanthesmooth,flatcerebralcortexesofcontrolmice.AnjenChennandChristopherWalsh,"RegulationofCerebralCorticalSizebyControlofCellCycleExitinNeuralPrecursors,"Science297(July2002):365–69.A2003comparisonofcerebral-cortexgene-expressionprofilesforhumans,chimpanzees,andrhesusmacaquesshowedadifferenceofexpressioninonlyninety-onegenesassociatedwithbrainorganizationandcognition.Thestudyauthorsweresurprisedtofindthat90percentofthesedifferencesinvolvedupregulation(higheractivity).SeeM.Cacaresetal.,"ElevatedGeneExpressionLevelsDistinguishHumanfromNon-humanPrimateBrains,"ProceedingsoftheNationalAcademyofSciences100.22(October28,2003):13030–35.However,UniversityofCalifornia-IrvineCollegeofMedicineresearchershavefoundthatgraymatterinspecificregionsinthebrainismorerelatedtoIQthanisoverallbrainsizeandthatonlyabout6percentofallthegraymatterinthebrainappearsrelatedtoIQ.Thestudyalsodiscoveredthatbecausetheseregionsrelatedtointelligencearelocatedthroughoutthebrain,asingle"intelligencecenter,"suchasthefrontallobe,isunlikely.See"HumanIntelligenceDeterminedbyVolumeandLocationofGrayMatterTissueinBrain,"UniversityofCalifornia–Irvinenewsrelease(July19,2004),http://today.uci.edu/news/release_detail.asp?key=1187.A2004studyfoundthathumannervoussystemgenesdisplayedacceleratedevolutioncomparedwithnonhumanprimatesandthatallprimateshadacceleratedevolutioncomparedwithothermammals.SteveDorusetal.,"AcceleratedEvolutionofNervousSystemGenesintheOriginofHomosapiens,"Cell119(December29,2004):1027–40.Indescribingthisfinding,theleadresearcher,BruceLahn,states,"Humansevolvedtheircognitiveabilitiesnotduetoafewaccidentalmutations,butratherfromanenormousnumberofmutationsacquiredthroughexceptionallyintenseselectionfavoringmorecomplexcognitiveabilities."CatherineGianaro,UniversityofChicagoChronicle24.7(January6,2005).AsinglemutationtothemusclefibergeneMYH16hasbeenproposedasonechangeallowinghumanstohavemuchlargerbrains.Themutationmadeancestralhumans'jawsweaker,sothathumansdidnotrequirethebrain-sizelimitingmuscleanchorsfoundinothergreatapes.Stedmanetal.,"MyosinGeneMutationCorrelateswithAnatomicalChangesintheHumanLineage,"Nature428(March25,2004):415–18.

33.RobertA.FreitasJr.,"ExploratoryDesigninMedicalNanotechnology:AMechanicalArtificialRedCell,"ArtificialCells,BloodSubstitutes,andImmobil.Biotech.26(1998):411–30;http://www.foresight.org/Nanomedicine/Respirocytes.html;seealsotheNanomedicineArtGalleryimages(http://www.foresight.org/Nanomedicine/Gallery/Species/Respirocytes.html)andaward-winninganimation(http://www.phleschbubble.com/album/beyondhuman/respirocyte01.htm)oftherespirocytes.

34.FogletsaretheconceptionofthenanotechnologypioneerandRutgersprofessorJ.StorrsHall.Hereisasnippetofhisdescription:"Nanotechnologyisbasedontheconceptoftiny,self-replicatingrobots.TheUtilityFogisaverysimpleextensionoftheidea:Suppose,insteadofbuildingtheobjectyouwantatombyatom,thetinyrobots[foglets]linkedtheirarmstogethertoformasolidmassintheshapeoftheobjectyouwanted?Then,whenyougottiredofthatavant-gardecoffeetable,therobotscouldsimplyshiftaroundalittleandyou'dhaveanelegantQueenAnnepieceinstead."J.StorrsHall,"WhatIWanttoBeWhenIGrowUp,IsaCloud,"Extropy,

Quarters3and4,1994.PublishedonKurzweilAI.netJuly6,2001:http://www.KurzweilAI.net/foglets.SeealsoJ.StorrsHall,"UtilityFog:TheStuffThatDreamsAreMadeOf,"inNanotechnology:MolecularSpeculationsonGlobalAbundance,B.C.Crandall,ed.(Cambridge,Mass.:MITPress,1996).PublishedonKurzweilAI.netJuly5,2001:http://www.KurzweilAI.net/utilityfog.

35.SherryTurkle,ed.,"EvocativeObjects:ThingsWeThinkWith,"forthcoming.36.Seethe"ExponentialGrowthofComputing"figureinchapter2(p,70).Projectingthedouble

exponentialgrowthoftheprice-performanceofcomputationtotheendofthetwenty-firstcentury,onethousanddollars'worthofcomputationwillprovide1060calculationspersecond(cps).Aswewilldiscussinchapter2,threedifferentanalysesoftheamountofcomputingrequiredtofunctionallyemulatethehumanbrainresultinanestimateof1015cps.Amoreconservativeestimate,whichassumesthatitwillbenecessarytosimulateallofthenonlinearitiesineverysynapseanddendrite,resultsinanestimateof1019cpsforneuromorphicemulationofthehumanbrain.Eventakingthemoreconservativefigure,wegetafigureof1029

cpsfortheapproximately1010humans.Thus,the1060cpsthatcanbepurchasedforonethousanddollarscirca2099willrepresent1031(tenmilliontrilliontrillion)humancivilizations.

37.TheinventionofthepowerloomandtheothertextileautomationmachinesoftheearlyeighteenthcenturydestroyedthelivelihoodsofthecottageindustryofEnglishweavers,whohadpasseddownstablefamilybusinessesforhundredsofyears.Economicpowerpassedfromtheweavingfamiliestotheownersofthemachines.Aslegendhasit,ayoungandfeeblemindedboynamedNedLuddbroketwotextilefactorymachinesoutofsheerclumsiness.Fromthatpointon,wheneverfactoryequipmentwasfoundtohavemysteriouslybeendamaged,anyonesuspectedoffoulplaywouldsay,"ButNedLudddidit."In1812thedesperateweaversformedasecretsociety,anurbanguerrillaarmy.Theymadethreatsanddemandsoffactoryowners,manyofwhomcomplied.Whenaskedwhotheirleaderwas,theyreplied,"Why,GeneralNedLudd,ofcourse."AlthoughtheLuddites,astheybecameknown,initiallydirectedmostoftheirviolenceagainstthemachines,aseriesofbloodyengagementseruptedlaterthatyear.ThetoleranceoftheTorygovernmentfortheLudditesended,andthemovementdissolvedwiththeimprisonmentandhangingofprominentmembers.Althoughtheyfailedtocreateasustainedandviablemovement,theLudditeshaveremainedapowerfulsymbolofoppositiontoautomationandtechnology.

38.Seenote34above.

ChapterTwo:ATheoryofTechnologyEvolution:TheLawofAcceleratingReturns

1.JohnSmart,Abstractto"UnderstandingEvolutionaryDevelopment:AChallengeforFuturists,"presentationtoWorldFuturistSocietyannualmeeting,Washington,D.C.,August3,2004.

2.ThatepochaleventsinevolutionrepresentincreasesincomplexityisTheodoreModis'sview.SeeTheodoreModis,"ForecastingtheGrowthofComplexityandChange,"TechnologicalForecastingandSocialChange69.4(2002),http://ourworld.compuserve.com/homepages/tmodis/TedWEB.htm.

3.Compressingfilesisakeyaspectofbothdatatransmission(suchasamusicortextfileovertheInternet)anddatastorage.Thesmallerthefileis,thelesstimeitwilltaketotransmitandthelessspaceitwillrequire.ThemathematicianClaudeShannon,oftencalledthefatherofinformationtheory,definedthebasictheoryofdatacompressioninhispaper"AMathematicalTheoryofCommunication,"TheBellSystemTechnicalJournal27(July–October1948):379–423,623–56.Datacompressionispossiblebecauseoffactorssuchasredundancy(repetition)andprobabilityofappearanceofcharactercombinationsindata.Forexample,silenceinan

audiofilecouldbereplacedbyavaluethatindicatesthedurationofthesilence,andlettercombinationsinatextfilecouldbereplacedwithcodedidentifiersinthecompressedfile.Redundancycanberemovedbylosslesscompression,asShannonexplained,whichmeansthereisnolossofinformation.Thereisalimittolosslesscompression,definedbywhatShannoncalledtheentropyrate(compressionincreasesthe"entropy"ofthedata,whichistheamountofactualinformationinitasopposedtopredeterminedandthuspredictabledatastructures).Datacompressionremovesredundancyfromdata;losslesscompressiondoesitwithoutlosingdata(meaningthattheexactoriginaldatacanberestored).Alternatively,lossycompression,whichisusedforgraphicsfilesorstreamingvideoandaudiofiles,doesresultininformationloss,thoughthatlossisoftenimperceptibletooursenses.Mostdata-compressiontechniquesuseacode,whichisamappingofthebasicunits(orsymbols)inthesourcetoacodealphabet.Forexample,allthespacesinatextfilecouldbereplacedbyasinglecodewordandthenumberofspaces.Acompressionalgorithmisusedtosetupthemappingandthencreateanewfileusingthecodealphabet;thecompressedfilewillbesmallerthantheoriginalandthuseasiertotransmitorstore.Herearesomeofthecategoriesintowhichcommonlossless-compressiontechniquesfall:

·Run-lengthcompression,whichreplacesrepeatingcharacterswithacodeandavaluerepresentingthenumberofrepetitionsofthatcharacter(examples:Pack-BitsandPCX).

·Minimumredundancycodingorsimpleentropycoding,whichassignscodesonthebasisofprobability,withthemostfrequentsymbolsreceivingtheshortestcodes(examples:Huffmancodingandarithmeticcoding).

·Dictionarycoders,whichuseadynamicallyupdatedsymboldictionarytorepresentpatterns(examples:Lempel-Ziv,Lempel-Ziv-Welch,andDEFLATE).

·Block-sortingcompression,whichreorganizescharactersratherthanusingacodealphabet;run-lengthcompressioncanthenbeusedtocompresstherepeatingstrings(example:Burrows-Wheelertransform).

·Predictionbypartialmapping,whichusesasetofsymbolsintheuncompressedfiletopredicthowoftenthenextsymbolinthefileappears.

4.MurrayGell-Mann,"WhatIsComplexity?"inComplexity,vol.1(NewYork:JohnWileyand

Sons,1995).5.Thehumangeneticcodehasapproximatelysixbillion(about1010)bits,notconsideringthe

possibilityofcompression.Sothe1027bitsthattheoreticallycanbestoredinaone-kilogramrockisgreaterthanthegeneticcodebyafactorof1017.Seenote57belowforadiscussionofgenomecompression.

6.Ofcourse,ahuman,whoisalsocomposedofanenormousnumberofparticles,containsanamountofinformationcomparabletoarockofsimilarweightwhenNOTES509weconsiderthepropertiesofalltheparticles.Aswiththerock,thebulkofthisinformationisnotneededtocharacterizethestateoftheperson.Ontheotherhand,muchmoreinformationisneededtocharacterizeapersonthanarock.

7.Seenote175inchapter5foranalgorithmicdescriptionofgeneticalgorithms.8.Humans,chimpanzees,gorillas,andorangutansareallincludedinthescientificclassification

ofhominids(familyHominidae).Thehumanlineageisthoughttohavedivergedfromitsgreataperelativesfivetosevenmillionyearsago.ThehumangenusHomowithintheHominidaeincludesextinctspeciessuchasH.erectusaswellasmodernman(H.sapiens).Inchimpanzeehands,thefingersaremuchlongerandlessstraightthaninhumans,andthethumbisshorter,weaker,andnotasmobile.Chimpscanflailwithastickbuttendtolosetheirgrip.Theycannotpinchhardbecausetheirthumbsdonotoverlaptheirindexfingers.Inthemodernhuman,thethumbislonger,andthefingersrotatetowardacentralaxis,soyoucantouchallthetipsofyourfingerstothetipofyourthumb,aqualitythatiscalledfull

opposability.Theseandotherchangesgavehumanstwonewgrips:theprecisionandpowergrips.EvenprehominoidhominidssuchastheAustralopithecinefromEthiopiacalledLucy,whoisthoughttohavelivedaroundthreemillionyearsago,couldthrowrockswithspeedandaccuracy.Sincethen,scientistsclaim,continualimprovementsinthehand'scapacitytothrowandclub,alongwithassociatedchangesinotherpartsofthebody,haveresultedindistinctadvantagesoverotheranimalsofsimilarsizeandweight.SeeRichardYoung,"EvolutionoftheHumanHand:TheRoleofThrowingandClubbing,"JournalofAnatomy202(2003):165–74;FrankWilson,TheHand:HowItsUseShapestheBrain,Language,andHumanCulture(NewYork:Pantheon,1998).

9.TheSantaFeInstitutehasplayedapioneeringroleindevelopingconceptsandtechnologyrelatedtocomplexityandemergentsystems.OneoftheprincipaldevelopersofparadigmsassociatedwithchaosandcomplexityisStuartKauffman.Kauffman'sAtHomeintheUniverse:TheSearchfortheLawsofSelf-OrganizationandComplexity(Oxford:OxfordUniversityPress,1995)looks"attheforcesfororderthatlieattheedgeofchaos."InhisbookEvolutionofComplexitybyMeansofNaturalSelection(Princeton:PrincetonUniversityPress,1988),JohnTylerBonnerasksthequestions"Howisitthataneggturnsintoanelaborateadult?Howisitthatabacterium,givenmanymillionsofyears,couldhaveevolvedintoanelephant?"JohnHollandisanotherleadingthinkerfromtheSantaFeInstituteintheemergingfieldofcomplexity.HisbookHiddenOrder:HowAdaptationBuildsComplexity(Reading,Mass.:Addison-Wesley,1996)includesaseriesoflecturesthathepresentedattheSantaFeInstitutein1994.SeealsoJohnH.Holland,Emergence:FromChaostoOrder(Reading,Mass.:Addison-Wesley,1998)andMitchellWaldrop,Complexity:TheEmergingScienceattheEdgeofOrderandChaos(NewYork:Simon&Schuster,1992).

10.Thesecondlawofthermodynamicsexplainswhythereisnosuchthingasaperfectenginethatusesalltheheat(energy)producedbyburningfueltodowork:someheatwillinevitablybelosttotheenvironment.Thissameprincipleofnatureholdsthatheatwillflowfromahotpantocoldairratherthaninreverse.Italsopositsthatclosed("isolated")systemswillspontaneouslybecomemoredisorderedovertime—thatis,theytendtomovefromordertodisorder.Moleculesinicechips,forexample,arelimitedintheirpossiblearrangements.Soacupoficechipshaslessentropy(disorder)thanthecupofwatertheicechipsbecomewhenleftatroomtemperature.Therearemanymorepossiblemoleculararrangementsintheglassofwaterthanintheice;greaterfreedomofmovementequalshigherentropy.Anotherwaytothinkofentropyisasmultiplicity.Themorewaysthatastatecouldbeachieved,thehigherthemultiplicity.Thus,forexample,ajumbledpileofbrickshasahighermultiplicity(andhigherentropy)thananeatstack.

11.MaxMorearticulatestheviewthat"advancingtechnologiesarecombiningandcross-fertilizingtoaccelerateprogressevenfaster."MaxMore,"Track7TechVectorstoTakeAdvantageofTechnologicalAcceleration,"ManyWorlds,August1,2003.

12.Formoreinformation,seeJ.J.Emersonetal.,"ExtensiveGeneTrafficontheMammalianXChromosome,"Science303.5657(January23,2004):537–40,http://www3.uta.edu/faculty/betran/science2004.pdf;NicholasWade,"YChromosomeDependsonItselftoSurvive,"NewYorkTimes,June19,2003;andBruceT.LahnandDavidC.Page,"FourEvolutionaryStrataontheHumanXChromosome,"Science286.5441(October29,1999):964–67,http://inside.wi.mit.edu/page/Site/Page%20PDFs/Lahn_and_Page_strata_1999.pdf.Interestingly,thesecondXchromosomeingirlsisturnedoffinaprocesscalledXinactivationsothatthegenesononlyoneXchromosomeareexpressed.ResearchhasshownthattheXchromosomefromthefatheristurnedoffinsomecellsandtheXchromosomefromthemotherinothercells.

13.HumanGenomeProject,"InsightsLearnedfromtheSequence,"http://www.ornl.gov/sci/techresources/Human_Genome/project/journals/insights.html.Even

thoughthehumangenomehasbeensequenced,mostofitdoesnotcodeforproteins(theso-calledjunkDNA),soresearchersarestilldebatinghowmanygeneswillbeidentifiedamongthethreebillionbasepairsinhumanDNA.Currentestimatessuggestlessthanthirtythousand,thoughduringtheHumanGenomeProjectestimatesrangedashighasonehundredthousand.See"HowManyGenesAreintheHumanGenome?"(http://www.ornl.gov/sci/techresources/Human_Genome/faq/genenumber.shtml)andElizabethPennisi,"ALowNumberWinstheGeneSweepPool,"Science300.5625(June6,2003):1484.

14.NilesEldredgeandthelateStephenJayGouldproposedthistheoryin1972(N.EldredgeandS.J.Gould,"PunctuatedEquilibria:AnAlternativetoPhyleticGradualism,"inT.J.M.Schopf,ed.,ModelsinPaleobiology[SanFrancisco:Freeman,Cooper],pp.82–115).IthassparkedheateddiscussionsamongpaleontoloNOTES511gistsandevolutionarybiologistseversince,thoughithasgraduallygainedacceptance.Accordingtothistheory,millionsofyearsmaypasswithspeciesinrelativestability.Thisstasisisthenfollowedbyaburstofchange,resultinginnewspeciesandtheextinctionofold(calleda"turnoverpulse"byElisabethVrba).Theeffectisecosystemwide,affectingmanyunrelatedspecies.EldredgeandGould'sproposedpatternrequiredanewperspective:"Fornobiascanbemoreconstrictingthaninvisibility—andstasis,inevitablyreadasabsenceofevolution,hadalwaysbeentreatedasanon-subject.Howodd,though,todefinethemostcommonofallpalaeontologicalphenomenaasbeyondinterestornotice!"S.J.GouldandN.Eldredge,"PunctuatedEquilibriumComesofAge,"Nature366(November18,1993):223–27.SeealsoK.Sneppenetal.,"EvolutionAsaSelf-OrganizedCriticalPhenomenon,"ProceedingsoftheNationalAcademyofSciences92.11(May23,1995):5209–13;ElisabethS.Vrba,"EnvironmentandEvolution:AlternativeCausesoftheTemporalDistributionofEvolutionaryEvents,"SouthAfricanJournalofScience81(1985):229–36.

15.AsIwilldiscussinchapter6,ifthespeedoflightisnotafundamentallimittorapidtransmissionofinformationtoremoteportionsoftheuniverse,thenintelligenceandcomputationwillcontinuetoexpandexponentiallyuntiltheysaturatethepotentialofmatterandenergytosupportcomputationthroughouttheentireuniverse.

16.Biologicalevolutioncontinuestobeofrelevancetohumans,however,inthatdiseaseprocessessuchascancerandviraldiseasesuseevolutionagainstus(thatis,cancercellsandvirusesevolvetocounteractspecificcountermeasuressuchaschemotherapydrugsandantiviralmedicationsrespectively).Butwecanuseourhumanintelligencetooutwittheintelligenceofbiologicalevolutionbyattackingdiseaseprocessesatsufficientlyfundamentallevelsandbyusing"cocktail"approachesthatattackadiseaseinseveralorthogonal(independent)waysatonce.

17.AndrewOdlyzko,"InternetPricingandtheHistoryofCommunications,"AT&TLabsResearch,revisedversionFebruary8,2001,http://www.dtc.umn.edu/~odlyzko/doc/history.communications1b.pdf.

18.CellularTelecommunicationsandInternetAssociation,Semi-AnnualWirelessIndustrySurvey,June2004,http://www.ctia.orglresearch_statistics/index.cfm/AID/l0030.

19.Electricity,telephone,radio,television,mobilephones:FCC,www.fcc.gov/Bureaus/Common_Carrier/Notices/2000/fc00057a.xls.HomecomputersandInternetuse:EricC.Newburger,U.S.CensusBureau,"HomeComputersandInternetUseintheUnitedStates:August2000"(September2001),http://www.census.gov/prod/2001pubs/p23-207.pdf.Seealso"TheMillenniumNotebook,"Newsweek,April13,1998,p.14.

20.Theparadigm-shiftrate,asmeasuredbytheamountoftimerequiredtoadoptnewcommunicationstechnologies,iscurrentlydoubling(thatis,theamountoftimeformassadoption—definedasbeingusedbyaquarteroftheU.S.population—isbeingcutinhalf)everynineyears.Seealsonote21.

21.The"MassUseofInventions"chartinthischapteronp.50showsthatthetimerequiredforadoptionby25percentoftheU.S.populationsteadilydeclinedoverthepast130years.Forthetelephone,35yearswererequiredcomparedto31fortheradio—areductionof11percent,or

0.58percentperyearinthe21yearsbetweenthesetwoinventions.Thetimerequiredtoadoptaninventiondropped0.60percentperyearbetweentheradioandtelevision,1.0percentperyearbetweentelevisionandthePC,2.6percentperyearbetweenthePCandthemobilephone,and7.4percentperyearbetweenthemobilephoneandtheWorldWideWeb.Massadoptionoftheradiobeginningin1897required31years,whiletheWebrequiredamere7yearsafteritwasintroducedin1991—areductionof77percentover94years,oranaveragerateof1.6percentreductioninadoptiontimeperyear.Extrapolatingthisratefortheentiretwentiethcenturyresultsinanoverallreductionof79percentforthecentury.Atthecurrentrateofreducingadoptiontimeof7.4percenteachyear,itwouldtakeonly20yearsattoday'srateofprogresstoachievethesamereductionof79percentthatwasachievedinthetwentiethcentury.Atthisrate,theparadigm-shiftratedoubles(thatis,adoptiontimesarereducedby50percent)inabout9years.Overthetwenty-firstcentury,elevendoublingsoftheratewillresultinmultiplyingtherateby211,toabout2,000timestheratein2000.Theincreaseinratewillactuallybegreaterthanthisbecausethecurrentratewillcontinuetoincreaseasitsteadilydidoverthetwentiethcentury.

22.Datafrom1967–1999,Inteldata,seeGordonE.Moore,"OurRevolution,"http://www.sia-online.orgfdownloads/Moore.pdf.Datafrom2000–2016,InternationalTechnologyRoadmapforSemiconductors(ITRS)2002Updateand2004Update,http://public.itrs.net/Files/2002Update/2002Update.pdfandhttp://www.itrs.net/Common/2004Update/2004_00_Overview.pdf.

23.TheITRSDRAMcostisthecostperbit(packagedmicrocents)atproduction.Datafrom1971–2000:VLSIResearchInc.Datafrom2001–2002:ITRS,2002Update,Table7a,Cost-Near-TermYears,p.172.Datafrom2003–2018:ITRS,2004Update,Tables7aand7b,Cost-Near-TermYears,pp.20–21.

24.IntelandDataquestreports(December2002),seeGordonE.Moore,"OurRevolution,"http://www.sia-online.org/downloads/Moore.pdf.

25.RandallGoodall,D.Fandel,andH.Huffet,"Long-TermProductivityMechanismsoftheSemiconductorIndustry,"NinthInternationalSymposiumonSiliconMaterialsScienceandTechnology,May12–17,2002,Philadelphia,sponsoredbytheElectrochemicalSociety(ECS)andInternationalSematech.

26.Datafrom1976–1999:E.R.Berndt,E.R.Dulberger,andN.J.Rappaport,"PriceandQualityofDesktopandMobilePersonalComputers:AQuarterCenturyofHistory,"July17,2000,http://www.nber.org/~confer/2000/si2000/berndt.pdf.Datafrom2001–2016:ITRS,2002Update,On-ChipLocalClockinTable4c:PerformanceandPackageChips:FrequencyOn-ChipWiringLevels-Near-TermYears,p.167.

27.Seenote26forclockspeed(cycletimes)andnote24forcostpertransistor.28.Inteltransistorsonmicroprocessors:MicroprocessorQuickReferenceGuide,IntelResearch,

http://www.intel.com/pressroom/kits/quickrefyr.htm.SeealsoSiliconResearchAreas,IntelResearch,http://www.intel.comlresearch/silicon/mooreslaw.htm.

29.DatafromIntelCorporation.SeealsoGordonMoore,"NoExponentialIsForever...butWeCanDelay'Forever,'"presentedattheInternationalSolidStateCircuitsConference(lSSCC),February10,2003,ftp://download.intel.com/researchlsilicon/Gordon_Moore_ISSCC_021003.pdf.

30.SteveCullen,"SemiconductorIndustryOutlook,"InStat/MDR,reportno.IN0401550SI,April2004,http://www.instat.com/abstract.asp?id=68&SKU=IN0401550SI.

31.WorldSemiconductorTradeStatistics,http://wsts.www5.kcom.at.32.BureauofEconomicAnalysis,U.S.DepartmentofCommerce,

http://www.bea.gov/bea/dn/home/gdp.htm.33.Seenotes22–24and26–30.34.InternationalTechnologyRoadmapforSemiconductors,2002update,InternationalSematech.35."25YearsofComputerHistory,"http://www.compros.com/timeline.html;LinleyGwennap,

"BirthofaChip,"BYTE(December1996),http://www.byte.com/art/9612/sec6/art2.htm;"TheCDC6000SeriesComputer,"http://www.moorecad.com/standardpascal/cdc6400.html;"AChronologyofComputerHistory,"http://www.cyberstreet.comlhcs/museum/chron.htm;MarkBrader,"AChronologyofDigitalComputingMachines(to1952),"http://www.davros.org/misc/chronology.html;KarlKempf,"ElectronicComputersWithintheOrdnanceCorps,"November1961,http://ftp.arl.mil/~mike/comphist/61ordnance/index.html;KenPolsson,"ChronologyofPersonalComputers,"http://www.islandnet.com/~kpolsson/comphist;"TheHistoryofComputingatLosAlamos,"http://bang.lanl.gov/video/sunedu/computer/comphist.html(requirespassword);theMachineRoom,http://www.machine-room.org;MindMachineWebMuseum,http://www.userwww.sfsu.edu/~hl/mmm.html;HansMoravec,computerdata,http://www.frc.ri.cmu.edu/~hpm/book97/ch3/processor.list;"PCMagazineOnline:FifteenYearsofPCMagazine,"http://www.pcmag.com/article2/0,1759,23390,00.asp;StanAugarten,BitbyBit:AnIllustratedHistoryofComputers(NewYork:TicknorandFields,1984);InternationalAssociationofElectricalandElectronicsEngineers(IEEE),AnnalsoftheHistoryoftheComputer9.2(1987):150–53and16.3(1994):20;HansMoravec,MindChildren:TheFutureofRobotandHumanIntelligence(Cambridge,Mass.:HarvardUniversityPress,1988);ReneMoreau,TheComputerComesofAge(Cambridge,Mass.:MITPress,1984).

36.Theplotsinthischapterlabeled"LogarithmicPlot"aretechnicallysemilogarithmicplotsinthatoneaxis(time)isonalinearscale,andtheotheraxisisonalogarithmicscale.However,Iamcallingtheseplots"logarithmicplots"forsimplicity.

37.Seetheappendix,"TheLawofAcceleratingReturnsRevisited,"whichprovidesamathematicalderivationofwhytherearetwolevelsofexponentialgrowth(thatis,exponentialgrowthovertimeinwhichtherateoftheexponentialgrowth—theexponent—isitselfgrowingexponentiallyovertime)incomputationalpowerasmeasuredbyMIPSperunitcost.

38.HansMoravec,"WhenWillComputerHardwareMatchtheHumanBrain?"JournalofEvolutionandTechnology1(1998),http://www.jetpress.org/volumel/moravec.pdf.

39.Seenote35above.40.AchievingthefirstMIPSper$1,000tookfrom1900to1990.We'renowdoublingthenumber

ofMIPSper$1,000inabout400days.Becausecurrentpriceperformanceisabout2,000MIPSper$1,000,weareaddingprice-performanceattherateof5MIPSperday,or1MIPSaboutevery5hours.

41."IBMDetailsBlueGeneSupercomputer,"CNETNews,May8,2003,http://news.com.com/2100-1008_3-1000421.html.

42.SeeAlfredNorthWhitehead,AnIntroductiontoMathematics(London:WilliamsandNorgate,1911),whichhewroteatthesametimeheandBertrandRussellwereworkingontheirseminalthree-volumePrincipiaMathematica.

43.Whileoriginallyprojectedtotakefifteenyears,"theHumanGenomeProjectwasfinishedtwoandahalfyearsaheadoftimeand,at$2.7billioninFY1991dollars,significantlyunderoriginalspendingprojections":http://www.ornl.gov/sci/techresources/Human_Genome/project/50yr/press4_2003.shtml.

44.HumanGenomeProjectInformation,http://www.ornl.gov/sci/techresources/Human_Genome/project/privatesector.shtml;StanfordGenomeTechnologyCenter,http://sequence-www.stanford.edu/group/techdev/auto.html;NationalHumanGenomeResearchInstitute,http://www.genome.gov;TabithaPowledge,"HowManyGenomesAreEnough?"Scientist,November17,2003,http://www.biomedcentral.com/news/20031117/07.

45.DatafromNationalCenterforBiotechnologyInformation,"GenBankStatistics,"revisedMay4,2004,http://www.ncbi.nlm.nih.gov/Genbank/genbankstats.html.

46.Severeacuterespiratorysyndrome(SARS)wassequencedwithinthirty-onedaysofthevirusbeingidentifiedbytheBritishColumbiaCancerAgencyandtheAmericanCentersforDiseaseControl.Thesequencingfromthetwocentersdifferedbyonlytenbasepairsoutoftwenty-nine

thousand.ThisworkidentifiedSARSasacoronavirus.Dr.JulieGerberding,directoroftheCDC,calledthequicksequencing"ascientificachievementthatIdon'tthinkhasbeenparalleledinourhistory."SeeK.Philipkoski,"SARSGeneSequenceUnveiled,"WiredNews,April15,2003,http://www.wired.com/news/medtech/0,1286,58481.00.html?tw=wn_story_related.Incontrast,theeffortstosequenceHIVbeganinthe1980s.HIV1andHIV2werecompletelysequencedin2003and2002respectively.NationalCenterforBiotechnologyInformation,http://www.ncbi.nlm.nih.gov/genomes/framik.cgi?db=genome&gi=12171;HIVSequenceDatabasemaintainedbytheLosAlamosNationalLaboratory,http://www.hiv.lanl.gov/content/hiv-db/HTML/outline.html.

47.MarkBrader,"AChronologyofDigitalComputingMachines(to1952),"http://www.davros.org/misc/chronology.html;RichardE.Matick,ComputerStorageSystemsandTechnology(NewYork:JohnWileyandSons,1977);UniversityofCambridgeComputerLaboratory,EDSAC99,http://www.cl.cam.ac.uk/UoCCL/misc/EDSAC99/statistics.html;MaryBellis,"InventorsoftheModernComputer:TheHistoryoftheUNIVACComputer—J.PresperEckertandJohnMauchly,"http://inventors.about.com/library/weekly/aa062398.htm;"InitialDateofOperationofComputingSystemsintheUSA(1950–1958),"compiledfrom1968OECDdata,http://members.iinet.net.au/~dgreen/timeline.html;DouglasJones,"FrequentlyAskedQuestionsabouttheDECPDP-8computer,"ftp://rtfrn.mit.edu/pub/usenet/alt.sys.pdp8/PDP-8_Frequently_Asked_Questions_%28posted_every_other_month%29;ProgrammedDataProcessor-1Handbook,DigitalEquipmentCorporation(1960–1963),http://www.dbit.com/greeng3/pdp1/pdp1.html#INTRODUCTION;JohnWalker,"TypicalUNIVAC®1108Prices:1968,"http://www.fourmilab.ch/documents/univac/config1108.html;JackHarper,"LISP1.5fortheUnivac1100Mainframe,"http://www.frobenius.com/univac.htm;Wikipedia,"DataGeneralNova,"http://www.answers.com/topic/data-general-nova;DarrenBrewer,"ChronologyofPersonalComputers1972–1974,"http://uk.geocities.com/magoos_universe/comp1972.htm;www.pricewatch.com;http://www.jc-news.com/parse.cgi?news/pricewatch/raw/pw-010702;http://www.jc-news.com/parse.cgi?news/pricewatch/raw/pw-020624;http://www.pricewatch.com(11/17/04);http://sharkyextreme.com/guidesIWMPG/article.php/10706_2227191_2;Byteadvertisements,September1975–March1998;PCComputingadvertisements,March1977–April2000.

48.Seagate,"Products,".http://www.seagate.com/cda/products/discsales/index;Byteadvertisements,1977–1998;PCComputingadvertisements,March1999;EditorsofTime-LifeBooks,UnderstandingComputers:MemoryandStorage,rev.ed.(NewYork:WarnerBooks,1990);"HistoricalNotesabouttheCostofHardDriveStorageSpace,"http://www.alts.net/ns1625/winchest.html;"IBM305RAMACComputerwithDiskDrive,"http://www.cedmagic.com/history/ibm-305-ramac.html;JohnC.McCallum,"DiskDrivePrices(1955–2004),"http://www.jcmit.com/diskprice.htm.

49.JamesDeRose,TheWirelessDataHandbook(St.Johnsbury,Vt.:Quantrum,1996);FirstMileWireless,http://www.firstmilewireless.coml;J.B.Miles,"WirelessLANs,"GovernmentComputerNews18.28(April30,1999),http://www.gcn.com/vo118_no28/guide/514-1.html;WirelessWeek(April14,1997),http://www.wirelessweek.com/toc/4%2F14%2F1997;OfficeofTechnologyAssessment,"WirelessTechnologiesandtheNationalInformationInfrastructure,"September1995,http://infoventures.com/emf/federal/ota/ota95-tc.html;SignalLake,"BroadbandWirelessNetworkEconomicsUpdate,"January14,2003,http://www.signallake.com/publications/broadbandupdate.pdf;BridgeWaveCommunicationscommunication,http://www.bridgewave.com/050604.htm.

50.InternetSoftwareConsortium(http://www.isc.org),ISCDomainSurvey:NumberofInternetHosts,http://www.isc.org/ds/host-count-history.html.

51.Ibid.52.AveragetrafficonInternetbackbonesintheU.S.duringDecemberofeachyearisusedto

estimatetrafficfortheyear.A.M.Odlyzko,"InternetTrafficGrowth:Sourcesand

Implications,"OpticalTransmissionSystemsandEquipmentforWDMNetworkingII,B.B.Dingel,W.Weiershausen,A.K.Dutta,andK.-I.Sato,eds.,Proc.SPIE(TheInternationalSocietyforOpticalEngineering)5247(2003):1–15,http://www.dtc.umn.edu/~odlyzko/doc/oft.internet.growth.pdf;datafor2003–2004values:e-mailcorrespondencewithA.M.Odlyzko.

53.DaveKristula,"TheHistoryoftheInternet"(March1997,updateAugust2001),http://www.davesite.com/webstation/net-history.shtml;RobertZakon,"Hobbes'InternetTimelinev8.0,"http://www.zakon.org/robert/internet/timeline;ConvergeNetworkDigest,December5,2002,http://www.convergedigest.com/Daily/daily.asp?vn=v9n229&fecha=December%2005,%202002;V.Cerf,"Cerf'sUp,"2004,http://global.mci.com/de/resources/cerfs_up/.

54.H.C.Nathansonetal.,"TheResonantGateTransistor,"IEEETransactionsonElectronDevices14.3(March1967):117–33;LarryJ.Hornbeck,"128x128DeformableMirrorDevice,"IEEETransactionsonElectronDevices30.5(April1983):539–43;J.StorrsHall,"NanocomputersandReversibleLogic,"Nanotechnology5(July1994):157–67;V.V.Aristovetal.,"ANewApproachtoFabricationofNanostructures,"Nanotechnology6(April1995):35–39;C.Montemagnoetal.,"ConstructingBiologicalMotorPoweredNanomechanicalDevices,"Nanotechnology10(1999):225–31,http://www.foresight.org/Conferences/MNT6/Papers/Montemagno/;CelesteBiever,"Tiny'Elevator'MostComplexNanomachineYet,"NewScientist.comNewsService,March18,2004,http://www.newscientist.com/article.ns?id=dn4794.

55.ETCGroup,"FromGenomestoAtoms:TheBigDown,"p.39,http://www.etcgroup.org/documents/TheBigDown.pdf.

56.Ibid.,p.41.57.Althoughitisnotpossibletodeterminepreciselytheinformationcontentinthegenome,

becauseoftherepeatedbasepairsitisclearlymuchlessthanthetotaluncompresseddata.Herearetwoapproachestoestimatingthecompressedinformationcontentofthegenome,bothofwhichdemonstratethatarangeofthirtytoonehundredmillionbytesisconservativelyhigh.

1.Intermsoftheuncompresseddata,therearethreebillionDNArungsinthehumangenetic

code,eachcodingtwobits(sincetherearefourpossibilitiesforeachDNAbasepair).Thus,thehumangenomeisabout800millionbytesuncompressed.ThenoncodingDNAusedtobecalled"junkDNA,"butitisnowclearthatitplaysanimportantroleingeneexpression.However,itisveryinefficientlycoded.Foronething,therearemassiveredundancies(forexample,thesequencecalled"ALU"isrepeatedhundredsofthousandsoftimes),whichcompressionalgorithmscantakeadvantageof.Withtherecentexplosionofgeneticdatabanks,thereisagreatdealofinterestincompressinggeneticdata.Recentworkonapplyingstandarddatacompressionalgorithmstogeneticdataindicatesthatreducingthedataby90percent(forbit-perfectcompression)isfeasible:HisahikoSatoetal.,"DNADataCompressioninthePostGenomeEra,"GenomeInformatics12(2001):512–14,http://www.jsbi.org/journal/GIW01/GIW01P130.pdf.Thuswecancompressthegenometoabout80millionbyteswithoutlossofinformation(meaningwecanperfectlyreconstructthefull800-million-byteuncompressedgenome).Nowconsiderthatmorethan98percentofthegenomedoesnotcodeforproteins.Evenafterstandarddatacompression(whicheliminatesredundanciesandusesadictionarylookupforcommonsequences),thealgorithmiccontentofthenoncodingregionsappearstoberatherlow,meaningthatitislikelythatwecouldcodeanalgorithmthatwouldperformthesamefunctionwithfewerbits.However,sincewearestillearlyintheprocessofreverseengineeringthegenome,wecannotmakeareliableestimateofthisfurtherdecreasebasedonafunctionallyequivalentalgorithm.Iamusing,therefore,arangeof30to100millionbytesofcompressedinformationinthegenome.Thetoppartofthisrangeassumesonlydatacompressionandnoalgorithmicsimplification.

Onlyaportion(althoughthemajority)ofthisinformationcharacterizesthedesignofthebrain.

2.Anotherlineofreasoningisasfollows.Thoughthehumangenomecontainsaround3billionbases,onlyasmallpercentage,asmentionedabove,codesforproteins.Bycurrentestimates,thereare26,000genesthatcodeforproteins.Ifweassumethosegenesaverage3,000basesofusefuldata,thoseequalonlyapproximately78millionbases.AbaseofDNArequiresonlytwobits,whichtranslatetoabout20millionbytes(78millionbasesdividedbyfour).Intheprotein-codingsequenceofagene,each"word"(codon)ofthreeDNAbasestranslatesintooneaminoacid.Thereare,therefore,43(64)possiblecodoncodes,eachconsistingofthreeDNAbases.Thereare,however,only20aminoacidsusedplusastopcodon(nullaminoacid)outofthe64.Therestofthe43codesareusedassynonymsofthe21usefulones.Whereas6bitsarerequiredtocodefor64possiblecombinations,onlyabout4.4(log221)bitsarerequiredtocodefor21possibilities,asavingsof1.6outof6bits(about27percent),bringingusdowntoabout15millionbytes.Inaddition,somestandardcompressionbasedonrepeatingsequencesisfeasiblehere,althoughmuchlesscompressionispossibleonthisprotein-codingportionoftheDNAthanintheso-calledjunkDNA,whichhasmassiveredundancies.thiswillbringthefigureprobablybelow12millionbytes.However,nowwehavetoaddinformationforthenoncodingportionoftheDNAthatcontrolsgeneexpression.AlthoughthisportionoftheDNAcomprisesthebulkofthegenome,itappearstohavealowlevelofinformationcontentandisrepletewithmassiveredundancies.Estimatingthatitmatchestheapproximately12millionbytesofprotein-codingDNA,weagaincometoapproximately24millionbytes.Fromthisperspective,anestimateof30to100millionbytesisconservativelyhigh.

58.Continuousvaluescanberepresentedbyfloating-pointnumberstoanydesireddegreeof

accuracy.Afloating-pointnumberconsistsoftwosequencesofbits.One"exponent"sequencerepresentsapowerof2.The"base"sequencerepresentsafractionof1.Byincreasingthenumberofbitsinthebase,anydesireddegreeofaccuracycanbeachieved.

59.StephenWolfram,ANewKindofScience(Champaign,Ill.:WolframMedia,2002).60.EarlyworkonadigitaltheoryofphysicswasalsopresentedbyFrederickW.Kantor,

InformationMechanics(NewYork:JohnWileyandSons,1977).LinkstoseveralofKantor'spaperscanbefoundathttp://w3.execnet.com/kantor/pm00.htm(l997);http://w3.execnet.com/kantor/1b2p.htm(l989);andhttp://w3.execnet.com/kantor/ipoim.htm(l982).Alsoseeathttp://www.kx.com/listbox/k/msg05621.html.

61.KonradZuse,"RechnenderRaum,"ElektronischeDatenverarbeitung,1967,vol.8,pp.336–44.KonradZuse'sbookonacellularautomaton-baseduniversewaspublishedtwoyearslater:RechnenderRaum,SchriftenzurDatenverarbeitung(Braunschweig,Germany:FriedrichVieweg&Sohn,1969).Englishtranslation:CalculatingSpace,MITTechnicalTranslationAZT-70-164-GEMIT,February1970.MITProjectMAC,Cambridge,MA02139.PDF.

62.EdwardFredkinquotedinRobertWright,"DidtheUniverseJustHappen?"AtlanticMonthly,April1988,29–44,http://digitalphysics.org/Publications/Wri88a/html.

63.Ibid.64.ManyofFredkin'sresultscomefromstudyinghisownmodelofcomputation,which

explicitlyreflectsanumberoffundamentalprinciplesofphysics.SeetheclassicarticleEdwardFredkinandTommasoToffoli,"ConservativeLogic,"InternationalJournalofTheoreticalPhysics21.3–4(l982):219–53,http://www.digitalphilosophy.org/download_documents/ConservativeLogic.pdf.Also,asetofconcernsaboutthephysicsofcomputationanalyticallysimilartothoseofFredkin'smaybefoundinNormanMargolus,"PhysicsandComputation,"Ph.D.thesis,MIT/LCS/TR-415,MITLaboratoryforComputerScience,1988.

65.IdiscussedNorbertWienerandEdFredkin'sviewofinformationasthefundamentalbuilding

blockforphysicsandotherlevelsofrealityinmy1990book,TheAgeofIntelligentMachines.Thecomplexityofcastingallofphysicsintermsofcomputationaltransformationsprovedtobeanimmenselychallengingproject,butFredkinhascontinuedhisefforts.Wolframhasdevotedaconsiderableportionofhisworkoverthepastdecadetothisnotion,apparentlywithonlylimitedcommunicationwithsomeoftheothersinthephysicscommunitywhoarealsopursuingtheidea.Wolfram'sstatedgoal"isnottopresentaspecificultimatemodelforphysics,"butinhis"NoteforPhysicists"(whichessentiallyequatestoagrandchallenge),Wolframdescribesthe"featuresthat[he]believe[s]suchamodelwillhave"(ANewKindofScience,pp.1043–65,http://www.wolframscience.com/nksonline/page-1043c-text).InTheAgeofIntelligentMachines,Idiscuss"thequestionofwhethertheultimatenatureofrealityisanalogordigital"andpointoutthat"aswedelvedeeperanddeeperintobothnaturalandartificialprocesses,wefindthenatureoftheprocessoftenalternatesbetweenanaloganddigitalrepresentationsofinformation."Asanillustration,Idiscussedsound.Inourbrains,musicisrepresentedasthedigitalfiringofneuronsinthecochlea,representingdifferentfrequencybands.Intheairandinthewiresleadingtoloudspeakers,itisananalogphenomenon.Therepresentationofsoundonacompactdiscisdigital,whichisinterpretedbydigitalcircuits.Butthedigitalcircuitsconsistofthresholdedtransistors,whichareanalogamplifiers.Asamplifiers,thetransistorsmanipulateindividualelectrons,whichcanbecountedandare,therefore,digital,butatadeeperlevelelectronsaresubjecttoanalogquantum-fieldequations.Atayetdeeperlevel,FredkinandnowWolframaretheorizingadigital(computational)basistothesecontinuousequations.Itshouldbefurthernotedthatifsomeoneactuallydoessucceedinestablishingsuchadigitaltheoryofphysics,wewouldthenbetemptedtoexaminewhatsortsofdeepermechanismsareactuallyimplementingthecomputationsandlinksofthecellularautomata.Perhapsunderlyingthecellularautomatathatruntheuniverseareyetmorebasicanalogphenomena,which,liketransistors,aresubjecttothresholdsthatenablethemtoperformdigitaltransactions.Thus,establishingadigitalbasisforphysicswillnotsettlethephilosophicaldebateastowhetherrealityisultimatelydigitaloranalog.Nonetheless,establishingaviablecomputationalmodelofphysicswouldbeamajoraccomplishment.Sohowlikelyisthis?Wecaneasilyestablishanexistenceproofthatadigitalmodelofphysicsisfeasible,inthatcontinuousequationscanalwaysbeexpressedtoanydesiredlevelofaccuracyintheformofdiscretetransformationsondiscretechangesinvalue.Thatis,afterall,thebasisforthefundamentaltheoremofcalculus.However,expressingcontinuousformulasinthiswayisaninherentcomplicationandwouldviolateEinstein'sdictumtoexpressthings"assimplyaspossible,butnosimpler."Sotherealquestioniswhetherwecanexpressthebasicrelationshipsthatweareawareofinmoreelegantterms,usingcellular-automataalgorithms.Onetestofanewtheoryofphysicsiswhetheritiscapableofmakingverifiablepredictions.Inatleastoneimportantway,thatmightbeadifficultchallengeforacellularautomata-basedtheorybecauselackofpredictabilityisoneofthefundamentalfeaturesofcellularautomata.Wolframstartsbydescribingtheuniverseasalargenetworkofnodes.Thenodesdonotexistin"space,"butratherspace,asweperceiveit,isanillusioncreatedbythesmoothtransitionofphenomenathroughthenetworkofnodes.Onecaneasilyimaginebuildingsuchanetworktorepresent"naive"(Newtonian)physicsbysimplybuildingathree-dimensionalnetworktoanydesireddegreeofgranularity.Phenomenasuchas"particles"and"waves"thatappeartomovethroughspacewouldberepresentedby"cellulargliders,"whicharepatternsthatareadvancedthroughthenetworkforeachcycleofcomputation.FansofthegameLife(whichisbasedoncellularautomata)willrecognizethecommonphenomenonofglidersandthediversityofpatternsthatcanmovesmoothlythroughacellular-automatonnetwork.Thespeedoflight,then,istheresultoftheclockspeedofthecelestialcomputer,sincegliderscanadvanceonlyonecellpercomputationalcycle.Einstein'sgeneralrelativity,whichdescribesgravityasperturbationsinspaceitself,asifourthree-dimensionalworldwerecurvedinsomeunseenfourthdimension,isalso

straightforwardtorepresentinthisscheme.Wecanimagineafour-dimensionalnetworkandcanrepresentapparentcurvaturesinspaceinthesamewaythatonerepresentsnormalcurvaturesinthree-dimensionalspace.Alternatively,thenetworkcanbecomedenserincertainregionstorepresenttheequivalentofsuchcurvature.Acellular-automataconceptionprovesusefulinexplainingtheapparentincreaseinentropy(disorder)thatisimpliedbythesecondlawofthermodynamics.Wehavetoassumethatthecellular-automataruleunderlyingtheuniverseisaclass4rule(seemaintext)—otherwisetheuniversewouldbeadullplaceindeed.Wolfram'sprimaryobservationthataclass4cellularautomatonquicklyproducesapparentrandomness(despiteitsdeterminateprocess)isconsistentwiththetendencytowardrandomnessthatweseeinBrownianmotionandthatisimpliedbythesecondlaw.Specialrelativityismoredifficult.ThereisaneasymappingfromtheNewtonianmodeltothecellularnetwork.ButtheNewtonianmodelbreaksdowninspecialrelativity.IntheNewtonianworld,ifatrainisgoingeightymilesperhourandyoudrivealongitonaparallelroadatsixtymilesperhour,thetrainwillappeartopullawayfromyouattwentymilesperhour.Butintheworldofspecialrelativity,ifyouleaveEarthatthreequartersofthespeedoflight,lightwillstillappeartoyoutomoveawayfromyouatthefullspeedoflight.Inaccordancewiththisapparentlyparadoxicalperspective,boththesizeandsubjectivepassageoftimefortwoobserverswillvarydependingontheirrelativespeed.Thus,ourfixedmappingofspaceandnodesbecomesconsiderablymorecomplex.Essentially,eachobserverneedshisorherownnetwork.However,inconsideringspecialrelativity,wecanessentiallyapplythesameconversiontoour"Newtonian"networkaswedotoNewtonianspace.However,itisnotclearthatweareachievinggreatersimplicityinrepresentingspecialrelativityinthisway.Acellular-noderepresentationofrealitymayhaveitsgreatestbenefitinNOTES521understandingsomeaspectsofthephenomenonofquantummechanics.Itcouldprovideanexplanationfortheapparentrandomnessthatwefindinquantumphenomena.Consider,forexample,thesuddenandapparentlyrandomcreationofparticle-antiparticlepairs.Therandomnesscouldbethesamesortofrandomnessthatweseeinclass4cellularautomata.Althoughpredetermined,thebehaviorofclass4automatacannotbeanticipated(otherthanbyrunningthecellularautomata)andiseffectivelyrandom.Thisisnotanewview.It'sequivalenttothe"hiddenvariables"formulationofquantummechanics,whichstatesthattherearesomevariablesthatwecannototherwiseaccessthatcontrolwhatappearstoberandombehaviorthatwecanobserve.Thehidden-variablesconceptionofquantummechanicsisnotinconsistentwiththeformulasforquantummechanics.Itispossiblebutisnotpopularwithquantumphysicistsbecauseitrequiresalargenumber.ofassumptionstoworkoutinaveryparticularway.However,Idonotviewthisasagoodargumentagainstit.Theexistenceofouruniverseisitselfveryunlikelyandrequiresmanyassumptionstoallworkoutinaverypreciseway.Yethereweare.Abiggerquestionis,Howcouldahidden-variablestheorybetested?Ifbasedoncellular-automata-likeprocesses,thehiddenvariableswouldbeinherentlyunpredictable,evenifdeterministic.Wewouldhavetofindsomeotherwayto"unhide"thehiddenvariables.Wolfram'snetworkconceptionoftheuniverseprovidesapotentialperspectiveonthephenomenonofquantumentanglementandthecollapseofthewavefunction.Thecollapseofthewavefunction,whichrendersapparentlyambiguouspropertiesofaparticle(forexample,itslocation)retroactivelydetermined,canbeviewedfromthecellular-networkperspectiveastheinteractionoftheobservedphenomenonwiththeobserveritself.Asobservers,wearenotoutsidethenetworkbutexistinsideit.Weknowfromcellularmechanicsthattwoentitiescannotinteractwithoutbothbeingchanged,whichsuggestsabasisforwave-functioncollapse.Wolframwrites,"Iftheuniverseisanetwork,thenitcaninasenseeasilycontainthreadsthatcontinuetoconnectparticlesevenwhentheparticlesgetfarapartintermsofordinaryspace."Thiscouldprovideanexplanationforrecentdramaticexperimentsshowingnonlocalityofactioninwhichtwo"quantumentangled"particlesappeartocontinuetoactinconcertwith

eachothereventhoughseparatedbylargedistances.Einsteincalledthis"spookyactionatadistance"andrejectedit,althoughrecentexperimentsappeartoconfirmit.Somephenomenafitmoreneatlyintothiscellularautomata-networkconceptionthanothers.Someofthesuggestionsappearelegant,butasWolfram's"NoteforPhysicists"makesclear,thetaskoftranslatingallofphysicsintoaconsistentcellular-automata-basedsystemisdauntingindeed.Extendinghisdiscussiontophilosophy,Wolfram"explains"theapparentphenomenonoffreewillasdecisionsthataredeterminedbutunpredictable.Since522NOTESI:1'.IIIfthereisnowaytopredicttheoutcomeofacellularprocesswithoutactuallyrunningtheprocess,andsincenosimulatorcouldpossiblyrunfasterthantheuniverseitself,thereisthereforenowaytoreliablypredicthumandecisions.Soeventhoughourdecisionsaredetermined,thereisnowaytopreidentifywhattheywillbe.However,thisisnotafullysatisfactoryexaminationoftheconcept.Thisobservationconcerningthelackofpredictabilitycanbemadefortheoutcomeofmostphysicalprocesses—suchaswhereapieceofdustwillfallontheground.Thisviewtherebyequateshumanfreewillwiththerandomdescentofapieceofdust.Indeed,thatappearstobeWolfram'sviewwhenhestatesthattheprocessinthehumanbrainis"computationallyequivalent"tothosetakingplaceinprocessessuchasfluidturbulence.Someofthephenomenainnature(forexample,clouds,coastlines)arecharacterizedbyrepetitivesimpleprocessessuchascellularautomataandfractals,butintelligentpatterns(suchasthehumanbrain)requireanevolutionaryprocess(oralternatively,thereverseengineeringoftheresultsofsuchaprocess).Intelligenceistheinspiredproductofevolutionandisalso,inmyview,themostpowerful"force"intheworld,ultimatelytranscendingthepowersofmindlessnaturalforces.Insummary,Wolfram'ssweepingandambitioustreatisepaintsacompellingbutultimatelyoverstatedandincompletepicture.Wolframjoinsagrowingcommunityofvoicesthatmaintainthatpatternsofinformation,ratherthanmatterandenergy,representthemorefundamentalbuildingblocksofreality.Wolframhasaddedtoourknowledgeofhowpatternsofinformationcreatetheworldweexperience,andIlookforwardtoaperiodofcollaborationbetweenWolframandhiscolleaguessothatwecanbuildamorerobustvisionoftheubiquitousroleofalgorithmsintheworld.Thelackofpredictabilityofclass4cellularautomataunderliesatleastsomeoftheapparentcomplexityofbiologicalsystemsanddoesrepresentoneoftheimportantbiologicalparadigmsthatwecanseektoemulateinourtechnology.Itdoesnotexplainallofbiology.Itremainsatleastpossible,however,thatsuchmethodscanexplainallofphysics.IfWolfram,oranyoneelseforthatmatter,succeedsinformulatingphysicsintermsofcellular-automataoperationsandtheirpatterns,Wolfram'sbookwillhaveearneditstitle.Inanyevent,Ibelievethebooktobeanimportantworkofontology.

66.Rule110statesthatacellbecomeswhiteifitspreviouscolorwas,anditstwoneighborsare,allblackorallwhite,orifitspreviouscolorwaswhiteandthetwoneighborsareblackandwhite,respectively;otherwise,thecellbecomesblack.

67.Wolfram,NewKindofScience,p.4,http://www.wolframscience.com/nksonline/page-4-text.68.Notethatcertaininterpretationsofquantummechanicsimplythattheworldisnotbasedon

deterministicrulesandthatthereisaninherentquantumrandomnesstoeveryinteractionatthe(small)quantumscaleofphysicalreality.

69.Asdiscussedinnote57above,theuncompressedgenomehasaboutsixbillionbitsofinformation(orderofmagnitude=1010bits),andthecompressedgenomeisabout30to100millionbytes.Someofthisdesigninformationapplies,ofcourse,tootherorgans.Evenassumingallof100millionbytesappliestothebrain,wegetaconservativelyhighfigureof109bitsforthedesignofthebraininthegenome.Inchapter3,Idiscussanestimatefor"humanmemoryonthelevelofindividualinterneuronalconnections,"including"theconnectionpatternsandneurotransmitterconcentrations"of1018(billionbillion)bitsinamaturebrain.This

isaboutabillion(109)timesmoreinformationthanthatinthegenomewhichdescribesthebrain'sdesign.Thisincreasecomesaboutfromtheself-organizationofthebrainasitinteractswiththeperson'senvironment.

70.Seethesections"Disdisorder"and"TheLawofIncreasingEntropyVersustheGrowthofOrder"inmybookTheAgeofSpiritualMachines:WhenComputersExceedHumanIntelligence(NewYork:Viking,1999),pp.30–33.

71.Auniversalcomputercanacceptasinputthedefinitionofanyothercomputerandthensimulatethatothercomputer.Thisdoesnotaddressthespeedofsimulation,whichmightberelativelyslow.

72.C.GeoffreyWoods,"CrossingtheMidline,"Science304.5676(June4,2004):1455–56;StephenMatthews,"EarlyProgrammingoftheHypothalamo-Pituitary-AdrenalAxis,"TrendsinEndocrinologyandMetabolism13.9(November1,2002):373–80;JustinCrowleyandLawrenceKatz,"EarlyDevelopmentofOcularDominanceColumns,"Science290.5495(November17,2000):1321–24;AnnaPennetal.,"CompetitionintheRetinogenicu1atePatterningDrivenbySpontaneousActivity,"Science279.5359(March27,1998):2108–12.

73.ThesevencommandsofaTuringmachineare:(1)ReadTape,(2)MoveTapeLeft,(3)MoveTapeRight,(4)Write0ontheTape,(5)Write1ontheTape,(6)JumptoAnotherCommand,and(7)Halt.

74.Inwhatisperhapsthemostimpressiveanalysisinhisbook,WolframshowshowaTuringmachinewithonlytwostatesandfivepossiblecolorscanbeauniversalTuringmachine.Forfortyyears,we'vethoughtthatauniversalTuringmachinehadtobemorecomplexthanthis.AlsoimpressiveisWolfram'sdemonstrationthatrule110iscapableofuniversalcomputation,giventherightsoftware.Ofcourse,universalcomputationbyitselfcannotperformusefultaskswithoutappropriatesoftware.

75.The"nor"gatetransformstwoinputsintooneoutput.Theoutputof"nor"istrueifandonlyifneitherAnorBistrue.

76.Seethesection"AnorB:TheBasisofIntelligence?"inTheAgeofIntelligentMachines(Cambridge,Mass.:MITPress,1990),pp.152–57,http://www.KurzweilAl.net/meme/frame.html?m=12.

77.UnitedNationsEconomicandSocialCommissionforAsiaandthePacific,"RegionalRoadMapTowardsanInformationSocietyinAsiaandthePacific,"ST/ESCAP/2283,http://www.unescap.org/publications/detail.asp?id=771;EconomicandSocialCommissionforWesternAsia,"RegionalProfileoftheInformationSocietyinWesternAsia,"October8,2003,http://www.escwa.org.lb/information/publications/ictd/docs/ictd-03-11-e.pdf;JohnEnger,"AsiaintheGlobalInformationEconomy:TheRiseofRegion-States,TheRoleofTelecommunications,"presentationattheInternationalConferenceonSatelliteandCableTelevisioninChineseandAsianRegions,CommunicationArtsResearchInstitute,FuIenCatholicUniversity,June4–6,1996.

78.See"The3by5Initiative,"FactSheet274,December2003,http://www.who.int/mediacentre/factsheets/2003/fs274/en/print.html.

79.Technologyinvestmentsaccountedfor76percentof1998venture-capitalinvestments($10.1billion)(PricewaterhouseCoopersnewsrelease,"VentureCapitalInvestmentsRise24PercentandSetRecordat$14.7Billion,PricewaterhouseCoopersFinds,"February16,1999).In1999,technology-basedcompaniescornered90percentofventure-capitalinvestments($32billion)(PricewaterhouseCoopersnewsrelease,"VentureFundingExplosionContinues:AnnualandQuarterlyInvestmentRecordsSmashed,AccordingtoPricewaterhouseCoopersMoneyTreeNationalSurvey,"February14,2000).Venture-capitallevelscertainlydroppedduringthehigh-techrecession;butinjustthesecondquarterof2003,softwarecompaniesaloneattractedcloseto$1billion(PricewaterhouseCoopersnewsrelease,"VentureCapitalInvestmentsStabilizeinQ22003,"July29,2003).In1974inallU.S.manufacturingindustriesforty-twofirmsreceivedatotalof$26.4millioninventure-capitaldisbursements(in1974dollars,or$81millionin1992

dollars).SamuelKortumandJoshLerner,"AssessingtheContributionofVentureCapitaltoInnovation,"RANDJournalofEconomics31.4(Winter2000):674–92,http://econ.bu.edu/kortum/rje_Winter'00_Kortum.pdf.AsPaulGompersandJoshLernersay,"Inflowstoventurecapitalfundshaveexpandedfromvirtuallyzerointhemid-1970s....,"GompersandLerner,TheVentureCapitalCycle,(Cambridge,Mass.:MITPress,1999).SeealsoPaulGompers,"VentureCapital,"inB.EspenEckbo,ed.,HandbookofCorporateFinance:EmpiricalCorporateFinance,intheHandbooksinFinanceseries(Holland:Elsevier,forthcoming),chapter11,2005,http://mba.tuck.dartmouth.edu/pages/faculty/espen.eckbo/PDFs/Handbookpdf/CH11-VentureCapital.pdf.

80.Anaccountofhow"neweconomy"technologiesaremakingimportanttransformationsto"oldeconomy"industries:JonathanRauch,"TheNewOldEconomy:Oil,Computers,andtheReinventionoftheEarth,"AtlanticMonthly,January3,2001.

81.U.S.DepartmentofCommerce,BureauofEconomicAnalysis(http://www.bea.doc.gov),usethefollowingsiteandselectTable1.1.6:http://www.bea.doc.gov/bealdn/nipaweb/SelectTable.asp?Selected=N.

82.U.S.DepartmentofCommerce,BureauofEconomicAnalysis,http://www.bea.doc.gov.Datafor1920–1999:PopulationEstimatesProgram,PopulationDivision,U.S.CensusBureau,"HistoricalNationalPopulationEstimates:July1,1900toJulyI,1999,"http://www.census.gov/popest/archivesl1990s/popdockest.txt;datafor2000–2004:http://www.census.gov/popest/states/tables/NST-EST2004-01.pdf

83."TheGlobalEconomy:FromRecoverytoExpansion,"ResultsfromGlobalEconomicProspects2005:Trade,RegionalismandProsperity(WorldBank,2004),http://globaloutlook.worldbank.org/globaloutlook/outside/globalgrowth.aspx;"WorldBank:2004EconomicGrowthLiftsMillionsfromPoverty,"VoiceofAmericaNews,http://www.voanews.com/english/2004-11-17-voa41.cfrn.

84.MarkBilsandPeterKlenow,"TheAccelerationinVarietyGrowth,"AmericanEconomicReview91.2(May2001):274–80,http://www.klenow.com/Acceleration.pdf.

85.Seenotes84,86,and87.86.U.S.DepartmentofLabor,BureauofLaborStatistics,newsreport,June3,2004.Youcan

generateproductivityreportsathttp://www.bls.gov/bls/productivity.htm.87.BureauofLaborStatistics,MajorSectorMultifactorProductivityIndex,Manufacturing

Sector:OutputperHourAllPersons(1996=100),http://data.bls.gov/PDQ/outside.jsp?survey=mp(RequiresJavaScript:select"Manufacturing,""OutputPerHourAllPersons,"andstartingyear1949),orhttp://data.bls.gov/cgi-bin/srgate(useseries"MPU300001,""AllYears,"andFormat2).

88.GeorgeM.Scalise,SemiconductorIndustryAssociation,in"LuncheonAddress:TheIndustryPerspectiveonSemiconductors,"2004ProductivityandCyclicalityinSemiconductors:Trends,Implications,andQuestions—ReportofaSymposium(2004)(NationalAcademiesPress,2004),p.40,http://www.nap.edu/openbook/0309092744/html/index.html.

89.DatafromKurzweilAppliedIntelligence,nowpartofScanSoft(formerlyKurzweilComputerProducts).

90.eMarketer,"E-Businessin2003:HowtheInternetIsTransformingCompanies,Industries,andtheEconomy—aReviewinNumbers,"February2003;"USB2CE-CommercetoTop$90Billionin2003,"April30,2003,http://www.emarketer.com/Article.aspx?1002207;and"WorldwideB2BE-CommercetoSurpass$1TrillionByYear'sEnd,"March19,2003,http://www.emarketer.com/Article.aspx?1002125.

91.Thepatentsusedinthischartare,asdescribedbytheU.S.PatentandTrademarkOffice,"patentsforinventions,"alsoknownas"utility"patents.TheU.S.PatentandTrademarkOffice,TableofAnnualU.S.PatentActivity,http://www.uspto.gov/web/offices/ac/ido/oeip/taf/h_counts.htm.

92.ThedoublingtimeforIT'sshareoftheeconomyistwenty-threeyears.U.S.Departmentof

Commerce,EconomicsandStatisticsAdministration,"TheEmergingDigitalEconomy,"figure2,http://www.technology.gov/digeconomy/emerging.htm.

93.ThedoublingtimeforU.S.educationexpenditurespercapitaistwenty-threeyears.NationalCenterforEducationStatistics,DigestofEducationStatistics,2002,http://nces.ed.gov/pubs2003/digest02/tables/dt030.asp.

94.TheUnitedNationsestimatedthatthetotalglobalequitymarketcapitalizationin2000wasthirty-seventrilliondollars.UnitedNations,"GlobalFinanceProfile,"ReportoftheHigh-LevelPanelofFinancingforDevelopment,June2001,http://www.un.org/reports/financing/profile.htm.Ifourperceptionoffuturegrowthratesweretoincrease(comparedtocurrentexpectations)byanannualcompoundedrateofaslittleas2percent,andconsideringanannualdiscountrate(fordiscountingfuturevaluestoday)of6percent,thenconsideringtheincreasedpresentvalueresultingfromonlytwentyyearsofcompoundedanddiscountedfuture(additional)growth,presentvaluesshouldtriple.Asthesubsequentdialoguepointsout,thisanalysisdoesnottakeintoconsiderationthelikelyincreaseinthediscountratethatwouldresultfromsuchaperceptionofincreasedfuturegrowth.

ChapterThree:AchievingtheComputationalCapacityoftheHumanBrain1.GordonE.Moore,"CrammingMoreComponentsontoIntegratedCircuits,"Electronics38.8

(April19,1965):114–17,ftp://download.intel.com/research/silicon/moorespaper.pdf.2.Moore'sinitialprojectioninthis1965paperwasthatthenumberofcomponentswould

doubleeveryyear.In1975thiswasrevisedtoeverytwoyears.However,thismorethandoublesprice-performanceeverytwoyearsbecausesmallercomponentsrunfaster(becausetheelectronicshavelessdistancetotravel).Sooverallprice-performance(forthecostofeachtransistorcycle)hasbeencomingdownbyhalfabouteverythirteenmonths.

3.PaoloGarginiquotedinAnnStefforaMutschler,"Moore'sLawHeretoStay,"ElectronicsWeekly.com,July14,2004,http://www.electronicsweekly.co.uk/articles/article.asp?liArticleID=36829.SeealsoTomKrazit,"IntelPreparesforNext20YearsofChipMaking,"Computerworld,October25,2004,http://www.computerworld.com/hardwaretopics/hardware/story/0,10801,96917,00.html.

4.MichaelKanellos,"'High-rise'ChipsSneakonMarket,"CNETNews.com,July13,2004,http://zdnet.com.com/2100-1103-5267738.html.

5.BenjaminFulford,"ChipmakersAreRunningOutofRoom:TheAnswerMightLiein3-D,"Forbes.com,July22,2002,http://www.forbes.com/forbes/2002/0722/173_print.html.

6.NTTnewsrelease,"Three-DimensionalNanofabricationUsingElectronBeamLithography,"February2,2004,http://www.ntt.co.jp/news/news04e/0402/040202.html.

7.LászlóForróandChristianSchonenberger,"CarbonNanotubes,MaterialsfortheFuture,"EurophysicsNews32.3(200l),http://www.europhysicsnews.com/full/09/article3/article3.html.Alsoseehttp://www.research.ibm.com/nanoscience/nanotubes.htmlforanoverviewofnanotubes.

8.MichaelBernstein,AmericanChemicalSocietynewsrelease,"High-SpeedNanotubeTransistorsCouldLeadtoBetterCellPhones,FasterComputers,"April27,2004,http://www.eurekalert.org/pub_releases/2004-04/acs-nt042704.php.

9.Iestimateananotube-basedtransistorandsupportingcircuitryandconnectionsrequireapproximatelyaten-nanometercube(thetransistoritselfwillbeafractionofthis),or103cubicnanometers.Thisisconservative,sincesingle-wallednanotubesareonlyonenanometerindiameter.Oneinch=2.54centimeters=2.54Î107nanometers.Thus,a1-inchcube=2.543Î1021=1.6Î1022cubicnanometers.Soaone-inchcubecouldprovide1.6Î1019transistors.Witheachcomputerrequiringapproximately107transistors(whichisamuchmorecomplexapparatusthanthatcomprisingthecalculationsinahumaninterneuronalconnection),wecan

supportabout1012(onetrillion)parallelcomputers.Ananotubetransistor-basedcomputerat1012calculationspersecond(basedonBurke'sestimate)givesusaspeedestimateof1024cpsfortheone-inchcubeofnanotubecircuitry.AlsoseeBernstein,"High-SpeedNanotubeTransistors."Withanestimateof1016cpsforfunctionalemulationofthehumanbrain(seediscussionlaterinthischapter),thisgivesusabout100million(108)humanbrainequivalents.Ifweusethemoreconservative1019cpsestimateneededforneuromorphicsimulation(simulatingeverynonlinearityineveryneuralcomponent;seesubsequentdiscussioninthischapter),aone-inchcubeofnanotubecircuitrywouldprovideonlyonehundredthousandhuman-brainequivalents.

10."Onlyfouryearsagodidwemeasureforthefirsttimeanyelectronictransportthroughananotube.Now,weareexploringwhatcanbedoneandwhatcannotintermsofsingle-moleculedevices.Thenextstepwillbetothinkabouthowtocombinetheseelementsintocomplexcircuits,"saysoneoftheauthors,CeesDekker,ofHenkW.Ch.Postmaetal.,"CarbonNanotubeSingle-ElectronTransistorsatRoomTemperature,"Science293.5527(July6,2001):76–129,describedintheAmericanAssociationfortheAdvancementofSciencenewsrelease,"Nano-transistorSwitcheswithJustOneElectronMayBeIdealforMolecularComputers,ScienceStudyShows,"http://www.eurekalert.org/pub_releases/2001-07/aaft-nsw062901.php.

11.TheIBMresearcherssolvedaprobleminnanotubefabrication.Whencarbonsootisheatedtocreatethetubes,alargenumberofunusablemetallictubesarecreatedalongwiththesemiconductortubessuitablefortransistors.Theteamincludedbothtypesofnanotubesinacircuitandthenusedelectricalpulsestoshattertheundesirableones-afarmoreefficientapproachthancherry-pickingthedesirabletubeswithanatomic-forcemicroscope.MarkK.Anderson,"MegaStepsTowardtheNanochip,"WiredNews,April27,2001,athttp://www.wired.com/news/technology/0,1282,43324,00.html,referringtoPhilipG.Collins,MichaelS.Arnold,andPhaedonAvouris,"EngineeringCarbonNanotubesandNanotubeCircuitsUsingElectricalBreakdown,"Science292.5517(April27,2001):706–9.

12."Acarbonnanotube,whichlookslikerolledchickenwirewhenexaminedattheatomiclevel,istensofthousandsoftimesthinnerthanahumanhair,yetremarkablystrong."UniversityofCaliforniaatBerkeleypressrelease,"ResearchersCreateFirstEverIntegratedSiliconCircuitwithNanotubeTransistors,"January5,2004,http://www.berkeley.edu/news/media/releases/2004/01/05_nano.shtml,referringtoYu-ChihTsengetal.,"MonolithicIntegrationofCarbonNanotubeDeviceswithSiliconMOSTechnology,"NanoLetters4.1(2004):123–27,http://pubs.acs.org/cgi-bin/sample.cgi/nalefd/2004/4/i01/pdf/nl0349707.pdf.

13.R.ColinJohnson,"IBMNanotubesMayEnableMolecular-ScaleChips,"EETimes,April26,2001,http://eetimes.com/article/showArticle.jhtml?articleId=10807704.

14.AviAviramandMarkA.Ratner,"MolecularRectifiers,"ChemicalPhysicsLetters(November15,1974):277–83,referredtoinCharlesM.Lieber,"TheIncredibleShrinkingCircuit,"ScientificAmerican(September2001),athttp://www.sciam.comandhttp://www-mcg.uni-r.de/downloads/lieber.pdf.Thesingle-moleculerectifierdescribedinAviramandRatnercouldpasscurrentpreferentiallyineitherdirection.

15.WillKnight,"SingleAtomMemoryDeviceStoresData,"NewScientist.com,September10,2002,http://www.newscientist.com/news/news.jsp?id=ns99992775,referringtoR.Bennewitzetal.,"AtomicScaleMemoryataSiliconSurface,"Nanotechnology13(July4,2002):499–502.

16.Theirtransistorismadefromindiumphosphideandindiumgalliumarsenide.UniversityofIllinoisatUrbana-Champaignnewsrelease,"IllinoisResearchersCreateWorld'sFastestTransistor—Again,"http://www.eurekalert.org/pub_releases/2003-11/uoia-irc110703.php.

17.MichaelR.Diehletal.,"Self-AssembledDeterministicCarbonNanotubeWiringNetworks,"AngewandteChemieInternationalEdition41.2(2002):353–56;C.P.Collieretal.,"ElectronicallyConfigurableMolecular-BasedLogicGates,"Science285.5426(July1999):

391–94.Seehttp://www.its.caltech.edu/~heathgrp/papers/Paperfiles/2002/diehlangchemint.pdfandhttp://www.cs.duke.edu/~thl/papers/Heath.Switch.pdf.

18.The"rosettenanotubes"designedbythePurdueteamcontaincarbon,nitrogen,hydrogen,andoxygen.Therosettesself-assemblebecausetheirinteriorsarehydrophobicandtheirexteriorsarehydrophilic;therefore,toprotecttheirinsidesfromwater,therosettesstackintonanotubes."Thephysicalandchemicalpropertiesofourrosettenanotubescannowbemodifiedalmostatwillthroughanoveldial-inapproach,"accordingtoleadresearcherHichamFenniri.R.ColinJohnson,"PurdueResearchersBuildMade-to-OrderNanotubes,"EETimes,October24,2002,http://www.eetimes.com/article/showArticle.jhtml?articleId=18307660;H.Fennirietal.,"EntropicallyDrivenSelf-AssemblyofMultichannelRosetteNanotubes,"ProceedingsoftheNationalAcademyofSciences99,supp1.2(April30,2002):6487–92;Purduenewsrelease,"AdaptableNanotubesMakeWayforCustom-BuiltStructures,Wires,"http://news.uns.purdue.edu/UNS/html4ever/020311.Fenniri.scaffold.html.SimilarworkhasbeendonebyscientistsintheNetherlands:GaiaVince,"Nano-TransistorSelf-AssemblesUsingBiology,"NewScientist.com,November20,2003,http://www.newscientist.com/news/news.jsp?id=ns99994406.

19.LizKalaugher,"LithographyMakesaConnectionforNanowireDevices,"June9,2004,http://www.nanotechweb.org/articles/news/3/6/6/1,referringtoSongJinetal.,"ScalableInterconnectionandIntegrationofNanowireDevicesWithoutRegistration,"NanoLetters4.5(2004):915–19.

20.ChaoLietal.,"MultilevelMemoryBasedonMolecularDevices,"AppliedPhysicsLetters84.11(March15,2004):1949–51.Alsoseehttp://www.technologyreview.com/articles/rnb_051304.asp?p=1.Seealsohttp://nanolab.usc.edu/PDF%5CAPL84-1949.pdf.

21.GaryStix,"NanoPatterning,"ScientificAmerican(February9,2004),http://www.sciam.com/print_version.cfm?articleID=000170D6-C99F-l01E-861F83414B7F0000;MichaelKanellos,"IBMGetsChipCircuitstoDrawThemselves,"CNETNews.com,http://zdnet.com.com/2100-1103-5114066.html.Seealsohttp://www.nanopolis.net/news_ind.php?type_id=3.

22.IBMisworkingonchipsthatautomaticallyreconfigureasneeded,suchasbyaddingmemoryoraccelerators."Inthefuture,thechipyouhavemaynotbethechipyoubought,"saidBernardMeyerson,chieftechnologist,IBMSystemsandTechnologyGroup.IBMpressrelease,"IBMPlansIndustry'sFirstOpenlyCustomizableMicroprocessor,"http://www.ibm.com/investor/press/mar-2004/31-03-04-1.phtml.

23.BBCNews,"'Nanowire'BreakthroughHailed,"April1,2003,http://news.bbc.co.uk/1/hi/sci/tech/2906621.stm.PublishedarticleisThomasScheibeletal.,"ConductingNanowiresBuiltbyControlledSelf-AssemblyofAmyloidFibersandSelectiveMetalDeposition,"ProceedingsoftheNationalAcademyofSciences100.8(April15,2003):4527–32,publishedonlineApril2,2003,http://www.pnas.org/cgi/content/full/100/8/4527.

24.DukeUniversitypressrelease,"DukeScientists'Program'DNAMoleculestoSelfAssembleintoPatternedNanostructures,"http://www.eurekalert.org/pub_releases/2003-09/du-ds092403.php,referringtoHaoYanetal.,"DNA-TemplatedSelf-AssemblyofProteinArraysandHighlyConductiveNanowires,"Science301.5641(September26,2003):1882–84.Seealsohttp://www.phy.duke.edu/~gleb/Pdf_FILES/DNA_science.pdf.

25.Ibid.26.Hereisanexampleoftheproceduretosolvewhat'scalledthetraveling-salespersonproblem.

Wetrytofindanoptimalrouteforahypotheticaltraveleramongmultiplecitieswithouthavingtovisitacitymorethanonce.Onlycertaincitypairsareconnectedbyroutes,sofindingtherightpathisnotstraightforward.Tosolvethetraveling-salespersonproblem,mathematicianLeonardAdlemanoftheUniversityofSouthernCaliforniaperformedthefollowingsteps:

1.GenerateasmallstrandofDNAwithauniquecodeforeachcity.2.Replicateeachsuchstrand(oneforeachcity)trillionsoftimesusingPCR.3.Next,putthepoolsofDNA(oneforeachcity)togetherinatesttube.Thisstepuses

DNA'saffinitytolinkstrandstogether.Longerstrandswillformautomatically.Eachsuchstrandrepresentsapossiblerouteofmultiplecities.Thesmallstrandsrepresentingeachcitylinkupwitheachotherinarandomfashion,sothereisnomathematicalcertaintythatalinkedstrandrepresentingthecorrectanswer(sequenceofcities)willbeformed.However,thenumberofstrandsissovastthatitisvirtuallycertainthatatleastonestrand—andprobablymillions—willbeformedthatrepresentsthecorrectanswer.

Thenextstepsusespeciallydesignedenzymestoeliminatethetrillionsofstrandsthatrepresentwronganswers,leavingonlythestrandsrepresentingthecorrectanswer:4.Usemoleculescalled"primers"todestroythoseDNAstrandsthatdonotstartwiththestartcity,aswellasthosethatdonotendwiththeendcity;thenreplicatethesurvivingstrands,usingPCR.

4.UseanenzymereactiontoeliminatethoseDNAstrandsthatrepresentatravelpathgreaterthanthetotalnumberofcities.

5.Useanenzymereactiontodestroythosestrandsthatdonotincludecity1.Repeatforeachofthecities.

6.Now,eachofthesurvivingstrandsrepresentsthecorrectanswer.Replicatethesesurvivingstrands(usingPCR)untiltherearebillionsofsuchstrands.

7.Usingatechniquecalledelectrophoresis,readouttheDNAsequenceofthesecorrectstrands(asagroup).Thereadoutlookslikeasetofdistinctlines,whichspecifiesthecorrectsequenceofcities.

SeeL.M.Adleman,"MolecularComputationofSolutionstoCombinatorialProblems,"Science266(1994):1021–24.

27.CharlesChoi,"DNAComputerSetsGuinnessRecord,"http://www.upi.com/view.cfm?StoryID=20030224-045551-7398r.SeealsoY.Benensonetal.,"DNAMoleculeProvidesaComputingMachinewithBothDataandFuel,"ProceedingsoftheNationalAcademyofSciences100.5(March4,2003):2191–96,availableathttp://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=12601148;Y.Benensonetal.,"AnAutonomousMolecularComputerforLogicalControlofGeneExpression,"Nature429.6990(May27,2004):423–29(publishedonline,April28,2004),availableathttp://www.wisdom.weizmann.ac.il/~udi/ShapiroNature2004.pdf.

28.StanfordUniversitynewsrelease,"'Spintronics'CouldEnableaNewGenerationofElectronicDevices,PhysicistsSay,"http://www.eurekalert.org/pub_releases/2003-08/su-ce080803.php,referringtoShuichiMurakami,NaotoNagaosa,andShou-ChengZhang,"DissipationlessQuantumSpinCurrentatRoomTemperature,"Science301.5638(September5,2003):1348–51.

29.CelesteBiever,"Silicon-BasedMagnetsBoostSpintronics,"NewScientist.com,March22,2004,http://www.newscientist.com/news/news.jsp?id=ns99994801,referringtoStevePearton,"Silicon-BasedSpintronics,"NatureMaterials3.4(April2004):203–4.

30.WillKnight,"DigitalImageStoredinSingleMolecule,"NewScientist.com,December1,2002,http://www.newscientist.com/news/news.jsp?id=ns99993129,referringtoAnatolyK.Khitrin,VladimirL.Ermakov,andB.M.Fung,"NuclearMagneticResonanceMolecularPhotography,"JournalofChemicalPhysics117.15(October15,2002):6903-{5.

31.Reuters,"ProcessingattheSpeedofLight,"WiredNews,http://www.wired.com/news/technology/0,1282,61009,00.html.

32.Todate,thelargestnumbertobefactoredisoneof512bits,accordingtoRSASecurity.33.StephanGuldeetal.,"ImplementationoftheDeutsch-IozsaAlgorithmonanIon-Trap

QuantumComputer,"Nature421(January2,2003):48–50.Seehttp://heart-

c704.uibk.ac.at/Papers/Nature03-Gulde.pdf.34.Sincewearecurrentlydoublingtheprice-performanceofcomputationeachyear,afactorofa

thousandrequirestendoublings,ortenyears.Butwearealso(slowly)decreasingthedoublingtimeitself,sotheactualfigureiseightyears.

35.Eachsubsequentthousandfoldincreaseisitselfoccurringataslightlyfasterrate.Seethepreviousnote.

36.HansMoravec,"RiseoftheRobots,"ScientificAmerican(December1999):124–35,http://www.sciam.comandhttp://www.frc.ri.cmu.edu/~hpm/project.archive/robot.papers/1999/SciAm.scan.html.MoravecisaprofessorattheRoboticsInstituteatCarnegieMellonUniversity.HisMobileRobotLaboratoryexploreshowtousecameras,sonars,andothersensorstogiverobots3-Dspatialawareness.Inthe1990s,hedescribedasuccessionofrobotgenerationsthatwould"essentially[be]ouroff-spring,byunconventionalmeans.Ultimately,Ithinkthey'reontheirownandthey'lldothingsthatwecan'timagineorunderstand—youknow,justthewaychildrendo"(NovaOnlineinterviewwithHansMoravec,October1997,http://www.pbs.org/wgbh/nova/robots/moravec.html).HisbooksMindChildren:TheFutureofRobotandHumanIntelligenceandRobot:MereMachinetoTranscendentMindexplorethecapabilitiesofthecurrentandfuturerobotgenerations.Disclosure:TheauthorisaninvestorinandontheboardofdirectorsofMoravec'sroboticscompany,Seegrid.

37.AlthoughinstructionspersecondasusedbyMoravecandcalculationspersecondareslightlydifferentconcepts,thesearecloseenoughforthepurposesoftheseorder-of-magnitudeestimates.Moravecdevelopedthemathematicaltechniquesforhisrobotvisionindependentofbiologicalmodels,butsimilarities(betweenMoravec'salgorithmsandthoseperformedbiologically)werenotedafterthefact.Functionally,Moravec'scomputationsre-createwhatisaccomplishedintheseneuralregions,socomputationalestimatesbasedonMoravec'salgorithmsareappropriateindeterminingwhatisrequiredtoachievefunctionallyequivalenttransformations.

38.LloydWatts,"Event-DrivenSimulationofNetworksofSpikingNeurons,"seventhNeuralInformationProcessingSystemsFoundationConference,1993;LloydWatts,"TheMode-CouplingLiouville-GreenApproximationforaTwo-DimensionalCochlearModel,"JournaloftheAcousticalSocietyofAmerica108.5(November2000):2266–71.WattsisthefounderofAudience,Inc.,whichisdevotedtoapplyingfunctionalsimulationofregionsofthehumanauditorysystemtoapplicationsinsoundprocessing,includingcreatingawayofpreprocessingsoundforautomatedspeech-recognitionsystems.Formoreinformation,seehttp://www.lloydwatts.com/neuroscience.shtml.Disclosure:TheauthorisanadvisertoAudience.

39.U.S.PatentApplication20030095667,U.S.PatentandTrademarkOffice,May22,2003.40.TheMedtronicMiniMedclosed-loopartificialpancreascurrentlyinhumanclinicaltrialsis

returningencouragingresults.Thecompanyhasannouncedthatthedeviceshouldbeonthemarketwithinthenextfiveyears.Medtronicnewsrelease,"MedtronicSupportsJuvenileDiabetesResearchFoundation'sRecognitionofArtificialPancreasasaPotential'Cure'forDiabetes,"March23,2004,http://www.medtronic.com/newsroom/news_2004323a.html.Suchdevicesrequireaglucosesensor,aninsulinpump,andanautomatedfeedbackmechanismtomonitorinsulinlevels(InternationalHospitalFederation,"ProgressinArtificialPancreasDevelopmentforTreatingDiabetes,"http://www.hospitalmanagement.net/informer/technology/tech10).Rocheisalsointheracetoproduceanartificialpancreasby2007.Seehttp://www.roche.com/pages/downloads/science/pdf/rtdcmannh02-6.pdf.

41.Anumberofmodelsandsimulationshavebeencreatedbasedonanalysesofindividualneuronsandinterneuronalconnections.TomasoPoggiowrites,"Oneviewoftheneuronisthatitismorelikeachipwiththousandsoflogical-gates-equivalentsratherthanasinglethreshold

element,"TomasoPoggio,privatecommunicationtoRayKurzweil,January2005.SeealsoT.PoggioandC.Koch,"SynapsesThatComputeMotion,"ScientificAmerican256(1987):46–52.C.KochandT.Poggio,"BiophysicsofComputationalSystems:Neurons,Synapses,andMembranes,"inSynapticFunction,G.M.Edelman,W.E.Gall,andW.M.Cowan,eds.(NewYork:JohnWileyandSons,1987),pp.637–97.Anothersetofdetailedneuron-levelmodelsandsimulationsisbeingcreatedattheUniversityofPennsylvania'sNeuroengineeringResearchLabbasedonreverseengineeringbrainfunctionattheneuronlevel.Dr.LeifFinkel,headofthelaboratory,says,"Rightnowwe'rebuildingacellular-levelmodelofasmallpieceofvisualcortex.It'saverydetailedcomputersimulationwhichreflectswithsomeaccuracyatleastthebasicoperationsofrealneurons.[MycolleagueKwabenaBoahen]hasachipthataccuratelymodelstheretinaandproducesoutputspikesthatcloselymatchrealretinae."Seehttp://nanodot.org/article.pl?sid=0l/12/18/1552221.Reviewsoftheseandothermodelsandsimulationsattheneuronlevelindicatethatanestimateof103calculationsperneuraltransaction(asingletransactioninvolvingsignaltransmissionandresetonasingledendrite)isareasonableupperbound.Mostsimulationsuseconsiderablylessthanthis.

42.PlansforBlueGene/L,thesecondgenerationofBlueGenecomputers,wereannouncedinlate2001.Thenewsupercomputer,plannedtobefifteentimesfasterthantoday'ssupercomputersandonetwentieththesize,isbeingbuiltjointlybytheNationalNuclearSecurityAgency'sLawrenceLivermoreNationalLaboratoryandIBM.In2002,IBMannouncedthatopen-sourceLinuxhadbeenchosenastheoperatingsystemforthenewsupercomputers.ByJuly2003,theinnovativeprocessorchipsforthesupercomputer,whicharecompletesystemsonchips,wereinproduction."BlueGene/Lisaposterchildforwhatispossiblewiththesystem-on-a-chipconcept.Morethan90percentofthischipwasbuiltfromstandardblocksinourtechnologylibrary,"accordingtoPaulCoteus,oneofthemanagersoftheproject(TimothyMorgan,"IBM'sBlueGene/LShowsOffMinimalistServerDesign,"TheFourHundred,http://www.midrangeserver.com/tfh/tfh120103-story05.html).ByJune2004,theBlueGene/Lprototypesystemsappearedforthefirsttimeonthelistoftoptensupercomputers.IBMpressrelease,"IBMSurgesPastHPtoLeadinGlobalSupercomputing,"http://www.research.ibm.com/bluegene.

43.Thistypeofnetworkisalsocalledpeer-to-peer,many-to-many,and"multihop,"Init,nodesinthenetworkcanbeconnectedtoalltheothernodesortoasubset,andtherearemultiplepathsthroughmeshednodestoeachdestination.Thesenetworksarehighlyadaptableandself-organizing."Thesignatureofameshnetworkisthatthereisnocentralorchestratingdevice.Instead,eachnodeisoutfittedwithradiocommunicationsgearandactsasarelaypointforothernodes."SebastianRupley,"Wireless:MeshNetworks,"PCMagazine,July1,2003,http://www.pcmag.com/article2/0,1759,1139094,00.asp;RobertPoor,"WirelessMeshNetworks,"SensorsOnline,February2003,http://www.sensorsmag.com/articles/0203/38/main.shtml;TomasKragandSebastianBuettrich,"WirelessMeshNetworking,"O'ReillyWirelessDevCenter,January22,2004,http://www.oreillynet.com/pub/a/wirelessl2004/01/22/wirelessmesh.html.

44.CarverMead,founderofmorethantwenty-fivecompaniesandholderofmorethanfiftypatents,ispioneeringthenewfieldofneuromorphicelectronicsystems,circuitsmodeledonthebrainandnervoussystem.SeeCarverA.Mead,"NeuromorphicElectronicSystems,"IEEEProceedings78.10(October1990):1629–36.Hisworkledtothecomputertouchpadandthecochlearchipusedindigitalhearingaids.His1999start-upcompanyFoveonmakesanalogimage-sensorsthatimitatethepropertiesoffilm.

45.EdwardFredkin,"APhysicist'sModelofComputation,"ProceedingsoftheTwenty-sixthRecontredeMoriond,TextsofFundamentalSymmetries(1991):283–97,

http://digitalphilosophy.org/physicists_model.htm.46.GeneFrantz,"DigitalSignalProcessingTrends,"IEEEMicro20.6(November/December

2000):52–59,http://csdl.computer.org/comp/mags/mi/2000/06/m6052abs.htm.47.In2004Intelannounceda"righthandturn"switchtowarddual-core(morethanoneprocessor

onachip)architectureafterreachinga"thermalwall"(or"powerwall")causedbytoomuchheatfromever-fastersingleprocessors:http://www.intel.com/employee/retiree/circuit/righthandturn.htm.

48.R.Landauer,"IrreversibilityandHeatGenerationintheComputingProcess,"IBMJournalofResearchDevelopment5(1961):183–91,http://www.research.ibm.com/journal/rd/053/ibmrd0503C.pdf.

49.CharlesH.Bennett,"LogicalReversibilityofComputation,"IBMJournalofResearchDevelopment17(1973):525–32,http://www.research.ibm.com/journal/rd/176/ibmrd1706G.pdf;CharlesH.Bennett,"TheThermodynamicsofComputation-aReview,"InternationalJournalofTheoreticalPhysics21(1982):905–40;CharlesH.Bennett,"Demons,Engines,andtheSecondLaw,"ScientificAmerican257(November1987):108–16.

50.EdwardFredkinandTommasoToffoli,"ConservativeLogic,"InternationalJournalofTheoreticalPhysics21(1982):219–53,http://digitalphilosophy.org/download_documents/ConservativeLogic.pdf.EdwardFredkin,"APhysicist'sModelofComputation,"ProceedingsoftheTwenty-sixthRecontredeMoriond,TestsofFundamentalSymmetries(1991):283–97,http://www.digitalphilosophy.org/physicists_model.htm.

51.Knight,"DigitalImageStoredinSingleMolecule,"referringtoKhitrinetaI.,"NuclearMagneticResonanceMolecularPhotography";seenote30above.

52.Tenbillion(1010)humansat1019cpseachis1029cpsforallhumanbrains;1042cpsisgreaterthanthisbytentrillion(1013).

53.Fredkin,"Physicist'sModelofComputation";seenotes45and50above.54.TwosuchgatesaretheInteractionGate,atwo-input,four-outputuniversal,reversible-logic

gate

andtheFeynmanGate,atwo-input,three-outputreversible,universal-logicgate.

Bothimagesarefromibid.,p.7.

55.Ibid.,p.8.56.C.1.Seitzetal.,"Hot-ClocknMOS,"Proceedingsofthe1985ChapelHillConferenceon

VLSI(Rockville,Md.:ComputerSciencePress,1985),pp.1–17,http://caltechcstr.library.caltech.edu/archive/00000365;RalphC.Merkle,"ReversibleElectronicLogicUsingSwitches,"Nanotechnology4(1993):21–40;S.G.YounisandT.F.Knight,"PracticalImplementationofChargeRecoveringAsymptoticZeroPowerCMOS,"Proceedingsofthe1993SymposiumonIntegratedSystems(Cambridge,Mass.:MITPress,1993),pp.234–50.

57.HiawathaBray,"YourNextBattery,"BostonGlobe,November24,2003,http://www.boston.com/business/technology/articles/2003/11/24/your_next_battery.

58.SethLloyd,"UltimatePhysicalLimitstoComputation,"Nature406(2000):1047–54.EarlyworkonthelimitsofcomputationwasdonebyHansJ.Bremermannin1962:HansJ.Bremermann,"OptimizationThroughEvolutionandRecombination,"inM.C.Yovits,C.T.Jacobi,c.D.Goldstein,eds.,Self-OrganizingSystems(Washington,nc.SpartanBooks,1962),pp.93–106.In1984RobertA.FreitasJr.builtonBremermann'sworkinRobertA.FreitasJr.,"Xenopsychology,"Analog104(April1984):41–53,http://www.rfreitas.com/Astro/Xenopsychology.htm#SentienceQuotient.

59.pÎmaximumenergy(1017kgÎmeter2/second2)/(6.6Î10–34)joule-seconds=~5Î1050operations/second.

60.5Î1050cpsisequivalentto5Î1021(5billiontrillion)humancivilizations(eachrequiring1029cps).

61.Tenbillion(1010)humansat1016cpseachis1026cpsforhumancivilization.So5Î1050cpsisequivalentto5Î1024(5trilliontrillion)humancivilizations.

62.Thisestimatemakestheconservativeassumptionthatwe'vehadtenbillionhumansforthepasttenthousandyears,whichisobviouslynotthecase.Theactualnumberofhumanshasbeenincreasinggraduallyoverthepasttoreachabout6.1billionin2000.Thereare3Î107secondsinayear,and3Î1011secondsintenthousandyears.So,usingtheestimateof1026cpsforhumancivilization,humanthoughtovertenthousandyearsisequivalenttocertainlynomorethan3Î1037calculations.Theultimatelaptopperforms5Î1050calculationsinonesecond.Sosimulatingtenthousandyearsoftenbillionhumans'thoughtswouldtakeitabout10–13seconds,whichisoneten-thousandthofananosecond.

63.AndersSandberg,"ThePhysicsoftheInformationProcessingSuperobjects:DailyLifeAmongtheJupiterBrains,"JournalofEvolution&Technology5(December22,1999),http://www.transhumanist.com/volume5/Brains2.pdf.

64.Seenote62above;1042cpsisafactorof10–8lessthan1050cps,sooneten-thousandthofananosecondbecomes10microseconds.

65.Seehttp://e-drexler.com/p/04/04/0330drexPubs.htmlforalistofDrexler'spublicationsandpatents.

66.Attherateof$1012and1026cpsperthousanddollars($103),weget1035cpsperyearinthemid-2040s.Theratioofthistothe1026cpsforallofthebiologicalthinkinginhumancivilizationis109(onebillion).

67.In1984RobertA.Freitasproposedalogarithmicscaleof"sentiencequotient"(SQ)basedonthecomputationalcapacityofasystem.Inascalethatrangesfrom–70to50,humanbrainscomeoutat13.TheCray1supercomputercomesoutat9.Freitas'ssentiencequotientisbasedontheamountofcomputationperunitmass.AveryfastcomputerwithasimplealgorithmwouldcomeoutwithahighSQ.ThemeasureIdescribeforcomputationinthissectionbuildsonFreitas'sSQandattemptstotakeintoconsiderationtheusefulnessofthecomputation.Soifasimplercomputationisequivalenttotheoneactuallybeingrun,thenwebasethecomputationalefficiencyontheequivalent(simpler)computation.Alsoinmymeasure,thecomputationneedstobe"useful."RobertA.FreitasJr.,"Xenopsychology,"Analog104(April1984):41–53,http://www.rfreitas.comfAstro/Xenopsychology.htm#SentienceQuotient.

68.Asaninterestingaside,engravingsonthesideofsmallrocksdidinfactrepresentanearlyformofcomputerstorage.Oneoftheearliestformsofwrittenlanguage,cuneiform,whichwasdevelopedinMesopotamiacirca3000B.C.,usedpictorialmarkingsonstonestostoreinformation.Agriculturalrecordsweremaintainedascuneiformmarkingsonstonesplacedintrays,andorganizedinrowsandcolumns.Thesemarkedstoneswereessentiallythefirstspreadsheet.Onesuchcuneiformstonerecordisaprizedartifactinmycollectionofhistoricalcomputers.

69.Onethousand(103)bitsislessthanthetheoreticalcapacityoftheatomsinthestonetostoreinformation(estimatedat1027bits)byafactorof10–24.

70.1cps(100cps)islessthanthetheoreticalcomputingcapacityoftheatomsinthestone(estimatedat1042cps)byafactorof10–42.

71.EdgarBuckingham,"JetPropulsionforAirplanes,"NACAreportno.159,inNinthAnnualReportofNACA-1923(Washington,D.C.:NACA,1924),pp.75–90.Seehttp://naca.larc.nasa.gov/reports/1924/naca-report-159/.

72.BelleDume,"MicroscopyMovestothePicoscale,"PhysicsWeb,June10,2004,http://physicsweb.org/artide/news/8/6/6,referringtoStefanHembacher,FranzJ.Giessibl,andIochenMannhart,"ForceMicroscopywithLight-AtomProbes,"Science305.5682(July16,2004):380–83.Thisnew"higherharmonic"forcemicroscope,developedbyUniversityofAugsburgphysicists,usesasinglecarbonatomasaprobeandhasaresolutionthatisatleastthreetimesbetterthanthatoftraditionalscanningtunnelingmicroscopes.Howitworks:asthetungstentipoftheprobeismadetooscillateatsubnanometeramplitudes,theinteractionbetweenthetipatomandthecarbonatomproduceshigherharmoniccomponentsintheunderlyingsinusoidal-wavepattern.Thescientistsmeasuredthesesignalstoobtainanultrahigh-resolutionimageofthetipatomthatshowedfeaturesjust77picometers(thousandthsofananometer)across.

73.HenryFountain,"NewDetectorMayTestHeisenberg'sUncertaintyPrinciple,"NewYorkTimes,July22,2003.

74.MitchJacoby,"ElectronMovesinAttoseconds,"ChemicalandEngineeringNews82.25(June21,2004):5,referringtoPeterAbbamonteetal.,"ImagingDensityDisturbancesinWaterwitha41.3-AttosecondTimeResolution,"PhysicalReviewLetters92.23(June11,2004):237–401.

75.S.K.Lamoreauxand1.R.Torgerson,"NeutronModerationintheOkloNaturalReactorandtheTimeVariationofAlpha,"PhysicalReviewD69(2004):121701–6,http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=PRVDAQ000069000012121701000001&idtype=cvips&gifs=yes;EugenieS.Reich,"SpeedofLightMayHaveChangedRecently,"NewScientist,June30,2004,

http://www.newscientist.com!news/news.jsp?id=ns99996092.76.CharlesChoi,"ComputerProgramtoSendDataBackinTime,"UPI,October1,2002,

http://www.upi.com/view.efm?StoryID=20021001-125805-3380r;ToddBrun,"ComputerswithClosedTimelikeCurvesCanSolveHardProblems,"FoundationofPhysicsLetters16(2003):245–53.Electronicedition,September11,2002,http://arxiv.org/PS_cache/gr-qc/pdf/0209/0209061.pdf.

ChapterFour:AchievingtheSoftwareofHumanIntelligence:

HowtoReverseEngineertheHumanBrain1.LloydWatts,"VisualizingComplexityintheBrain,"inD.FogelandC.Robinson,eds.,

ComputationalIntelligence:TheExpertsSpeak(Piscataway,N.J.:IEEEPress/Wiley,2003),http://www.lloydwatts.com/wcci.pdf.

2.J.G.Taylor,B.Horwitz,andK.J.Friston,"TheGlobalBrain:ImagingandModeling,"NeuralNetworks13,specialissue(2000):827.

3.NeilA.Busis,"NeurosciencesontheInternet,"http://www.neuroguide.com;"NeuroscientistsHaveBetterToolsontheBrain,"BioITBulletin,http://www.bio-it.world.com/news/041503_report2345.html;"BrainProjectstoReapDividendsforNeurotechFirms,"NeurotechReports,http://www.neurotechreports.com/pages/brainprojects.html.

4.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section4.8.6,"NoninvasiveNeuroelectricMonitoring"(Georgetown,Tex.:LandesBioscience,1999),pp.115–16,http://www.nanomedicine.com/NMI/4.8.6.htm.

5.Chapter3analyzedthisissue;seethesection"TheComputationalCapacityoftheHumanBrain."

6.Speech-recognitionresearchanddevelopment,KurzweilAppliedIntelligence,whichIfoundedin1982,nowpartofScanSoft(formerlyKurzweilComputerProducts).

7.LloydWatts,U.S.PatentApplication,U.S.PatentandTrademarkOffice,20030095667,May22,2003,"ComputationofMulti-sensorTimeDelays."Abstract:"Determiningatimedelaybetweenafirstsignalreceivedatafirstsensorandasecondsignalreceivedatasecondsensorisdescribed.Thefirstsignalisanalyzedtoderiveapluralityoffirstsignalchannelsatdifferentfrequenciesandthesecondsignalisanalyzedtoderiveapluralityofsecondsignalchannelsatdifferentfrequencies.Afirstfeatureisdetectedthatoccursatafirsttimeinoneofthefirstsignalchannels.Asecondfeatureisdetectedthatoccursatasecondtimeinoneofthesecondsignalchannels.Thefirstfeatureismatchedwiththesecondfeatureandthefirsttimeiscomparedtothesecondtimetodeterminethetimedelay."SeealsoNabilH.Farhat,U.S.PatentApplication20040073415,U.S.PatentandTrademarkOffice,April15,2004,"DynamicalBrainModelforUseinDataProcessingApplications."

8.Iestimatethecompressedgenomeataboutthirtytoonehundredmillionbytes(seenote57forchapter2);thisissmallerthantheobjectcodeforMicrosoftWordandmuchsmallerthanthesourcecode.SeeWord2003systemrequirements,October20,2003,http://www.microsoft.com/office/word/prodinfo/sysreq.mspx.

9.Wikipedia,http://en.wikipedia.org/wiki/Epigenetics.10.Seenote57inchapter2forananalysisoftheinformationcontentinthegenome,whichI

estimatetobe30to100millionbytes,thereforelessthan109bits.Seethesection"HumanMemoryCapacity"inchapter3(p.126)formyanalysisoftheinformationinahumanbrain,estimatedat1018bits.

11.MarieGustafssonandChristianBalkenius,"UsingSemanticWebTechniquesforValidationofCognitiveModelsagainstNeuroscientificData,"AILS04Workshop,SAIS/SSLSWorkshop(SwedishArtificialIntelligenceSociety;SwedishSocietyforLearningSystems),April15–16,2004,Lund,Sweden,www.lucs.lu.se/People/Christian.Balkenius/PDF/Gustafsson.Balkenius.2004.pdf.

12.Seediscussioninchapter3.Inoneusefulreference,whenmodelingneuronbyneuron,TomasoPoggioandChristofKochdescribetheneuronassimilartoachipwiththousandsoflogicalgates.SeeT.PoggioandC.Koch,"SynapsesThatComputeMotion,"ScientificAmerican256(1987):46–52.AlsoC.KochandT.Poggio,"BiophysicsofComputationalSystems:Neurons,Synapses,andMembranes,"inSynapticFunction,G.M.Edelman,W.E.Gall,andW.M.Cowan,eds.(NewYork:JohnWileyandSons,1987),pp.637–97.

13.OnMead,seehttp://www.technology.gov/Medal/2002/bios/Carver_A._Mead.pdf.CarverMead,AnalogVLSIandNeuralSystems(Reading,Mass.:Addison-Wesley,1986).

14.Seenote172inchapter5foranalgorithmicdescriptionofaself-organizingneuralnetandnote175inchapter5foradescriptionofaself-organizinggeneticalgorithm.

15.SeeGaryDudleyetal.,"AutonomicSelf-HealingSystemsinaCross-ProductITEnvironment,"ProceedingsoftheIEEEInternationalConferenceonAutonomicComputing,NewYorkCity,May17–19,2004,http://csdl.computer.org/comp/proceedings/icac/2004/2114/00121140312.pdf;"AboutIBMAutonomicComputing,"http://www-3.ibm.com/autonomic/about.shtml;andRicTelford,"TheAutonomicComputingArchitecture,"April14,2004,http://www.dcs.st-andrews.ac.uk/undergrad/current/dates/disclec/2003–2/RicTelfordDistinguished2.pdf.

16.ChristineA.SkardaandWalterJ.Freeman,"ChaosandtheNewScienceoftheBrain,"ConceptsinNeuroscience1.2(1990):275–85.

17.C.GeoffreyWoods,"CrossingtheMidline,"Science304.5676(June4,2004):1455–56;StephenMatthews,"EarlyProgrammingoftheHypothalamo-Pituitary-AdrenalAxis,"TrendsinEndocrinologyandMetabolism13.9(November1,2002):373–80;JustinCrowleyandLawrenceKatz,"EarlyDevelopmentofOcularDominanceColumns,"Science290.5495(November17.2000):1321–24;AnnaPennetal.,"CompetitionintheRetinogeniculatePatterningDrivenbySpontaneousActivity,"Science279.5359(March27,1998):2108–12;M.V.Johnstonetal.,"SculptingtheDevelopingBrain,"AdvancesinPediatrics48(2001):1–38;P.LaCerraandR.Bingham,"TheAdaptiveNatureoftheHumanNeurocognitiveArchitecture:AnAlternativeModel,"ProceedingsoftheNationalAcademyofSciences95(September15,1998):11290–94.

18.Neuralnetsaresimplifiedmodelsofneuronsthatcanself-organizeandsolveproblems.Seenote172inchapter5foranalgorithmicdescriptionofneuralnets.Geneticalgorithmsaremodelsofevolutionusingsexualreproductionwithcontrolledmutationrates.Seenote175inchapter5foradetaileddescriptionofgeneticalgorithms.Markovmodelsareproductsofamathematicaltechniquethataresimilarinsomerespectstoneuralnets.

19.Aristotle,TheWorksofAristotle,trans.W.D.Ross(Oxford:ClarendonPress,1908–1952(see,inparticular,Physics);seealsohttp://www.encyclopedia.com/html/section/aristotl_philosophy.asp.

20.E.D.Adrian,TheBasisofSensation:TheActionofSenseOrgans(London:Christophers,1928).

21.A.L.HodgkinandA.F.Huxley,"ActionPotentialsRecordedfromInsideaNerveFibre,"Nature144(1939):710–12.

22.A.L.HodgkinandA.F.Huxley,"AQuantitativeDescriptionofMembraneCurrentandItsApplicationtoConductionandExcitationinNerve,"JournalofPhysiology117(1952):500–544.

23.W.S.McCullochandW.Pitts,"ALogicalCalculusoftheIdeasImmanentinNervousActivity,"BulletinofMathematicalBiophysics5(1943):115-33.Thisseminalpaperisadifficultonetounderstand.Foraclearintroductionandexplanation,see"AComputerModeloftheNeuron,"theMindProject,IllinoisStateUniversity,http://www.mind.ilstu.edu/curriculum/perception/mpneuron1.html.

24.Seenote172inchapter5foranalgorithmicdescriptionofneuralnets.25.E.SalinasandP.Thier,"GainModulation:AMajorComputationalPrincipleoftheCentral

NervousSystem,"Neuron27(2000):15–21.26.K.M.O'CravenandR.L.Savoy,"VoluntaryAttentionCanModulatefMRIActivityin

HumanMT/MST,"InvestigationalOphthalmologicalVisionScience36(1995):S856(supp.).27.MarvinMinskyandSeymourPapert,Perceptrons(Cambridge,Mass.:MITPress,1969).28.FrankRosenblatt,CornellAeronauticalLaboratory,"ThePerceptron:AProbabilisticModel

forInformationStorageandOrganizationintheBrain,"PsychologicalReview65.6(1958):386–408;seeWikipedia,http://en.wikipedia.org/wiki/Perceptron.

29.O.Sporns,G.Tononi,andG.M.Edelman,"ConnectivityandComplexity:TheRelationshipBetweenNeuroanatomyandBrainDynamics,"NeuralNetworks13.8–9(2000):909–22.

30.R.H.Hahnloseretal.,"DigitalSelectionandAnalogueAmplificationCoexistinaCortex-InspiredSiliconCircuit,"Nature405.6789(June22,2000):947–51;"MITandBellLabsResearchersCreateElectronicCircuitThatMimicstheBrain'sCircuitry,"MITNews,June21,2000,http://web.mit.edu/newsoffice/nr/2000/machinebrain.html.

31.ManuelTrajtenberg,EconomicAnalysisofProductInnovation:TheCaseofCTScanners(Cambridge,Mass.:HarvardUniversityPress,1990);MichaelH.Priebe,Ph.D.,president,CEO,NEUROMEDGmbH;P-M.L.Robitaille,A.M.Abduljalil,andA.Kangarlu,"UltraHighResolutionImagingoftheHumanHeadat8Tesla:2Kx2KforY2K,"JournalofComputerAssistedTomography24.1(January–February2000):2–8.

32.Seong-GiKim,"ProgressinUnderstandingFunctionalImagingSignals,"ProceedingsoftheNationalAcademyofSciences100.7(April1,2003):3550–52,http://www.pnas.org/cgi/content/full/100/7/3550.SeealsoSeong-GiKimetal.,"LocalizedCerebralBloodFlowResponseatSubmillimeterColumnarResolution,"ProceedingsoftheNationalAcademyofSciences98.19(September11,2001):10904–9,httpi//www.pnas.org/cgi/content/abstract/98/19/10904.

33.K.K.Kwongetal.,"DynamicMagneticResonanceImagingofHumanBrainActivityDuringPrimarySensoryStimulation,"ProceedingsoftheNationalAcademyofSciences89.12(June15,1992):5675–79.

34.C.S.RoyandC.S.Sherrington,"OntheRegulationoftheBloodSupplyoftheBrain,"JournalofPhysiology11(1890):85–105.

35.M.I.Posneretal.,"LocalizationofCognitiveOperationsintheHumanBrain,"Science240.4859(June17,1988):1627–31.

36.F.M.Mottaghyetal.,"FacilitationofPictureNamingafterRepetitiveTranscranialMagneticStimulation,"Neurology53.8(November10,1999):1806–12.

37.DaithíÓhAnluain,"TMS:TwilightZoneScience?"WiredNews,April18,2002,http://wired.com/news/medtech/0,1286,51699.00.html.

38.LawrenceOsborne,"SavantforaDay,"NewYorkTimesMagazine,June22,2003,availableathttp://www.wireheading.com/brainstim/savant.html.

39.BruceH.McCormick,"BrainTissueScannerEnablesBrainMicrostructureSurveys,"Neurocomputing44–46(2002):1113-18;BruceH.McCormick,"DesignofaBrainTissueScanner,"Neurocomputing26-27(1999):1025-32;BruceH.McCormick,"DevelopmentoftheBrainTissueScanner,"BrainNetworksLaboratoryTechnicalReport,TexasA&MUniversityDepartmentofComputerScience,CollegeStation,Tex.,March18,2002,http://research.cs.tamu.edu/bnl/pubs/McC02.pdf.

40.LeifFinkeletal.,"Meso-scaleOpticalBrainImagingofPerceptualLearning,"UniversityofPennsylvaniagrant2000–01737(2000).

41.E.CallawayandR.Yuste,"StimulatingNeuronswithLight,"CurrentOpinionsinNeurobiology12.5(October2002):587–92.

42.B.L.SabatiniandK.Svoboda,"AnalysisofCalciumChannelsinSingleSpinesUsingOpticalFluctuationAnalysis,"Nature408.6812(November30,2000):589–93.

43.JohnWhitfield,"LasersOperateInsideSingleCells,"News@nature.com,October6,2003,http://www.nature.com/nsu/030929/030929-12.html(subscriptionrequired).Mazur'slab:http://mazur-www.harvard.edu/research/.JasonM.SamondsandA.B.Bonds,"FromAnotherAngle:DifferencesinCorticalCodingBetweenFineandCoarseDiscriminationofOrientation,"JournalofNeurophysiology91(2004):1193–1202.

44.RobertA.FreitasJr.,Nanomedicine,vol.2A,Biocompatibility,section15.6.2,"BloodstreamIntrusiveness"(Georgetown,Tex.:LandesBioscience,2003),pp.157–59,http://www.nanomedicine.com/NMIIA/15.6.2.htm.

45.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section7.3,"CommunicationNetworks"(Georgetown,Tex.:LandesBioscience,1999),pp.186–88,http://www.nanomedicine.com/NMI/7.3.htm.

46.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section9.4.4.3,"IntercellularPassage"(Georgetown,Tex.:LandesBioscience,1999),pp.320–21,http://www.nanomedicine.com/NMI/9.4.4.3.htm#p2.

47.KeithL.Black,M.D.,andNagendraS.Ningaraj,"ModulationofBrainTumorCapillariesforEnhancedDrugDeliverySelectivelytoBrainTumor,"CancerControl11.3(May/June2004):165–73,http://www.moffitt.usf.edu/pubs/ccj/v11n3/pdf/165.pdf.

48.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section4.1,"NanosensorTechnology"(Georgetown,Tex.:LandesBioscience,1999),p.93,http://www.nanomedicine.com/NMI/4.1.htm.

49.ConferenceonAdvancedNanotechnology(http://www.foresight.orglConferences/AdvNan02004/index.html),NanoBioTechCongressandExhibition(http://www.nanobiotec.de/),NanoBusinessTrendsinNanotechnology(http://www.nanoevent.com/),andNSTINanotechnologyConferenceandTradeShow(http://www.nsti.org/events.html).

50.PeterD.Kramer,ListeningtoProzac(NewYork:Viking,1993).51.LeDoux'sresearchisonthebrainregionsthatdealwiththreateningstimuli,ofwhichthe

centralplayeristheamygdala,analmond-shapedregionofneuronslocatedatthebaseofthebrain.Theamygdalastoresmemoriesofthreateningstimuliandcontrolsresponseshavingtodowithfear.MITbrainresearcherTomasoPoggiopointsoutthat"synapticplasticityisonehardwaresubstratumforlearningbutitmaybeimportanttoemphasizethatlearningismuchmorethanmemory."SeeT.PoggioandE.Bizzi,"GeneralizationinVisionandMotorControl,"Nature431(2004):768-74.SeealsoE.Benson,"TheSynapticSelf'APAOnline,November2002,http://www.apa.org/monitor/nov02/synaptic.html.

52.AnthonyJ.Bell,"LevelsandLoops:TheFutureofArtificialIntelligenceandNeuroscience,"PhilosophicalTransactionsoftheRoyalSocietyofLondonB354.1352(December29,1999):2013–20,http://www.cnl.salk.edu/~tony/ptrsl.pdf.

53.PeterDayanandLarryAbbott,TheoreticalNeuroscience:ComputationalandMathematicalModelingofNeuralSystems(Cambridge,Mass.:MITPress,2001).

54.D.O.Hebb,TheOrganizationofBehavior:ANeuropsychologicalTheory(NewYork:Wiley,1949).

55.MichaelDomjanandBarbaraBurkhard,ThePrinciplesofLearningandBehavior,3ded.(PacificGrove,Calif.:Brooks/Cole,1993).

56.J.QuintanaandJ.M.Fuster,"FromPerceptiontoAction:TemporalIntegrativeFunctionsofPrefrontalandParietalNeurons,"CerebralCortex9.3(April-May1999):213–21;W.F.Asaad,G.Rainer,andE.K.Miller,"NeuralActivityinthePrimatePrefrontalCortexDuringAssociativeLearning,"Neuron21.6(December1998):1399–1407.

57.G.G.Turrigianoetal.,"Activity-DependentScalingofQuantalAmplitudeinNeocorticalNeurons,"Nature391.6670(February26,1998):892–96;R.J.O'Brienetal.,"Activity-DependentModulationofSynapticAMPAReceptorAccumulation,"Neuron21.5(November1998):1067–78.

58.From"ANewWindowtoViewHowExperiencesRewiretheBrain,"HowardHughesMedicalInstitute(December19,2002),http://www.hhmi.org/news/svoboda2.html.SeealsoJ.T.Trachtenbergetal.,"Long-TerminVivoImagingofExperience-DependentSynapticPlasticityinAdultCortex,"Nature420.6917(December2002):788–94,http://cpmcnet.columbia.edu/dept/physio/physi02/Trachtenberg_NATURE.pdf;andKarenZita

andKarelSvoboda,"Activity-DependentSynaptogenesisintheAdultMammalianCortex,"Neuron35.6(September2002):1015–17,http://svobodalab.cshl.edu/reprints/2414zito02neur.pdf.

59.Seehttp://whyfiles.org/184make_memory/4.html.Formoreinformationonneuronalspinesandmemory,seeJ.Grutzendleretal.,"Long-TermDendriticSpineStabilityintheAdultCortex,"Nature420.6917(Dec.19–26,2002):812–16.

60.S.R.YoungandE.W.Rubel,"EmbryogenesisofArborizationPatternandTypographyofIndividualAxonsinN.LaminarisoftheChickenBrainStem,"JournalofComparativeNeurology254.4(December22,1986):425–59.

61.ScottMakeig,"SwartzCenterforComputationalNeuroscienceVisionOverview,"http://www.sccn.ucsd.edu/VisionOverview.html.

62.D.H.HubelandT.N.Wiesel,"BinocularInteractioninStriateCortexofKittensRearedwithArtificialSquint,"JournalofNeurophysiology28.6(November1965):1041–59.

63.JeffreyM.SchwartzandSharonBegley,TheMindandtheBrain:NeuroplasticityandthePowerofMentalForce(NewYork:ReganBooks,2002).SeealsoC.Xerri,M.Merzenichetal.,"ThePlasticityofPrimarySomatosensoryCortexParallelingSensorimotorSkillRecoveryfromStrokeinAdultMonkeys,"TheJournalofNeurophysiology,79.4(April1980):2119–48.SeealsoS.Begley,"SurvivaloftheBusiest,"WallStreetJournal,October11,2002,http://webreprints.djreprints.com/606120211414.html.

64.PaulaTallaletal.,"LanguageComprehensioninLanguage-LearningImpairedChildrenImprovedwithAcousticallyModifiedSpeech,"Science271(January5,1996):81–84.PaulaTallalisBoardofGovernorsProfessorofNeuroscienceandcodirectoroftheCMBN(CenterforMolecularandBehavioralNeuroscience)atRutgersUniversity,andcofounderanddirectorofSCIL(ScientificLearningCorporation);seehttp://www.cmbn.rutgers.edu/faculty/tallal.html.SeealsoPaulaTallal,"LanguageLearningImpairment:IntegratingResearchandRemediation,"NewHorizonsforLearning4.4(August–September1998),http://www.newhorizons.org/neuro/tallal.htm,A.Pascual-Leone,"TheBrainThatPlaysMusicandIsChangedbyIt,"AnnalsoftheNewYorkAcademyofSciences930(June2001):315–29.Seealsonote63above.

65.F.A.Wilson,S.P.Scalaidhe,andP.S.Goldman-Rakic,"DissociationofObjectandSpatialProcessingDomainsinPrimatePrefrontalCortex."Science260.5116(June25,1993):1955–58.

66.C.Buechel,J.T.Coull,andK.J.Friston,"ThePredictiveValueofChangesinEffectiveConnectivityforHumanLearning,"Science283.5407(March5,1999):1538–41.

67.Theyproduceddramaticimagesofbraincellsformingtemporaryandpermanentconnectionsinresponsetovariousstimuli,illustratingstructuralchangesbetweenneuronsthat,manyscientistshavelongbelieved,takeplacewhenwestorememories."PicturesRevealHowNerveCellsFormConnectionstoStoreShort-andLong-TermMemoriesinBrain,"UniversityofCalifornia,SanDiego,November29,2001,http://ucsdnews.ucsd.edu/newsrel/science/mcceli.htm;M.A.Colicosetal.,"RemodelingofSynapticActionInducedbyPhotoconductiveStimulation,"Cell107.5(November30,2001):605–16.Videolink:http://www.qflux.net/NeuroStim01.rm,NeuralSiliconInterface—QuantumFlux.

68.S.LowelandW.Singer,"SelectionofIntrinsicHorizontalConnectionsintheVisualCortexbyCorrelatedNeuronalActivity,"Science255.5041(January10,1992):209–12.

69.K.Sietal.,"ANeuronalIsoformofCPEBRegulatesLocalProteinSynthesisandStabilizesSynapse-SpecificLong-TermFacilitationinAplysia,"Cell115.7(December26,2003):893–904;K.Si,S.Lindquist,andE.R.Kandel,"ANeuronalIsoformoftheAplysiaCPEBHasPrion-LikeProperties,"Cell115.7(December26,2003):879–91.TheseresearchershavefoundthatCPEBmayhelpformandpreservelong-termmemoriesbyundergoingshapechangesinsynapsessimilartodeformationsofprions(proteinfragmentsimplicatedinmad-cowdiseaseandotherneurologicillnesses).Thestudysuggeststhatthisproteindoesitsgoodworkwhileinaprionstate,contradictingawidelyheldbeliefthataproteinthathasprionactivityistoxicorat

leastdoesn'tfunctionproperly.Thisprionmechanismmayalsohaverolesinareassuchascancermaintenanceandorgandevelopment,suspectsEricR.Kandel,UniversityProfessorofphysiologyandcellbiophysics,psychiatry,biochemistry,andmolecularbiophysicsatColumbiaUniversityandwinnerofa2000NobelPrizeforMedicine.SeeWhiteheadInstitutepressrelease,http://www.wi.mit.edu/nap/features/nap_feature_memory.html.

70.M.C.Andersonetal.,"NeuralSystemsUnderlyingtheSuppressionofUnwantedMemories,"Science303.5655(January9,2004):232-35.Thefindingscouldencouragethedevelopmentofnewwaysforpeopletoovercometraumatizingmemories.KeayDavidson,"StudySuggestsBrainIsBuilttoForget:MRIsinStanfordExperimentsIndicateActiveSuppressionofUnneededMemories,"SanFranciscoChronicle,January9,2004,http://www.sfgate.com/cgi-bin/article.cgi?file=/c/a/2004/01/09/FORGET.TMP&type=science.

71.DieterC.Lieetal.,"NeurogenesisintheAdultBrain:NewStrategiesforCNSDiseases,"AnnualReviewofPharmacologyandToxicology44(2004):399–421.

72.H.vanPraag,G.Kempermann,andF.H.Gage,"RunningIncreasesCellProliferationandNeurogenesisintheAdultMouseDentateGyrus,"NatureNeuroscience2.3(March1999):266–70.

73.MinskyandPapert,Perceptrons.74.RayKurzweil,TheAgeofSpiritualMachines(NewYork:Viking,1999),p.79.75.Basisfunctionsarenonlinearfunctionsthatcanbecombinedlinearly(byaddingtogether

multipleweighted-basisfunctions)toapproximateanynonlinearfunction.PougetandSnyder,"ComputationalApproachestoSensorimotorTransformations,"NatureNeuroscience3.11Supplement(November2000):1192–98.

76.T.Poggio,"ATheoryofHowtheBrainMightWork,"inProceedingsofColdSpringHarborSymposiaonQuantitativeBiology4(ColdSpringHarbor,N.Y.:ColdSpringHarborLaboratoryPress,1990),899–910.AlsoseeT.PoggioandE.Bizzi,"GeneralizationinVisionandMotorControl,"Nature431(2004):768–74.

77.R.LlinasandJ.P.Welsh,"OntheCerebellumandMotorLearning,"CurrentOpinioninNeurobiology3.6(December1993):958–65;E.CourchesneandG.Allen,"PredictionandPreparation,FundamentalFunctionsoftheCerebellum,"LearningandMemory4.1(May–June1997):1-35;J.M.Bower,"ControlofSensoryDataAcquisition,"InternationalReviewofNeurobiology41(1997):489–513.

78.J.VoogdandM.Glickstein,"TheAnatomyoftheCerebellum,"TrendsinNeuroscience21.9(September1998):370–75;JohnC.Eccles,MasaoIto,andJánosSzentágothai,TheCerebellumasaNeuronalMachine(NewYork:Springer-Verlag,1967);MasaoIto,TheCerebellumandNeuralControl(NewYork:Raven,1984).

79.N.Bernstein,TheCoordinationandRegulationofMovements(NewYork:PergamonPress,1967).

80.U.S.OfficeofNavalResearchpressrelease,"Boneless,Brainy,andAncient,"September26,2001,http://www.eurekalert.org/pub_releases/2001-11/oonr-bba112601.php;theoctopusarm"couldverywellbethebasisofnext-generationroboticarmsforundersea,space,aswellasterrestrialapplications."

81.S.GrossbergandR.W.Paine,"ANeuralModelofCortico-CerebellarInteractionsDuringAttentiveImitationandPredictiveLearningofSequentialHandwritingMovements,"NeuralNetworks13.8-9(October-November2000):999-1046.

82.VoogdandGlickstein,"AnatomyoftheCerebellum";Eccles,Ito,andSzentágothai,CerebellumasaNeuronalMachine;Ito,CerebellumandNeuralControl;R.Llinas,inHandbookofPhysiology,vol.2,TheNervousSystem,ed.V.B.Brooks(Bethesda,Md.:AmericanPhysiologicalSociety,1981),pp.831–976.

83.J.L.Raymond,S.G.Lisberger,andM.D.Mauk,"TheCerebellum:ANeuronalLearningMachine?"Science272.5265(May24,1996):1126–31;J.J.KimandR.F.Thompson,"CerebellarCircuitsandSynapticMechanismsInvolvedinClassicalEyeblinkConditioning,"TrendsinNeuroscience20.4(April1997):177–81.

84.Thesimulationincluded10,000granulecells,900Golgicells,500mossyfibercells,20Purkinjecells,and6nucleuscells.

85.J.F.Medinaetal.,"TimingMechanismsintheCerebellum:TestingPredictionsofaLarge-ScaleComputerSimulation,"JournalofNeuroscience20.14(July15,2000):5516–25;DeanBuonomanoandMichaelMauk,"NeuralNetworkModeloftheCerebellum:TemporalDiscriminationandtheTimingofMotorReponses,"NeuralComputation6.1(1994):38–55.

86.Medinaetal.,"TimingMechanismsintheCerebellum."87.CarverMead,AnalogVLSIandNeuralSystems(Boston:Addison-WesleyLongman,1989).88.LloydWatts,"VisualizingComplexityintheBrain,"inComputationalIntelligence:The

ExpertsSpeak,D.FogelandC.Robinson,eds.(Hoboken,N.J.:IEEEPress/Wiley,2003),pp.45–56,http://www.lloydwatts.com/wcci.pdf.

89.Ibid.90.Seehttp://www.lloydwatts.com/neuroscience.shtml.NanoComputerDreamTeam,"TheLaw

ofAcceleratingReturns,PartII,"http://nanocomputer.org/index.cfm?content=90&Menu=19.91.Seehttp://info.med.yale.edu/bbs/faculty/she_go.html.92.GordonM.Shepherd,ed.,TheSynapticOrganizationoftheBrain,4thed.(NewYork:

OxfordUniversityPress,1998),p.vi.93.E.Young,"CochlearNucleus,"inibid.,pp.121–58.94.TomYin,"NeuralMechanismsofEncodingBinauralLocalizationCuesintheAuditory

Brainstem,"inD.Oertel,R.Fay,andA.Popper,eds.,IntegrativeFunctionsintheMammalianAuditoryPathway(NewYork:Springer-Verlag,2002),pp.99–159.

95.JohnCasseday,ThanePremouw,andEllenCovey,"TheInferiorColliculus:AHubfortheCentralAuditorySystem,"inOertel,Fay,andPopper,IntegrativeFunctionsintheMammalianAuditoryPathway,pp.238–318.

96.DiagrambyLloydWatts,http://www.lloydwatts.com/neuroscience.shtml,adaptedfromE.Young,"CochlearNucleus"inG.Shepherd,ed.,TheSynapticOrganizationoftheBrain,4thed.(NewYork:OxfordUniversityPress,2003[firstpublished1998]),pp.121–58;D.OertelinD.Oertel,R.Fay,andA.Popper,eds.,IntegrativeFunctionsintheMammalianAuditoryPathway(NewYork:Springer-Verlag,2002),pp.1–5;JohnCasseday,T.Fremouw,andE.Covey,"InferiorColliculus"inibid.;J.LeDoux,TheEmotionalBrain(NewYork:Simon&Schuster,1997);J.RauscheckerandB.Tian,"MechanismsandStreamsforProcessingof'What'and'Where'inAuditoryCortex,"ProceedingsoftheNationalAcademyofSciences97.22:11800–11806.Brainregionsmodeled:

Cochlea:Senseorganofhearing.Thirtythousandfibersconvertmotionofthestapesinto

spectrotemporalrepresentationsofsound.MC:Multipolarcells.Measurespectralenergy.GBC:Globularbushycells.Relayspikesfromtheauditorynervetothelateralsuperior

olivarycomplex(includesLSOandMSO).Encodingoftimingandamplitudeofsignalsforbinauralcomparisonoflevel.

SBC:Sphericalbushycells.Providetemporalsharpeningoftimeofarrival,asapreprocessorforinterauraltime-differencecalculation(differenceintimeofarrivalbetweenthetwoears,usedtotellwhereasoundiscomingfrom).

OC:Octopuscells.Detectionoftransients.DCN:Dorsalcochlearnucleus.Detectionofspectraledgesandcalibratingfornoise

levels.VNTB:Ventralnucleusofthetrapezoidbody.Feedbacksignalstomodulateouterhair-

cellfunctioninthecochlea.VNLL,PON:Ventralnucleusofthelaterallemniscus;peri-olivarynuclei:processing

transientsfromthe0C.MSO:Medialsuperiorolive.Computinginterauraltimedifference.

LSO:Lateralsuperiorolive.Alsoinvolvedincomputinginterauralleveldifference.ICC:Centralnucleusoftheinferiorcolliculus.Thesiteofmajorintegrationofmultiple

representationsofsound.ICx:Exteriornucleusoftheinferiorcolliculus.Furtherrefinementofsoundlocalization.SC:Superiorcolliculus.Locationofauditory/visualmerging.MGB:Medialgeniculatebody.Theauditoryportionofthethalamus.LS:Limbicsystem.Comprisingmanystructuresassociatedwithemotion,memory,

territory,etcetera.AC:Auditorycortex.

97.M.S.Humayunetal.,"HumanNeuralRetinalTransplantation,"InvestigativeOphthalmology

andVisualScience41.10(September2000):3100–3106.98.InformationScienceandTechnologyColloquiumSeries,May23,2001,

http://isandtcolloq.gsfc.nasa.gov/spring2001/speakers/poggio.html.99.Kah-KaySungandTomasoPoggio,"Example-BasedLearningforView-BasedHumanFace

Detection,"IEEETransactionsonPatternAnalysisandMachineIntelligence20.1(1998):39–51,http://portal.acm.org/citation.cfm?id=275345&dl=ACM&coll=GUIDE.

100.MaximilianRiesenhuberandTomasoPoggio,"ANoteonObjectClassRepresentationandCategoricalPerception,"CenterforBiologicalandComputationalLearning,MIT,AIMemo1679(1999),ftp://publications.ai.mit.edu/ai-publications/pdf/AIM-1679.pdf.

101.K.Tanaka,"InferoternporalCortexandObjectVision,"AnnualReviewofNeuroscience19(1996):109-39;AnujMohan,"ObjectDetectioninImagesbyComponents,"CenterforBiologicalandComputationalLearning,MIT,AIMemo1664(1999),http://citeseer.ist.psu.edu/cache/papers/cs/12185/ftp:zSzzSzpublications.ai.mit.eduzSzai-publicationszSz1500–1999zSzAIM-1664.pdf/mohan99object.pdf;AnujMohan,ConstantinePapageorgiou,andTomasoPoggio,"Example-BasedObjectDetectioninImagesbyComponents,"IEEETransactionsonPatternAnalysisandMachineIntelligence23.4(April2001),http://cbcl.mit.edu/projects/cbd/publications/ps/mohan-ieee.pdf;B.Heisele,T.Poggio,andM.Pontil,"FaceDetectioninStillGrayImages,"ArtificialIntelligenceLaboratory,MIT,TechnicalReportAIMemo1687(2000).AlsoseeBerndHeisele,ThomasSerre,andStanleyBilesch,"Component-BasedApproachtoFaceDetection,"ArtificialIntelligenceLaboratoryandtheCenterforBiologicalandComputationalLearning,MIT(2001),http://www.ai.mit.edulresearch/abstracts/abstracts2001/vision-applied-to-people/03heisele2.pdf.

102.D.VanEssenandJ.Gallant,"NeuralMechanismsofFormandMotionProcessinginthePrimateVisualSystem,"Neuron13.1(July1994):1–10.

103.ShimonUllman,High-LevelVision:ObjectRecognitionandVisualCognition(Cambridge,Mass.:MITPress,1996);D.Mumford,"OntheComputationalArchitectureoftheNeocortex.II.TheRoleofCorticocorticalLoops,"BiologicalCybernetics66.3(1992):241–51;R.RaoandD.Ballard,"DynamicModelofVisualRecognitionPredictsNeuralResponsePropertiesintheVisualCortex,"NeuralComputation9.4(May15,1997):721–63.

104.B.RoskaandF.Werblin,"VerticalInteractionsAcrossTenParallel,StackedRepresentationsintheMammalianRetina,"Nature410.6828(March29,2001):583–87;UniversityofCalifornia,Berkeley,newsrelease,"EyeStripsImagesofAllbutBareEssentialsBeforeSendingVisualInformationtoBrain,UCBerkeleyResearchShows,"March28,2001,www.berkeley.edu/news/media/releases/200l/03/28_wers1.html.

105.HansMoravecandScottFriedmanhavefoundedaroboticscompanycalledSeegridbasedonMoravec'sresearch.Seewww.Seegrid.com.

106.M.A.MahowaldandC.Mead,"TheSiliconRetina,"ScientificAmerican264.5(May1991):76–82.

107.Specifically,alow-passfilterisappliedtoonereceptor(suchasaphotoreceptor).Thisismultipliedbythesignaloftheneighboringreceptor.Ifthisisdoneinbothdirectionsandtheresultsofeachoperationsubtractedfromzero,wegetanoutputthatreflectsthedirectionof

movement.108.OnBerger,seehttp://www.usc.edu/dept/engineering/CNE/faculty/Berger.html.109."TheWorld'sFirstBrainProsthesis,"NewScientist177.2386(March15,2003):4,

http://www.newscientist.com/news/news.jsp?id=ns99993488.110.CharlesChoi,"Brain-MimickingCircuitstoRunNavyRobot,"UPI,June7,2004,

http://www.upi.com/view.cfm?StoryID=20040606-103352-6086r.111.GiacomoRizzolattietal.,"FunctionalOrganizationofInferiorArea6intheMacaque

Monkey.II.AreaF5andtheControlofDistalMovements,"ExperimentalBrainResearch71.3(1998):491–507.

112.M.A.Arbib,"TheMirrorSystem,Imitation,andtheEvolutionofLanguage,"inKerstinDautenhahnandChrystopherL.Nehaniv,eds.,ImitationinAnimalsandArtifacts(Cambridge,Mass.:MITPress,2002).

113.MarcD.Hauser,NoamChomsky,andW.TecumsehFitch,"TheFacultyoflanguage:WhatIsIt,WhoHasIt,andHowDidItEvolve?"Science298(November2002):1569–79,www.wjh.harvard.edu/~mnkylab/publications/languagespeech/Hauser,Chomsky,Fitch.pdf.

114.DanielC.Dennett,FreedomEvolves(NewYork:Viking,2003).115.SeeSandraBlakeslee,"Humanity?MaybeIt'sAllintheWiring,"NewYorkTimes,December

11,2003,http://www.nytimes.com/2003112/09/science/09BRAI.html?ex=1386306000&en=294f5e91dd262a1a&ei=5007&partner=USERLAND.

116.AntonioR.Damasio,Descartes'Error:Emotion,ReasonandtheHumanBrain(NewYork:Putnam,1994).

117.M.P.Maheretal.,"MicrostructuresforStudiesofCulturedNeuralNetworks,"MedicalandBiologicalEngineeringandComputing37.1(January1999):110–18;JohnWrightetal.,"TowardsaFunctionalMEMSNeurowellbyPhysiologicalExperimentation,"TechnicalDigest,ASME,1996InternationalMechanicalEngineeringCongressandExposition,Atlanta,November1996,DSC(DynamicSystemsandControlDivision),vol.59,pp.333–38.

118.W.FrenchAnderson,"GeneticsandHumanMalleability,"HastingsCenterReport23.20(January/February1990):1.

119.RayKurzweil,"AWagerontheTuringTest:WhyIThinkIWillWin,"KurzweilAI.net,April9,2002,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0374.html.

120.RobertA.FreitasJr.proposesafuturenanotechnology-basedbrain-uploadingsystemthatwouldeffectivelybeinstantaneous.AccordingtoFreitas(personalcommunication,January2005),"Aninvivofibernetworkasproposedinhttp://www.nanomedicine.com/NMI/7.3.1.htmcanhandle1018bits/secofdatatraffic,capaciousenoughforreal-timebrain-statemonitoring.Thefibernetworkhasa30cm3volumeandgenerates4–6wattswasteheat,bothsmallenoughforsafeinstallationina1400cm325-watthumanbrain.Signalstravelatmostafewmetersatnearlythespeedoflight,sotransittimefromsignaloriginationatneuronsitesinsidethebraintotheexternalcomputersystemmediatingtheuploadare~0.00001msecwhichisconsiderablylessthantheminimum~5msecneurondischargecycletime.Neuron-monitoringchemicalsensorslocatedonaverage~2micronsapartcancapturerelevantchemicaleventsoccurringwithina~5msectimewindow,sincethisistheapproximatediffusiontimefor,say,asmallneuropeptideacrossa2-microndistance(http://www.nanomedicine.com/NMII/Tables/3.4.jpg).Thushumanbrainstatemonitoringcanprobablybeinstantaneous,atleastonthetimescaleofhumanneuralresponse,inthesenseof'nothingofsignificancewasmissed.'"

121.M.C.Diamondetal.,"OntheBrainofaScientist:AlbertEinstein,"ExperimentalNeurology88(1985):198–204.

ChapterFive:GNR:ThreeOverlappingRevolutions1.SamuelButler(1835–1902),"DarwinAmongtheMachines,"ChristChurchPress,June13,

1863(republishedbyFestingJonesin1912inTheNotebooksofSamuelButler).

2.PeterWeibel,"VirtualWorlds:TheEmperor'sNewBodies,"inArsElectronica:FacingtheFuture,ed.TimothyDruckery(Cambridge,Mass.:MITPress,1999),pp.207–23;availableonlineathttp://www.aec.atlen/archiv_files/19902/EI990b_009.pdf.

3.JamesWatsonandFrancisCrick,"MolecularStructureofNucleicAcids:AStructureforDeoxyriboseNucleicAcid,"Nature171.4356(April23,1953):737–38,http://www.nature.comlnature/dna50/watsoncrick.pdf.

4.RobertWaterstonquotedin"ScientistsRevealCompleteSequenceofHumanGenome,"CBCNews,April14,2003,http://www.cbc.ca/story/science/national/2003/04/14/genome030414.html.

5.Seechapter2,note57.6.TheoriginalreportsofCrickandWatson,whichstillmakecompellingreadingtoday,maybe

foundinJamesA.Peters,ed.,ClassicPapersinGenetics(EnglewoodCliffs,N.J.:Prentice-Hall,1959).AnexcitingaccountofthesuccessesandfailuresthatledtothedoublehelixisgiveninJ.D.Watson,TheDoubleHelix:APersonalAccountoftheDiscoveryoftheStructureofDNA(NewYork:Atheneum,1968).Nature.comhasacollectionofCrick'spapersavailableonlineathttp://www.nature.com/nature/focus/crick/index.html.

7.MorislavRadmanandRichardWagner,"TheHighFidelityofDNADuplication,"ScientificAmerican259.2(August1988):40–46.

8.ThestructureandbehaviorofDNAandRNAaredescribedinGaryFelsenfeld,"DNA,"andJamesDarnell,"RNA,"bothinScientificAmerican253.4(October1985),p.58–67and68–78respectively.

9.MarkA.IoblingandChrisTyler-Smith,"TheHumanYChromosome:AnEvolutionaryMarkerComesofAge,"NatureReviewsGenetics4(August2003):598–612;HelenSkaletskyetal.,"TheMale-SpecificRegionoftheHumanYChromosomeIsaMosaicofDiscreteSequenceClasses,"Nature423(June19,2003):825–37.

10.Misformedproteinsareperhapsthemostdangeroustoxinofall.Researchsuggeststhatmisfoldedproteinsmaybeattheheartofnumerousdiseaseprocessesinthebody.SuchdiversediseasesasAlzheimer'sdisease,Parkinson'sdisease,thehumanformofmad-cowdisease,cysticfibrosis,cataracts,anddiabetesareallthoughttoresultfromtheinabilityofthebodytoadequatelyeliminatemisfoldedproteins.Proteinmoleculesperformthelion'sshareofcellularwork.ProteinsaremadewithineachcellaccordingtoDNAblueprints.Theybeginaslongstringsofaminoacids,whichmustthenbefoldedintoprecisethree-dimensionalconfigurationsinordertofunctionasenzymes,transportproteins,etcetera.Heavy-metaltoxinsinterferewithnormalfunctionoftheseenzymes,furtherexacerbatingtheproblem.Therearealsogeneticmutationsthatpredisposeindividualstomisformed-proteinbuildup.Whenprotofibrilsbegintosticktogether,theyformfilaments,fibrils,andultimatelylargerglobularstructurescalledamyloidplaque.Untilrecentlytheseaccumulationsofinsolubleplaquewereregardedasthepathologicagentsforthesediseases,butitisnowknownthattheprotofibrilsthemselvesaretherealproblem.Thespeedwithwhichaprotofibrilisturnedintoinsolubleamyloidplaqueisinverselyrelatedtodiseaseprogression.ThisexplainswhysomeindividualsarefoundtohaveextensiveaccumulationofplaqueintheirbrainsbutnoevidenceofAlzheimer'sdisease,whileothershavelittlevisibleplaqueyetextensivemanifestationsofthedisease.Somepeopleformamyloidplaquequickly,whichprotectsthemfromfurtherprotofibrildamage.Otherindividualsturnprotofibrilsintoamyloidplaquelessrapidly,allowingmoreextensivedamage.Thesepeoplealsohavelittlevisibleamyloidplaque.SeePerHammarström,FrankSchneider,andJeffreyW.Kelly,"Trans-SuppressionofMisfoldinginanAmyloidDisease,"Science293.5539(September28,2001):2459–62.

11.AfascinatingaccountofthenewbiologyisgiveninHoraceF.Judson,TheEighthDayofCreation:TheMakersoftheRevolutioninBiology(Woodbury,N.Y.:CSHLPress,1996).

12.RaymondKurzweilandTerryGrossman,M.D.,FantasticVoyage:LiveLongEnoughtoLiveForever(NewYork:Rodale,2004).Seehttp://www.Fantastic-Voyage.netand

http://www.RayandTerry.com.13.RaymondKurzweil,The10%SolutionforaHealthyLife:HowtoEliminateVirtuallyAllRisk

ofHeartDiseaseandCancer(NewYork:CrownBooks,1993).14.KurzweilandGrossman,FantasticVoyage."Ray&Terry'sLongevityProgram"isarticulated

throughoutthebook.15.Thetestfor"biologicalage,"calledtheH-scantest,includestestsforauditoryreactiontime,

highestaudiblepitch,vibrotactilesensitivity,visual-reactiontime,muscle-movementtime,lung(forcedexpiratory)volume,visual-reactiontimewithdecision,muscle-movementtimewithdecision,memory(lengthofsequence),alternativebutton-tappingtime,andvisualaccommodation.TheauthorhadthistestdoneatFrontierMedicalInstitute(Grossman'shealthandlongevityclinic),http://www.FMIClinic.com.ForinformationontheH-scantest,seeDiagnosticandLabTesting,LongevityInstitute,Dallas,http://www.lidhealth.com/diagnostic.html.

16.KurzweilandGrossman,FantasticVoyage,chapter10:"Ray'sPersonalProgram."17.Ibid.18.AubreyD.N.J.deGrey,"TheForeseeabilityofRealAnti-AgingMedicine:Focusingthe

Debate,"ExperimentalGerontology38.9(September2003):927–34;AubreyD.N.J.deGrey,"AnEngineer'sApproachtotheDevelopmentofRealAnti-AgingMedicine,"ScienceofAging,Knowledge,Environment1(2003):AubreyD.N.J.deGreyetal.,"IsHumanAgingStillMysteriousEnoughtoBeLeftOnlytoScientists?"BioEssays24.7(July2002):667–76.

19.AubreyD.N.J.deGrey,ed.,StrategiesforEngineeredNegligibleSenescence:WhyGenuineControlofAgingMayBeForeseeable,AnnalsoftheNewYorkAcademyofSciences,vol.1019(NewYork:NewYorkAcademyofSciences,June2004).

20.Inadditiontoprovidingthefunctionsofdifferenttypesofcells,twootherreasonsforcellstocontroltheexpressionofgenesareenvironmentalcuesanddevelopmentalprocesses.Evensimpleorganismssuchasbacteriacanturnonandoffthesynthesisofproteinsdependingonenvironmentalcues.E.coli,forexample,canturnoffthesynthesisofproteinsthatallowittocontrolthelevelofnitrogengasfromtheairwhenthereareother,lessenergy-intensivesourcesofnitrogeninitsenvironment.Arecentstudyof1,800strawberrygenesfoundthattheexpressionof200ofthosegenesvariedduringdifferentstagesofdevelopment.E.Marshall,"AnArrayofUses:ExpressionPatternsinStrawberries,Ebola,TB,andMouseCells,"Science286.5439(1999):445.

21.Alongwithaprotein-encodingregion,genesincluderegulatorysequencescalledpromotersandenhancersthatcontrolwhereandwhenthatgeneisexpressed.Promotersofgenesthatencodeproteinsaretypicallylocatedimmediately"upstream"ontheDNA.Anenhanceractivatestheuseofapromoter,therebycontrollingtherateofgeneexpression.Mostgenesrequireenhancerstobeexpressed.Enhancershavebeencalled"themajordeterminantofdifferentialtranscriptioninspace(celltype)andtime";andanygivengenecanhaveseveraldifferentenhancersiteslinkedtoit(S.F.Gilbert,DevelopmentalBiology,6thed.[Sunderland,Mass.:SinauerAssociates,2000];availableonlineatwww.ncbi.nlm.nih.gov/books/bv.fcgi?call=bv.View..ShowSection&rid=.0BpKYEB-SPfx18nm8QOxH).Bybindingtoenhancerorpromoterregions,transcriptionfactorsstartorrepresstheexpressionofagene.Newknowledgeoftranscriptionfactorshastransformedourunderstandingofgeneexpression.PerGilbertinthechapter"TheGeneticCoreofDevelopment:DifferentialGeneExpression":"Thegeneitselfisnolongerseenasanindependententitycontrollingthesynthesisofproteins.Rather,thegenebothdirectsandisdirectedbyproteinsynthesis.NatalieAnger(1992)haswritten,'AseriesofdiscoveriessuggeststhatDNAismorelikeacertaintypeofpolitician,surroundedbyaflockofproteinhandlersandadvisorsthatmustvigorouslymassageit,twistitand,onoccasion,reinventitbeforethegrandblueprintofthebodycanmakeanysenseatall.'"

22.BobHolmes,"GeneTherapyMaySwitchOffHuntington's,"March13,2003,http://www.newscientist.com/news/news.jsp?id=ns99993493."Emergingasapowerfultoolforreversegeneticanalysis,RNAiisrapidlybeingappliedtostudythefunctionofmanygenesassociatedwithhumandisease,inparticularthoseassociatedwithoncogenesisandinfectiousdisease."J.C.Cheng,T.B.Moore,andK.M.Sakamoto,"RNAInterferenceandHumanDisease,"MolecularGeneticsandMetabolism80.1–2(October2003):121–28.RNAiisa"potentandhighlysequence-specificmechanism,"1.Zhang,D.K.Fogg,andD.M.Waisman,"RNAInterference-MediatedSilencingoftheS100A10GeneAttenuatesPlasminGenerationandInvasivenessofColo222ColorectaCancerCells,"JournalofBiologicalChemistry279.3(January16,2004):2053–62.

23.Eachchipcontainssyntheticoligonucleotidesthatreplicatesequencesthatidentifyspecificgenes."Todeterminewhichgeneshavebeenexpressedinasample,researchersisolatemessengerRNAfromtestsamples,convertittocomplementaryDNA(cDNA),tagitwithfluorescentdye,andrunthesampleoverthewafer.EachtaggedcDNAwillsticktoanoligowithamatchingsequence,lightingupaspotonthewaferwherethesequenceisknown.Anautomatedscannerthendetermineswhicholigoshavebound,andhencewhichgeneswereexpressed...."E.Marshall,"Do-It-YourselfGeneWatching,"Science286.5439(October15,1999):444–47.

24.Ibid.25.J.RosamondandA.Allsop,"HarnessingthePoweroftheGenomeintheSearchforNew

Antibiotics,"Science287.5460(March17,2000):1973–76.26.T.R.Golubetal.,"MolecularClassificationofCancer:ClassDiscoveryandClassPrediction

byGeneExpressionMonitoring,"Science286.5439(October15,1999):531–37.27.Ibid.,asreportedinA.Berns,"Cancer:GeneExpressioninDiagnosis,"Nature403(February

3,2000):491–92.Inanotherstudy,1percentofthegenesstudiedshowedreducedexpressioninagedmuscles.Thesegenesproducedproteinsassociatedwithenergyproductionandcellbuilding,soareductionmakessensegiventheweakeningassociatedwithage.Geneswithincreasedexpressionproducedstressproteins,whichareusedtorepairdamagedDNAorproteins.J.Marx,"ChippingAwayattheCausesofAging,"Science287.5462(March31,2000):2390.Asanotherexample,livermetastasesareacommoncauseofcolorectalcancer.Thesemetastasesresponddifferentlytotreatmentdependingontheirgeneticprofile.Expressionprofilingisanexcellentwaytodetermineanappropriatemodeoftreatment.J.C.Sungetal."GeneticHeterogeneityofColorectalCancerLiverMetastases,"JournalofSurgicalResearch114.2(October2003):251.Asafinalexample,researchershavehaddifficultyanalyzingtheReed-SternbergcellofHodgkin'sdiseasebecauseofitsextremerarityindiseasedtissue.Expressionprofilingisnowprovidingaclueregardingtheheritageofthiscell.J.Cossmanetal.,"Reed-SternbergCellGenomeExpressionSupportsaB-CellLineage,"Blood94.2(July15,1999):411–16.

28.T.Uelandetal.,"GrowthHormoneSubstitutionIncreasesGeneExpressionofMembersoftheIGFFamilyinCorticalBonefromWomenwithAdultOnsetGrowthHormoneDeficiency—RelationshipwithBoneTurn-Over,"Bone33.4(October2003):638–45.

29.R.Lovett,"ToxicologistsBraceforGenomicsRevolution,"Science289.5479(July28,2000):536–37.

30.Genetransfertosomaticcellsaffectsasubsetofcellsinthebodyforaperiodoftime.Itistheoreticallypossiblealsotoaltergeneticinformationineggandsperm(germ-line)cells,forthepurposeofpassingonthosechangestothenextgenerations.Suchtherapyposesmanyethicalconcernsandhasnotyetbeenattempted."GeneTherapy,"Wikipedia,http://en.wikipedia.org/wiki/Gene_therapy.

31.Genesencodeproteins,whichperformvitalfunctionsinthehumanbody.Abnormalormutatedgenesencodeproteinsthatareunabletoperformthosefunctions,resultingingeneticdisordersanddiseases.Thegoalofgenetherapyistoreplacethedefectivegenessothatnormalproteinsareproduced.Thiscanbedoneinanumberofways,butthemosttypicalwayistoinsertatherapeuticreplacementgeneintothepatient'stargetcellsusingacarriermoleculecalledavector."Currently,themostcommonvectorisavirusthathasbeengeneticallyalteredtocarrynormalhumanDNA.Viruseshaveevolvedawayofencapsulatinganddeliveringtheirgenestohumancellsinapathogenicmanner.Scientistshavetriedtotakeadvantageofthiscapabilityandmanipulatethevirusgenometoremovethedisease-causinggenesandinserttherapeuticgenes"(HumanGenomeProject,"GeneTherapy,"http://www.ornl.gov/TechResources/Human_Genome/medicine/genetherapy.html).SeetheHumanGenomeProjectsiteformoreinformationaboutgenetherapyandlinks.Genetherapyisanimportantenoughareaofresearchthattherearecurrentlysixscientificpeer-reviewedgene-therapyjournalsandfourprofessionalassociationsdedicatedtothistopic.

32.K.R.Smith,"GeneTransferinHigherAnimals:TheoreticalConsiderationsandKeyConcepts,"JournalofBiotechnology99.1(October9,2002):1–22.

33.AnilAnanthaswamy,"UndercoverGenesSlipintotheBrain,"March20,2003,http://www.newscientist.com/news/news.jsp?id=ns99993520.

34.A.E.Treziseetal.,"InVivoGeneExpression:DNAElectrotransfer,"CurrentOpinioninMolecularTherapeutics5.4(August2003):397-404.

35.SylviaWestphal,"DNANanoballsBoostGeneTherapy,"May12,2002,http://www.newscientist.com/news/news.jsp?id=ns99992257.

36.L.Wu,M.Johnson,andM.Sato,"TranscriptionallyTargetedGeneTherapytoDetectandTreatCancer,"TrendsinMolecularMedicine9.10(October2003):421–29.

37.S.Westphal,"VirusSynthesizedinaFortnight,"November14,2003,http://www.newscientist.com/news/news.jsp?id=ns99994383.

38.G.Chiesa,"RecombinantApolipoproteinA-I(Milano)InfusionintoRabbitCarotidArtery

RapidlyRemovesLipidfromFattyStreaks,"CirculationResearch90.9(May17,2002):974–80;P.K.Shahetal.,"High-DoseRecombinantApolipoproteinA-I(Milano)MobilizesTissueCholesterolandRapidlyReducesPlaqueLipidandMacrophageContentinApolipoproteine-DeficientMice,"Circulation103.25(June26,2001):3047–50.

39.S.E.Nissenetal.,"EffectofRecombinantApoA-IMilanoonCoronaryAtherosclerosisinPatientswithAcuteCoronarySyndromes:ARandomizedControlledTrial,"JAMA290.17(November5,2003):2292–2300.

40.Arecentphase2studyreported"markedlyincreasedHDLcholesterollevelsandalsodecreasedLDLcholesterollevels,"M.E.Brousseauetal.,"EffectsofanInhibitorofCholesterylEsterTransferProteinonHDLCholesterol,"NewEnglandJournalofMedicine350.15(April8,2004):1505–15,http://content.nejm.org/cgi/content/abstract/350/15/1505.Globalphase3trialsbeganinlate2003.InformationonTorcetrapibisavailableonthePfizersite:http://www.pfizer.com/are/investors_reports/annual_2003/review/p2003ar14_15.htm.

41.O.J.Finn,"CancerVaccines:BetweentheIdeaandtheReality,"NatureReviews:Immunology3.8(August2003):630–41;R.C.KennedyandM.H.Shearer,"ARoleforAntibodiesinTumorImmunity,"InternationalReviewsofImmunology22.2(March–April2003):141–72.

42.T.F.GretenandE.M.Jaffee,"CancerVaccines,"JournalofClinicalOncology17.3(March1999):1047–60.

43."Cancer'Vaccine'ResultsEncouraging,"BBCNews,January8,2001,http://news.bbc.co.uk/2/hi/health/1102618.stm,reportingonresearchbyE.M.Jaffeeetal.,"NovelAllogeneicGranulocyte-MacrophageColony-StimulatingFactor-SecretingTumorVaccineforPancreaticCancer:APhaseITrialofSafetyandImmuneActivation,"JournalofClinicalOncology19.1(January1,2001):145–56.

44.JohnTravis,"FusedCellsHoldPromiseofCancerVaccines,"March4,2000,http://www.sciencenews.org/articles/20000304/fob3.asp,referringtoD.W.Kufe,"Smallpox,PolioandNowaCancerVaccine?"NatureMedicine6(March2000):252–53.

45.J.D.Lewis,B.D.Reilly,andR.K.Bright,"Tumor-AssociatedAntigens:FromDiscoverytoImmunity,"InternationalReviewsofImmunology22.2(March–April2003):81–112.

46.T.Boehmetal.,"AntiangiogenicTherapyofExperimentalCancerDoesNotInduceAcquiredDrugResistance,"Nature390.6658(November27,1997):404–7.

47.AngiogenesisFoundation,"UnderstandingAngiogenesis,"http://www.angio.org/understanding/content_understanding.html;L.K.LassiterandM.A.Carducci,"EndothelinReceptorAntagonistsintheTreatmentofProstateCancer,"SeminarsinOncology30.5(October2003):678–88.Foranexplanationoftheprocess,seetheNationalCancerInstituteWebsite,"UnderstandingAngiogenesis,"http://press2.nci.nih.gov/sciencebehind/angiogenesis/angio02.htm.

48.I.B.Roninson,"TumorCellSenescenceinCancerTreatment,"CancerResearch63.11(June1,2003):2705–15;B.R.Daviesetal.,"ImmortalizationofHumanOvarianSurfaceEpitheliumwithTelomeraseandTemperature-SensitiveSV40LargeTAntigen,"ExperimentalCellResearch288.2(August15,2003):390–402.

49.SeealsoR.C.WoodruffandJ.N.ThompsonJr.,"TheRoleofSomaticandGermlineMutationsinAgingandaMutationInteractionModelofAging,"JournalofAnti-AgingMedicine6.1(Spring2003):29–39.Seealsonotes18and19.

50.AubreyD.N.J.deGrey,"TheReductiveHotspotHypothesisofMammalianAging:MembraneMetabolismMagnifiesMutantMitochondrialMischief,"EuropeanJournalofBiochemistry269.8(April2002):2003–9;P.F.Chinneryetal.,"AccumulationofMitochondrialDNAMutationsinAgeing,Cancer,andMitochondrialDisease:IsThereaCommonMechanism?"Lancet360.9342(October26,2002):1323–25;A.D.deGrey,"MitochondrialGeneTherapy:AnArenafortheBiomedicalUseofInteins,"TrendsinBiotechnology18.9(September2000):394–99.

51."Thenotionof'vaccinating'individualsagainstaneurodegenerativedisordersuchas

Alzheimer'sdiseaseisamarkeddeparturefromclassicalthinkingaboutmechanismandtreatment,andyettherapeuticvaccinesforbothAlzheimer'sdiseaseandmultiplesclerosishavebeenvalidatedinanimalmodelsandareintheclinic.Suchapproaches,however,havethepotentialtoinduceunwantedinflammatoryresponsesaswellastoprovidebenefit"(H.L.WeinerandD.J.Selkoe,"InflammationandTherapeuticVaccinationinCNSDiseases,"Nature420.6917[December19–26,2002]:879-84).TheseresearchersshowedthatavaccineintheformofnosedropscouldslowthebraindeteriorationofAlzheimer's.H.L.Weineretal.,"NasalAdministrationofAmyloid-betaPeptideDecreasesCerebralAmyloidBurdeninaMouseModelofAlzheimer'sDisease,"AnnalsofNeurology48.4(October2000):567–79.

52.S.Vasan,P.Foiles,andH.Founds,"TherapeuticPotentialofBreakersofAdvancedGlycationEndProduct-ProteinCrosslinks,"ArchivesofBiochemistryandBiophysics419.1(November1,2003):89–96;D.A.Kass,"GettingBetterWithoutAGE:NewInsightsintotheDiabeticHeart,"CirculationResearch92.7(April18,2003):704–6.

53.S.Graham,"MethuselahWormRemainsEnergeticforLife,"October27,2003,www.sciam.com/article.cfm?chanID=sa003&articleID=000C601F-8711-1F99-86FB83414B7F0156.

54.RonWeiss'shomepageatPrincetonUniversity(http://www.princeton.edu/~rweiss)listshispublications,suchas"GeneticCircuitBuildingBlocksforCellularComputation,Communications,andSignalProcessing,"NaturalComputing,anInternationalJournal2.1(January2003):47–84.

55.S.L.Garfinkel,"BiologicalComputing,"TechnologyReview(May–June2000),http://static.highbeam.com/t/technologyreview/may012000/biologicalcomputing.

56.Ibid.SeealsothelistofcurrentresearchontheMITMediaLabWebsite,http://www.media.mit.edu/research/index.html.

57.Hereisonepossibleexplanation:"Inmammals,femaleembryoshavetwoX-chromosomesandmaleshaveone.Duringearlydevelopmentinfemales,oneoftheX'sandmostofitsgenesarenormallysilencedorinactivated.Thatway,theamountofgeneexpressioninmalesandfemalesisthesame.Butinclonedanimals,oneX-chromosomeisalreadyinactivatedinthedonatednucleus.Itmustbereprogrammedandthenlaterinactivatedagain,whichintroducesthepossibilityoferrors."CBCNewsonlinestaff,"GeneticDefectsMayExplainCloningFailures,"May27,2002,http://www.cbc.ca/stories/2002/05/27/cloning_errors020527.ThatstoryreportsonF.Xueetal.,"AberrantPatternsofXChromosomeInactivationinBovineClones,"NatureGenetics31.2(June2002):216–20.

58.RickWeiss,"CloneDefectsPointtoNeedfor2GeneticParents,"WashingtonPost,May10,1999,http://www.gene.ch/genet/1999/Jun/msg00004.html.

59.A.Baguisietal.,"ProductionofGoatsbySomaticCellNuclearTransfer,"NatureBiotechnology5(May1999):456–61.FormoreinformationonthepartnershipbetweenGenzymeTransgenicsCorporation,LouisianaStateUniversity,andTuftsUniversitySchoolofMedicinethatproducedthiswork,seetheApril27,1999,pressrelease,"GenzymeTransgenicsCorporationAnnouncesFirstSuccessfulCloningofTransgenicGoat,"http://www.transgenics.com/pressreleases/pr042799.html.

60.LubaVangelova,"TrueorFalse?ExtinctionIsForever,"Smithsonian,June2003,http://www.smithsonianmag.com/smithsonian/issues03/jun03/phenomena.html.

61.J.B.GurdonandA.Colman,"TheFutureofCloning,"Nature402.6763(December16,1999):743–46;GregoryStockandJohnCampbell,eds.,EngineeringtheHumanGermline:AnExplorationoftheScienceandEthicsofAlteringtheGenesWePasstoOurChildren(NewYork:OxfordUniversityPress,2000).

62.AstheScrippsResearchInstitutepointsout,"Theabilitytodedifferentiateorreverselineage-committedcellstomultipotentprogenitorcellsmightovercomemanyoftheobstaclesassociatedwithusingESCsandadultstemcellsinclinicalapplications(inefficientdifferentiation,rejectionofallogeniccells,efficientisolationandexpansion,etc.).Withanefficientdedifferentiationprocess,itisconceivablethathealthy,abundantandeasilyaccessible

adultcellscouldbeusedtogeneratedifferenttypesoffunctionalcellsfortherepairofdamagedtissuesandorgans"(http://www.scripps.edu/chem/ding/sciences.htm).

Thedirectconversionofonedifferentiatedcelltypeintoanother—aprocessreferredtoastransdifferentiation—wouldbebeneficialforproducingisogenic[patient'sown]cellstoreplacesickordamagedcellsortissue.Adultstemcellsdisplayabroaderdifferentiationpotentialthananticipatedandmightcontributetotissuesotherthanthoseinwhichtheyreside.Assuch,theycouldbeworthytherapeuticagents.Recentadvancesintransdifferentiationinvolvenucleartransplantation,manipulationofcellcultureconditions,inductionofectopicgeneexpressionanduptakeofmoleculesfromcellularextracts.Theseapproachesopenthedoorstonewavenuesforengineeringisogenicreplacementcells.Toavoidunpredictabletissuetransformation,nuclearreprogrammingrequirescontrolledandheritableepigeneticmodifications.ConsiderableeffortsremaintounravelthemolecularprocessesNOTES557underlyingnuclearreprogrammingandevaluatestabilityofthechangesinreprogrammedcells.

QuotedfromP.CollasandAnne-MarlHakelien,"TeachingCellsNewTricks,"TrendsinBiotechnology21.8(August2003):354–61;P.Collas,"NuclearReprogramminginCell-FreeExtracts,"PhilosophicalTransactionsoftheRoyalSocietyofLondon,B358.1436(August29,2003):1389–95.

63.Researchershaveconvertedhumanlivercellstopancreascellsinthelaboratory:JonathanSlacketal.,"ExperimentalConversionofLivertoPancreas,"CurrentBiology13.2(January2003):105–15.Researchersreprogrammedcellstobehavelikeothercellsusingcellextracts;forexample,skincellswerereprogrammedtoexhibitT-cellcharacteristics.Anne-MariHakelienetal.,"ReprogrammingFibroblaststoExpressT-CellFunctionsUsingCellExtracts,"NatureBiotechnology20.5(May2002):460–66;Anne-MariHakelienandP.Collas,"NovelApproachestoTransdifferentiation,"CloningStemCells4.4(2002):379–87.SeealsoDavidToshandJonathanM.W.Slack,"HowCellsChangeTheirPhenotype,"NatureReviewsMolecularCellBiology3.3(March2002):187–94.

64.Seethedescriptionoftranscriptionfactorsinnote21,above.65.R.P.Lanzaetal.,"ExtensionofCellLife-SpanandTelomereLengthinAnimalsCloned

fromSenescentSomaticCells,"Science288.5466(April28,2000):66–9.SeealsoJ.C.Ameisen,"OntheOrigin,Evolution,andNatureofProgrammedCellDeath:ATimelineofFourBillionYears,"CellDeathandDifferentiation9.4(April2002):367–93;Mary-EllenShay,"TransplantationWithoutaDonor,"Dream:TheMagazineofPossibilities(Children'sHospital,Boston),Fall2001.

66.In2000theImmuneToleranceNetwork(http://www.immunetolerance.org),aprojectoftheNationalInstitutesofHealth(NIH)andtheJuvenileDiabetesFoundation,announcedamulticenterclinicaltrialtoassesstheeffectivenessofislettransplantation.Accordingtoaclinical-trialresearchsummary(JamesShapiro,"Campath-IfiandOne-YearTemporarySirolimusMaintenanceMonotherapyinClinicalIsletTransplantation,"http://www.immunetolerance.org/public/clinical/islet/trials/shapiro2.html),"ThistherapyisnotsuitableforallpatientswithTypeIdiabetes,eveniftherewerenolimitationinisletsupply,becauseofthepotentiallong-termrisksofcancer,life-threateninginfectionsanddrugside-effectsrelatedtotheantirejectiontherapy.Iftolerance[indefinitegraftfunctionwithoutaneedforlong-termdrugstopreventrejection]couldbeachievedatminimalup-frontrisk,thenislettransplantcouldbeusedsafelyearlierinthecourseofdiabetes,andeventuallyinchildrenatthetimeofdiagnosis."

67."LabGrownSteaksNearingMenu,"http://www.newscientist.com/news/news.jsp?id=ns99993208,includesdiscussionoftechnicalissues.

68.Thehalvingtimeforfeaturesizesisfiveyearsineachdimension.Seediscussioninchapter2.69.AnanalysisbyRobertA.FreitasJr.indicatesthatreplacing10percentofaperson'sredblood

cellswithroboticrespirocyteswouldenableholdingone'sbreathforaboutfourhours,whichisabout240timeslongerthanoneminute(aboutthelengthoftimefeasiblewithallbiologicalredbloodcells).Sincethisincreasederivesfromreplacingonly10percentoftheredbloodcells,therespirocytesarethousandsoftimesmoreeffective.

70.Nanotechnologyis"thorough,inexpensivecontrolofthestructureofmatterbasedonmolecule-by-moleculecontrolofproductsandbyproducts;theproductsandprocessesofmolecularmanufacturing,includingmolecularmachinery"(EricDrexlerandChrisPeterson,UnboundingtheFuture:TheNanotechnologyRevolution[NewYork:WilliamMorrow,1991]).Accordingtotheauthors:

Technologyhasbeenmovingtowardgreatercontrolofthestructureofmatterformillennia....[P]astadvancedtechnologies—microwavetubes,lasers,superconductors,satellites,robots,andthelike—havecometricklingoutoffactories,atfirstwithhighpricetagsandnarrowapplications.Molecularmanufacturing,though,willbemorelikecomputers:aflexibletechnologywithahugerangeofapplications.Andmolecularmanufacturingwon'tcometricklingoutofconventionalfactoriesascomputersdid;itwillreplacefactoriesandreplaceorupgradetheirproducts.Thisissomethingnewandbasic,notjustanothertwentieth-centurygadget.Itwillariseoutoftwentieth-centurytrendsinscience,butitwillbreakthetrend-linesintechnology,economics,andenvironmentalaffairs.[chap.1]

DrexlerandPetersonoutlinethepossiblescopeoftheeffectsoftherevolution:efficientsolarcells"ascheapasnewspaperandastoughasasphalt,"molecularmechanismsthatcankillcoldvirusesinsixhoursbeforebiodegrading,immunemachinesthatdestroymalignantcellsinthebodyatthepushofabutton,pocketsupercomputers,theendoftheuseoffossilfuels,spacetravel,andrestorationoflostspecies.AlsoseeE.Drexler,EnginesofCreation(NewYork:AnchorBooks,1986).TheForesightInstitutehasausefullistofnanotechnologyFAQs(http://www.foresight.org/NanoRev/FIFAQ1.html)andotherinformation.OtherWebresourcesincludetheNationalNanotechnologyInitiative(http://www.nano.gov),http://nanotechweb.org,Dr.RalphMerkle'snanotechnologypage(http://www.zyvex.com/nano).andNanotechnology,anonlinejournal(http://www.iop.org/EJ/journal/0957-4484).Extensivematerialonnanotechnologycanbefoundontheauthor'sWebsiteathttp://www.kurzweilAI.net/meme/frame.html?m=18.

71.RichardP.Feynman,"There'sPlentyofRoomattheBottom,"AmericanPhysicalSocietyannualmeeting,Pasadena,California,1959;transcriptathttp://www.zyvex.com/nanotech/feynman.html.

72.JohnvonNeumann,TheoryofSelf-ReproducingAutomata,A.W.Burks,ed.(Urbana:UniversityofIllinoisPress,1966).

73.ThemostcomprehensivesurveyofkinematicmachinereplicationisRobertA.FreitasJr.andRalphC.Merkle,KinematicSelf-ReplicatingMachines(Georgetown,Tex.:LandesBioscience,2004),http://www.MolecularAssembler.com/KSRM.htm.

74.K.EricDrexler,EnginesofCreation,andK.EricDrexler,Nanosystems:MolecularMachinery,Manufacturing,andComputation(NewYork:WileyInterscience,1992).

75.Seethediscussionofnanotubecircuitryinchapter3,includingtheanalysisofthepotentialofnanotubecircuitryinnote9ofthatchapter.

76.K.EricDrexlerandRichardE.Smalley,"Nanotechnology:DrexlerandSmalleyMaketheCaseforandAgainst'MolecularAssemblers,'"ChemicalandEngineeringNews,November30,2003,http://pubs.acs.org/cen/coverstory/8148/8148counterpoint.html.

77.RalphC.Merkle,"AProposed'Metabolism'foraHydrocarbonAssembler,"Nanotechnology8(December1997):149–62,http://www.iop.org/EJ/abstract/0957-4484/8/4/001orhttp://www.zyvex.com/nanotech/hydroCarbonMetabolism.html.SeealsoRalphC.Merkle,"BindingSitesforUseinaSimpleAssembler,"Nanotechnology8(1997):23–28,

http://www.zyvex.com/nanotech/bindingSites.html;RalphC.Merkle,"ANewFamilyofSixDegreeofFreedomPositionalDevices,"Nanotechnology8(1997):47–52,http://www.zyvex.com/nanotech/6dof.html;RalphC.Merkle,"CasinganAssembler,"Nanotechnology10(1999):315–22,http://www.zyvex.com/nanotech/casing;RobertA.FreitasJr.,"ASimpleToolforPositionalDiamondMechanosynthesis,andItsMethodofManufacture,"U.S.ProvisionalPatentApplicationNo.60/543,802,filedFebruary11,2004,processdescribedinlectureathttp://www.MolecularAssembler.com/Papers/PathDiamMolMfg.htm;RalphC.MerkleandRobertA.FreitasJr.,"TheoreticalAnalysisofaCarbon-CarbonDimerPlacementToolforDiamondMechanosynthesis,"JournalofNanoscienceandNanotechnology3(August2003):319–24,http://www.rfreitas.com/Nano/JNNDimerTool.pdf;RobertA.FreitasJr.andRalphC.Merkle,"Merkle-FreitasHydrocarbonMolecularAssembler;inKinematicSelf-ReplicatingMachines,section4.11.3(Georgetown,Tex.:LandesBioscience,2004),pp.130–35,http://www.MolecularAssembler.com/KSRM/4.11.3.htm.

78.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section6.3.4.5,"ChemoelectricCells"(Georgetown,Tex.:LandesBioscience,1999),pp.152–54,http://www.nanomedicine.com/NMI/6.3.4.5.htm;RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section6.3.4.4,"GlucoseEngines"(Georgetown,Tex.:LandesBioscience,1999),pp.149-52,http://www.nanomedicine.com/NMI/6.3.4.4.htm;K.EricDrexler,Nanosystems:MolecularMachinery,Manufacturing,andComputation,section16.3.2,"AcousticPowerandControl"(NewYork:WileyInterscience,1992),pp.472–76.SeealsoRobertA.FreitasJr.andRalphC.Merkle,KinematicSelf-ReplicatingMachines,appendixB.4,"AcousticTransducerforPowerandControl"(Georgetown,Tex.:LandesBioscience,2004),pp.225–33,http://www.MolecularAssembler.com/KSRM/AppB.4.htm.

79.ThemostcomprehensivesurveyoftheseproposalsmaybefoundinRobertA.FreitasJr.andRalphC.Merkle,KinematicSelf-ReplicatingMachines,chapter4,"MicroscaleandMolecularKinematicMachineReplicators"(Georgetown,Tex.:LandesBioscience,2004),pp.89–144,http://www.MolecularAssembler.com/KSRM/4.htm.

80.Drexler,Nanosystems,p.441.81.ThemostcomprehensivesurveyoftheseproposalsmaybefoundinRobertA.FreitasJr.and

RalphC.Merkle,KinematicSelf-ReplicatingMachines,chapter4,"MicroscaleandMolecularKinematicMachineReplicators"(Georgetown,Tex.:LandesBioscience,2004),pp.89–144,http://www.MolecularAssembler.com/KSRM/4.htm.

82.T.R.Kelly,H.DeSilva,andR.A.Silva,"UnidirectionalRotaryMotioninaMolecularSystem,"Nature401.6749(September9,1999):150–52.

83.CarloMontemagnoandGeorgeBachand,"ConstructingNanomechanicalDevicesPoweredbyBiomolecularMotors,"Nanotechnology10(1999):225–31;GeorgeD.BachandandCarloD.Montemagno,"ConstructingOrganic/InorganicNEMSDevicesPoweredbyBiomolecularMotors,"BiomedicalMicrodevices2.3(June2000):179–84.

84.N.Koumuraetal.,"Light-DrivenMonodirectionalMolecularRotor,"Nature401.6749(September9,1999):152–55.

85.BerkeleyLab,"AConveyorBeltfortheNano-Age,"April28,2004,http://www.lbl.gov/Science-Articles/Archive/MSD-conveyor-belt-for-nanoage.html.

86."Study:Self-ReplicatingNanomachinesFeasible,"June2,2004,http://www.smalltimes.com/document_display.cfm?section_id=53&document_id=8007,reportingonTihamerToth-Pejel,"ModelingKinematicCellularAutomata,"April30,2004,http://www.niac.usra.edu/files/studies/final_report/pdf/883Toth-Fejel.pdf.

87.W.U.Dittmer,A.Reuter,andF.C.Simmel,"ADNA-BasedMachineThatCanCyclicallyBindandReleaseThrombin,"AngewandteChemieInternationalEdition43(2004):3550–53.

88.ShipingLiaoandNadrianC.Seeman,"TranslationofDNASignalsintoPolymerAssemblyInstructions,"Science306(December17,2004):2072–74,http://www.sciencemag.org/cgi/reprint/306/5704/2072.pdf.

89.ScrippsResearchInstitute,"Nano-origami,"February11,2004,

http://www.eurekalert.org/pub_releases/2004-02/sri-n021004.php.90.JennyHogan,"DNARobotTakesItsFirstSteps,"May6,2004,

http://www.newscientist.com/news/news.jsp?id=ns99994958,reportingonNadrianSeemanandWilliamSherman,"APreciselyControlledDNABipedWalkingDevice,"NanoLetters4.7(July2004):1203–7.

91.HelenPearson,"ConstructionBugsFindTinyWork,"NatureNews,July11,2003,http://www.nature.com/news/2003/030707/full/030707-9.html.

92.RichardE.Smalley,"Nanofallacies:OfChemistry,LoveandNanobots,"ScientificAmerican285.3(September2001):76–77;subscriptionrequiredforthislink:http://www.sciamdigital.com/browse.cfm?sequencenameCHAR=item2&methodnameCHAR=resource_getitembrowse&interfacenameCHAR=browse.cfm&ISSUEID_CHAR=6A628AB3-17A5-4374-B100-3185A0CCC86&ARTICLEID_CHAR=F90C4210-C153-4B2F-83A1-28F2012B637&sc=I100322.

93.Seethebibliographyofreferencesinnotes108and109below.SeealsoDrexler,Nanosystems,forhisproposal.Forsampleconfirmations,seeXiaoYanChang,MartinPerry,JamesPeploski,DonaldL.Thompson,andLionelM.Raff,"TheoreticalStudiesofHydrogen-AbstractionReactionsfromDiamondandDiamondlikeSurfaces,"JournalofChemicalPhysics99(September15,1993):4748–58.SeealsoL.J.LauhonandW.Ho,"InducingandObservingtheAbstractionofaSingleHydrogenAtominBimolecularReactionwithaScanningTunnelingMicroscope,"JournalofPhysicalChemistry105(2000):3987–92;G.AllisandK.EricDrexler,"DesignandAnalysisofaMolecularToolforCarbonTransferinMechanosynthesis,"JournalofComputationalandTheoreticalNanoscience2.1(March–April2005,inpress).

94.LeaWinerman,"HowtoGrabanAtom,"PhysicalReviewFocus,May2,2003,http://focus.aps.org/story/v11/st19,reportingonNoriakiOyabu,"MechanicalVerticalManipulationofSelectedSingleAtomsbySoftNanoindentationUsingaNearContactAtomicForceMicroscope,"PhysicalReviewLetters90.17(May2,2003):176102.

95.RobertA.FreitasJr.,"TechnicalBibliographyforResearchonPositionalMechanosynthesis,"ForesightInstituteWebsite,December16,2003,http://foresight.org/stage2/mechsynthbib.html.

96.Seeequationandexplanationonp.3ofRalphC.Merkle,"That'sImpossible!HowGoodScientistsReachBadConclusions,"http://www.zyvex.com/nanotech/impossible.html.

97."ThusΔXcisjust~5%ofthetypicalatomicelectronclouddiameterof~0.3nm,imposingonlyamodestadditionalconstraintonthefabricationandstabilityofnanomechanicalstructures.(Eveninmostliquidsattheirboilingpoints,eachmoleculeisfreetomoveonly~0.07nmfromitsaverageposition.)"RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section2.1,"IsMolecularManufacturingPossible?"(Georgetown,Tex.:LandesBioscience,1999),p.39,http://www.nanomedicine.com/NMI/2.1.htm#p9.

98.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section6.3.4.5,"ChemoelectricCells"(Georgetown,Tex.:LandesBioscience,1999),pp.152–54,http://www.nanomedicine.com/NMI/6.3.4.5.htm.

99.MontemagnoandBachand,"ConstructingNanomechanicalDevicesPoweredbyBiomolecularMotors."

100.OpenletterfromForesightchairmanK.EricDrexlertoNobellaureateRichardSmalley,http://www.foresight.org/NanoRev/Letter.html,andreprintedhere:http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0560.html.ThefullstorycanbefoundatRayKurzweil,"TheDrexler-SmalleyDebateonMolecularAssembly,"http://www.KurzweilAI.net/meme/frame.htrnl?main=/articles/art0604.html.

101.K.EricDrexlerandRichardE.Smalley,"Nanotechnology:DrexlerandSmalleyMaketheCaseforandAgainst'MolecularAssemblers,'"Chemical&EngineeringNews81.48(Dec.1,2003):37–42,http://pubs.acs.org/cen/coverstory/8148/8148counterpoint.html.

102.A.ZaksandA.M.Klibanov,"EnzymaticCatalysisinOrganicMediaat100DegreesC,"Science224.4654(June15,1984):1249–51.

103.PatrickBailey,"UnravelingtheBigDebateAboutSmallMachines,"BetterHumans,August16,.2004,http://www.betterhumans.com/Features/Reports/report.aspx?articleID=2004-08-16-1.

104.CharlesB.Musgraveetal.,"TheoreticalStudiesofaHydrogenAbstractionToolforNanotechnology,"Nanotechnology2(October1991):187–95;MichaelPageandDonaldW.Brenner,"HydrogenAbstractionfromaDiamondSurface:AbinitioQuantumChemicalStudywithConstrainedIsobutaneasaModel,"JournaloftheAmericanChemicalSociety113.9(1991):3270–74;XiaoYanChang,MartinPerry,JamesPeploski,DonaldL.Thompson,andLionelM.Raff,"TheoreticalStudiesofHydrogen-AbstractionReactionsfromDiamondandDiamond-likeSurfaces,"JournalofChemicalPhysics99(September15,1993):4748–58;J.W.Lyding,K.Hess,G.C.Abeln,etal.,"UHV-STMNanofabricationandHydrogen/DeuteriumDesorptionfromSiliconSurfaces:ImplicationsforCMOSTechnology,"AppliedSurfaceScience132(1998):221;http://www.hersam-group.northwestern.edu/publications.html;E.T.Foleyetal.,"CryogenicUHV-STMStudyofHydrogenandDeuteriumDesorptionfromSilicon(100),"PhysicalReviewLetters80(1998):1336–39,http://prola.aps.org/abstract/PRL/v80/i6/p1336_1;L.J.LauhonandW.Ho,"InducingandObservingtheAbstractionofaSingleHydrogenAtominBimolecularReactionwithaScanningTunnelingMicroscope,"JournalofPhysicalChemistry105(2000):3987–92.

105.StephenP.WalchandRalphC.Merkle,"TheoreticalStudiesofDiamondMechanosynthesisReactions,"Nanotechnology9(September1998):285–96;FedorN.Dzegilenko,DeepakSrivastava,andSubhashSaini,"SimulationsofCarbonNanotubeTipAssistedMechano-ChemicalReactionsonaDiamondSurface,"Nanotechnology9(December1998):325–30;RalphC.MerkleandRobertA.FreitasJr.,"TheoreticalAnalysisofaCarbon-CarbonDimerPlacementToolforDiamondMechanosynthesis,"JournalofNanoscienceandNanotechnology3(August2003):319–24,http://www.rfreitas.com/Nano/DimerTool.htm;JingpingPeng,RobertA.FreitasJr.,andRalphC.Merkle,"TheoreticalAnalysisofDiamondMechano-Synthesis.PartI.StabilityofC2MediatedGrowthofNanocrystallineDiamondC(1lO)Surface,"JournalofComputationalandTheoreticalNanoscience1(March2004):62–70,http://www.molecularassembler.com/JCTNPengMar04.pdf;DavidJ.Mann,JingpingPeng,RobertA.FreitasJr.,andRalphC.Merkle,"TheoreticalAnalysisofDiamondMechanoSynthesis.PartII.C2MediatedGrowthofDiamondC(`lO)SurfaceviaSi/Ge-TriadamantaneDimerPlacementTools,"JournalofComputationalandTheoreticalNanoscience1(March2004),71–80,http://www.molecularassembler.com/JCTNMannMar04.pdf.

106.Theanalysisofthehydrogenabstractiontoolandcarbondepositiontoolshasinvolvedmanypeople,including:DonaldW.Brenner,TahirCagin,RichardJ.Colton,K.EricDrexler,FedorN.Dzegilenko,RobertA.FreitasJr.,WilliamA.GoddardIII,J.A.Harrison,CharlesB.Musgrave,RalphC.Merkle,MichaelPage,JasonK.Perry,SubhashSaini,O.A.Shenderova,SusanB.Sinnott,DeepakSrivastava,StephenP.Walch,andCarterT.White.

107.RalphC.Merkle,"AProposed'Metabolism'foraHydrocarbonAssembler,"Nanotechnology8(December1997):14–62,http://www.iop.org/EJ/abstraet/0957-4484/8/4/001orhttp://www.zyvex.com/nanotech/hydroCarbonMetabolism.html.

108.Ausefulbibliographyofreferences:RobertA.FreitasJr.,"TechnicalBibliographyforResearchonPositionalMechanosynthesis,"ForesightInstituteWebsite,December16,2003,http://foresight.org/stage2/mechsynthbib.html;WilsonHoandHyojuneLee,"SingleBondFormationandCharacterizationwithaScanningTunnelingMicroscope,"Science286.5445(November26,1999):1719–22,http://www.physics.uci.edu/~wilsonho/stm-iets.html;K.EricDrexler,Nanosystems,chapter8;RalphMerkle,"Proposed'Metabolism'foraHydrocarbonAssembler";Musgraveetal.,"TheoreticalStudiesofaHydrogenAbstractionToolforNanotechnology";MichaelPageandDonaldW.Brenner,"HydrogenAbstractionfromaDiamondSurface:AbinitioQuantumChemicalStudywithConstrainedIsobutaneasaModel,"JournaloftheAmericanChemicalSociety113.9(1991):3270–74;D.W.Brenneretal.,

"SimulatedEngineeringofNanostructures,"Nanotechnology7(September1996):161–67,http://www.zyvex.com/nanotech/nano4/brennerPaper.pdf;S.P.Walch,W.A.GoddardIII,andRalphMerkle,"TheoreticalStudiesofReactionsonDiamondSurfaces,"FifthForesightConferenceonMolecularNanotechnology,1997,http://www.foresight.org/Conferences/MNT05/Abstracts/Walcabst.html;StephenP.WalchandRalphC.Merkle,"TheoreticalStudiesofDiamondMechanosynthesisReactions,"Nanotechnology9(September1998):285–96;FedorN.Dzegilenko,DeepakSrivastava,andSubhashSaini,"SimulationsofCarbonNanotubeTipAssistedMechano-ChemicalReactionsonaDiamondSurface,"Nanotechnology9(December1998):325–30;J.W.Lydingetal.,"UHV-STMNanofabricationandHydrogen/DeuteriumDesorptionfromSiliconSurfaces:ImplicationsforCMOSTechnology,"AppliedSurfaceScience132(1998):221,http://www.hersam-group.northwestem.edu/publications.html;E.T.Foleyetal.,"CryogenicUHV-STMStudyofHydrogenandDeuteriumDesorptionfromSilicon(100),"PhysicalReviewLetters80(1998):1336–39,http://prola.aps.org/abstract/PRL/v80/i6/p1336_1;M.C.Hersam,G.CAbeln,andJ.W.Lyding,"AnApproachforEfficientlyLocatingandElectricallyContactingNanostructuresFabricatedviaUHV-STMLithographyonSi(100),"MicroelectronicEngineering47(1999):235–37;L.J.LauhonandW.Ho,"InducingandObservingtheAbstractionofaSingleHydrogenAtominBimolecularReactionwithaScanningTunnelingMicroscope,"JournalofPhysicalChemistry105(2000):3987–92,http://www.physics.uci.edu/~wilsonho/stm-iets.html.

109.EricDrexler,"DrexlerCounters,"firstpublishedonKurzweilAI.netonNovember1,2003:http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0606.html.SeealsoK.EricDrexler,Nanosystems:MolecularMachinery,Manufacturing,andComputation(NewYork:WileyInterscience,1992),chapter8;RalphC.Merkle,"ForesightDebatewithScientificAmerican"(1995),http://www.foresight.org/SciAmDebate/SciAmResponse.html;WilsonHoandHyojuneLee,"SingleBondFormationandCharacterizationwithaScanningTunnelingMicroscope,"Science286.5445(November26,1999):1719–22,http://www.physics.uci.edu/~wilsonho/stm-iets.html;K.EricDrexler,DavidForrest,RobertA.FreitasJr.,J.StorrsHall,NeilJacobstein,TomMcKendree,RalphMerkle,andChristinePeterson,"OnPhysics,Fundamentals,andNanorobots:ARebuttaltoSmalley'sAssertionthatSelf-ReplicatingMechanicalNanorobotsAreSimplyNotPossible:ADebateAboutAssemblers"(2001),http://www.imm.org/SciAmDebate2/smalley.html.

110.Seehttp://pubs.acs.org/cen/coverstory/8148/8148counterpoint.html;http://www.kurzweilAI.net/meme/frame.html?main=/articles/art0604.html?

111.D.Maysingeretal.,"BlockCopolymersModifytheInternalizationofMicelle-IncorporatedProbesintoNeuralCells,"BiochimicaetBiophysicaActa1539.3(June20,2001):205–17;R.Savicetal.,"MicellarNanocontainersDistributetoDefinedCytoplasmicOrganelles,"Science300.5619(April25,2003):615–18.

112.T.Yamadaetal.,"NanopartidesfortheDeliveryofGenesandDrugstoHumanHepatocytes,"NatureBiotechnology21.8(August2003):885–90.PublishedelectronicallyJune29,2003.Abstract:http://www.nature.com/cgi-taf/DynaPage.taf?file=/nbt/journal/v21/n8/abs/nbt843.html.ShortpressreleasefromNature:http://www.nature.com/nbt/press_release/nbt0803.html.

113.RichardsGraysonetal.,"ABioMEMSReview:MEMSTechnologyforPhysiologicallyIntegratedDevices,"IEEEProceedings92(2004):6–21;RichardsGraysonetal.,"MolecularReleasefromaPolymericMicroreservoirDevice:InfluenceofChemistry,PolymerSwelling,andLoadingonDevicePerformance,"JournalofBiomedicalMaterialsResearch69A.3(June1,2004):502–12.

114.D.PatrickO'Nealetal.,"Photo-thermalTumorAblationinMiceUsingNearInfrared-AbsorbingNanoparticles,"CancerLetters209.2(June25,2004):171–76.

115.InternationalEnergyAgency,fromanR.E.Smalleypresentation,"Nanotechnology,theS&TWorkforce,Energy&Prosperity,"p.12,presentedatPCAST(President'sCouncilofAdvisorsonScienceandTechnology),Washington,D.C.,March3,2003,

http://www.ostp.gov/PCASTIPCAST%203-3-03%20R%20Smalley%20Slides.pdf;alsoathttp://cohesion.rice.edu/NaturalSciences/Smalley/emplibrary/PCAST%20March%203,%202003.ppt.

116.Smalley,"Nanotechnology,theS&TWorkforce,Energy&Prosperity."117."FutureGen—ASequestrationandHydrogenResearchInitiative,"U.S.Departmentof

Energy,OfficeofFossilEnergy,February2003,http://www.fossil.energy.gov/programs/powersystems/futuregen/futuregen_factsheet.pdf.

118.Drexler,Nanosystems,pp.428,433.119.BarnabyJ.Feder,"ScientistatWork/RichardSmalley:SmallThoughtsforaGlobalGrid,"

NewYorkTimes,September2,2003;thefollowinglinkrequiressubscriptionorpurchase:http://query.nytimes.com/gst/abstract.html?res=F30C17FC3D5C0C718CDDA00894DB404482.

120.InternationalEnergyAgency,fromSmalley,"Nanotechnology,theS&TWorkforce,Energy&Prosperity,"p.12.

121.AmericanCouncilfortheUnitedNationsUniversity,MillenniumProjectGlobalChallenge13:http://www.acunu.org/millennium/ch-13.html.

122."WirelessTransmissioninEarth'sEnergyFuture,"EnvironmentNewsService,November19,2002,reportingonJeromeC.GlennandTheodoreJ.Gordonin"2002StateoftheFuture,"AmericanCouncilfortheUnitedNationsUniversity(August2002).

123.Disclosure:theauthorisanadvisertoandinvestorinthiscompany.124."NECUnveilsMethanol-FueledLaptop,"AssociatedPress,June30,2003,

http://www.siliconvalley.com/mld/siliconvalley/news/6203790.htm.reportingonNECpressrelease,"NECUnveilsNotebookPCwithBuilt-InFuelCell,"June30,2003,http://www.nec.co.jp/press/en/0306/3002.html.

125.TonySmith,"ToshibaBoffinsPrepLaptopFuelCell,"TheRegister,March5,2003,http://www.theregister.co.uk/2003/03/05/toshiba_boffins_prep_laptop_fuel;YoshikoHara,"ToshibaDevelopsMatchbox-SizedFuelCellforMobilePhones,"EETimes,June24,2004,http://www.eet.com/article/showArticle.jhtml?articleId=22101804,reportingonToshibapressrelease,"ToshibaAnnouncesWorld'sSmallestDirectMethanolFuelCellwithEnergyOutputof100Milliwats,"http://www.toshiba.com/taec/press/dmfc04_222.shtml.

126.KarenLurie,"HydrogenCars,"ScienceCentralNews,May13,2004,http://www.sciencentral.com/articles/view.php3?language=english&type=article&article_id=218392247.

127.LouiseKnapp,"BoozetoFuelGadgetBatteries,"WiredNews,April2,2003,http://www.wired.com/news/gizmos/0,1452,58119,00.html,andSt.LouisUniversitypressrelease,"PoweredbyYourLiquorCabinet,NewBiofuelCellCouldReplaceRechargeableBatteries,"March24,2003,http.//www.slu.edu/readstory/newsinfo/2474,reportingonNickAkersandShelleyMinteer,"TowardstheDevelopmentofaMembraneElectrodeAssembly,"presentedattheAmericanChemicalSocietynationalmeeting,Anaheim,Calif.(2003).

128."BiofuelCellRunsonMetabolicEnergytoPowerMedicalImplants,"NatureOnline,November12,2002,http://www.nature.com/news/2002/021111/full/021111-1.html,reportingonN.Mano,F.Mao,andA.Heller,"AMiniatureBiofuelCellOperatinginaPhysiologicalBuffer,"JournaloftheAmericanChemicalSociety124(2002):12962–63.

129."PowerfromBloodCouldLeadto'HumanBatteries,'"FairfaxDigital,August4,2003,http://www.smh.com.au/articles/2003/08/03/1059849278131.html?oneclick=true.Readmoreaboutthemicrobialfuelcellshere:http://www.geobacter.org/research/microbial/.MatsuhikoNishizawa'sBioMEMslaboratorydiagramsamicro-biofuelcell:http://www.biomems.mech.tohoku.ac.jp/research_e.html.Thisshortarticledescribesworkonanimplantable,nontoxicpowersourcethatnowcanproduce0.2watts:http://www.iol.co.za/index.php?se_id=l&click_id=31&art_id=qw111596760144B215.

130.MikeMartin,"Pace-SettingNanotubesMayPowerMicro-Devices,"NewsFactor,February27,2003,http://physics.iisc.ernet.in/~asood/Pace-Setting%20Nanotubes%20May%20Power%20Micro-Devices.htm.

131."Finally,itispossibletoderivealimittothetotalplanetaryactivenanorobotmassby

consideringtheglobalenergybalance.TotalsolarinsolationreceivedattheEarth'ssurfaceis~1.75Î1017watts(IEarth~1370W/m2±0.4%atnormalincidence),"RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section6.5.7,"GlobalHypsithermalLimit"(Georgetown,Tex.:LandesBioscience,1999),pp.175-76,http://www.nanomedicine.com/NMI/6.5.7.htm#p1.

132.Thisassumes10billion(1010)persons,apowerdensityfornanorobotsofaround107wattspercubicmeter,ananorobotsizeofonecubicmicron,andapowerdrawofabout10picowatts(10-11watts)pernanorobot.Thehypsithermallimitof1016wattsimpliesabout10kilogramsofnanorobotsperperson,or1016nanorobotsperperson.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section6.5.7"GlobalHypsithermalLimit"(Georgetown,Tex.:LandesBioscience,1999),pp.175-76,http://www.nanomedicine.com/NMI/6.5.7.htm#p4.

133.Alternatively,nanotechnologycanbedesignedtobeextremelyenergyefficientinthefirstplacesothatenergyrecapturewouldbeunnecessary,andinfeasiblebecausetherewouldberelativelylittleheatdissipationtorecapture.Inaprivatecommunication(January2005),RobertA.FreitasJr.writes:"Drexler(Nanosystems:396)claimsthatenergydissipationmayintheorybeaslowasEdiss~0.1MJ/kg'ifoneassumesthedevelopmentofasetofmechanochemicalprocessescapableoftransformingfeedstockmoleculesintocomplexproductstructuresusingonlyreliable,nearlyreversiblesteps:0.1MJ/kgofdiamondcorrespondsroughlytotheminimumthermalnoiseatroomtemperature(e.g.,kT~4zJ/atomat298K)."

134.AlexisDeVos,EndoreversibleThermodynamicsofSolarEnergyConversion(London:OxfordUniversityPress,1992),p.103.

135.R.D.SchallerandV.1.Klimov,"HighEfficiencyCarrierMultiplicationinPbSeNanocrystals:ImplicationsforSolarEnergyConversion,"PhysicalReviewLetters92.18(May7,2004):186601.

136.NationalAcademiesPress,CommissiononPhysicalSciences,Mathematics,andApplications,HarnessingLight:OpticalScienceandEngineeringforthe21stCentury,(Washington,D.C.:NationalAcademyPress,1998),p.166,http://books.nap.edu/books/0309059917/html/166.html.

137.MattMarshall,"WorldEventsSparkInterestinSolarCellEnergyStart-ups,"MercuryNews,August15,2004,http://www.konarkatech.com/news_articles_082004/b-silicon_valley.phpandhttp://www.nanosolar.com/cache/merc081504.htm.

138.JohnGartner,"NASASpacesonEnergySolution,"WiredNews,June22,2004,http://www.wired.com/news/technology/0,1282,63913,00.html.SeealsoArthurSmith,"TheCaseforSolarPowerfromSpace,"http://www.lispace.org/articles/SSPCase.html.

139."TheSpaceElevatorPrimer,"SpacewardFoundation,http://www.elevator2010.org/site/primer.html.

140.KennethChang,"ExpertsSayNewDesktopFusionClaimsSeemMoreCredible,"NewYorkTimes,March3,2004,http://www.rpi.edu/web/News/nytlahey3.html,reportingonR.P.Taleyarkhan,"AdditionalEvidenceofNuclearEmissionsDuringAcousticCavitation,"PhysicalReviewE:Statistical,Nonlinear,andSoftMatterPhysics69.3,pt.2(March2004):036109.

141.TheoriginalPonsandFleischmanmethodofdesktopcoldfusionusingpalladiumelectrodesisnotdead.Ardentadvocateshavecontinuedtopursuethetechnology,andtheDepartmentofEnergyannouncedin2004thatitwasconductinganewformalreviewoftherecentresearchinthisfield.ToniFeder,"DOEWarmstoColdFusion,"PhysicsToday(April2004),http://www.physicstoday.org/vol-57/iss-4/p27.html.

142.AkiraFujishima,TataN.Rao,andDonaldA.Tryk,"TitaniumDioxidePhotocatalysis,"JournalofPhotochemistryandPhotobiologyC:PhotochemistryReview1(June29,2000):1–21;PrashantV.Kamat,RebeccaHuehn,andRoxanaNicolaescu,"A'SenseandShoot'ApproachforPhotocatalyticDegradationofOrganicContaminantsinWater,"JournalofPhysicalChemistryB106(January31,2002):788–94.

143.A.G.Panovetal.,"PhotooxidationofTolueneandp-XyleneinCation-ExchangedZeolitesX,Y,ZSM-5,andBeta:TheRoleofZeolitePhysicochemicalPropertiesinProductYieldand

Selectivity,"JournalofPhysicalChemistryB104(June22,2000):5706–14.144.GaborA.SomorjaiandKeithMcCrea,"RoadmapforCatalysisScienceinthe21stCentury:A

PersonalViewofBuildingtheFutureonPastandPresentAccomplishments,"AppliedCatalysisA:General222.1–2(2001):3–18,LawrenceBerkeleyNationalLaboratorynumber3.LBNL-48555,http://www.cchem.berkeley.edu/~gasgrp/2000.html(publication877).SeealsoZhao,Lu,andMillar,"AdvancesinmesoporousmolecularsieveMCM-41,"Industrial&EngineeringChemistryResearch35(1996):2075–90,http://cheed.nus.edu.sg/~chezxs/Zhao/publication/1996_2075.pdf.

145.NTSC/NSETreport,NationalNanotechnologyInitiative:TheInitiativeandItsImplementationPlan,July2000,http://www.nano.gov/html/res/nni2.pdf.

146.Wei-xianZhang,Chuan-BaoWang,andHsing-LungLien,"TreatmentofChlorinatedOrganicContaminantswithNanoscaleBimetallicParticles,"CatalysisToday40(May14,1988):387–95.

147.R.Q.LongandR.T.Yang,"CarbonNanotubesasSuperiorSorbentforDioxinRemoval,"JournaloftheAmericanChemicalSociety123.9(2001):2058–59.

148.RobertA.Freitas,Jr."DeathIsanOutrage!"presentedattheFifthAlcorConferenceonExtremeLifeExtension,NewportBeach,California,November16,2002,http://www.rfreitas.com/Nano/DeathIsAnOutrage.htm.

149.Forexample,thefifthannualBIOMEMSconference,June2003,SanJose,http://www.knowledgepress.com/events/11201717.htm.

150.Firsttwovolumesofaplannedfour-volumeseries:RobertA.FreitasJr.,Nanomedicine,vol.I,BasicCapabilities(Georgetown,Tex.:LandesBioscience,1999);Nanomedicine,vol.IIA,Biocompatibility(Georgetown,Tex.:LandesBioscience,2003);http://www.nanomedicine.com.

151.RobertA.FreitasJr.,"ExploratoryDesigninMedicalNanotechnology:AMechanicalArtificialRedCell,"ArtificialCells,BloodSubstitutes,andImmobilizationBiotechnology26(1998):411–30,http://www.foresight.org.Nanomedicine/Respirocytes.html.

152.RobertA.FreitasJr.,"Microbivores:ArtificialMechanicalPhagocytesusingDigestandDischargeProtocol,"Zyvexpreprint,March2001,http://www.rfreitas.com/Nano/Microbivores.htm;RobertA.FreitasJr.,"Microbivores:ArtificialMechanicalPhagocytes,"ForesightUpdateno.44,March31,2001,pp.11–13,http://www.imm.org/Reports/Rep025.html;seealsomicrobivoreimagesattheNanomedicineArtGallery,http://www.foresight.org/Nanomedicine/Gallery/Species/Microbivores.html.

153.RobertA.FreitasJr.,Nanomedicine,vol.I,BasicCapabilities,section9.4.2.5"NanomechanismsforNatation"(Georgetown,Tex.:LandesBioscience,1999),pp.309–12,http://www.nanomedicine.com/NMI/9.4.2.5.htm.

154.GeorgeWhitesides,"Nanoinspiration:TheOnceandFutureNanomachine,"ScientificAmerican285.3(September16,2001):78–83.

155."AccordingtoEinstein'sapproximationforBrownianmotion,after1secondhaselapsedatroomtemperatureafluidicwatermoleculehas,onaverage,diffusedadistanceof~50microns(~400,000moleculardiameters)whereasal-rnicronnanorobotimmersedinthatsamefluidhasdisplacedbyonly~0.7microns(only~0.7devicediameter)duringthesametimeperiod.ThusBrownianmotionisatmostaminorsourceofnavigationalerrorformotilemedicalnanorobots,"SeeK.EricDrexleretal.,"ManyFutureNanomachines:ARebuttaltoWhitesides'AssertionThatMechanicalMolecularAssemblersAreNotWorkableandNotaConcern,"aDebateaboutAssemblers,InstituteforMolecularManufacturing,2001,http://www.imm.org/SciAmDebate2/whitesides.html.

156.TejalA.Desai,"MEMS-BasedTechnologiesforCellularEncapsulation,"AmericanJournalofDrugDelivery1.1(2003):3–11,abstractavailableathttp://www.ingentaconnect.com/search/expand?pub=infobike://adis/add/2003/00000001/00000001/art00001.

157.AsquotedbyDouglasHofstadterinGödel,Escher,Bach:AnEternalGoldenBraid(NewYork:BasicBooks,1979).

158.Theauthorrunsacompany,FATKAT(FinancialAcceleratingTransactionsbyKurzweilAdaptiveTechnologies),whichappliescomputerizedpatternrecognitiontofinancialdatatomakestock-marketinvestmentdecisions,http://www.FatKat.com.

159.Seediscussioninchapter2onprice-performanceimprovementsincomputermemoryandelectronicsingeneral.

160.RunawayAIreferstoascenariowhere,asMaxMoredescribes,"superintelligentmachines,initiallyharnessedforhumanbenefit,soonleaveusbehind."MaxMore,"Embrace,Don'tRelinquish,theFuture,"http://www.KurzweilAI.net/articles/art0106.html?printable=1.SeealsoDamienBroderick'sdescriptionofthe"SeedAI":"Aself-improvingseedAIcouldrunglaciallyslowlyonalimitedmachinesubstrate.Thepointis,solongasithasthecapacitytoimproveitself,atsomepointitwilldosoconvulsively,burstingthroughanyarchitecturalbottleneckstodesignitsownimprovedhardware,maybeevenbuildit(ifit'sallowedcontroloftoolsinafabricationplant)."DamienBroderick,"TearingTowardtheSpike,"presentedat"AustraliaattheCrossroads?ScenariosandStrategiesfortheFuture"(April31–May2,2000),publishedonKurzweilAI.netMay7,2001,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0173.html.

161.DavidTalbot,"LordoftheRobots,"TechnologyReview(April2002).162.HeatherHavensteinwritesthatthe"inflatednotionsspawnedbysciencefictionwritersabout

theconvergenceofhumansandmachinestarnishedtheimageofAIinthe1980sbecauseAIwasperceivedasfailingtoliveuptoitspotential."HeatherHavenstein,"SpringComestoAIWinter:AThousandApplicationsBloominMedicine,CustomerService,EducationandManufacturing,"Computerworld,February14,2005,http://www.computerworld.com/softwaretopics/software/story/0,10801,99691,00.html.Thistarnishedimageledto"AIWinter,"definedas"atermcoinedbyRichardGabrielforthe(circa1990–94?)crashofthewaveofenthusiasmfortheAIlanguageLispandAIitself,followingaboominthe1980s."DuaneRettigwrote:"...companiesrodethegreatAIwaveintheearly80's,whenlargecorporationspouredbillionsofdollarsintotheAIhypethatpromisedthinkingmachinesin10years.Whenthepromisesturnedouttobeharderthanoriginallythought,theAIwavecrashed,andLispcrashedwithitbecauseofitsassociationwithAI.WerefertoitastheAIWinter."DuaneRettigquotedin"AIWinter,"http://c2.com/cgi/wiki?AiWinter.

163.TheGeneralProblemSolver(GPS)computerprogram,writtenin1957,wasabletosolveproblemsthroughrulesthatallowedtheGPStodivideaproblem'sgoalsintosubgoals,andthencheckifobtainingaparticularsubgoalwouldbringtheGPSclosertosolvingtheoverallgoal.Intheearly1960sThomasEvanwroteANALOGY,a"program[that]solvesgeometric-analogyproblemsoftheformA:B::C:?takenfromIQtestsandcollegeentranceexams."BoichoKokinovandRobertM.French,"ComputationalModelsofAnalogy-Making,"inL.Nadel,ed.,EncyclopediaofCognitiveScience,vol.1(London:NaturePublishingGroup,2003),pp.113–18.SeealsoA.Newell,J.C.Shaw,andH.A.Simon,"ReportonaGeneralProblem-SolvingProgram,"ProceedingsoftheInternationalConferenceonInformationProcessing(Paris:UNESCOHouse,1959),pp.256–64;ThomasEvans,"AHeuristicProgramtoSolveGeometric-AnalogyProblems,"inM.Minsky,ed.,SemanticInformationProcessing(Cambridge,Mass.:MITPress,1968).

164.SirArthurConanDoyle,"TheRed-HeadedLeague,"1890,availableathttp://www.eastoftheweb.com/short-stories/UBooks/RedHead.shtml.

165.V.Yuetal.,"AntimicrobialSelectionbyaComputer:ABlindedEvaluationbyInfectiousDiseasesExperts,"JAMA242.12(1979):1279–82.

166.GaryH.Anthes,"ComputerizingCommonSense,"Computerworld,April8,2002,http://www.computerworld.com/news/2002/story/0,11280,69881,00.html.

167.KristenPhilipkoski,"NowHere'saReallyBigIdea,"WiredNews,November25,2002,http://www.wired.com/news/technology/0,1282,56374,00.html,reportingonDarrylMacer,"TheNextChallengeIstoMaptheHumanMind,"Nature420(November14,2002):121;seealsoadescriptionoftheprojectathttp://www.biol.tsukuba.ac.jp/~macer/index.html.

168.ThomasBayes,"AnEssayTowardsSolvingaProblemintheDoctrineofChances,"publishedin1763,twoyearsafterhisdeathin1761.

169.SpamBayesspamfilter,http://spambayes.sourceforge.net.170.LawrenceR.Rabiner,"ATutorialonHiddenMarkovModelsandSelectedApplicationsin

SpeechRecognition,"ProceedingsoftheIEEE77(1989):257–86.ForamathematicaltreatmentofMarkovmodels,seehttp://jedlik.phy.bme.hu/~gerjanos/HMM/node2.html.

171.KurzweilAppliedIntelligence(KAI),foundedbytheauthorin1982,wassoldin1997for$100millionandisnowpartofScanSoft(formerlycalledKurzweilComputerProducts,theauthor'sfirstcompany,whichwassoldtoXeroxin1980),nowapubliccompany.KAIintroducedthefirstcommerciallymarketedlarge-vocabularyspeech-recognitionsystemin1987(KurzweilVoiceReport,withaten-thousand-wordvocabulary).

172.Hereisthebasicschemaforaneuralnetalgorithm.Manyvariationsarepossible,andthedesignerofthesystemneedstoprovidecertaincriticalparametersandmethods,detailedbelow.

Creatinganeural-netsolutiontoaprobleminvolvesthefollowingsteps:

·Definetheinput.·Definethetopologyoftheneuralnet(i.e.,thelayersofneuronsandtheconnectionsbetweentheneurons).

·Traintheneuralnetonexamplesoftheproblem.·Runthetrainedneuralnettosolvenewexamplesoftheproblem.·Takeyourneural-netcompanypublic.

Thesesteps(exceptforthelastone)aredetailedbelow:TheProblemInputTheprobleminputtotheneuralnetconsistsofaseriesofnumbers.Thisinputcanbe:

·Inavisualpattern-recognitionsystem,atwo-dimensionalarrayofnumbersrepresentingthepixelsofanimage;or

·Inanauditory(e.g.,speech)recognitionsystem,atwo-dimensionalarrayofnumbersrepresentingasound,inwhichthefirstdimensionrepresentsparametersofthesound(e.g.,frequencycomponents)andtheseconddimensionrepresentsdifferentpointsintime;or

·Inanarbitrarypattern-recognitionsystem,ann-dimensionalarrayofnumbersrepresentingtheinputpattern.

DefiningtheTopologyTosetuptheneuralnet,thearchitectureofeachneuronconsistsof:

·Multipleinputsinwhicheachinputis"connected"toeithertheoutputofanotherneuron,oroneoftheinputnumbers.

·Generally,asingleoutput,whichisconnectedeithertotheinputofanotherneuron(whichisusuallyinahigherlayer),ortothefinaloutput.

SetUptheFirstLayerofNeurons

·CreateN0neuronsinthefirstlayer.Foreachoftheseneurons,"connect"eachofthemultipleinputsoftheneuronto"points"(i.e.,numbers)intheprobleminput.Theseconnectionscanbedeterminedrandomlyorusinganevolutionaryalgorithm(seebelow).

·Assignaninitial"synapticstrength"toeachconnectioncreated.Theseweightscanstartoutallthesame,canbeassignedrandomly,orcanbedeterminedinanotherway(seebelow).

SetUptheAdditionalLayersofNeuronsSetupatotalofMlayersofneurons.Foreachlayer,setuptheneuronsinthatlayer.

Forlayeri:

·CreateNineuronsinlayer.,Foreachoftheseneurons,"connect"eachofthemultipleinputsoftheneurontotheoutputsoftheneuronsinlayeri-1(seevariationsbelow).

·Assignaninitial"synapticstrength"toeachconnectioncreated.Theseweightscanstartoutallthesame,canbeassignedrandomly,orcanbedeterminedinanotherway(seebelow).

·TheoutputsoftheneuronsinlayerMaretheoutputsoftheneuralnet(seevariationsbelow).

TheRecognitionTrialsHowEachNeuronWorksOncetheneuronissetup,itdoesthefollowingforeachrecognitiontrial:

·Eachweightedinputtotheneuroniscomputedbymultiplyingtheoutputoftheotherneuron(orinitialinput)thattheinputtothisneuronisconnectedtobythesynapticstrengthofthatconnection.

·Alloftheseweightedinputstotheneuronaresummed.·Ifthissumisgreaterthanthefiringthresholdofthisneuron,thenthisneuronisconsideredtofireanditsoutputis1.Otherwise,itsoutputis0(seevariationsbelow).

DotheFollowingforEachRecognitionTrialForeachlayer,fromlayer,tolayerM:Foreachneuroninthelayer:

·Sumitsweightedinputs(eachweightedinput=theoutputoftheotherneuron[orinitialinput]thattheinputtothisneuronisconnectedtomultipliedbythesynapticstrengthofthatconnection).

·Ifthissumofweightedinputsisgreaterthanthefiringthresholdforthisneuron,settheoutputofthisneuron=1,otherwisesetitto0.

ToTraintheNeuralNet

·Runrepeatedrecognitiontrialsonsampleproblems.·Aftereachtrial,adjustthesynapticstrengthsofalltheinterneuronalconnectionstoimprovetheperformanceoftheneuralnetonthistrial(seethediscussionbelowonhowtodothis).

·Continuethistraininguntiltheaccuracyrateoftheneuralnetisnolongerimproving(i.e.,reachesanasymptote).

KeyDesignDecisionsInthesimpleschemaabove,thedesignerofthisneural-netalgorithmneedstodetermineattheoutset:

·Whattheinputnumbersrepresent.·Thenumberoflayersofneurons.·Thenumberofneuronsineachlayer.(Eachlayerdoesnotnecessarilyneedtohavethesamenumberofneurons.)

·Thenumberofinputstoeachneuronineachlayer.Thenumberofinputs(i.e.,interneuronalconnections)canalsovaryfromneurontoneuronandfromlayerto

layer.·Theactual"wiring"(i.e.,theconnections).Foreachneuronineachlayer,thisconsistsofalistofotherneurons,theoutputsofwhichconstitutetheinputstothisneuron.Thisrepresentsakeydesignarea.Thereareanumberofpossiblewaystodothis:

(i)Wiretheneuralnetrandomly;or(ii)Useanevolutionaryalgorithm(seebelow)todetermineanoptimalwiring;or(iii)Usethesystemdesigner'sbestjudgmentindeterminingthewiring.

·Theinitialsynapticstrengths(i.e.,weights)ofeachconnection.Thereareanumberofpossiblewaystodothis:

(i)Setthesynapticstrengthstothesamevalue;or(ii)Setthesynapticstrengthstodifferentrandomvalues;or(iii)Useanevolutionaryalgorithmtodetermineanoptimalsetofinitialvalues;or(iv)Usethesystemdesigner'sbestjudgmentindeterminingtheinitialvalues.

·Thefiringthresholdofeachneuron.·Theoutput.Theoutputcanbe:

(i)theoutputsoflayerMofneurons;or(ii)theoutputofasingleoutputneuron,theinputsofwhicharetheoutputsofthe

neuronsinlayerM;or(iii)afunctionof(e.g.,asumof)theoutputsoftheneuronsinlayerM;or(iv)anotherfunctionofneuronoutputsinmultiplelayers.

·Howthesynapticstrengthsofalltheconnectionsareadjustedduringthetrainingofthisneuralnet.Thisisakeydesigndecisionandisthesubjectofagreatdealofresearchanddiscussion.Thereareanumberofpossiblewaystodothis:

(i)Foreachrecognitiontrial,incrementordecrementeachsynapticstrengthbya

(generallysmall)fixedamountsothattheneuralnet'soutputmorecloselymatchesthecorrectanswer.Onewaytodothisistotrybothincrementinganddecrementingandseewhichhasthemoredesirableeffect.Thiscanbetime-consuming,soothermethodsexistformakinglocaldecisionsonwhethertoincrementordecrementeachsynapticstrength.

(ii)Otherstatisticalmethodsexistformodifyingthesynapticstrengthsaftereachrecognitiontrialsothattheperformanceoftheneuralnetonthattrialmorecloselymatchesthecorrectanswer.Notethatneural-nettrainingwillworkeveniftheanswerstothetrainingtrialsarenotallcorrect.Thisallowsusingreal-worldtrainingdatathatmayhaveaninherenterrorrate.Onekeytothesuccessofaneuralnet-basedrecognitionsystemistheamountofdatausedfortraining.Usuallyaverysubstantialamountisneededtoobtainsatisfactoryresults.Justlikehumanstudents,theamountoftimethataneuralnetspendslearningitslessonsisakeyfactorinitsperformance.

VariationsManyvariationsoftheabovearefeasible:

·Therearedifferentwaysofdeterminingthetopology.Inparticular,theinterneuronalwiringcanbeseteitherrandomlyorusinganevolutionaryalgorithm.

·Therearedifferentwaysofsettingtheinitialsynapticstrengths.·Theinputstotheneuronsinlayer,donotnecessarilyneedtocomefromtheoutputsoftheneuronsinlayeri-1.Alternatively,theinputstotheneuronsineachlayercancome

fromanylowerlayeroranylayer.·Therearedifferentwaystodeterminethefinaloutput.·Themethoddescribedaboveresultsinan"allornothing"(1or0)firingcalledanonlinearity.Thereareothernonlinearfunctionsthatcanbeused.Commonlyafunctionisusedthatgoesfrom0to1inarapidbutmoregradualfashion.Also,theoutputscanbenumbersotherthan0and1.

·Thedifferentmethodsforadjustingthesynapticstrengthsduringtrainingrepresentkeydesigndecisions.

Theaboveschemadescribesa"synchronous"neuralnet,inwhicheachrecognitiontrialproceedsbycomputingtheoutputsofeachlayer,startingwithlayer,throughlayerM.Inatrueparallelsystem,inwhicheachneuronisoperatingindependentlyoftheothers,theneuronscanoperate"asynchronously"(thatis,independently).Inanasynchronousapproach,eachneuronisconstantlyscanningitsinputsandfireswheneverthesumofitsweightedinputsexceedsitsthreshold(orwhateveritsoutputfunctionspecifies).

173.Seechapter4foradetaileddiscussionofbrainreverseengineering.Asoneexampleoftheprogression,S.J.Thorpewrites:"Wehavereallyonlyjustbegunwhatwillcertainlybealongtermprojectaimedatreverseengineeringtheprimatevisualsystem.Forthemoment,wehaveonlyexploredsomeverysimplearchitectures,involvingessentiallyjustfeed-forwardarchitecturesinvolvingarelativelysmallnumbersoflayers....Intheyearstocome,wewillstrivetoincorporateasmanyofthecomputationaltricksusedbytheprimateandhumanvisualsystemaspossible.Moretothepoint,itseemsthatbyadoptingthespikingneuronapproach,itwillsoonbepossibletodevelopsophisticatedsystemscapableofsimulatingverylargeneuronalnetworksinrealtime."S.J.Thorpeetal.,"ReverseEngineeringoftheVisualSystemUsingNetworksofSpikingNeurons,"ProceedingsoftheIEEE2000InternationalSymposiumonCircuitsandSystemsIV(IEEEPress),pp.405–8,http://www.sccn.ucsd.edu/~arno/mypapers/thorpe.pdf.

174.T.Schoenaueretal.write:"Overthepastyearsahugediversityofhardwareforartificialneuralnetworks(ANN)hasbeendesigned....Todayonecanchoosefromawiderangeofneuralnetworkhardware.Designsdifferintermsofarchitecturalapproaches,suchasneurochips,acceleratorboardsandmulti-boardneurocomputers,aswellasconcerningthepurposeofthesystem,suchastheANNalgorithm(s)andthesystem'sversatility....Digitalneurohardwarecanbeclassifiedbythe:[sic]systemarchitecture,degreeofparallelism,typicalneuralnetworkpartitionperprocessor,inter-processorcommunicationnetworkandnumericalrepresentation."T.Schoenauer,A.Jahnke,u.Roth,andH.Klar,"DigitalNeurohardware:PrinciplesandPerspectives,"inProc.NeuronaleNetzeinderAnwendung—NeuralNetworksinApplicationsNN'98,Magdeburg,invitedpaper(February1998):101–6,http://bwrc.eecs.berkeley.edu/People/kcamera/neural/papers/schoenauer98digital.pdf.SeealsoYihuaLiao,"NeuralNetworksinHardware:ASurvey"(2001),http://ailab.das.ucdavis.edu/~yihua/research/NNhardware.pdf.

175.Hereisthebasicschemaforagenetic(evolutionary)algorithm.Manyvariationsarepossible,andthedesignerofthesystemneedstoprovidecertaincriticalparametersandmethods,detailedbelow.

THEEVOLUTIONARYALGORITHMCreateNsolution"creatures."Eachonehas:

·Ageneticcode:asequenceofnumbersthatcharacterizeapossiblesolutiontotheproblem.Thenumberscanrepresentcriticalparameters,stepstoasolution,rules,etc.

Foreachgenerationofevolution,dothefollowing:

·DothefollowingforeachoftheNsolutioncreatures:

(i)Applythissolutioncreature'ssolution(asrepresentedbyitsgeneticcode)totheproblem,orsimulatedenvironment.

(ii)Ratethesolution.·PicktheLsolutioncreatureswiththehighestratingstosurviveintothenextgeneration.

·Eliminatethe(N–L)nonsurvivingsolutioncreatures.·Create(N–L)newsolutioncreaturesfromtheLsurvivingsolutioncreaturesby:(i)MakingcopiesoftheLsurvivingcreatures.Introducesmallrandomvariations

intoeachcopy;or(ii)Creatingadditionalsolutioncreaturesbycombiningpartsofthegeneticcode

(using"sexual"reproduction,orotherwisecombiningportionsofthechromosomes)fromtheLsurvivingcreatures;or

(iii)Doingacombinationof(i)and(ii),·Determinewhetherornottocontinueevolving:Improvement=(highestratinginthisgeneration)–(highestratinginthepreviousgeneration).IfImprovement<ImprovementThreshold,thenwe'redone.

·Thesolutioncreaturewiththehighestratingfromthelastgenerationofevolutionhasthebestsolution.Applythesolutiondefinedbyitsgeneticcodetotheproblem.

KeyDesignDecisionsInthesimpleschemaabove,thedesignerneedstodetermineattheoutset:

·Keyparameters:NL

·Improvementthreshold·Whatthenumbersinthegeneticcoderepresentandhowthesolutioniscomputedfromthegeneticcode.

·AmethodfordeterminingtheNsolutioncreaturesinthefirstgeneration.Ingeneral,theseneedonlybe"reasonable"attemptsatasolution.Ifthesefirst-generationsolutionsaretoofarafield,theevolutionaryalgorithmmayhavedifficultyconvergingonagoodsolution.Itisoftenworthwhiletocreatetheinitialsolutioncreaturesinsuchawaythattheyarereasonablydiverse.Thiswillhelppreventtheevolutionaryprocessfromjustfindinga"locally"optimalsolution.

·Howthesolutionsarerated.·Howthesurvivingsolutioncreaturesreproduce.

VariationsManyvariationsoftheabovearefeasible.Forexample:

·Theredoesnotneedtobeafixednumberofsurvivingsolutioncreatures(L)fromeachgeneration.Thesurvivalrule(s)canallowforavariablenumberofsurvivors.

·Theredoesnotneedtobeafixednumberofnewsolutioncreaturescreatedineachgeneration(N–L).Theprocreationrulescanbeindependentofthesizeofthepopulation.Procreationcanberelatedtosurvival,therebyallowingthefittestsolutioncreaturestoprocreatethemost.

·Thedecisionastowhetherornottocontinueevolvingcanbevaried.Itcanconsidermorethanjustthehighest-ratedsolutioncreaturefromthemostrecentgeneration(s).Itcanalsoconsideratrendthatgoesbeyondjustthelasttwogenerations.

176.SamWilliams,"WhenMachinesBreed,"August12,2004,

http://www.salon.com/tech/feature/2004/08/12/evolvable_hardware/index_np.html.177.Hereisthebasicscheme(algorithmdescription)ofrecursivesearch.Manyvariationsare

possible,andthedesignerofthesystemneedstoprovidecertaincriticalparametersandmethods,detailedbelow.

THERECURSIVEALGORITHMDefineafunction(program)"PickBestNextStep."Thefunctionreturnsavalueof"SUCCESS"(we'vesolvedtheproblem)or"FAILURE"(wedidn'tsolveit).IfitreturnswithavalueofSUCCESS,thenthefunctionalsoreturnsthesequenceofstepsthatsolvedtheproblem.PICKBESTNEXTSTEPdoesthefollowing:

·Determineiftheprogramcanescapefromcontinuedrecursionatthispoint.Thisbullet,andthenexttwobulletsdealwiththisescapedecision.First,determineiftheproblemhasnowbeensolved.SincethiscalltoPickBestNextStepprobablycamefromtheprogramcallingitself,wemaynowhaveasatisfactorysolution.Examplesare:

(i)Inthecontextofagame(forexample,chess),thelastmoveallowsustowin

(suchascheckmate).(ii)Inthecontextofsolvingamathematicaltheorem,thelaststepprovesthe

theorem.(iii)Inthecontextofanartisticprogram(forexample,acomputerpoetorcomposer),

thelaststepmatchesthegoalsforthenextwordornote.

Iftheproblemhasbeensatisfactorilysolved,theprogramreturnswithavalueof"SUCCESS"andthesequenceofstepsthatcausedthesuccess.

·Iftheproblemhasnotbeensolved,determineifasolutionisnowhopeless.Examplesare:

(i)Inthecontextofagame(suchaschess),thismovecausesustolose(checkmate

fortheotherside).(ii)Inthecontextofsolvingamathematicaltheorem,thisstepviolatesthetheorem.(iii)Inthecontextofanartisticcreation,thisstepviolatesthegoalsforthenextword

ornote.

Ifthesolutionatthispointhasbeendeemedhopeless,theprogramreturnswithavalueof"FAILURE."

·Iftheproblemhasbeenneithersolvednordeemedhopelessatthispointofrecursiveexpansion,determinewhetherornottheexpansionshouldbeabandonedanyway.Thisisakeyaspectofthedesignandtakesintoconsiderationthelimitedamountofcomputertimewehavetospend.Examplesare:

(i)Inthecontextofagame(suchaschess),thismoveputsoursidesufficiently

"ahead"or"behind."Makingthisdeterminationmaynotbestraightforwardandistheprimarydesigndecision.However,simpleapproaches(suchasaddinguppiecevalues)canstillprovidegoodresults.Iftheprogramdeterminesthatoursideissufficientlyahead,thenPickBestNextStepreturnsinasimilarmannertoadeterminationthatoursidehaswon(thatis,withavalueof"SUCCESS").Iftheprogramdeterminesthatoursideissufficientlybehind,thenPickBestNextStepreturnsinasimilarmannertoadeterminationthatoursidehaslost(thatis,withavalueof"FAILURE").

(ii)Inthecontextofsolvingamathematicaltheorem,thisstepinvolvesdetermining

ifthesequenceofstepsintheproofisunlikelytoyieldaproof.Ifso,thenthispathshouldbeabandoned,andPickBestNextStepreturnsinasimilarmannertoadeterminationthatthisstepviolatesthetheorem(thatis,withavalueof"FAILURE").Thereisno"soft"equivalentofsuccess.Wecan'treturnwithavalueof"SUCCESS"untilwehaveactuallysolvedtheproblem.That'sthenatureofmath.

(iii)Inthecontextofanartisticprogram(suchasacomputerpoetorcomposer),thisstepinvolvesdeterminingifthesequenceofsteps(suchasthewordsinapoem,notesinasong)isunlikelytosatisfythegoalsforthenextstep.Ifso,thenthispathshouldbeabandoned,andPickBestNextStepreturnsinasimilarmannertoadeterminationthatthisstepviolatesthegoalsforthenextstep(thatis,withavalueof"FAILURE").

·IfPickBestNextStephasnotreturned(becausetheprogramhasneitherdeterminedsuccessnorfailurenormadeadeterminationthatthispathshouldbeabandonedatthispoint),thenwehavenotescapedfromcontinuedrecursiveexpansion.Inthiscase,wenowgeneratealistofallpossiblenextstepsatthispoint.Thisiswheretheprecisestatementoftheproblemcomesin:

(i)Inthecontextofagame(suchaschess),thisinvolvesgeneratingallpossible

movesfor"our"sideforthecurrentstateoftheboard.Thisinvolvesastraightforwardcodificationoftherulesofthegame.

(ii)Inthecontextoffindingaproofforamathematicaltheorem,thisinvolveslistingthepossibleaxiomsorpreviouslyprovedtheoremsthatcanbeappliedatthispointinthesolution.

(iii)Inthecontextofacyberneticartprogram,thisinvolveslistingthepossiblewords/notes/linesegmentsthatcouldbeusedatthispoint.

Foreachsuchpossiblenextstep:

(i)Createthehypotheticalsituationthatwouldexistifthisstepwereimplemented.

Inagame,thismeansthehypotheticalstateoftheboard.Inamathematicalproof,thismeansaddingthisstep(forexample,axiom)totheproof.Inanartprogram,thismeansaddingthisword/note/linesegment.

(ii)NowcallPickBestNextSteptoexaminethishypotheticalsituation.Thisis,ofcourse,wheretherecursioncomesinbecausetheprogramisnowcallingitself.

(iii)IftheabovecalltoPickBestNextStepreturnswithavalueof"SUCCESS,"thenreturnfromthecalltoPickBestNextStep(thatwearenowin)alsowithavalueof"SUCCESS."Otherwiseconsiderthenextpossiblestep.

IfallthepossiblenextstepshavebeenconsideredwithoutfindingastepthatresultedinareturnfromthecalltoPickBestNextStepwithavalueof"SUCCESS,"thenreturnfromthiscalltoPickBestNextStep(thatwearenowin)withavalueof"FAILURE."EndofPICKBESTNEXTSTEPIftheoriginalcalltoPickBestNextMovereturnswithavalueof"SUCCESS,"itwillalsoreturnthecorrectsequenceofsteps:

(i)Inthecontextofagame,thefirststepinthissequenceisthenextmoveyou

shouldmake.(ii)Inthecontextofamathematicalproof,thefullsequenceofstepsistheproof.(iii)Inthecontextofacyberneticartprogram,thesequenceofstepsisyourworkof

art.

IftheoriginalcalltoPickBestNextStepreturnswithavalueof"FAILURE,"thenyouneedtogobacktothedrawingboard.

KeyDesignDecisionsInthesimpleschemaabove,thedesigneroftherecursivealgorithmneedstodeterminethefollowingattheoutset:

·ThekeytoarecursivealgorithmisthedeterminationinPickBestNextStepofwhentoabandontherecursiveexpansion.Thisiseasywhentheprogramhasachievedclearsuccess(suchascheckmateinchessortherequisitesolutioninamathorcombinatorialproblem)orclearfailure.Itismoredifficultwhenaclearwinorlosshasnotyetbeenachieved.Abandoningalineofinquirybeforeawell-definedoutcomeisnecessarybecauseotherwisetheprogrammightrunforbillionsofyears(oratleastuntilthewarrantyonyourcomputerrunsout).

·Theotherprimaryrequirementfortherecursivealgorithmisastraightforwardcodificationoftheproblem.Inagamelikechess,that'seasy.Butinothersituations,acleardefinitionoftheproblemisnotalwayssoeasytocomeby.

178.SeeKurzweilCyberArt,http://www.KurzweilCyberArt.com.forfurtherdescriptionofRay

Kurzweil'sCyberneticPoetandtodownloadafreecopyoftheprogram.SeeU.S.PatentNo.6,647,395,"PoetPersonalities,"inventors:RayKurzweilandJohnKeklak.Abstract:"Amethodofgeneratingapoetpersonalityincludingreadingpoems,eachofthepoemscontainingtext,generatinganalysismodels,eachoftheanalysismodelsrepresentingoneofthepoemsandstoringtheanalysismodelsinapersonalitydatastructure.Thepersonalitydatastructurefurtherincludesweights,eachoftheweightsassociatedwitheachoftheanalysismodels.Theweightsincludeintegervalues."

179.BenGoertzel:TheStructureofIntelligence(NewYork:Springer-Verlag,1993);TheEvolvingMind(GordonandBreach,1993);ChaoticLogic(Plenum,1994);FromComplexitytoCreativity(Plenum,1997).ForalinktoBenGoertzel'sbooksandessays,seehttp://www.goertzel.orglwork.html.

180.KurzweilAI.net(http://www.KurzweilAI.net)provideshundredsofarticlesbyonehundred"bigthinkers"andotherfeatureson"acceleratingintelligence."Thesiteoffersafreedailyorweeklynewsletteronthelatestdevelopmentsintheareascoveredbythisbook.Tosubscribe,enteryoure-mailaddress(whichismaintainedinstrictconfidenceandisnotsharedwithanyone)onthehomepage.

181.JohnGosney,BusinessCommunicationsCompany,"ArtificialIntelligence:BurgeoningApplicationsinIndustry,"June2003,http://www.bccresearch.com/comm/G275.html.

182.KathleenMelymuka,"GoodMorning,Dave...,"Computerworld;November11,2002,http://www.computerworld.com/industrytopics/defense/story/0,10801,75728,00.html.

183.JTRSTechnologyAwarenessBulletin,August2004,http://jtrs.army.mil/sections/technicalinformation/fset_technical.html?tech_aware_2004-8.

184.OtisPort,MichaelArndt,andJohnCarey,"SmartTools,"Spring2003,http://www.businessweek.com/bw50/content/mar2003/a3826072.htm.

185.WadeRoush,"ImmobotsTakeControl:FromPhotocopierstoSpaceProbes,MachinesInjectedwithRoboticSelf-AwarenessAreReliableProblemSolvers,"TechnologyReview(December2002–January2003),http://www.occm.de/roush1202.pdf.

186.JasonLohnquotedinNASAnewsrelease"NASA'Evolutionary'SoftwareAutomaticallyDesignsAntenna,"http://www.nasa.gov/lb/centers/ames/news/releases/2004/04_55AR.html.

187.RobertRoyBritt,"AutomaticAstronomy:NewRoboticTelescopesSeeandThink,"June4,2003,http://www.space.com/businesstechnology/technology/automated_astronomy_030604.html.

188.H.KeithMelton,"SpiesintheDigitalAge,"

http://www.cnn.com/SPECIALS/cold.war/experience/spies/melton.essay.189."UnitedTherapeutics(UT)isabiotechnologycompanyfocusedondevelopingchronic

therapiesforlife-threateningconditionsinthreetherapeuticareas:cardiovascular,oncologyandinfectiousdiseases"(http://www.unither.com).KurzweilTechnologiesisworkingwithUTtodeveloppatternrecognition-basedanalysisfromeither"Holter"monitoring(twenty-four-hourrecordings)or"Event"monitoring(thirtydaysormore).

190.KristenPhilipkoski,"AMapThatMapsGeneFunctions,"WiredNews,May28,2002,http://www.wired.com/news/medtech/0,1286,52723,00.html.

191.JenniferOuellette,"BioinformaticsMovesintotheMainstream,"TheIndustrialPhysicist(October–November2003),http://www.sciencemasters.com/bioinformatics.pdf.

192.Port,Arndt,andCarey,"SmartTools."193."ProteinPatternsinBloodMayPredictProstateCancerDiagnosis,"NationalCancerInstitute,

October15,2002,http://www.nci.nih.gov/newscenter/ProstateProteomics,reportingonEmanuelF.Petricoinetal.,"SerumProteomicPatternsforDetectionofProstateCancer,"JournaloftheNationalCancerInstitute94(2002):1576–78.

194.CharleneLaino,"NewBloodTestSpotsCancer,"December13,2002,http://my.webmd.com/content/Article/56/65831.htm;EmanuelF.PetricoinIIIetal.,"UseofProteomicPatternsinSerumtoIdentifyOvarianCancer,"Lancet359.9306(February16,2002):572–77.

195.ForinformationofTriPath'sFocalPoint,see"MakeaDiagnosis,"Wired,October2003,http://www.wired.com/wired/archive/10.03/everywhere.html?pg=5.MarkHagland,"Doctors'Orders,"January2003,http://www.healthcare-informatics.com/issues/2003/01_03/cpoe.htm.

196.RossD.Kingetal.,"FunctionalGenomicHypothesisGenerationandExperimentationbyaRobotScientist,"Nature427(January15,2004):247–52.

197.Port,Arndt,andCarey,"SmartTools."198."FutureRouteReleasesAI-BasedFraudDetectionProduct,"August18,2004,

http://www.finextra.com/fullstory.asp?id=12365.199.JohnHackett,"ComputersAreLearningtheBusiness,"CollectionsWorld,April24,2001,

http://www.creditcollectionsworld.com/news/042401_2.htm.200."InnovativeUseofArtificialIntelligence,MonitoringNASDAQforPotentialInsiderTrading

andFraud,"AAAIpressrelease,July30,2003,http://www.aaai.org/Pressroom/Releases/release-03-0730.html.

201."AdaptiveLearning,FlytheBrainySkies,"WiredNews,March2002,http://www.wired.com/wired/rchive/10.03/everywhere.html?pg=2.

202."IntroductiontoArtificialIntelligence,"EL629,MaxwellAirForceBase,AirUniversityLibrarycourse,http://www.au.af.mil/au/aul/school/acsc/ai02.htm.SamWilliams,"Computer,HealThyself,"Salon.com,July12,2004,http://www.salon.com/tech/feature/2004/07/12/self_healing_computing/index_np.html.

203.Seehttp://www.Seegrid.com.Disclosure:TheauthorisaninvestorinSeegridandamemberofitsboardofdirectors.

204.NoHandsAcrossAmericaWebsite,http://cart.frc.ri.cmu.edu/users/hpm/project.archive/reference.file/nhaa.html,and"CarnegieMellonResearchersWillProveAutonomousDrivingTechnologiesDuringa3,000Mile,Hands-off-the-WheelTripfromPittsburghtoSanDiego,"CarnegieMellonpressrelease,http://www-2.cs.cmu.edu/afs/cs/user/tjochem/www/nhaa/official_press_release.html;RobertJ.Derocher,"AlmostHuman,"September2001,http://www.insight-mag.com/insight/01/09/col-2-pt-1-ClickCulture.htm.

205."SearchandRescueRobots,"AssociatedPress,September3,2004,http://www.smh.com.au/articles/2004/09/02/1093939058792.html?oneclick=true.

206."FromFactoidstoFacts,"Economist,August26,2004,http://www.economist.com/science/displayStory.cfm?story_id=3127462.

207.JoeMcCool,"VoiceRecognition,ItPaystoTalk,"May2003,

http://www.bcs.org/BCS/Products/Publications/JournalsAndMagazines/ComputerBulletin/OnlineArchive/may03/voicerecognition.htm.208.JohnGartner,"FinallyaCarThatTalksBack,"WiredNews,September2,2004,

http://www.wired.com/news/autotech/0,2554,64809,00.html?tw=wn_14techhead.209."ComputerLanguageTranslationSystemRomancestheRosettaStone,"InformationSciences

Institute,USCSchoolofEngineering(July24,2003),http://www.usc.edu/isinews/stories/102.html.

210.TorstenReilquotedinStevenJohnson,"DarwininaBox,"Discover24.8(August2003),http://www.discover.com/issues/aug-03/departments/feattech/.

211."LetSoftwareCatchtheGameforYou,"July3,2004,http://www.newscientist.com/news/news.jsp?id=ns99996097.

212.MichelleDelio,"BreedingRaceCarstoWin,"WiredNews,June18,2004,http://www.wired.com/news/autotech/0,2554,63900,00.html.

213.MarvinMinsky,TheSocietyofMind(NewYork:Simon&Schuster,1988).214.HansMoravec,"WhenWillComputerHardwareMatchtheHumanBrain?"Journalof

EvolutionandTechnology1(1998).215.RayKurzweil,TheAgeofSpiritualMachines(NewYork:Viking,1999),p.156.216.Seechapter2,notes22and23,ontheInternationalTechnologyRoadmapfor

Semiconductors.217."TheFirstTuringTest,"http://www.loebner.net/Prizef/loebner-prize.html.218.DouglasR.Hofstadter,"ACoffeehouseConversationontheTuringTest,"May1981,

includedinRayKurzweil,TheAgeofIntelligentMachines(Cambridge,Mass.:MITPress,1990),pp.80–102,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0318.html.

219.RayKurzweil,"WhyIThinkIWillWin,"andMitchKapor,"WhyIThinkIWillWin,"rules:http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0373.html;Kapor:http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0412.html;Kurzweil:http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0374.html;Kurzweil"finalword":http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0413.html.

220.EdwardA.Feigenbaum,"SomeChallengesandGrandChallengesforComputationalIntelligence,"JournaloftheAssociationforComputingMachinery50(January2003):32–40.

221.Accordingtotheserialendosymbiosistheoryofeukaryoticevolution,theancestorsofmitochondria(thestructuresincellsthatproduceenergyandhavetheirowngeneticcodecomprisingthirteengenesinhumans)wereoriginallyindependentbacteria(thatis,notpartofanothercell)similartotheDaptobacterbacteriaoftoday."SerialEndosymbiosisTheory,"http://encyclopedia.thefreedictionary.com/Serial%20endosymbiosis%20theory.

ChapterSix:TheImpact...1.Donovan,"SeasonoftheWitch,"SunshineSuperman(1966).2.Reasonsforthereductioninfarmworkforceincludethemechanizationthatlessenedtheneed

foranimalandhumanlabor,theeconomicopportunitiesthatwerecreatedinurbanareasduringWorldWarII,andthedevelopmentofintensivefarmingtechniquesthatrequiredlesslandforcomparableyields.U.S.DepartmentofAgriculture,NationalAgriculturalStatisticsService,TrendsinU.S.Agriculture,http://www.usda.gov/nass/pubs/trends/farmpopulation.htm.Computer-assistedproduction,just-in-timeproduction(whichresultsinlowerinventory),andoffshoringmanufacturingtoreducecostsaresomeofthemethodsthathavecontributedtothelossoffactoryjobs.SeeU.S.DepartmentofLabor,Futurework:TrendsandChallengesofWorkinthe21stCentury,http://www.dol.gov/asp/programs/history/herman/reports/futurework/report.htm.

3.Forexample,seeNatashaVita-More,"TheNew[Human]GenrePrimo[First]Posthuman,"paperdeliveredatCiber@RTConference,Bilbao,Spain,April2004,http://www.natasha.cc/paper.htm.

4.RashidBashirsummarizesin2004:

Muchprogresshasalsobeenmadeintherapeuticmicro-andnanotechnology....Somespecificexamplesinclude(i)silicon-basedimplantabledevicesthatcanbeelectricallyactuatedtoopenanorificefromwhichpreloadeddrugscanbereleased,(ii)silicondevicesfunctionalizedwithelectricallyactuatedpolymerswhichcanactasavalveormuscletoreleasepreloadeddrugs,(iii)silicon-basedmicro-capsuleswithnano-porousmembranesforthereleaseofinsulin,(iv)allpolymer(orhydrogel)particleswhichcanbepreloadedwithdrugsandthenforcedtoexpanduponexposuretospecificenvironmentalconditionssuchaschangeinpHandreleasetheloadeddrug,(v)metalnano-particlescoatedwithrecognitionproteins,wheretheparticlescanbeheatedwithexternalopticalenergyandcanlocallyheatanddamageunwantedcellsandtissue,etc.

R.Bashir,"BioMEMS:State-of-the-ArtinDetection,OpportunitiesandProspects,"AdvancedDrugDeliveryReviews56.11(September22,2004):1565–86.Reprintavailableathttps://engineering.purdue.edu/LIBNA/pdf/publications/BioMEMS%20review%20ADDR%20final.pdf.SeealsoRichardGraysonetal.,"ABioMEMSReview:MEMSTechnologyforPhysiologicallyIntegratedDevices,"IEEEProceedings92(2004):6–21.

5.ForactivitiesoftheInternationalSocietyforBioMEMSandBiomedicalNanotechnology,seehttp://www.bme.ohio-state.edu/isb.BioMEMSconferencesarealsolistedontheSPIEWebsite,http://www.spie.org/Conferences.

6.Researchersusedagoldnanoparticletomonitorbloodsugarindiabetics.Y.Xiaoetal.,"'PluggingintoEnzymes':NanowiringofRedoxEnzymesbyaGoldNanoparticle,"Science299.5614(March21,2003):1877–81.AlsoseeT.A.Desaietal.,"AbstractNanoporousMicrosystemsforIsletCellReplacement,"AdvancedDrugDeliveryReviews56.11(September22,2004):1661–73.

7.A.Grayson,etal.,"Multi-pulseDrugDeliveryfromaResorbablePolymericMicrochipDevice,"NatureMaterials2(2003):767–72.

8.Q.BaiandK.D.Wise,"Single-UnitNeuralRecordingwithActiveMicroelectrodeArrays,"IEEETransactionsonBiomedicalEngineering48.8(August2001):911–20.SeethediscussionofWise'sworkinJ.DeGaspari,"Tiny,Tuned,andUnattached,"MechanicalEngineering(July2001),http://www.memagazine.org/backissues/july01/features/tinytune/tinytune.html;K.D.Wise,"TheComingRevolutioninWirelessIntegratedMicroSystems,"DigestInternationalSensorConference2001(InvitedPlenary),Seoul,October2001.Onlineversion(January13,2004):http://www.stanford.edu/class/ee392s/Stanford392S-kw.pdf.

9."'Microbots'HuntDownDisease,"BBCNews,June13,2001,http://news.bbc.co.uk/l/hi/health/1386440.stm.Themicromachinesarebasedoncylindricalmagnets;seeK.Ishiyama,M.Sendoh,andK.I.Arai,"MagneticMicromachinesforMedicalApplications,"JournalofMagnetismandMagneticMaterials242–45,part1(April2002):41–46.

10.SeeSandiaNationalLaboratoriespressrelease,"Pac-Man-LikeMicrostructureInteractswithRedBloodCells,"August15,2001,http://www.sandia.gov/medialNewsRel/NR2001/gobbler.htm.Foranindustrytradearticleinresponse,seeD.Wilson,"MicroteethHaveaBigBite,"August17,2001,http://www.e4engineering.com/item.asp?ch=e4_home&type=Features&id=42543.

11.SeeFreitas'sbooksNanomedicine,vol.1,BasicCapabilities(Georgetown,Tex.:LandesBioscience,1999),andNanomedicine,vol.2A,Biocompatibility(Georgetown,Tex.:LandesBioscience,2003),bothfreelyavailableonlineathttp://www.nanomedicine.com.AlsoseetheForesightInstitute's"Nanomedicine"pagebyRobertFreitas,whichlistshiscurrenttechnicalworks(http://www.foresight.org/Nanomedicine/index.html#MedNanoBots).

12.RobertA.FreitasJr.,"ExploratoryDesigninMedicalNanotechnology:AMechanicalArtificialRedCell,"ArtificialCells,BloodSubstitutes,andImmobilizationBiotechnology26

(1998):411–30,http://www.foresight.org/Nanomedicine/Respirocytes.html..13.RobertA.FreitasJr.,"Clottocytes:ArtificialMechanicalPlatelets,"ForesightUpdateno.41,

June30,2000,pp.9–11,http://www.imm.org/Reports/Rep018.html.14.RobertA.FreitasJr.,"Microbivores:ArtificialMechanicalPhagocytes,"ForesightUpdate

no.44,March31,2001,pp.11–13,http://www.imm.org/Reports/Rep025.htmlorhttp://www.KurzweilAI.net/meme/frame.html?main==/articles/art0453.html.

15.RobertA.FreitasJr.,"TheVasculoidPersonalAppliance,"ForesightUpdateno.48,March31,2002,pp.10–12,http://www.imm.org/Reports/Rep031.html;fullpaper:RobertA.FreitasJr.andChristopherJ.Phoenix,"Vasculoid:APersonalNanomedicalAppliancetoReplaceHumanBlood,"JournalofEvolutionandTechnology11(April2002),http://www.jetpress.org/volume11/vasculoid.html.

16.CarloMontemagnoandGeorgeBachand,"ConstructingNanomechanicalDevicesPoweredbyBiomolecularMotors,"Nanotechnology10(September1999):225–31;"BiofuelCellRunsonMetabolicEnergytoPowerMedicalImplants,"Natureonline,Nov.12,2002,http://www.nature.com/news/2002/021111/full/021111-1.html,reportingonN.Mano,F.Mao,andA.Heller,"AMiniatureBiofuelCellOperatinginaPhysiologicalBuffer,"JournaloftheAmericanChemicalSociety124(2002):12962–63;CarloMontemagnoetal.,"Self-AssembledMicrodevicesDrivenbyMuscle,"NatureMaterials4.2(February2005):180–84,publishedelectronically(January16,2005).

17.SeetheLawrenceLivermoreNationalLaboratoryWebsite(http://www.llnl.gov)forupdatedinformationaboutthisinitiative,alongwiththeMedtronicMiniMedWebsite,http://www.minimed.com/corpinfo/index.shtml.

18."Directbrain-to-braincommunication...seem[s]morelikethestuffofHollywoodmoviesthanofgovernmentreports—buttheseareamongtheadvancesforecastinarecentreportbytheU.S.NationalScienceFoundationandDepartmentofCommerce."G.Brumfiel,"FuturistsPredictBodySwapsforPlanetHops,"Nature418(July25,2002):359.Deepbrainstimulation,bywhichelectriccurrentfromimplantedelectrodesinfluencesbrainfunction,isanFDA-approvedneuralimplantforParkinson'sdiseaseandisbeingtestedforotherneurologicaldisorders.SeeAIAbbott,"BrainImplantsShowPromiseAgainstObsessiveDisorder,"Nature419(October17,2002):658,andB.Nuttinetal.,"ElectricalStimulationinAnteriorLimbsofInternalCapsulesinPatientswithObsessive-CompulsiveDisorder,"Lancet354.9189(October30,1999):1526.

19.SeetheRetinalImplantProjectWebsite(http://www.bostonretinalimplant.org),whichcontainsarangeofresourcesincludingrecentpapers.Onesuchpaperis:R.J.Jensenetal.,"ThresholdsforActivationofRabbitRetinalGanglionCellswithanUltrafine,ExtracellularMicroelectrode,"InvestigativeOphthalmalogyandVisualScience44.8(August2003):3533–43.

20.TheFDAapprovedtheMedtronicimplantforthispurposein1997foronlyonesideofthebrain;itwasapprovedforbothsidesofthebrainonJanuary14,2002.S.Snider,"FDAApprovesExpandedUseofBrainImplantforParkinson'sDisease,"U.S.FoodandDrugAdministration,FDATalkPaper,January14,2002,http://www.fda.gov/bbs/topics/ANSWERS/2002/ANS01130.html.Themostrecentversionsprovideforsoftwareupgradesfromoutsidethepatient.

21.Medtronicalsomakesanimplantforcerebralpalsy.SeeS.Hart,"BrainImplantQuellsTremors,"ABCNews,December23,1997,http://nasw.org/users/hart/subhtml/abcnews.html.AlsoseetheMedtronicWebsite,http://www.medtronic.com.

22.GuntherZeckandPeterFromherz,"NoninvasiveNeuroelectronicInterfacingwithSynapticallyConnectedSnailNeuronsImmobilizedonaSemiconductorChip,"ProceedingsoftheNationalAcademyofSciences98.18(August28,2001):10457–62.

23.SeeR.ColinJohnson,"ScientistsActivateNeuronswithQuantumDots,"EETimes,December4,2001,http://www.eetimes.com/story/OEG20011204S0068.Quantumdotscanalsobeusedforimaging;seeM.Dahanetal.,"DiffusionDynamicsofGlycineReceptorsRevealedbySingle-QuantumDotTracking,"Science302.5644(October17,2003):442–45;J.K.Jaiswal

andS.M.Simon,"PotentialsandPitfallsofFluorescentQuantumDotsforBiologicalImaging,"TrendsinCellBiology14.9(September2004):497–504.

24.S.Shohametal.,"Motor-CorticalActivityinTetraplegics,"Nature413.6858(October25,2001):793.FortheUniversityofUtahnewsrelease,see"AnEarlyStepTowardHelpingtheParalyzedWalk,"October24,2001,http://www.utah.edu/news/releases/01/oct/spinal.html.

25.StephenHawking'sremarks,whichweremistranslatedbyFocus,werequotedinNickPatonWalsh,"AlterOurDNAorRobotsWillTakeOver,WarnsHawking,"Observer,September2,2001,http://observer.guardian.co.uk/uk_news/story/0,6903,545653,00.html.ThewidelyreportedmistranslationimpliedthatHawkingwaswarningagainstdevelopingsmarter-than-humanmachineintelligence.Infact,hewasadvocatingthatwehastentocloselinksbetweenbiologicalandnonbiologicalintelligence.HawkingprovidedtheexactquotestoKurzweilAI.net("HawkingMisquotedonComputersTakingOver,"September13,2001,http://www.KurzweilAI.net/news/frame.html?main=news_single.html?id%3D495).

26.Seenote34inchapter1.27.Oneexample,NomadforMilitaryApplications,hasbeenproducedbyMicrovision,a

companybasedinBothell,Washington.Seehttp://www.microvision.com/nomadmilitary/index.html.

28.OlgaKharif,"YourLapelIsRinging,"BusinessWeek,June21,2004.29.LailaWeir,"High-TechHearingBypassesEars,"WiredNews,September16,2004,

http://www.wired.com/news/technology/0,1282,64963,00.html?tw=wn_tophead_4.30.HypersonicSoundtechnology,http://www.atcsd.com/tl_hss.html;AudioSpotlight,

http://www.holosonics.com/technology.html.31.PhillipF.ScheweandBenStein,AmericanInstituteofPhysicsBulletinofPhysicsNews236

(August7,1995),http://www.aip.org/enews/physnews/1995/physnews.236.htm.SeealsoR.WeisandP.Fromherz,"FrequencyDependentSignal-TransferinNeuron-Transistors,"PhysicalReviewE55(1997):877–89.

32.Seenote18above.AlsoseeJ.O.Winteretal.,"RecognitionMoleculeDirectedInterfacingBetweenSemiconductorQuantumDotsandNerveCells,"AdvancedMaterials13(November2001):1673–77;I.WillnerandB.Willner,"BiomaterialsIntegratedwithElectronicElements:EnRoutetoBioelectronics,"TrendsinBiotechnology19(June2001):222–30;DeborahA.Fitzgerald,"BridgingtheGapwithBioelectronics,"Scientist16.6(March18,2002):38.

33.RobertFreitasprovidesananalysisofthisscenario:RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section7.4.5.4,"CellMessageModification"(Georgetown,Tex.:LandesBioscience,1999),pp.194-96,http://www.nanomedicine.com/NMI/7.4.5.4.htm#p5,andsection7.4.5.6,"OutmessagingtoNeurons,"pp.196–97,http://www.nanomedicine.comINMI/7.4.5.6.htm#p2.

34.FordescriptionsoftheRamonaproject,includingvideosofthevirtual-realitypresentationattheTEDconferenceandabehind-the-scenes"MakingofRamona"video,see"AllAboutRamona,"http://www.KurzweilAI.net/meme/frame.html?m=9.

35.I.Friedetal.,"ElectricCurrentStimulatesLaughter,"Nature391.6668(February12,1998):650.SeeRayKurzweil,TheAgeofSpiritualMachines(NewYork:Viking,1999).

36.RobertA.FreitasJr.,Nanomedicine,vol.1,BasicCapabilities,section7.3,"CommunicationNetworks"(Georgetown,Tex.:LandesBioscience,1999),pp.186–88,http://www.nanomedicine.com/NMI/7.3.htm.

37.AllenKurzweil,TheGrandComplication:ANovel(NewYork:Hyperion,2002);AllenKurzweil,ACaseofCuriosities(NewYork:HarvestBooks,2001).AllenKurzweilismyfirstcousin.

38.AsquotedinAubreydeGrey,"EngineeringNegligibleSenescence:RationalDesignofFeasible,ComprehensiveRejuvenationBiotechnology,"KronosInstituteSeminarSeries,February8,2002.PowerPointpresentationavailableathttp://www.gen.cam.ac.uk/sens/sensov.ppt.

39.RobertA.FreitasJr.,"DeathIsanOutrage!"presentationatthefifthAlcorConferenceon

ExtremeLifeExtension,NewportBeach,Calif.,November16,2002,http://www.rfreitas.com/Nano/DeathIsAnOutrage.htm,publishedonKurzweilAI.netJanuary9,2003:http://www.KurzweilAI.net/articles/art0536.html.

40.Cro-magnon,"30yearsorless,oftenmuchless...":http://anthro.palomar.edu/homo2/sapiens_culture.htm.Egypt:JacJ.JanssenquotedinBrettPalmer,"PlayingtheNumbersGame,"inSkepticalReview,publishedonlineMay5,2004,athttp://www.theskepticalreview.com/palmer/numbers.html.Europe1400:GregoryClark,TheConquestofNature:ABriefEconomicHistoryoftheWorld(PrincetonUniversityPress,forthcoming,2005),chapter5,"MortalityintheMalthusianEra,"http://www.econ.ucdavis.edu/faculty/gclark/GlobalHistory/Global%20History-5.pdf.1800:JamesRiley,RisingLifeExpectancy:AGlobalHistory(Cambridge,U.K.:CambridgeUniversityPress,2001),pp.32–33.1900:http://www.cdc.gov/nchs/datalhus/tables/2003/03hus027.pdf.

41.ThemuseumwasoriginallylocatedinBostonandisnowinMountainView,Calif.(http://www.computerhistory.org).

42.LymanandKahleonlong-termstorage:"Whilegoodpaperlasts500years,computertapeslast10.Whilethereareactiveorganizationstomakecopies,wewillkeepourinformationsafe,wedonothaveaneffectivemechanismtomake500yearcopiesofdigitalmaterials...."PeterLymanandBrewsterKahle,"ArchivingDigitalCulturalArtifacts:OrganizinganAgendaforAction,"D-LibMagazine,July–August1998.StewartBrandwrites:"Behindeveryhotnewworkingcomputerisatrailofbodiesofextinctcomputers,extinctstoragemedia,extinctapplications,extinctfiles.SciencefictionwriterBruceSterlingreferstoourtimeas'theGoldenAgeofdeadmedia,mostofthemwiththeworkinglifespanofapackofTwinkles,"StewartBrand,"WrittenontheWind,"CivilizationMagazine,November1998("01998"inLongNowterminology),availableonlineathttp://www.longnow.org/10klibrary/library.htm.

43.DARPA'sInformationProcessingTechnologyOffice'sprojectinthisveiniscalledLifeLog,http://www.darpa.mil/ipto/Programs/lifelog;seealsoNoahShachtman,"ASpyMachineofDARPA'sDreams,"WiredNews,May20,2003,http://www.wired.com/news/business/0,1367,58909,00.html;GordonBell'sproject(forMicrosoft)isMyLifeBits,http://research.microsoft.com/research/barc/MediaPresence/MyLifeBits.aspx;fortheLongNowFoundation,seehttp://longnow.org.

44.BergeronisassistantprofessorofanesthesiologyatHarvardMedicalSchoolandtheauthorofsuchbooksasBioinformaticsComputing,BiotechIndustry:AGlobal,Economic,andFinancingOverview,andTheWirelessWebandHealthcare.

45.TheLongNowFoundationisdevelopingonepossiblesolution:theRosettaDisk,whichwillcontainextensivearchivesoftextinlanguagesthatmaybelostinthefarfuture.Theyplantouseauniquestoragetechnologybasedonatwo-inchnickeldiskthatcanstoreupto350,000pagesperdisk,withanestimatedlifeexpectancyof2,000to10,000years.SeetheLongNowFoundation,LibraryIdeas,http://longnow.org/10klibrary/10kLibConference.htm.

46.JohnA.Parmentola,"ParadigmShiftingCapabilitiesforArmyTransformation,"invitedpaperpresentedattheSPIEEuropeanSymposiumonOptics/PhotonicsinSecurityandDefence,October25–28,2004;availableelectronicallyatBridge34.3(Fall2004),http://www.nae.edu/NAE/bridgecom.nsf/weblinks/MKEZ-65RLTA?OpenDocument.

47.FredBayles,"High-techProjectAimstoMakeSuper-soldiers,"USAToday,May23,2003,http://www.usatoday.com/news/nation/2003-05-22-nanotech-usat_x.htm;seetheInstituteforSoldierNanotechnologiesWebsite,http://web.mit.edu/isn;SarahPutnam,"ResearchersToutOpportunitiesinNanotech,"MITNewsOffice,October9,2002,http://web.mit.edu/newsoffice/2002/cdc-nanotech-1009.html.

48.RonSchafer,"RoboticstoPlayMajorRoleinFutureWarfighting,"http://www.jfcom.mil/newslink/storyarchive/2003/pa072903.htm;Dr.RussellRichards,

"UnmannedSystems:ABigPlayerforFutureForces?"UnmannedEffectsWorkshopattheAppliedPhysicsLaboratory,JohnsHopkinsUniversity,Baltimore,July29–August1,2003.

49.JohnRhea,"NASARobotinFormofSnakePlannedtoPenetrateInaccessibleAreas,"MilitaryandAerospaceElectronics,November2000,http://mae.pennnet.com/Articles/Article_Display.cfm?Section=Archives&Subsection=Display&ARTICLE_ID=86890.

50.LakshmiSandhana,"TheDroneArmiesAreComing,"WiredNews,August30,2002,http://www.wired.com/news/technology/0,1282,54728,00.html.SeealsoMarioGerla,KaixinXu,andAllenMoshfegh,"Minuteman:ForwardProjectionofUnmannedAgentsUsingtheAirborneInternet,"IEEEAerospaceConference2002,BigSky,Mont.,March2002:http://www.cs.ucla.edu/NRL/wireless/uploads/mgerla_aerospace02.pdf.

51.JamesKennedyandRussellC.Eberhart,withYuhuiShi,SwarmIntelligence(SanFrancisco:MorganKaufmann,2001),http://www.swarmintelligence.org/SIBook/SI.php.

52.WillKnight,"MilitaryRobotstoGetSwarmIntelligence,"April25,2003,http://www.newscientist.com/news/news.jsp?id=ns99993661.

53.Ibid.54.S.R.Whiteetal.,"AutonomicHealingofPolymerComposites,"Nature409(February15,

2001):794–97,http://www.autonomic.uiuc.edu/files/NaturePaper.pdf;KristinLeutwyler,"Self-HealingPlastics,"ScientificAmerican.com,February15,2001,http://www.sciam.com/article.cfm?articleID=000B307F-C71A-1C5AB882809EC588ED9F.

55.SueBaker,"PredatorMissileLaunchTestTotallySuccessful,"StrategicAffairs,April1,2001,http://www.stratmag.com/issueApr-l/page02.htm.

56.SeetheOpenCourseWarecourselistathttp://ocw.mit.edu/index.html.57.BrigitteBouissouquotedonMITOpenCourseWare'sadditionalquotespageat

http://ocw.mit.edu/OcwWeb/Global/AboutOCW/additionalquotes.htmandEricBender,"TeachLocally,EducateGlobally,"MITTechnologyReview,June2004,http://www.techreview.com/articles/04/06/bender0604.asp?p=1.

58.KurzweilEducationalSystems(http://www.Kurzweiledu.com)providestheKurzweil3000readingsystemforpeoplewithdyslexia.Itcanreadanybooktotheuserwhilehighlightingwhatisbeingreadonahigh-resolutionimageofthepage.Itincorporatesarangeoffeaturestoimprovethereadingskillsofusers.

59.AsquotedbyNatashaVita-More,"ArterationIdeas,"http://64.233.167.104/search?q=cache:QAnJsLcXHXUJ:www.extropy.com/ideas/journal/previous/1998/02-01.html+Arterati+on+ideas&hl=enandhttp://www.extropy.com/ideas/journal/previous/1998/02-01.html.

60.ChristineBoese,"TheScreen-Age:OurBrainsinourLaptops,"CNN.com,August2,2004.61.ThomasHobbes,Leviathan(1651).62.SethLloydandY.JackNg,"BlackHoleComputers,"ScientificAmerican,November2004.63.AlanM.MacRobert,"TheAllenTelescopeArray:SETI'sNextBigStep,"Sky&Telescope,

April2004,http://skyandtelescope.com/printable/resources/seti/article_256.asp.64.Ibid.65.Ibid.66.C.H.Townes,"AtWhatWavelengthShouldWeSearchforSignalsfromExtraterrestrial

Intelligence?"ProceedingsoftheNationalAcademyofSciencesUSA80(1983):1147–51.S.A.KingsleyinTheSearchforExtraterrestrialIntelligenceintheOpticalSpectrum,vol.2,S.A.KingsleyandG.A.Lemarchand,eds.(1996)Proc.WPIE2704:102–16.

67.N.S.Kardashev,"TransmissionofInformationbyExtraterrestrialCivilizations,"SovietAstronomy8.2(1964):217–20.SummarizedinGuillermoA.Lemarchand,"DetectabilityofExtraterrestrialTechnologicalActivities,"SETIQuest1:1,pp.3–13,http://www.coseti.org/lemarch1.htm.

68.FrankDrakeandDavaSobel,IsAnyoneOutThere?(NewYork:Dell,1994);CarlSaganandFrankDrake,"TheSearchforExtraterrestrialIntelligence,"ScientificAmerican(May1975):

80–89.ADrake-equationcalculatorcanbefoundathttp://www.activemind.com/Mysterious/Topics/SETI/drake_equation.html.

69.ManyofthedescriptionsoftheDrakeequationexpressfLasthefractionoftheplanet'slifeduringwhichradiotransmissiontakesplace,butthisshouldproperlybeexpressedasafractionofthelifeoftheuniverse,aswedon'treallycarehowlongthatplanethasbeenaround;rather,wecareaboutthedurationoftheradiotransmissions.

70.SethShostakprovided"anestimateofbetween10,000andonemillionradiotransmittersinthegalaxy."MarcusChown,"ETFirstContact'Within20Years,'"NewScientist183.2457(July24,2004).Availableonlineathttp://www.newscientist.com/article.ns?id=dn6189.

71.T.L.Wilson,"TheSearchforExtraterrestrialIntelligence,"Nature,February22,2001.72.Mostrecentestimateshavebeenbetweentenandfifteenbillionyears.In2002estimates

basedondatafromtheHubbleSpaceTelescopewerebetweenthirteenandfourteenbillionyears.AstudypublishedbyCaseWesternReserveUniversityscientistLawrenceKraussandDartmouthUniversity'sBrianChaboyerappliedrecentfindingsontheevolutionofstarsandconcludedthattherewasa95percentlevelofconfidencethattheageoftheuniverseisbetween11.2and20billionyears.LawrenceKraussandBrianChaboyer,"Irion,theMilkyWay'sRestlessSwarmsofStars,"Science299(January3,2003):60–62.RecentresearchfromNASAhasnarroweddowntheageoftheuniverseto13.7billionyearsplusorminus200million,http://map.gsfc.nasa.gov/m_mm/mr_age.html.

73.QuotedinEricM.Jones,"'WhereIsEverybody?':AnAccountofFermi'sQuestion,"LosAlamosNationalLaboratories,March1985,http://www.bayarea.net/~kins/AboutMe/Fermi_and_Teller/fermi_question.html.

74.First,considertheestimateof1042cpsfortheultimatecoldlaptop(asinchapter3).Wecanestimatethemassofthesolarsystemasbeingapproximatelyequaltothemassofthesun,whichis2x1030kilograms.Onetwentiethof1percentofthismassis1027kilograms.At1042cpsperkilogram,1027kilogramswouldprovide1069cps.Ifweusetheestimateof1050cpsfortheultimatehotlaptop,weget1077cps.

75.AndersSandberg,"ThePhysicsofInformationProcessingSuperobjects:DailyLifeAmongtheJupiterBrains,"JournalofEvolutionandTechnology5(December22,1999),http://www.jetpress.org/volume5/Brains2.pdf.

76.FreemanJohnDyson,"SearchforArtificialStellarSourcesofInfraredRadiation,"Science131(June3,1960):1667–68.

77.CitedinSandberg,"PhysicsofInformationProcessingSuperobjects."78.Therewere195.5billionunitsofsemiconductorchipsshippedin1994,433.5billionin2004.

JimFeldhan,president,SemicoResearchCorporation,http://www.semico.com.79.RobertFreitashasbeenaleadingadvocateofusingroboticprobes,especiallyself-replicating

ones.SeeRobertA.FreitasJr.,"InterstellarProbes:ANewApproachtoSETI,"J.BritishInterplanet.Soc.33(March1980):95–100,http://www.rfreitas.com/Astro/lnterstellarProbesJBIS1980.htm;RobertA.FreitasJr.,"ASelf-ReproducingInterstellarProbe,"J.BritishInterplanet.Soc.33(July1980):251–64,http://www.rfreitas.com/Astro/ReproJBISJuly1980.htm;FranciscoValdesandRobertA.FreitasJr.,"ComparisonofReproducingandNonreproducingStarprobeStrategiesforGalacticExploration,"J.BritishInterplanet.Soc.33(November1980):402–8,http://www.rfreitas.com/Astro/ComparisonReproNov1980.htm;RobertA.FreitasJr.,"DebunkingtheMythsofInterstellarProbes,"AstroSearch1(July–August1983):8–9,http://www.rfreitas.com/Astro/ProbeMyths1983.htm;RobertA.FreitasJr.,"TheCaseforInterstellarProbes,"J.BritishInterplanet.Soc.36(November1983):490–95,http://www.rfreitas.com/Astro/TheCaseForInterstellarProbes1983.htm.

80.M.Stenneretal.,"TheSpeedofInformationina'Fast-Light'OpticalMedium,"Nature425(October16,2003):695–98.SeealsoRaymondY.Chiaoetal.,"SuperluminalandParelectricEffectsinRubidiumVaporandAmmoniaGas,"QuantumandSemiclassicalOptics7(1995):

279.81.I.Marcikicetal.,"Long-DistanceTeleportationofQubitsatTelecommunication

Wavelengths,"Nature421(January2003):509–13;JohnRoach,"PhysicistsTeleportQuantumBitsoverLongDistance,"NationalGeographicNews,January29,2003;HerbBrody,"QuantumCryptography;in"10EmergingTechnologiesThatWillChangetheWorld,"MITTechnologyReview,February2003;N.Gisinetal.,"QuantumCorrelationswithMovingObservers,"QuantumOptics(December2003):51;QuantumCryptographyexhibit,ITUTelecomWorld2003,Geneva,Switzerland,October1,2003;SoraSong,"TheQuantumLeaper,"Time,March15,2004;MarkBuchanan,"Light'sSpookyConnectionsSetNewDistanceRecord,"NewScientist,June28,1997.

82.CharlesH.LineweaverandTamaraM.Davis,"MisconceptionsAbouttheBigBang,"ScientificAmerican,March2005.

83.A.EinsteinandN.Rosen,"TheParticleProblemintheGeneralTheoryofRelativity,"PhysicalReview48(1935):73.

84.J.A.Wheeler,"Geons,"PhysicalReview97(1955):511–36.85.M.S.Morris,K.S.Thorne,andU.Yurtsever,"Wormholes,TimeMachines,andtheWeak

EnergyCondition,"PhysicalReviewLetters61.13(September26,1988):1446–49;M.S.MorrisandK.S.Thorne,"WormholesinSpacetimeandTheirUseforInterstellarTravel:AToolforTeachingGeneralRelativity,"AmericanJournalofPhysics56.5(1988):395–412.

86.M.Visser,"Wormholes,BabyUniverses,andCausality,"PhysicalReviewD41.4(February15,1990):1116–24.

87.Sandberg,"PhysicsofInformationProcessingSuperobjects."88.DavidHochbergandThomasW.Kephart,"WormholeCosmologyandtheHorizonProblem,"

PhysicalReviewLetters70(1993):2665–68,http://prola.aps.org/abstract/PRL/v70/i18/p2665_1;D.HochbergandM.Visser,"GeometricStructureoftheGenericStaticTransversableWormholeThroat,"PhysicalReviewD56(1997):4745.

89.J.K.Webbetal.,"FurtherEvidenceforCosmologicalEvolutionoftheFineStructureConstant,"PhysicalReviewLetters87.9(August27,2001):091301;"WhenConstantsAreNotConstant,"PhysicsinAction(October2001),http://physicsweb.org/articles/world/14/10/4.

90.JoaoMagueijo,JohnD.Barrow,andHaavardBunesSandvik,"IsIteorIsItc?ExperimentalTestsofVaryingAlpha,"PhysicalLettersB549(2002):284–89.

91.JohnSmart,"AnsweringtheFermiParadox:ExploringtheMechanismsofUniversalTranscension,"http://www.transhumanist.com/Smart-Fermi.htm.Seealsohttp://singuIaritywatch.comandhisbiographyathttp://www.singuIaritywatch.com/bio_johnsmart.html.

92.JamesN.Gardner,Biocosm:TheNewScientificTheoryofEvolution:IntelligentLifeIstheArchitectoftheUniverse(Maui:InnerOcean,2003).

93.LeeSmolinin"Smolinvs.Susskind:TheAnthropicPrinciple,"Edge145,http://www.edge.org/documents/archive/edge145.html;LeeSmolin,"ScientificAlternativestotheAnthropicPrinciple,"http://arxiv.org/abs/hep-th/0407213.

94.Kurzweil,AgeofSpiritualMachines,pp.258–60.95.Gardner,Biocosm.96.S.W.Hawking,"ParticleCreationbyBlackHoles,"CommunicationsinMathematical

Physics43(1975):199–220.97.Theoriginalbetislocatedathttp://www.theory.caltech.edu/people/preskill/info_bet.html.

AlsoseePeterRodgers,"HawkingLosesBlackHoleBet,"PhysicsWorld,August2004,http://physicsweb.org/articles/news/8/7/11.

98.ToarriveatthoseestimatesLloydtooktheobserveddensityofmatter—aboutonehydrogenatompercubicmeter—andcomputedthetotalenergyintheuniverse.DividingthisfigurebythePlanckconstant,hegotabout1090cps.SethLloyd,"UltimatePhysicalLimitstoComputation,"Nature406.6799(August31,2000):1047–54.Electronicversions(version3

datedFebruary14,2000)availableathttp://arxiv.org/abs/quant-ph/9908043(August31,2000).Thefollowinglinkrequiresapaymenttoaccess:http://www.nature.com/cgi-taf/DynaPage.taf?file=/nature/journal/v406/n6799/full/4061047a0_fs.html&content_filetype=PDF.

99.JacobD.Bekenstein,"InformationintheHolographicUniverse:TheoreticalResultsaboutBlackHolesSuggestThattheUniverseCouldBeLikeaGiganticHologram,"ScientificAmerican289.2(August2003):58–65,http://www.sciam.com/article.cfm?articleID=000AF072-4891-1F0A-97AE80A84189EEDF.

ChapterSeven:IchbineinSingularitarian1.InJayW.Richardsetal.,AreWeSpiritualMachines?RayKurzweilvs.theCriticsofStrong

A.I.(Seattle:DiscoveryInstitute,2002),introduction,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0502.html.

2.RayKurzweilandTerryGrossman,M.D.,FantasticVoyage:LiveLongEnoughtoLiveForever(NewYork:RodaleBooks,2004).

3.Ibid.4.Ibid.5.MaxMoreandRayKurzweil,"MaxMoreandRayKurzweilontheSingularity,"February

26,2002,http://www.KurzweilAI.net/articles/art0408.html.6.Ibid.7.Ibid.8.ArthurMiller,AftertheFall(NewYork:Viking,1964).9.FromapaperreadtotheOxfordPhilosophicalSocietyin1959andthenpublishedas"Minds,

MachinesandGödel,"Philosophy36(1961):112–27.ItwasreprintedforthefirstofmanytimesinKennethSayreandFrederickCrosson,eds.,TheModelingofMind(NotreDame:UniversityofNotreDamePress,1963),pp.255–71.

10.MartineRothblatt,"Biocyberethics:ShouldWeStopaCompanyfromUnplugginganIntelligentComputer?"September28,2003,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0594.html(includeslinkstoaWebcastandtranscripts).

11.JaronLanier,"OneHalfofaManifesto,"Edge,http://www.edge.org/3rd_culture/lanier/lanier_index.html;seealsoJaronLanier,"One-HalfofaManifesto,"WiredNews,December2000,http://www.wired.com/wired/archive/8.12/lanier.html.

12.Ibid.13.NorbertWiener,Cybernetics:or,ControlandCommunicationintheAnimalandtheMachine

(Cambridge,Mass.:MITPress,1948).14."HowDoYouPersistWhenYourMoleculesDon't?"ScienceandConsciousnessReview1.1

(June2004),http://www.sci-con.org/articles/20040601.html.15.DavidJ.Chalmers,"FacingUptotheProblemofConsciousness,"JournalofConsciousness

Studies2.3(1995):200–219,http://jamaica.u.arizona.edu/~chalmers/papers/facing.html.16.HustonSmith,TheSacredUnconscious,videotape(TheWisdomFoundation,2001),

availableforsaleathttp://www.fonsvitae.com/sacredhuston.html.17.JerryA.Fodor,RePresentations:PhilosophicalEssaysontheFoundationsofCognitive

Science(Cambridge,Mass.:MITPress,1981).

ChapterEight:TheDeeplyIntertwinedPromiseandPerilofGNR1.BillMcKibben,"HowMuchIsEnough?TheEnvironmentalMovementasaPivotPointin

HumanHistory,"HarvardSeminaronEnvironmentalValues,October18,2000.2.Inthe1960s,theU.S.governmentconductedanexperimentinwhichitaskedthreerecently

graduatedphysicsstudentstobuildanuclearweaponusingonlypubliclyavailableinformation.

Theresultwassuccessful;thethreestudentsbuiltoneinaboutthreeyears(http://www.pimall.com/nais/nl/n.nukes.html).PlansforhowtobuildanatomicbombareavailableontheInternetandhavebeenpublishedinbookformbyanationallaboratory.In2002,theBritishMinistryofDefencereleasedmeasurements,diagrams,andprecisedetailsonbombbuildingtothePublicRecordOffice,sinceremoved(http://news.bbc.co.uk/1/hi/uk/1932702.stm).Notethattheselinksdonotcontainactualplanstobuildatomicweapons.

3."TheJohnStosselSpecial:YouCan'tSayThat!"ABCNews,March23,2000.4.ThereisextensiveinformationontheWeb,includingmilitarymanuals,onhowtobuild

bombs,weapons,andexplosives.Someofthisinformationiserroneous,butaccurateinformationonthesetopicscontinuestobeaccessibledespiteeffortstoremoveit.Congresspassedanamendment(theFeinsteinAmendment,SP419)toaDefenseDepartmentappropriationsbillinJune1997,banningthedisseminationofinstructionsonbuildingbombs.SeeAnneMarieHelmenstine,"HowtoBuildaBomb,"February10,2003,http://chemistry.about.com/library/weekly/aa021003a.htm.InformationontoxicindustrialchemicalsiswidelyavailableontheWebandinlibraries,asareinformationandtoolsforcultivatingbacteriaandvirusesandtechniquesforcreatingcomputervirusesandhackingintocomputersandnetworks.NotethatIdonotprovidespecificexamplesofsuchinformation,sinceitmightbehelpfultodestructiveindividualsandgroups.Irealizethatevenstatingtheavailabilityofsuchinformationhasthispotential,butIfeelthatthebenefitofopendialogueaboutthisissueoutweighsthisconcern.Moreover,theavailabilityofthistypeofinformationhasbeenwidelydiscussedinthemediaandothervenues.

5.RayKurzweil,TheAgeofIntelligentMachines(Cambridge,Mass.:MITPress,1990).6.KenAlibek,Biohazard(NewYork:RandomHouse,1999).7.RayKurzweil,TheAgeofSpiritualMachines(NewYork:Viking,1999).8.BillJoy,"WhytheFutureDoesn'tNeedUs,"Wired,April2000,

http://www.wired.com/wired/archive/8.04/joy.html.9.Handbooksongenesplicing(suchasA.J.Harwood,ed.,BasicDNAandRNAProtocols

[Totowa,N.J.:HumanaPress,1996])alongwithreagentsandkitsthatenablegenesplicingaregenerallyavailable.EvenifaccesstothesematerialswerelimitedintheWest,therearealargenumberofRussiancompaniesthatcouldprovideequivalentmaterials.

10.Foradetailedsummarysiteofthe"DarkWinter"simulation,see"DARKWINTER:ABioterrorismExerciseJune2001":http://www.biohazardnews.net/scen_smallpox.shtml.Forabriefsummary,see:http://www.homelandsecurity.org/darkwinter/index.cfm.

11.RichardPreston,"TheSpecterofaNewandDeadlierSmallpox,"NewYorkTimes,October14,2002,availableathttp://www.ph.ucla.edu/epi/bioter/specterdeadliersmallpox.html.

12.AlfredW.Crosby,America'sForgottenPandemic:TheInfluenzaof1918(NewYork:CambridgeUniversityPress,2003).

13."PowerfromBloodCouldLeadto'HumanBatteries,'"SydneyMorningHerald,August4,2003,http://www.smh.com.au/articles/2003/08/0311059849278131.html.Seenote129inchapter5.SeealsoS.C.Barton,J.Gallaway,andP.Atanassov,"EnzymaticBiofuelCellsforImplantableandMicroscaleDevices,"ChemicalReviews104.10(October2004):4867–86.

14.J.M.Hunthascalculatedthatthereare1.55Î1019kilograms(1022grams)oforganiccarbononEarth.Basedonthisfigure,andassumingthatall"organiccarbon"iscontainedinthebiomass(notethatthebiomassisnotclearlydefined,sowearetakingaconservativelybroadapproach),wecancomputetheapproximatenumberofcarbonatomsasfollows:Averageatomicweightofcarbon(adjustingforisotoperatios)=12.011.Carboninthebiomass=1.55Î1022grams/12.011=1.3Î1021mols.1.3Î1011Î6.02Î1023(Avogadro'snumber)=7.8Î1044carbonatoms.J.M.Hunt,PetroleumGeochemistryandGeology(SanFrancisco:W.H.Freeman,1979).

15.RobertA.FreitasJr.,"TheGrayGooProblem,"March20,2001,

http://www.KurzweilAI.net/articles/art0142.html.16."GrayGooIsaSmallIssue,"BriefingDocument,CenterforResponsibleNanotechnology,

December14,2003,http://crnano.org/BD-Goo.htm;ChrisPhoenixandMikeTreder,"SafeUtilizationofAdvancedNanotechnology,"CenterforResponsibleNanotechnology,January2003,http://crnano.org/safe.htm;K.EricDrexler,EnginesofCreation,chapter11,"EnginesofDestruction"(NewYork:AnchorBooks,1986),pp.171–90,http://www.foresight.org/EOC/EOC_Chaptec11.html;RobertA.FreitasJr.andRalphC.Merkle,KinematicSelf-ReplicatingMachines,section5.11,"ReplicatorsandPublicSafety"(Georgetown,Tex.:LandesBioscience,2004),pp.196–99,http://www.MolecularAssembler.com/KSRM/5.11.htm.andsection6.3.1,"MolecularAssemblersAreTooDangerous,"pp.204–6,http://www.MolecularAssembler.com/KSRM/6.3.1.htm;ForesightInstitute,"MolecularNanotechnologyGuidelines:DraftVersion3.7,"June4,2000,http://www.foresight.org/guidelines/.

17.RobertA.FreitasJr.,"GrayGooProblem"and"SomeLimitstoGlobalEcophagybyBiovorousNanoreplicators,withPublicPolicyRecommendations,"Zyvexpreprint,April2000,section8.4"MaliciousEcophagy"andsection6.0"EcophagicThermalPollutionLimits(ETPL),"http://www.foresight.org/NanoRev/Ecophagy.html.

18.NickD.Bostrom,"ExistentialRisks:AnalyzingHumanExtinctionScenariosandRelatedHazards,"May29,2001,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0194.html.

19.RobertKennedy,13Days(London:Macmillan,1968),p.110.20.InH.Putnam,"ThePlaceofFactsinaWorldofValues,"inD.HuffandO.Prewitt,eds.,The

NatureofthePhysicalUniverse(NewYork:JohnWiley,1979),p.114.21.GrahamAllison,NuclearTerrorism(NewYork:TimesBooks,2004).22.MartinI.Meltzer,"MultipleContactDatesandSARSIncubationPeriods,"Emerging

InfectiousDiseases10.2(February2004),http://www.cdc.gov/ncidod/EID/vol10no2/03-0426-G1.htm.

23.RobertA.FreitasJr.,"Microbivores:ArtificialMechanicalPhagocytesUsingDigestandDischargeProtocol,"Zyvexpreprint,March2001,http://www.rfreitas.com/Nano/Microbivores.htm,and"Microbivores:ArtificialMechanicalPhagocytes,"ForesightUpdateno.44,March31,2001,pp.11–13,http://www.imm.org/Reports/Rep025.html.

24.MaxMore,"TheProactionaryPrinciple,"May2004,http://www.maxmore.com/proactionary.htmandhttp://www.extropy.org/proactionaryprinciple.htm.Moresummarizestheproactionaryprincipleasfollows:

1.People'sfreedomtoinnovatetechnologicallyisvaluabletohumanity.The

burdenofproofthereforebelongstothosewhoproposerestrictivemeasures.Allproposedmeasuresshouldbecloselyscrutinized.

2.Evaluateriskaccordingtoavailablescience,notpopularperception,andallowforcommonreasoningbiases.

3.Giveprecedencetoamelioratingknownandproventhreatstohumanhealthandenvironmentalqualityoveractingagainsthypotheticalrisks.

4.Treattechnologicalrisksonthesamebasisasnaturalrisks;avoidunderweightingnaturalrisksandoverweightinghuman-technologicalrisks.Fullyaccountforthebenefitsoftechnologicaladvances.

5.Estimatethelostopportunitiesofabandoningatechnology,andtakeintoaccountthecostsandrisksofsubstitutingothercredibleoptions,carefullyconsideringwidelydistributedeffectsandfollow-oneffects.

6.Considerrestrictivemeasuresonlyifthepotentialimpactofanactivityhasboth

significantprobabilityandseverity.Insuchcases,iftheactivityalsogeneratesbenefits,discounttheimpactsaccordingtothefeasibilityofadaptingtotheadverseeffects.Ifmeasurestolimittechnologicaladvancedoappearjustified,ensurethattheextentofthosemeasuresisproportionatetotheextentoftheprobableeffects.

7.Whenchoosingamongmeasurestorestricttechnologicalinnovation,prioritizedecisioncriteriaasfollows:Giveprioritytoriskstohumanandotherintelligentlifeoverriskstootherspecies;givenon-lethalthreatstohumanhealthpriorityoverthreatslimitedtotheenvironment(withinreasonablelimits);giveprioritytoimmediatethreatsoverdistantthreats;preferthemeasurewiththehighestexpectationvaluebygivingprioritytomorecertainoverlesscertainthreats,andtoirreversibleorpersistentimpactsovertransientimpacts.

25.MartinRees,OurFinalHour:AScientist'sWarning:HowTerror,Error,andEnvironmental

DisasterThreatenHumankind'sFutureinThisCentury-onEarthandBeyond(NewYork:BasicBooks,2003).

26.ScottShane,DismantlingUtopia:HowInformationEndedtheSovietUnion(Chicago:IvanR.Dee,1994);seealsothereviewbyJamesA.Dornathttp://www.cato.org/pubs/journal/cj16n2-7.html.

27.SeeGeorgeDeWan,"DiaryofaColonialHousewife,"Newsday,2005,foroneaccountofthedifficultyofhumanlifeacoupleofcenturiesago:http://www.newsday.com/community/guide/lihistory/ny-history-hs331a,0,6101197.story.

28.JimOeppenandJamesW.Vaupel,"BrokenLimitstoLifeExpectancy,"Science296.5570(May10,2002):1029–31.

29.SteveBowmanandHelitBarel,WeaponsofMassDestruction:TheTerroristThreat,CongressionalResearchServiceReportforCongress,December8,1999,http://www.cnie.org/nle/crsreports/international/inter-75.pdf.

30.EliezerS.Yudkowsky,"CreatingFriendlyAI1.0,TheAnalysisandDesignofBenevolentGoalArchitectures"(2001),TheSingularityInstitute,http://www.singinst.org/CFAI/;EliezerS.Yudkowsky,"WhatIsFriendlyAI?"May3,2001,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0172.html.

31.TedKaczynski,"TheUnabomber'sManifesto,"May14,2001,http://www.KurzweilAI.netlmeme/frame.html?main=/articles/art0182.html.

32.BillMcKibben,Enough:StayingHumaninanEngineeredAge(NewYork:TimesBooks,2003).

33.Kaczynski,"TheUnabomber'sManifesto."34.ForesightInstituteandIMM,"ForesightGuidelinesonMolecularNanotechnology,"February

21,1999,http://www.foresight.org/guidelines/current.html;ChristinePeterson,"MolecularManufacturing:SocietalImplicationsofAdvancedNanotechnology,"April9,2003,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0557.html;ChrisPhoenixandMikeTreder,"SafeUtilizationofAdvancedNanotechnology,"January28,2003,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0547.html;RobertA.FreitasJr.,"TheGrayGooProblem,"KurzweilAI.net,20March2002,http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0142.html.

35.RobertA.FreitasJr.,privatecommunicationtoRayKurzweil,January2005.FreitasdescribeshisproposalindetailinRobertA.FreitasJr.,"SomeLimitstoGlobalEcophagybyBiovorousNanoreplicators,withPublicPolicyRecommendations."

36.RalphC.Merkle,"SelfReplicatingSystemsandLowCostManufacturing,"1994,http://www.zyvex.com/nanotech/selfRepNATO.html.

37.NeilKingJr.andTedBridis,"FBISystemCovertlySearchesE-mail,"WallStreetJournalOnline(July10,2000),http://zdnet.com.com/2100-11-522071.html?legacy=zdnn.

38.PatrickMoore,"TheBattleforBiotechProgress-GMCropsAreGoodfortheEnvironmentandHumanWelfare,"Greenspirit(February2004),http://www.greenspirit.com/logbook.cfm?

msid=62.39."GMOs:AreThereAnyRisks?"EuropeanCommission(October9,2001),

http://europa.eu.int/comm/research/biosociety/pdf/gmo_press_release.pdf.40.RoryCarroll,"ZambiansStarveAsFoodAidLiesRejected,"Guardian(October17,2002),

http://www.guardian.co.uk/gmdebate/Story/0,2763,813220,00.html.41.LarryThompson,"HumanGeneTherapy:HarshLessons,HighHopes,"FDAConsumer

Magazine(September–October2000),http://www.fda.gov/fdac/features/2000/500_gene.html.42.BillJoy,"WhytheFutureDoesn'tNeedUs."43.TheForesightGuidelines(ForesightInstitute,version4.0,October2004,

http://www.foresight.org/guidelines/current.html)aredesignedtoaddressthepotentialpositiveandnegativeconsequencesofnanotechnology.Theyareintendedtoinformcitizens,companies,andgovernments,andprovidespecificguidelinestoresponsiblydevelopnanotechnology-basedmolecularmanufacturing.TheForesightGuidelineswereinitiallydevelopedattheInstituteWorkshoponMolecularNanotechnologyResearchPolicyGuidelines,sponsoredbytheinstituteandtheInstituteforMolecularManufacturing(IMM),February19–21,1999.ParticipantsincludedJamesBennett,GregBurch,K.EricDrexler,NeilJacobstein,TanyaJones,RalphMerkle,MarkMiller,EdNiehaus,PatParker,ChristinePeterson,GlennReynolds,andPhilippeVanNedervelde.Theguidelineshavebeenupdatedseveraltimes.

44.MartineRothblatt,CEOofUnitedTherapeutics,hasproposedreplacingthismoratoriumwitharegulatoryregimeinwhichanewInternationalXenotransplantationAuthorityinspectsandapprovespathogen-freeherdsofgeneticallyengineeredpigsasacceptablesourcesofxenografts.Rothblatt'ssolutionalsohelpsstampoutroguexenograftsurgeonsbypromisingeachcountrythatjoinstheDCA,andhelpstoenforcetheruleswithinitsborders,afairshareofthepathogenfreexenograftsforitsowncitizenssufferingfromorganfailure.SeeMartineRothblatt,"YourLifeorMine:UsingGeoethicstoResolvetheConflictBetweenPublicandPrivateInterests,"inXenotransplantation(Burlington,Vt.:Ashgate,2004).Disclosure:IamontheboardofdirectorsofUnitedTherapeutics.

45.SeeSingularityInstitute,http://www.singinst.org.Alsoseenote30above.YudkowskyformedtheSingularityInstituteforArtificialIntelligence(SIAI)todevelop"FriendlyAI,"intendedto"createcognitivecontent,designfeatures,andcognitivearchitecturesthatresultinbenevolence"beforenear-humanorbetter-than-humanAlsbecomepossible.SIAIhasdevelopedTheSIAIGuidelinesonFriendlyAI:"FriendlyAI,"http://www.singinst.org/friendly/.BenGoertzelandhisArtificialGeneralIntelligenceResearchInstitutehavealsoexaminedissuesrelatedtodevelopingfriendlyAI;hiscurrentfocusisondevelopingtheNovamenteAIEngine,asetoflearningalgorithmsandarchitectures.PeterVoss,founderofAdaptiveA.I.,Inc.,hasalsocollaboratedonfriendly-AIissues:http://adaptiveai.com/.

46.IntegratedFuelCellTechnologies,http://ifctech.com.Disclosure:TheauthorisanearlyinvestorinandadvisertoIFCT.

47.NewYorkTimes,September23,2003,editorialpage.48.TheHouseCommitteeonScienceoftheU.S.HouseofRepresentativesheldahearingon

April9,2003,to"examinethesocietalimplicationsofnanotechnologyandH.R.766,theNanotechnologyResearchandDevelopmentActof2002."See"FullScienceCommitteeHearingontheSocietalImplicationsofNanotechnology,"http://www.house.gov/science/hearings/full03/index.htm,and"HearingTranscript,"http://commdocs.house.gov/committees/science/hsy86340.000/hsy86340_0f.htm.ForRayKurzweil'stestimony,seealsohttp://www.KurzweilAI.net/meme/frame.html?main=/articles/art0556.html.AlsoseeAmaraD.Angelica,"CongressionalHearingAddressesPublicConcernsAboutNanotech,"April14,2003,http://www.KurzweilAI.net/articles/art0558.html.

ChapterNine:ResponsetoCritics1.MichaelDenton,"OrganismandMachine,"inJayW.Richardsetal.,AreWeSpiritual

Machines?RayKurzweilvs.theCriticsofStrongA.I.(Seattle:DiscoveryInstitutePress,2002),http://www.KurzweilAI.net/meme/frame.html?main=/articles/art0502.html.

2.JaronLanier,"OneHalfofaManifesto,"Edge(September25,2000),http://www.edge.org/documents/archive/edge74.html.

3.Ibid.4.Seechapters5and6forexamplesofnarrowAInowdeeplyembeddedinourmodern

infrastructure.5.Lanier,"OneHalfofaManifesto."6.AnexampleisKurzweilVoice,developedoriginallybyKurzweilAppliedIntelligence.7.AlanG.Ganek,"TheDawningoftheAutonomicComputingEra,"IBMSystemsJournal

(March2003),http://www.findarticles.com/p/articles/mi_m0ISJ/is_1_42/ai_98695283/print.8.ArthurH.WatsonandThomasJ.McCabe,"StructuredTesting:ATestingMethodology

UsingtheCyclomaticComplexityMetric,"NISTspecialpublication500–35,ComputerSystemsLaboratory,NationalInstituteofStandardsandTechnology,1996.

9.MarkA.RichardsandGaryA.Shaw,"Chips,ArchitecturesandAlgorithms:ReflectionsontheExponentialGrowthofDigitalSignalProcessingCapability,"submittedtoIEEESignalProcessing,December2004.

10.JonBentley,"ProgrammingPearls,"CommunicationsoftheACM27.11(November1984):1087–92.

11.C.Eldering,M.L.Sylla,andJ.A.Eisenach,"IsThereaMoore'sLawforBandwidth,"IEEECommunications(October1999):117–21.

12.J.W.CooleyandJ.W.Tukey,"AnAlgorithmfortheMachineComputationofComplexFourierSeries,"MathematicsofComputation19(April1965):297–301.

13.Thereareanestimated100billionneuronswithanestimatedinterneuronalconnection"fanout"ofabout1,000,sothereareabout100trillion(1014)connections.Eachconnectionrequiresatleast70bitstostoreanIDforthetwoneuronsateitherendoftheconnection.Sothat'sapproximately1016bits.Eventheuncompressedgenomeisabout6billionbits(about1010),aratioofatleast106:1.Seechapter4.

14.RobertA.FreitasJr.,Nanomedicine,vol.I,BasicCapabilities,section6.3.4.2,"BiologicalChemomechanicalPowerConversion"(Georgetown,Tex.:LandesBioscience,1999),pp.147–48,http://www.nanomedicine.com/NMI/6.3.4.2.htm#p4;seeillustrationathttp://www.nanomedicine.com/NMI/Figures/6.2.jpg.

15.RichardDawkins,"WhyDon'tAnimalsHaveWheels?"SundayTimes,November24,1996,http://www.simonyi.ox.ac.uk/dawkins/WorldOfDawkins-archive/Dawkins/Work/Articles/1996-11-24wheels.shtml.

16.ThomasRay,"Kurzweil'sTuringFallacy,"inRichardsetal.,AreWeSpiritualMachines?17.Ibid.18.AnthonyJ.Bell,"LevelsandLoops:TheFutureofArtificialIntelligenceandNeuroscience,"

PhilosophicalTransactionsoftheRoyalSocietyofLondonB354(1999):2013–20,http://www.cnl.salk.edu/~tony/ptrsl.pdf.

19.Ibid.20.DavidDewey,"IntroductiontotheMandelbrotSet,"http://www.ddewey.net/mandelbrot.21.ChristofKochquotedinJohnHorgan,TheEndofScience(Reading,Mass.:Addison-Wesley,

1996).22.RogerPenrose,ShadowsoftheMind:ASearchfortheMissingScienceofConsciousness

(NewYork:OxfordUniversityPress,1996);StuartHameroffandRogerPenrose,"OrchestratedObjectiveReductionofQuantumCoherenceinBrainMicrotubules:The'OrchOR'ModelforConsciousness,"MathematicsandComputerSimulation40(1996):453–80,

http://www.quanturnconsciousness.org/penrosehameroff/orchOR.html.23.SanderOlson,"InterviewwithSethLloyd,"November17,2002,http://www.nano

magazine.com/i.php?id=2002_11_17.24.Bell,"LevelsandLoops."25.Seetheexponentialgrowthofcomputinggraphsinchapter2(pp.67,70).26.AlfredN.WhiteheadandBertrandRussell,PrincipiaMathematica,3vols.(Cambridge,U.K.:

CambridgeUniversityPress,1910,1912,1913).27.Gödel'sincompletenesstheoremfirstappearedinhis"UberformalunenscheiderbareSatzeder

PrincipiaMathematicaundverwandterSystemeI,"MonatsheftefürMathematikundPhysik38(1931):173–98.

28.AlanM.Turing,"OnComputableNumberswithanApplicationtotheEntscheidungsproblem,"ProceedingsoftheLondonMathematicalSociety42(1936):230-65.The"Entscheidungsproblem"isthedecisionorhaltingproblem—thatis,howtodetermineaheadoftimewhetheranalgorithmwillhalt(cometoadecision)orcontinueinaninfiniteloop.

29.Church'sversionappearedinAlonzoChurch,"AnUnsolvableProblemofElementaryNumberTheory,"AmericanJournalofMathematics58(1936):345–63.

30.ForanentertainingintroductoryaccountofsomeoftheimplicationsoftheChurch-Turingthesis,seeDouglasR.Hofstadter,Gödel;Escher,Bach:AnEternalGoldenBraid(NewYork:BasicBooks,1979).

31.Thebusy-beaverproblemisoneexampleofalargeclassofnoncomputablefunctions,asseeninTiborRado,"OnNoncomputableFunctions,"BellSystemTechnicalJournal41.3(1962):877–84.

32.Ray,"Kurzweil'sTuringFallacy."33.Lanier,"OneHalfofaManifesto."34.Ahuman,thatis,whoisnotasleepandnotinacomaandofsufficientdevelopment(thatis,

notaprebrainfetus)tobeconscious.35.JohnR.Searle,"IMarriedaComputer,"inRichardsetal.,AreWeSpiritualMachines?36.JohnR.Searle,TheRediscoveryoftheMind(Cambridge,Mass.:MITPress,1992).37.HansMoravec,LettertotheEditor,NewYorkReviewofBooks,

http://www.kurzweiltech.com/Searle/searle_response_letter.htm.38.JohnSearletoRayKurzweil,December15,1998.39.Lanier,"OneHalfofaManifesto."40.DavidBrooks,"GoodNewsAboutPoverty,"NewYorkTimesNovember27,2004,A35.41.HansMoravec,LettertotheEditor,NewYorkReviewofBooks,

http://www.kurzweiltech.com/Searle/searle_response_letter.htm.42.PatrickMoore,"TheBattleforBiotechProgress—GMCropsAreGoodfortheEnvironment

andHumanWelfare,"Greenspirit(February2004),http://www.greenspirit.com/logbook.cfm?msid=62.

43.JoelCutcher-Gershenfeld,privatecommunicationtoRayKurzweil,February2005.44.WilliamA.Dembski,"Kurzweil'sImpoverishedSpirituality,"inRichardsetal.,AreWe

SpiritualMachines?45.Denton,"OrganismandMachine."

Epilogue1.AsquotedinJamesGardner,"SelfishBiocosm,"Complexity5.3(January–February2000):

34–45.2.Inthefunctiony=1/x,ifx=0,thenthefunctionisliterallyundefined,butwecanshowthat

thevalueofyexceedsanyfinitenumber.Wecantransformy=1/xintox=1/ybyflippingthenominatoranddenominatorofbothsidesoftheequation.Soifwesetytoalargefinitenumber,thenwecanseethatxbecomesverysmallbutnotzero,nomatterhowbigygets.Sothevalueofyiny=1/xcanbeseentoexceedanyfinitevalueforyifx=0.Anotherwaytoexpressthisisthatwecanexceedanypossiblefinitevalueofybysettingxtobegreaterthan0butsmallerthan1dividedbythatvalue.

3.Withestimatesof1016cpsforfunctionalsimulationofthehumanbrain(seechapter3)andabout1010(undertenbillion)humanbrains,that's1026cpsforallbiologicalhumanbrains.So1090cpsexceedsthisbyafactorof1064.Ifweusethemoreconservativefigureof1019cps,whichIestimatedwasnecessarytosimulateeachnonlinearityineachneuroncomponent(dendrite,axon,andsoon),wegetafactorof1061.Atrilliontrilliontrilliontrilliontrillionis1060.

4.Seetheestimatesintheprecedingnote;1042cpsexceedsthisbyafactoroftenthousandtrillion(1016).

5.StephenJayGould,"Jove'sThunderbolts,"NaturalHistory103.10(October1994):6–12;chapter13inDinosaurinaHaystack:ReflectionsinNaturalHistory(NewYork:HarmonyBooks,1995).

[Indexomitted]