digital creativity what is generative art? - compart...

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [Friedrich Althoff Konsortium]On: 18 March 2011Access details: Access Details: [subscription number 907681676]Publisher RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Digital CreativityPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t714576173

What is generative art?Margaret A. Boden; Ernest A. Edmonds

Online publication date: 01 December 2010

To cite this Article Boden, Margaret A. and Edmonds, Ernest A.(2009) 'What is generative art?', Digital Creativity, 20: 1, 21— 46To link to this Article: DOI: 10.1080/14626260902867915URL: http://dx.doi.org/10.1080/14626260902867915

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

http://www.informaworld.com/smpp/title~content=t714576173

http://dx.doi.org/10.1080/14626260902867915

http://www.informaworld.com/terms-and-conditions-of-access.pdf

What is generative art?Margaret A. Boden1 and Ernest A. Edmonds2

1Centre for Cognitive Science, University of Sussex2Creativity & Cognition Studios, University of Technology, Sydney

[email protected]

Abstract

There are various forms of what’s sometimescalled generative art, or computer art. Thispaper distinguishes the major categoriesand asks whether the appropriate aestheticcriteria—and the locus of creativity—are thesame in each case.

Keywords: computer art, generative art,interactive art, computational creativity

1 Introduction

Since the late-1950s, an ever-diversifying setof novel art practices has arisen which are stilllittle known or discussed in aesthetics and arttheory. (For a wide variety of examples, see:Krueger 1991; Wilson 2002; Candy andEdmonds 2002; Whitelaw 2004; Woolf 2004;Popper 2007.) As Jon McCormack, one of theartists concerned, has put it,[M]uch of the innovation today is notachieved within the precious bubble of fineart, but by those who work in the industries ofpopular culture–computer graphics, film,music videos, games, robotics and the Internet(McCormack 2003, p. 5).

The “bubble of fine art” refers to a shiftingsocially accepted norm. Artists often workoutside the norm of their day, as famouslyillustrated by Marcel Duchamp and his ready-mades or John Cage and silence. The novelapproaches that McCormack mentions areclosely related, both theoretically and metho-dologically—so much so, that they are often alllumped together under one label: “computerart”, “electronic art”, or “generative art”. Oneaim of this paper is to clarify how they can bedistinguished.

Theoretically, this new art originated incybernetics and general systems theory. Theyoung painter Roy Ascott, later to be highlyinfluential in the field, identified the novelactivity as “a cybernetic vision” (1966/67).And the exceptionally creative cyberneticianGordon Pask was a key influence. For besidesproducing and/or imagining some of the firstartworks of this general type (in the 1950s), heprovided much of the theoretical impetus thatinspired the more philosophically mindedartists in the field (Boden 2006, p. 4.v.e).

Very soon, the “cybernetic vision” wasbolstered by ideas about structure and processdrawn from computer science. This paper’sco-author Ernest Edmonds, for instance, turned

Digital Creativity2009, Vol. 20, Nos. 1–2, pp. 21–46

DOI: 10.1080/14626260902867915

Downloaded By: [Friedrich Althoff Konsortium] At: 17:42 18 March 2011

from paintbrush and easel to the computer inthe1960s: he thought he could produce moreinteresting art in that way (see Section 3). Atmuch the same time, music and visual art wasproduced which reflected AI’s computationaltheories of mind. Indeed, Harold Cohen, arenowned abstract painter in 1960s London,deserted his previous working practices largelybecause he felt that doing computer art wouldhelp him to understand his own creativeprocesses (McCorduck 1991; Boden 2004,pp. 150–166, 314f.).

Over the past twenty years, this artistic fieldhas been inspired also by ideas about emer-gence, evolution, embodiment, and self-organisation. These concepts are borrowedfrom various areas of cognitive science and inparticular from artificial life (A-Life). However,the theoretical roots of A-Life reach back tomid-century cybernetics and automata theory(Boden 2006, p. 4.v.e, 15.iv-v). In short, thetheoretical wheel has turned full circle.

The methodological wheel, meanwhile, hasclimbed an ascending spiral. For the artpractices outside the bubble are grounded intechnologies for communication and infor-mation processing whose power and varietyhave burgeoned over the last half-century.(Often, this means that the customary lone artistis replaced by a team, some of whose membersmay be computer scientists and/or tele-engineers.)

Most of them rely heavily on digital com-puting and in particular on methods drawn fromAI/A-Life. Specifically, they have employedboth symbolic and connectionist computation,and—more recently—cellular automata, L-systems and evolutionary programming too.This is an ascending spiral, not a linear ascent,because two of those ‘recent’ methods wereforeseen (by John von Neumann) in 1950scybernetics, and all three had been mathemat-ically defined by the 1960s—but none could befruitfully explored, by artists or scientists, untilpowerful computers became available muchlater (Boden 2006, p. 15.v-vi).

The resulting artworks are highly diverse.They include music, sonics, the visual arts,video art, multimedia installations, virtualreality, kinetic sculpture, robotics, performanceart and text. And whereas some of theseoutside-the-bubble activities place ink or paintonto a surface, others involve desk-top VDUsor room-scale video-projection. Yet otherseschew the virtuality of cyberspace, construct-ing moving physical machines instead.

The labels attached to these new art formsvary and have not yet settled down into agenerally accepted taxonomy. The namespreferred by the artists involved include:generative art, computer art, digital art,computational art, process-based art, electronicart, software art, technological art andtelematics. All of those terms are commonlyused to denote the entire field–and (althoughdistinctions are sometimes drawn) they areoften treated as synonyms. In addition, thereare names for subfields: interactive art,evolutionary art, video art, media (andnew-media and multimedia) art, holographicart, laser art, virtual art, cyborg art, robotic art,telerobotics, net art . . . and more. Again, theextension of these labels is not always clear.

It’s partly for that reason that “a satisfactorycritical framework of new forms in art tech-nology has yet to be developed” (Candy andEdmonds 2002, p. 266). We hope that the dis-tinctions made in this paper may help towardssuch a framework. Mainly, however, we aim tooutline some philosophical problems that arisewith respect to the art that lies outsideMcCormack’s “precious bubble”.

In Section 2, we sketch the history andcurrent usage of some of the new labels withinthe artistic community concerned. Then, inSection 3, Edmonds gives a brief autobiogra-phical account of why he chose—and nowchooses—one name rather than another for hisown art.

Section 4 distinguishes various categories,focusing especially on computer art, generativeart, evolutionary art, robotic art and interactive

Boden and Edmonds

22

Digita

lCreativity,Vol.20,Nos.

1–2


art. The terms we define here use words that arealready being used by the artists in question.Indeed, we hope that our analysis may helpreaders to interpret these artists’ discussions oftheir work. But we are not aiming to offer areport of common usage because, as alreadyremarked, such usage is not consistent. Rather,our account is intended as a theoretical toolwith which to highlight certain distinctions thathave philosophical and/or aesthetic interest.

The philosophical issues concerned areindicated in Section 5. We’ll see thatjudgements concerning creativity, authorialresponsibility, agency, autonomy, authenticityand (sometimes) ontology are even moreproblematic outside the precious bubble thaninside it. To explore these matters fully wouldbe beyond the scope of this paper. But weidentify some pertinent questions and somepossible responses to them.

2 A terminological history

The terms ‘generative art’ and ‘computer art’have been used in tandem, and more or lessinterchangeably, since the very earliest days.For the first exhibition of computer art wascalled Generative Computergraphik (see thedescription of the event in Nake 2005). It washeld in Stuttgart in February 1965 and showedthe work of Georg Nees. Four years later heproduced the first PhD thesis on computer art,giving it the same title as the exhibition (Nees1969). That thesis was soon widely consultedby the small but growing community, harnes-sing the words generative and computertogether in its readers’ minds.

In November 1965 Nees showed again inStuttgart. On this occasion, the exhibition alsoincluded the computer graphics work of FriederNake. Both artists used the term ‘generative’.The word was used here to identify art that wasproduced from a computer program and, hence,was at least in part produced automatically. Inthat sense, the work of the graphic artist Michael

Noll that was exhibited in New York betweenthe two German shows was also generative.

Others pioneering the activities outsideMcCormack’s bubble also adopted the term.For example, when Manfred Mohr, who knewNake, started producing drawings with a com-puter program in 1968 he termed it ‘generativeart’. (Mohr still uses that description of hiswork and uses it in the same sense.) And thephilosopher Max Bense—who had composedthe manifesto for the original Stuttgart exhibi-ton of 1965—was writing about what he called‘generative aesthetics’ (Nake 1998). Alterna-tive tags were already being offered, however:an influential discussion by the art historianJack Burnham (1968), for instance, identifiedthe new work as ‘process art’.

In music, the use of computer systems toproduce work started very early on. By 1957Lejaren Hiller and Leonard Isaacson hadcreated the Illiac Suite for String Quartet(Hiller and Isaacson 1958) and in 1962 IannisXenakis completed the Stochastic Musicprogram. Xenakis published essays on for-malised music, including this program in thesame year (in English, Xenakis 1971) and wasa key figure in the development of computer-generated music. Probably as a result of thisearly start, the development of computer-generative music preceded that of computer-generative visual art.

Not all generative visual art involves com-puters. Pre-computer examples include suchclear cases as Kenneth Martin, whose 1949abstract painting used basic geometrical figures(squares, circles, diagrams) and rules of pro-portion (Martin 1951/1954). Later, his ‘Chanceand Order’ and ‘Chance, Order, Change’ seriescombined rule-driven generation with randomchoice. Whilst chance events–such as selectinga number out of a hat–determined the course ofthe work, everything else was determined bythe rules that Martin had devised. Other non-computer generative artists are identified inSection 4, where we’ll see that the generativeprocesses involved vary widely in type.

What is generative art?

23

Digita

lCreativity,

Vol.20,Nos.

1–2


Today, the term ‘generative art is stillcurrent within the relevant artistic community.Since 1998 a series of conferences have beenheld in Milan with that title (Generativeart.com) and Brian Eno has been influential inpromoting and using generative art methods(Eno 1996). Both in music and in visual art, theuse of the term has now converged on workthat has been produced by the activation of aset of rules and where the artist lets a computersystem take over at least some of thedecision-making (although, of course, theartist determines the rules).

Rules are at the heart of this type of art.But what computer scientists call rule-basedprogramming (e.g. Kowalski and Levy 1996)is not necessarily implied. The computer-artcommunity regards it as important that theartwork is generated from a set of specified rules,or constraints, rather than from a step-by-stepalgorithm. But the detailed implementationmethod (i.e. the specific computer system that’sbeing used) is not normally seen to be significant.

To understand this point, consider an over-simplified illustration of the programmingconcepts just mentioned. When a program iswritten in a step-by-step (algorithmic) way, theprogrammer instructs the computer to “do A”,then “do B”, then under certain conditions “doC”, otherwise “do D” and so on. When aprogrammer writes rules (constraints),however, they tell the computer that (forexample) “Z should always be bigger than Y”,“X must never equal W’, and so on—but theyleave it to the computer system to work outhow to apply those rules.

So in the step-by-step approach, the pro-grammer is explicitly directing the actions ofthe computer. In the rule-based approach, bycontrast, the translation from the specified rulesto computer actions is not immediately clear.To know just how the specification of a ruledetermines computer behaviour we need to beaware of the details of the computer system thatinterprets the rules—which even a professionalcomputer scientist may not know. Another way

of putting this is to say that the artist leaves thecomputer to do its own thing without knowingjust what it is that the computer will be doing.

One could argue that the art-community’sobjection to a step-by-step algorithm is more amatter of taste than anything else. For evenwhena programmer has written explicit step-by-stepcode, he or she does not necessarily–or evenusually–know the outcome. If they did, therewould be no bugs (except those due to typingmistakes and punctuation errors). After all, theearliest programming was often done in order tocalculate numerical values that were almostimpossible to obtain otherwise. Those earlyefforts were relatively simple: most programstoday are hugely more complex. So despite thedifferences between the two programmingapproaches described above, there is nodistinction at the most fundamental level. Bothtypes of program are unpredictable by theirprogrammer.

However, computer artists—and computerscientists, too—know from their own experi-ence that when writing algorithmic code thereis a ‘feel’ of fully controlling the computer.This ‘feel’ does not carry over to systems wherethey are specifying sets of rules, or constraints.

In other words, rule-driven systems appearto have a greater degree of autonomy, relativeto the conscious decisions of the human artist.That phenomenological fact is significantbecause autonomy is a concept that’s closelyconnected with art-making (see Section 5).This explains why computer artists are morecomfortable in speaking of ‘generative art’where the system is rule-driven, not algorith-mic—and why they usually avoid step-by-stepprogramming.

The concepts of generative art and(programmed) computer art were assimilatedright from the start. With the recent appearanceof art using methods drawn from A-Life (forexamples, see Whitelaw 2004; Tofts 2003;2005, pp. 80–103; Popper 2007, pp. 118–129), the label ‘generative art’, as used in thecommunity concerned, has acquired biological

Boden and Edmonds

24

Digita


1–2


overtones. In biology, the key word is commonin discussions of morphological developmentand growth in plants and animals, and inreferences to reproduction. One or both of thosemeanings is/are sometimes explicitly stressedby self-styled generative artists whose workfocuses on emergence, self-organisation and/orevolution. McCormack himself is one suchexample (e.g. Dorin and McCormack 2001;McCormack et al. 2004). Even so, the formal-mathematical sense remains a core aspect of thelabel’s meaning.

3 A terminologicalautobiography

Addendum by M.A.B.: The core of Edmonds’artistic oeuvre is an exploration of the possi-bilities inherent in colour as such. (As for hiscomputer-science oeuvre, that includes one ofthe first logic-programming languages.) Thesubtlety of his colours, and the sequence oftheir changes, cannot be conveyed here, andnor can their audio-aspects–but they are indi-cated on his website (www.ernestedmonds.-com). It’s worth noting that the curators of amajor Washington DC festival in 2007 cele-brating the 50th anniversary of the “Color-Field” painters—Mark Rothko, Clyfford Still,Kenneth Noland and the like—have chosen toinclude live showings of some of his recentwork in the Experimental Media series at theCorcoran Gallery and The Ellipse Arts Center.

I (that is, Ernest Edmonds) had been com-mitted to painting since my early-teenage years,strongly influenced by the constructivists.I started to use a computer in my artwork in1968, when I wrote a program that searched foran arrangement, a layout, of a given set ofpicture elements.

At the time, I was working on someproblems in logic that involved finding sets ofnumbers that displayed, or revealed, certainconditions. The problem was to show thatvarious axioms were independent by demon-strating that values existed that showed that no

one of them could be derived from the others.The details of what this means do not matterhere. Suffice it to say that I realised that I couldwrite a computer program that would search forsuch numbers, and I did so. This resulted in thepublication of my first research paper, whichwas in the Journal of Symbolic Logic(Edmonds 1969).

An interesting thing to notice is that thepaper makes no mention of computers orcomputer programs. The program had enabledme to find the required set of numbers. All I hadto do in the proof was to provide them. Whilstthe computer was used to help me solve aproblem in a non-computing domain, thecomputer itself was not part of the solution.

This is exactly how it was to turn out in myfirst use of the computer in art. I wrote aprogram to search for a visual layout whichsatisfied a set of conditions that I was able tospecify. As with the logic, the computer couldbe dispensed with once the problem wassolved, and the work itself neither used norreferred to a computer program at all. It is justthat I could not have made the piece, calledNineteen, without the use of the computer.

With respect to a more specific concern forgenerative art, a crucial step was actually madeconceptually in 1980. I realised that my long-standing interest in time could lead to a new kindof art, with the use of computers. Music, par-ticularly serial music, had been very influentialon my visual work; and I was fascinated by film.During the 1970s, I made some abstract films thehard way, by filming static abstract images andsplicing the film together from very many suchstill clips. I started working on what I came tocall video constructs. These are time-based,generative, abstract works (Edmonds 1988). Thegenerative rules at the core of these works were,and are, based on the rules that I started toformulate when making Nineteen and which Iused in my painting through the 1970s and on.

I did not show such a work publicly until1985, but in many ways that showing was theculmination of much of the work of the 1970s.


25

Digita

lCreativity,

Vol.20,Nos.

1–2


Fragments was the generative work shown atExhibiting Space, London, as part of my exhi-bition Duality and Co-Existence, in 1985. In2004, a restored version of Fragments wasissued as a limited edition DVD and shown inthe exhibition Australian Concrete Construc-tive Art (at the Conny Dietzschold Gallery,Sydney).

Why were generative processes interestingin my art? Well, there are many answers tothat question, but I will start with the mostpragmatic of them.

Making art is an iterative process. Onemakes a mark, for example, looks at it and thenmakes another (or not, of course). Art-makingis a continual process of conception, action andperception followed by re-conceptualisation–and so on. An important issue is that ofmatching the time and effort of the making (thelaying down of a mark, for example) with itsappropriate place in the cycle. Time-basedwork is very hard to make if each step has to beexplicitly constructed. Using the computer as agenerative engine magically removes thismajor problem. So, from a pragmatic point ofview, I constructed a generative engine to makemy time-based work so that the cycle ofmaking, evaluating and refining the work wasfaster and tighter.

My pragmatic answer to the question begsmany points, of course. The key underlyingissue is that of the order and use of rules in theworks in the first place. That makes the gen-erative solution possible. My motivations arenot ideological or political. I understand andam sympathetic with many computer-artistcolleagues who use rules and generativemethods in order to distance themselvesfrom the art-making, and who do this forideological reasons (see Boden and Paul Brownin preparation). For me, however, it is simply amatter of reducing the enormous decision spaceof art-making tosomething manageable.

Ever since Nineteen my work has involvedrules. Often, they were constraint sets, asdescribed in Section 2. But, with the advent of

the time-based work, they became generative.The distinction here is important.

The use of rules in art-making does notnecessarily imply that the art is generative inthe sense that I have used the word in the past.(Note that we define G-art in a different way inSection 4.) The function of the rules in theprocess is a crucial issue. Where the rules areconstraint sets, the art-making on the part of thehuman artist is as free, or almost as free, as ever.It is just that conditions are placed upon what issuccessful, whose results lie beyond the artist’sintuition. A satisfactory work must not violatethe constraints. As in mathematics or computerscience, the fact that the problem space(of making the artwork) is bounded by clearconstraints does not necessarily lead to anymethod or process for designing or making art.

In what I count as generative art, by con-trast, the rules must be constructive. Thereare parallels here with mathematical logic(Goodstein 1951). That is to say, they mustprovide or imply a specific process that canlead to the desired outcome. That is the definingfeature of generative art as I came to see it.Only if the rules are constructive does the artisthand over to the computer a significant elementof the decision-making that is at the core of theart-making.

There is no doubt that I am interested in therules themselves, not merely in what they mightgenerate. Early on, I saw that they were morethan just a convenience. The rules define theform. Think of a fugue, or of serial music.Think of perspective, or the golden section.This is the second answer to the question ofwhy I found generative processes artisticallyinteresting. Generative art enables the artist toconcentrate on the underlying rules themselves:the structures that define the artwork, as againstthe surface. (This is a position clearly associ-ated with distaste for decoration and ornament.)

It is all a little more complicated than this,however. In generative art that follows theevolutionary A-Life approach (see Section 4),for example, sometimes the ‘fitness function’ is

Boden and Edmonds

26

Digita


1–2


not automated. Sometimes a human operator,whether audience or artist, makes the fitnessdecision on the basis of who knows what: theirintuitive judgement, one might as well say. Insuch cases, the rules are not fully constructive,because their deployment requires the inter-vention of a human agent.

This compromise to the core concept ofgenerativity is normally accepted by computerartists, and I accept it also. It applies more widelythan in A-Life evolution. In general, it is a matterof enabling interaction to come into the picturewithout dropping the term ‘generative art’.

I was interested in interaction from thebeginning of my concern for the role of com-puters in my art. Indeed, I made interactivework as early as 1970. However, when I startedto use logic to generate time-based works in the1980s, I was not working in that way. Later on,and independently of evolutionary art (which Ihave never personally pursued), I made asimple extension to my generative approach inorder to accommodate interaction.

It did not involve a fitness function and nordid it involve any other explicit evaluation step.What I did was to introduce the notion of‘external data’. These were values, stored in thecomputer, which were set and changed fromtime to time as a result of various sensorsystems. For example, such an item could bethe room temperature. Alternatively, it could bethe degree of animation of people in front of theartwork. I made these external data available tothe generative process and wrote rules (con-straints) that took account of them. I saw thoserules as meta-rules, because the values detectedat a given time were used to select which rulesor rule-set should be used.

In my more recent work, I have added yetanother mechanism. Again, in principle it issimple enough. It is, in effect, a state vector thatcan be varied by inputs (such as sounds pickedup by a microphone and detected behaviours bythe audience), and which can be referred to justas the external data are. Thus, memory of a sortis made possible. Inputs or sensed data can

cause changes at first only in the vector, but inways that will influence the artwork’sbehaviour later–possibly much later (Edmonds2007). Because of this delay, which precludesinstantaneous effects on the system’sbehaviour, I talk about such works in terms of‘influence’ rather than ‘interaction’.

4 A taxonomy of generative art

Our taxonomy distinguishes eleven types ofart. We call them Ele-art, C-art, D-art, CA-art,G-art, CG-art, Evo-art, R-art, I-art, CI-art andVR-art. Some of these activities, having beenlocated within our classification, are thenignored. We pay most attention to variousforms of CG-art, because these raise the mostinteresting philosophical issues.

This ‘taxonomy’is a decidedly non-Linnaean structure. Quite apart from the factthat our definitions, like most definitions, admitborderline cases and even anomalous counter-examples, there’s no neat and tidy hierarchy ofgenus and species within which these eleventypes can be located. Although there are somepart-whole relations here, there are also untidyoverlappings.

One type of overlapping concerns linkswith more traditional, or familiar, categories ofart. Most cases of such art do not fall under ourclassification at all. But some of our concepts–namely: G-art, I-art, Evo-art and R-art–coverartworks both inside and outside McCormack’s“precious bubble” Admittedly, those which lieinside the bubble are relatively maverickexamples, as we’ll see. Indeed, some of them(produced by the ‘conceptual’ artists) werespecifically intended to undermine the notionof “fine art” in terms of which the bubble isdefined. For shorthand purposes, however, welocate all examples of non-computer art insidethe bubble.

A summary list of our definitions is given atthe end of this section. Meanwhile, we’llintroduce them one by one, giving illustrativeexamples of each category.


27

Digita

lCreativity,

Vol.20,Nos.

1–2


Let us start with electronic art, or Ele-art.This wide concept covers (df.) any artworkwhose production involves electrical engin-eering and/or electronic technology. So itranges from simple analogue devices of the1950s and 1960s such as Pask’s Musicolourand Colloquy (Pask 1971; Mallen 2005) andEdward Ihnatowicz’s kinetic sculpture SAM(Zivanovic 2005, p. 103)—all pioneeringexamples of interactive art, or I-art—to thehighly sophisticated man-robot integrationsrecently embodied by the performance artistStelarc (Smith 2005). And along the way, itcovers the whole of computer art and mediaart, including those examples which exploitthe advanced computational techniques ofvirtual reality.

Unlike mechanical art, such as Leonardo daVinci’s metal lion (who “after he had a whilewalked vp and downe, stoode still opening hisbreast, which was all full of Lillies and otherflowers of diuers sortes”—Marr forthcoming,p. 66), electronic art could not appear until themid-twentieth century. But, as the previousparagraph implies, the technologies concernedhave diversified richly since then. Accordingly,the highly inclusive label Ele-art is not veryinteresting for our purposes.

Surprisingly, perhaps, neither is the conceptof computer art, or C-art. By C-art, we mean(df.) art in whose productive process computersare involved. This concept is apt for generalart-historical purposes, because it covers everyexample that anyone might want to call com-puter art—including many that are commonlygiven other labels. It’s less useful for us here,however, for two reasons.

First, it includes analogue as well as digitalcomputers. Some of the earliest C-art workcombined digital methods with specially-builtanalogue devices. Ihnatowicz’ giraffe-likekinetic sculpture Senster is a case in point(Zivanovic 2005). As for analogue computersas such, these were used in the early days. Forexample, in visual arts by Ben Laposky’s workof the 1950s (Laposky 1969), and in the growth

of electronic music at the same time, famouslyencouraged by the invention of the Moogsynthesiser (Pinch and Trocco 2002).

Today, a few computer artists sometimesemploy analogue processes, namely electro-chemical reactions like those pioneered byPask. Some of their work, including Pask-inspired ‘sculptures’ by Richard Brown (2001,2006) and Andy Webster (2006-ongoing), willfeature in a 2007 Edinburgh exhibition on“Gordon Pask and his Maverick Machines”. (Inaddition, a video on this theme called TuningPask’s Ear has been shown in several Europeanart galleries: Webster and Bird 2002.) Butanalogue computers are another matter—andare very rarely used by artists today. Because ofthe huge flexibility that is afforded by thegeneral-purpose nature of digital computers, itis those machines which underlie most C art.Indeed, to speak of computer art is typically toassume that digital computers are being used.

In other words, computer art is (usually)tacitly classed as digital art, or D-art. D-art (df.)uses digital electronic technology of some sort.It includes not only artworks generated bycomputers but also digitally manipulable (buthuman produced) music and video. Commonusage sometimes treats ‘digital art’ and ‘com-puter art’ as near-synonyms. In our taxonomy,they are analytically distinct—with most, butnot quite all, C-art being included within D-art.(If the word ‘electronic’ were removed fromour definition, the nineteenth-centuryPointillistes would count as D-artists; for theirpictures were composed not of continuousbrush-strokes or colour-washes but ofmyriad individual spots of paint.)

D-art is more wide-ranging than mayappear at first sight. For instance, someC-artists use visual software that is intuitivelyanalogue and so relatively ‘natural’ to workwith. (One example is continuous vector-mapping, used instead of pixel-editing (Leggett2000)). But they are relying on methods/hardware that are digital at base. In fact, mostpeople who said today that they are using an

Boden and Edmonds

28

Digita


1–2


analogue method (i.e. an analogue virtualmachine) would actually be working on adigital computer, used to simulate an analoguecomputer.

Similarly, most ‘neural networks’ or con-nectionist systems, whether used by cognitivescientists or by computer artists, are actuallysimulated on von Neumann machines. That’strue, for instance, of Richard Brown’s interac-tive Mimetic Starfish, a millennial version ofthe Senster that was described by The Times in2000 as “the best bit of the entire [Millennium]dome”. The starfish was built by engineeringvisual imagery, not metal: it is a neural-networkbased virtual creature (an image projected ontoa marble table) that responds in extraordinarilylifelike ways to a variety of hand-movements.In short, digital technology reaches further thanone might think.

The second reason why the definition ofC-art given above is too catholic for ourpurposes is that it includes cases where thecomputer functions merely as a tool under theclose direction of the artist, rather like an extrapaintbrush or a sharper chisel. Artists in therelevant community sometimes speak of this as‘computer-aided’ or ‘computer-assisted’ art,contrasting it with what they call ‘compu-tational’ art—where the computer is more of aparticipant, or partner, in the art-making (e.g.Brown 2003, p. 1). We call this CA-art,wherein (df.) the computer is used as an aid(in principle, non-essential) in the art-makingprocess.

Consider video art and music videos, forinstance. These popular outside-the-bubbleactivities qualify as CA-art in our sense. For thehuman-originated images and/or music aredigitally stored and (usually) manipulated/transformed by the artist, using the computer asa tool. Other cases of CA-art include someone’sdoing a line drawing by hand on the computerscreen and then calling on a colouring programsuch as PhotoShop to produce a LimitedEdition of identical prints—or, for that matter, aunique image. This is an upmarket form of

painting-by-numbers, wherein the hues foreach area are chosen by the individual artist.Yet other examples include computer musicthat’s so called because it uses electronicsynthesisers and ‘virtual’ instruments.

In practice, the computer ‘aid’ may benecessary for the art-making. It’s impossible,for instance, to alter video-images in certainways except by using a computer. Similarly,some visual effects delivered by Photoshopcould not have been produced by using oils,water-colours, or gouache. And synthesisedcomputer music exploits sounds that had neverbeen heard before synthesisers were developed.Nevertheless, the computer is not essential inprinciple. The relevant visual/sonic effects arespecifically sought by the human artist andmight conceivably have been produced in someother way. Much as a species with iron-hardfinger nails would not need chisels, so our vocalcords (or wood, metal, or cats’ sinews . . . )might have been able to produce the soundsproduced by synthesisers.

The sub-class of C-art which interests us isthe type where the computer is not used as atool to effect some idea in the artist’s mind butis in a sense ( just what sense will be exploredin Section 5) partly responsible for coming upwith the idea itself. In other words, the C-artthat’s most relevant here is a form of generativeart, or G-art.

In G-art, (df.) the artwork is generated, atleast in part, by some process that is notunder the artist’s direct control. This is a verybroad definition. It does not specify theminimal size of the “part”. It does not laydown just what sort of generative process is inquestion. It does not say what counts as beingoutside the artist’s direct control. And it issilent on the extent (if any) to which theprocesses concerned may have been deliber-ately moulded by the artist before ‘losing’direct control. In short, our definition of G-artis largely intuitive. In general, it picks outcases of art-making in which personal controlis deliberately diminished, or even wholly


29

Digita

lCreativity,

Vol.20,Nos.

1–2


relinquished, and relatively impersonalprocesses take over.

Those impersonal processes vary greatly.They may be physical, psychological, socio-cultural, biological, or abstract (formal). And ifabstract, they may or may not be implementedin a computer.

For example, in the dice-music written byHaydn and Mozart the exact order of the pre-composed phrases was decided by throwing adie. Although a human threw the die volunta-rily, he/she could not influence, still lessdetermine, just how it fell. That was due topurely physical forces. Such forces alsoconstructed the various versions of BryanJohnson’s (1969) novel The unfortunates,published as 27 separate sections in a box: allbut the first and last were to be read in a randomorder, decided by shuffling or dice-throwing.One might even say, thanks to ‘t least in part’,that Jackson Pollock’s paintings examplifiedG-art grounded in physics. For although hecertainly was not throwing (still less, choosing)paint at random, he did not have directcontrol over the individual splashes—as hewould have done over marks made with apaintbrush.

Even more control was lost, or ratherdeliberately sacrificed, when Hans Haacke, inthe 1960s, began to exploit–and even to high-light–the physical behaviour of water/vapour/ice, of waves and of weather conditions. Hewanted to make “something which experi-ences, reacts to its environment, changes, isnonstable . . ., always looks different, the shapeof which cannot be predicted precisely . . .”(Lippard 1973, p. 38, 64f.). He saw these worksnot as art objects but as “‘systems’ of interde-pendent processes”—which evolve without theviewer’s interaction or “empathy”, so that theviewer is a mere “witness”. A few years later,Jan Dibbets placed eighty sticks in the sea, afew inches below the surface, and watchedthem oscillate in the water from fifty feetabove: “That”, he said, “was the work”(Lippard 1973, p. 59).

The Surrealists of the 1920s, by contrast,had exploited psychological processes–but of arelatively impersonal kind. Inspired by Freud,they engaged in automatic writing and paintedwhile in trance states, in order to prioritise theunconscious mind—which Andre Bretondeclared to be “by far the most important part[of our mental world]”. Indeed, Surrealism wasdefined by Breton as:Pure psychic automatism [sic] by which oneproposes to express . . . the actual functioningof thought, in the absence of any controlexerted by reason, exempt from all aestheticor moral preoccupations (Breton 1969).

The unconscious thought was taking place in aperson’s mind, to be sure, but voluntary choiceand personal “preoccupations” (i.e. the realityprinciple and ego-ideals) were not directing it.

More recently, the conceptual artist SolLeWitt was also recommending G-art when hesaid that art should be designed by some for-mulaic rule. The crucial idea, he said, “becomesa machine that makes the art,” where “all of theplanning and decisions are made beforehandand the execution is a perfunctory affair”(LeWitt, 1967, p. 824); once the plan has beenchosen, “The artist’s will is secondary to the[artmaking] process he initiates from idea tocompletion” (Lewitt, 1969, item 7, italicsadded). He even added that “His wilfulness mayonly be ego”. That artmaking process wasnevertheless psychological, in the sense that theimplications of his abstract rules were discov-ered not by computers but by consciousreasoning. Conscious rule-based reasoning(combined with chance) was used also in theG-art of Kenneth Martin, who–as remarked inSection 2—made some artistic choices bypicking numbers out of a hat.

Sociocultural processes—in the form of theUnited States postal system—produced DouglasHuebler’s artwork called 42nd Parallel. Here,items were posted from 14 different townsspanning 3,040 miles on latitude 42, all sent tothe Massachusetts town of Truro. The work,according to Huebler, was not the conception in

Boden and Edmonds

30

Digita


1–2


his mind, nor the posted items, nor even the actsof posting. Rather, it was the widespread patternof activity within the US postal system. But, hesaid, the work was “brought into its completeexistence” through documents: the certifiedpostal receipts (for sender and for receiver) and amap marked with ink to show the geographicalrelations between the 15 towns. Its nature asG-art is evident in his remarks “An inevitabledestiny is set in motion by the specific processselected to form such a work, freeing it fromfurther decisions on my part”, and “I like theidea that even as I eat, sleep, or play, the work ismoving towards its completion” (quoted inLippard 1973, p. 62).

The artist Hubert Duprat turned to biologyfor constructing the work of art. He put dra-gonfly larvae into an aquarium containing notpebbles and pondweed but tiny flakes of gold,plus a few small pearls, opals, and sapphires—and left them to ‘sculpt’ opulent little protectivecases, held together by caddis-silk (Duprat andBesson 1998). Some thirty years earlier, Haackehad experimented with the growth of grass andthe hatching of chickens (as well as with waterand weather), to make something “natural”,which “lives in time and makes the ‘spectator’experience time” (Lippard 1973, p. 38).

Others have even exploited physical andbiological de-generation to produce their G-art.The environmental sculptor Andy Goldsworthysometimes highlights effects caused by undir-ected physical change: in his gradually meltingice-sculptures, for example. And in GustavMetzger’s “auto-destructive” art (notorious forthe occasion on which an overnight gallerycleaner innocently threw Metzger’s bag ofrotting rubbish into the dustbin), the point of theexercise is to remind us of the deterioration thatawaits all human constructions—and humanbeings, too (Metzger 1965). The artwork isoriginally assembled by a human artist, but itattains its final form, and its significance, throughthe natural processes of damage and decay.

However, such inside-the-bubble (albeitunorthodox) cases are not what the new artists

normally have in mind when they refer to‘generative art’. Their phraseology is borrowedfrom mathematics and computer science, withwhich the maverick artists just named were notconcerned. These disciplines see generativesystems as sets of abstract rules that canproduce indefinitely many structures/formulaeof a given type and which (given the Church-Turing thesis) can in principle be implementedin a computer. The GA-community outside thebubble put this principle into practice. That is,their artmaking rests on processes generated byformal rules carried out by computers—asopposed to physical, biological, or psycho-logical processes, or abstractions personallydiscovered by conscious thought.

In other words, the instances of G-art whichmost concern us here are also instances of C-art. They are computer generated art: CG-art,for short.

A very strict definition of CG-art wouldinsist that (df.) the artwork results from somecomputer program being left to run by itself,with zero interference from the human artist.The artist (or a more computer-literate collab-orator) writes the program, but does not interactwith it during its execution. In effect, he/shecan go out for lunch while the program is left todo its own thing.

Such cases do exist. Cohen’s AARONprogram (see below) is one well-knownexample. Nevertheless, that definition is sostrict that it may be misleading. We saw inSection 3 that generative artists allow a‘compromise’ in the core concept, so as toinclude interactive art. This is such a prominentsubclass of what’s called generative art that,even though we are not aiming to capturecommon usage, it would be highly anomalousto exclude it.

To be sure, the definition given above doescover most interactive art, because it insists onzero interference from “the human artist”,rather than from “any human being, whetherartist or audience”. However, it would be veryeasy for readers to elide that distinction–which,


31

Digita

lCreativity,

Vol.20,Nos.

1–2


in any case, makes a questionable assumptionabout authorial responsibility (see Section 5).Moreover, the overly strict definition of CG-artexcludes those cases (also mentioned inSection 3) wherein artists rely on their intuitivejudgement to make selections during anartwork’s evolution.

We therefore prefer to define CG-art lesstidily, as art wherein (df.) the artwork resultsfrom some computer program being left to runby itself, with minimal or zero interference froma human being. The word “minimal”, ofcourse, is open to interpretation. It necessitatescareful attention to just what interference goeson, and by whom, in any particular case.(Attention need not be paid, however, to thedistinction between rule-based and step-by-stepprogramming explained in Section 2: ourdefinition of CG-art allows for both.)

Most of what people call ‘computer art’ isCG-art, in this sense. Indeed, the phrases ‘com-puter art’ and ‘generative art’ are often regardedas synonyms. Notice, however, that in our ter-minology not all C-art is CG-art. CA-art is not,because the computer is there used as a toolsubject to the artist’s hands-on control (and isof no more philosophical interest than apaintbrush or a chisel). That’s why we calledthis paper ‘What is generative art?’, ratherthan the seemingly synonymous ‘What iscomputer art?’

CG-art is intriguing on two counts. First,the generality and potential complexity ofcomputer programs means that the possiblespace of CG-artworks is huge, indeed infinite.Moreover, most of the structures in that spacewill be images/music which the unaidedhuman mind could not have generated, or evenimagined—as the artists themselves admit.

A sceptic might object that much the sameis true of a trumpet, or a cello: not even themost skilled stage-impressionists could mimicthese instruments plausibly. In short, humanartists often need help from machines.Trumpets, computers . . . what’s the difference?Well, one important difference has just been

mentioned, namely, the generality of digitalcomputers. In principle, these machines can(and do) offer us an entire symphony orchestra,and an infinite set of visual images and sculp-tural forms–indeed, an infinite range of virtualworlds. McCormack (2003, p. 7) goes so far asto compare this infinite space of possibilities,way beyond our comprehension, with theKantian sublime.

The second point is even more pertinent.Whereas there’s no interesting sense in which atrumpet, or a cello, can be “left to do its ownthing”, a computer certainly can. And it is partof the definition of CG-art that this happens. Aswe’ll see later, this aspect of CG-art raises sometricky problems concerning concepts such asautonomy, agency, creativity, authenticity andauthorial responsibility.

Especially well-known cases of CG-artinclude the successive versions of AARON.This is a drawing-and-colouring programdeveloped over the last forty years by the one-time abstract painter Cohen (1995, 2002), andexhibited at venues all around the world—including the Tate Gallery. It has shown clearprogression along various aesthetic dimen-sions. Indeed, Cohen (p.c.) describes the 2006version as a “world-class” colourist, whereas hehimself is merely a “first-rate” colourist: “Iwouldn’t have had the courage to use thosecolours”, he sometimes says. (At earlier stages,colouring-AARON mixed liquid dyes and used‘painting blocks’ of five different sizes to placethem on paper; the current version prints outcomputer-generated colours instead of usingliquids, but these colours too are ‘mixed’ at theprogram’s behest.)

An almost equally famous example isEmmy, a computer-musician developed overmuch the same period by the composer DavidCope (2001, 2006). This generates music in thestyles of many renowned composers, and veryconvincingly, too—see Hofstadter 2001, p. 38f.(Nevertheless, Cope has recently abandoned it,because of the prejudiced reactions of audi-ences: see Section 5.)

Boden and Edmonds

32

Digita


1–2


Both of those programs were based onmethods drawn from what the philosopher JohnHaugeland dubbed GOFAI, or Good Old-Fashioned AI (although the mature Emmy alsouses connectionist AI). More recent methodsfor constructing CG-artworks, as remarked inSection 1 include cellular automata, L-systemsand evolutionary programming—all widelyused in A-Life research.

Cellular automata are systems made up ofmany computational units, each following asmall set (usually, the same set) of simple rules.In such systems, surprising patterns can emergefrom the simple, anodyne, base. Further vari-ations ensue if another ‘level’ of rules is addedto the system. Examples in CG-art include thetesselated visual constructs within PaulBrown’s Sandlines and Infinite Permutations,and other works by Paul Brown (Whitelaw2004, pp. 148–153; Tofts 2005, p. 85f.;www.paul-brown.com).

L-systems are automatically branchingstructures, used by botanists to study plant formand physiology (Lindenmayer 1968; Prusin-kiewicz and Lindenmayer 1990; Prusinkiewicz2004). In the hands of CG-artists, they have led(for instance) to McCormack’s Turbulenceinstallation (McCormack 2004; Tofts 2005,p. 80ff.). This generates images of un-naturalyet lifelike vegetation growing in front of one’seyes–and in response to one’s actions (thusqualifying as interactive art). Another exampleis the use of a ‘swarm grammar’ based onL-systems to generate structures in (simulated)3D-space, comparable to the decentralised yetorganised constructions of social insects suchas termites (Jacob and von Mammen 2007).

As for evolutionary programming, this hasgiven rise to an important sub-class of CG-art:evolutionary art, or Evo-art. Examples includeKarl Sims’ Genetic Images and Galapogos(Sims 1991, 2007), plus many others(Whitelaw 2004, ch. 2). In Evo-art, the artworkis not produced by a computer program that hasremained unchanged since being written by theartist. Rather, the artwork is (df.) evolved by

processes of random variation and selectivereproduction that affect the art-generatingprogram itself.

Evo-art relies on programs that include self-modifying processes called genetic algorithms.To begin, a ‘population’ of near-identicalartworks–or, to be more precise, the mini-programs that generate them–is produced bythe computer. There can be any number: 9, or16, or even more. In aesthetic terms, thesefirst-generation artworks are boring at best andchaotic at worst. Next, each of these first-generation programs is altered (‘mutated’) inone or more ways, at random. Usually, thealterations are very slight. Now, some selectiveprocedure—the ‘fitness function’ (decided bythe artist/programmer)—is applied to choosethe most promising candidate/s for breedingthe next generation. And this process goes onrepeatedly, perhaps for hundreds of gener-ations. Provided that the mutations allowed arenot too fundamental (see Section 5), whatensues is a gradual evolutionary progresstowards the type of structure favoured by theartist.

Occasionally, the fitness function is fullyautomatic, being applied by the computer itself.(If so, there may be scores, or even hundreds, of‘siblings’ in a given generation.) This is a primeexample of the computer’s being ‘left to do itsown thing’. More usually (as remarked inSection 3), the selection is done hands-on bythe artist—or by some other human being: agallery-visitor, for instance—using intuitive,and often unverbalised, criteria. (If so, thepopulation-size rarely rises above 16, becausepeople cannot ‘take in’ more than a limitednumber of patterns at once.) In other words,and for reasons touched on in Section 5, there isusually no programmed fitness function. Insuch cases, the Evo-art also counts as I-art.

One might argue that our definition ofEvo-art is faulty, on the grounds that evolution-ary art need not involve a computer. It’s certainlytrue that the very earliest G-art works ofWilliam Latham, who later became famous


33

Digita

lCreativity,

Vol.20,Nos.

1–2


as a computer artist, were generated by repeateddice-throwing and hand-drawing. At that time,he had no idea that computers might be able todo the job for him—and do it better (Todd andLatham 1992). But that is highly unusual:virtually all art that’s produced by an iterativeprocess of random variation plus selection iscomputer-based. Indeed, we do not know of anynon-computerised example besides early-Latham. (Someone might suggest ClaudeMonet’s water-lily series: but although theseshowed gradual improvement by way of smallchanges, those changes were far from random.)Even Richard Dawkins’ simple “Biomorphs,which were hugely seminal for Evo-artists, werecomputer-generated (Dawkins 1986, pp. 55–74). We have therefore chosen to define Evo-artas a sub-class of CG-art, even though thisexcludes the early-Latham efforts.

Another sub-class of CG-art is robot art, orR-art. By R-art, we mean (df.) the constructionof robots for artistic purposes, where robots arephysical machines capable of autonomousmovement and/or communication.We hope weshall be forgiven for not attempting to define“artistic purposes”. As for “autonomous”, theword may be understood intuitively here. Atsome point, however, it should be consideredcarefully—not least, because the concept ofautonomy is closely connected also withagency and creativity (see Section 5).

Clearly, not all R-art is Ele-art. Indeed,R-art covers examples built many centuriesago. A few of these ancient machines couldmove as a whole from one place to another–such as Leonardo’s mechanical lion that“walked vp and downe” the room, or Daedalus’mercury-filled Venus which (according toAristotle’s De Anima) had to be tethered toprevent it from running away. Most, however,could move only their body-parts–like themoving statues of nymphs and satyrs in thegrotto fountains at St. Germain, whichenthused Rene Descartes as a young man.

Electronic R-art is highly varied (Whitelaw2004, ch. 4). It includes Ihnatowicz’s eerily

hypnotic Senster, Stelarc’s thought-provokingman-robot hybrids and Ken Goldberg’sTeleGarden—wherein living plants are wateredby a robot that is controlled by Everyman viathe Internet (Popper 2007, pp. 379–393).

In all those cases, only one robot isinvolved. Sometimes, however, groups ofinteracting (‘distributed’) robots areconstructed. Usually, such groups employ thetechniques of situated robotics, wherein themachines respond directly to specific environ-mental cues–here, including the behaviour ofother robots (Boden 2006, p. 13.iii.b and iii.d).Occasionally, they exploit self-organisingtechniques whereby the system graduallyreaches an equilibrium state. (Futuristic thoughthey may seem, both these methodologies werefirst used by mid-century cyberneticians: GreyWalter and Ross Ashby, respectively.) Oneexample of the latter type is Jane Prophet’s NetWork installation (Bird, d’Inverno and Prophet2007). One might think of this as a hi-techversion of Dibbets’ oscillating sticks. Butinstead of eighty ‘isolated’ sticks, placed belowthe surface of the sea, Net Work consists of2500 floating, and intercommunicating,buoys–each of which is colour-emitting andwave-sensitive. (More accurately, it will consistin 2500 such buoys: it has been tested in a 3X3miniature on the Thames, but is planned tosurround the pier at Herne Bay.)

Such mutually interacting robot-groups donot count as interactive art on our definition(given below), unless they are also capable ofinteracting with the human audience. Net Workdoes have that capability: the audience canaffect it by shining torchlight on the buoys, orby remote control over the Internet. Otherexamples of interactive (and interacting) robot-groups include Kenneth Rinaldo’s works inwhat he calls eco-technology. His R-art (andI-art) installation called TheFlock comprisesthree wire-and-vine robotic ‘arms’ suspendedfrom the ceiling, which interact with each otherand with the moving/speaking human audi-ence. Similarly, his Autopoiesis has fifteen

Boden and Edmonds

34

Digita


1–2


robot wire-frame ‘arms’ distributed around theroom, which sense the observer’s movementsand communicate with each other so as tocoordinate their behaviour in various ways.

This brings us to our ninth category: inter-active art. In this genre, the human audience isnot a passive observer but an active participant.Audiences are never wholly passive, of course,since art-appreciation involves active psycho-logical processes. Indeed, Marcel Duchamp(1957) went so far as to say:The creative act is not performed by the artistalone; the spectator brings the work incontact with the external world by decipher-ing and interpreting its inner qualificationand thus adds his contribution to the creativeact (Duchamp, 1957).Even for Duchamp, however, the spectator’s

contribution concerns only the work’s “inner”qualification (its role, he said, is “to determine[its] weight on the aesthetic scale”). The work’sperceptible nature—or, many would say, theartwork itself—does not change as a result. Ininteractive art, by contrast, it does.

In I-art, then, (df.) the form/content of theartwork is significantly affected by the beha-viour of the audience. And in CI-art (i.e. thecomputer-based varieties), (df.) the form/content of some CG-artwork is significantlyaffected by the behaviour of the audience.Again, we are speaking intuitively here:worries about just what counts as “the artwork”are left to the next section. The word “signifi-cantly” is needed, even though it is a hostage tointerpretative fortune, so as to exclude per-formance art—for performance is usuallysubtly affected by audience reception. As forthe word “behaviour”, this must be interpretedwith generosity. In CI-art it covers voluntaryactions (such as waving, walking and touchingthe computer screen), largely automatic yetcontrollable actions (such as the direction ofeye-gaze) and involuntary bodily movements(such as breathing). It even includes arcanephysical factors such as the radiation ofbody-heat.

(Occasionally, the ‘interaction’ involves notthe audience but the physical environment:aspects of the weather, for example. Such casesfall outside our definition, unless–which isusually the case—they also involve interactionwith the human audience.)

CI-art is generative art by definition. But itis not ‘generative’ in our strictest sense (above),as AARON is. For although the artist can go tolunch and leave the program to do its ownthing, the audience cannot. However, it quali-fies as CG-art in our broader sense, since theartist has handed over control of the final formof the artwork to the computer, in interactionwith some other human being. The degree ofcontrol attributable to the audience varies: theymay not realise that they are affecting theartwork, nor (if they do) just what behaviourleads to just which changes. We’ll see later thatthis variability is an important dimension in theaesthetics of CI-art.

I-art is not an entirely recent phenomenon:remember Haydn’s dice-music, for instance.But it became prominent in the mid-twentiethcentury. (This was often justified in politicalterms: I-art was seen as offering valuablehuman-human communication, in societieswhere the sense of community had beendiluted—Bishop 2006.) It was made possiblelargely by cyberneticians such as Pask applyingtheir theory of communicative feedback to art,and by the new electronic technology devel-oped in World War II.

That’s not to say that all these I-art effortswere examples of Ele-art. Many artists,indeed, eschewed such technology for(counter-cultural) ideological reasons: it wastoo strongly linked with the military-industrialcomplex. Even Ascott’s first I-art had nary anelectron in sight: it consisted of canvases withitems/images on them that could be continu-ally moved around by hand, so that the viewerof the resulting collages was their maker too.SAM and the Senster were early examples ofI-art that did use electronics. But, as we haveseen, they did not involve computers.


35

Digita

lCreativity,

Vol.20,Nos.

1–2


Today’s I-art, however, is overwhelminglycomputer-based. That’s because the generalityof digital computers enables them, in principle,to support an infinite variety of human-computer interactions.

The types of interaction explored in CI-artare already widely diverse—hence theinclusiveness of the term ‘behaviour’ in ourdefinition. The by-now-countless examplesrange from interactive CD-Roms viewed on adesk-top and altered (for instance) by touchingthe screen (Leggett and Michael 1996), toroom-sized video or VR installations—such asChrista Sommerer and Laurent Mignonneau’sTrans Plant. In this case, a jungle graduallyappears on the walls as the audience movesaround the enclosure: grass grows when theviewer walks, and trees and bushes when he/she stands still; the plants’ size, colour, andshape depend on the size and bodily attitudes ofthe human being; and the colour densitychanges as the person’s body moves slightlybackwards or forwards. Trans Plant is drivenby the viewer’s movements, but some CI-art-works are modified also, or instead, by thesound of human voices or footsteps. This isreminiscent of the Senster—but these compu-ter-generated changes are much more variedand complex than those which could beengineered in the 1960s by Ihnatowicz.

Sometimes, the relevant interactions involveon-line access to the Internet, automaticallyincorporating into the artwork items that happento be present on the world-wide-web at thatparticular moment. One example is The LivingRoom, another installation built by Sommererand Mignonneau. Unlike Trans Plant, thisCI-artwork does not undergo changes thatdepend systematically on what the viewer isdoing. Instead, random images and sounds,picked up from the Internet, appear in the roomas a result of the viewer’s movements andspeech.

It’s usual, as in that example, for the changein the CI-artwork (whether systematic or not) tobe near-simultaneous with the observer’s

triggering activity. In Edmonds’ most recentCI-art, however, the effects of the viewer’sbehaviour are delayed in time (see Section 3).Partly because of the lesser likelihood that theviewer will realise—and be able to control—what is going on, Edmonds speaks of ‘influ-ence’ rather than ‘interaction’ in these cases. Aswe’ll see in Section 5, whether mere ‘influence’can be aesthetically satisfying is controversialeven outside the precious bubble.

Certainly, mere influence, as againstinstantaneous interaction, would not be enoughfor our final category, namely virtual veality art,or VR-art. VR-art is the most advanced versionof CI-art (for examples, see Popper 2007, chs.4–6). Already foreseen in the mid-1960s, itwas not technologically possible until thelate-1980s (Boden 2006, p. 13.vi).

In VR-art, interaction leads to illusion–ofan especially compelling kind. In other words,(df.) the observer is immersed in a computer-generated virtual world, experiencing it andresponding to it as if it were real. We do notpretend that this definition is clear: just what isit for someone to experience/respond “as if itwere real”? Some relevant issues will beindicated in Section 5. Meanwhile, we’ll con-tinue to rely on readers’ intuitive understandingof such language.

Someone might want to argue that VR-artwas initiated centuries ago. For pseudo-realisticmimetic worlds have been depicted in variousforms of trompe l’oeuil (including ‘realistic’panoramas) for many centuries and evenappeared in some of the wall-paintings andarchitecture of Classical Rome. But there’s acrucial difference between the relevant aes-thetics in times ancient and modern. As OliverGrau (2003, p. 16) has pointed out, the“moment of aesthetic pleasure” in trompel’oeuil comes when the viewer consciouslyrealises that they are not experiencing reality. InVR-art, by contrast, the enjoyment lies infeeling as though one is really inhabiting, andmanipulating, an alternative world. The longerthe awareness of its unreality can be delayed,

Boden and Edmonds

36

Digita


1–2


the better. In other words, the experience of pastforms of mimetic art was based only on illu-sion, not on immersion. Although one can saythat the viewers were invited/deceived intoresponding to the art as if it were real, that “asif” was much less richly textured, much lesssustained, and therefore much less persuasive,than it is now.

(Cinema is a half-way house—Grau 2003,ch. 4. It often elicits an emotional/narrative‘immersion’ in the filmgoer and sometimes—using special screens and techniques—leads tonear-veridical experiences of inhabiting thecinematic world. These tend to exploit ourreflex bodily responses to visual images thatsuggest falling, or imminent collision: so roller-coasters, white-water-rafting and tigers leapingtowards us out of the screen are familiarfavourites. But there’s little psychologicalsubtlety in this ‘inhabitation’ and no detailedinteraction with the alternate world—still less,any physical manipulation of it.)

In general, VR-art aims to make theparticipants (often called ‘immersants’) feel asthough they are personally present in thecyberworld concerned. Normally, this world isvisual or audio-visual, being presented on aVDU screen or projected onto the walls/floorof a real-world room. (McCormack’s UniversalZoologies VR-artwork is an exception: here,the images/sounds are projected onto two large‘talking heads’, in an attempt to provide arealistic illusion of human conversation–Tofts2005, p. 81f.). But sometimes, VR-art leadsalso to convincing experiences of touch,pressure and motion by providing the observerwith special gloves and other equipment(Boden 2006, p. 13.vi). Sometimes, too, theobserver experiences utterly unreal capacities,such as being able to fly or to activate highlyunnatural causal chains within the virtualworld.

Even when the viewer is not presented withsuch shockingly unfamiliar experiences asthose, something about the virtual world will beperceptibly unlike the real world. And this is

deliberate. For VR-artists are not aiming toachieve a fully detailed mimesis: what wouldbe the point of that? Rather, they use near-mimesis to cast some aesthetically/concep-tually interesting light on our usual experiencesand assumptions. (Detailed mimesis may beappropriate for other purposes, of course: forinstance, a VR-brain used in training neuro-surgeons provides nicely realistic sensations oftouch and vision when the trainee’s virtualisedsurgical tool prods, pinches, or cuts differentparts of it.)

In sum, our eleven definitions are asfollows:1. Ele-art involves electrical engineering

and/or electronic technology.2. C-art uses computers as part of the art-

making process.3. D-art uses digital electronic technology of

some sort.4. CA-art uses the computer as an aid (in

principle, non-essential) in the art-makingprocess.

5. G-art works are generated, at least in part,by some process that is not under theartist’s direct control.

6. CG-art is produced by leaving a computerprogram to run by itself, with minimal orzero interference from a human being.NB: We chose to reject the stricterdefinition of CG-art (art produced by aprogram left to run by itself, with zerointerference from the human artist).

7. Evo-art is evolved by processes of randomvariation and selective reproduction thataffect the art-generating program itself.

8. R-art is the construction of robots forartistic purposes, where robots are physicalmachines capable of autonomous move-ment and/or communication.

9. In I-art, the form/content of the artwork issignificantly affected by the behaviour ofthe audience.

10. In CI-art, the form/content of someCG-artwork is significantly affected bythe behaviour of the audience.


37

Digita

lCreativity,

Vol.20,Nos.

1–2


11. In VR-art, the observer is immersed in acomputer-generated virtual world,experiencing it and responding to it as if itwere real.

5 Some questions for aesthetics

Various aesthetic and/or philosophical pro-blems arise with respect to CG-art in generaland others with respect to particular varieties ofit. None of these can be explored at length here.(We have discussed them elsewhere, however.Some answers to the questions raised below areoffered in: Boden 1999, 2004, 200613.iii.d–e and 16.viii.c, 2007, in press (a,b,c);Boden and Paul Brown in preparation; Cornockand Edmonds 1973; Costello and Edmonds2007; Edmonds 2006 in press; Muller,Edmonds, and Connell 2006.) Instead, thissection merely indicates the wide rangeof puzzles that attend the consideration ofgenerative art.

One obvious question can be put informallythus: Is it really the case that a computer canever ‘do its own thing’? Or is it always doingthe programmer’s (artist’s) thing, howeverindirectly?

To answer that question seriously requiresboth general philosophical argument andattention to specific aspects of particular CG-artexamples—in the underlying program, as wellas in the observable artwork. That sort ofattention is not appropriate in cases of G-artthat are not computer-based. For the physical,psychological, or biological processes in whichthey are grounded are not specifiable indetail—not even by scientists, let alone byartists. Computer programs, in contrast, are sospecifiable. That’s why one can make sensiblecomparisons between the extent to whichdifferent CG-art programs are or are not ‘underthe artist’s direct control’ and the extent towhich, and the points at which, they are subjectto “interference from a human being”.However, one can do this only if one knowssomething about how the program and/orinstallation works. Merely observing, or even

participating in, the resultant artwork is notenough.

Whether it appears to participants that theprogram/installation is independent, or auton-omous, is of course another question (onewhich may not be easy to answer, in practice).So too is the question whether it ever makessense to ascribe autonomy to a computer.Remember Edmonds’ remarks about rule-based programming in Section 3: does it reallymatter what the ‘feel’ of this activity is, if infact it is no less directive, no less determinate,than algorithmic programming? Irrespective ofthe artist’s phenomenology while writing theprogram, and/or of the participants’ phenom-enology when experiencing it, do some cat-egories of CG-art have more autonomy thanothers? What of Evo-art, for instance: does theself-modification involved, and the automaticselection (in cases where that happens), meanthat evo-programs are more autonomous than(say) AARON? With respect to AARON, canwe ascribe at least a minimal level of autonomyto the computer, given that Cohen has nohands-on control over what picture will bedrawn, or how?

Insofar as a program is ‘doing its ownthing’, does it take on the authorial responsi-bility? (Let us ignore the fact that ‘authorialresponsibility’ is often unclear here anyway,since most CG-art is produced by a team, not alone artist.)

For instance, did AARON generate thosemagnificent ‘world-class’ coloured drawings,or did Cohen do so? He admits, after all, that hehimself “wouldn’t have had the courage to usethose colours”. On the other hand, he says he ishappy that there will be more of “his” originalartworks appearing well after his death (Cohen2002). Is he right, or is he deluded? The answerwill depend not only on one’s philosophy of theself but also on one’s views as to whether anycomputer program can be seen as an author/artist. Douglas Hofstadter, for example, wouldbe content to ascribe the posthumous works toCohen himself (and would even deny that they

Boden and Edmonds

38

Digita


1–2


are in the fullest sense posthumous (Hofstadter2007)); if he was emotionally moved by them,he would also resist ascribing authorship to thecomputer (2001).

Does Evo-art leave more room, or less, forhuman authorship than AARON does? That is,does the artist’s choice of fitness functionsuffice to give him/her the authorial credit forwhatever artwork emerges after many gener-ations of random (i.e. undirected) change? Isthe credit greater, or less, if instead of relyingon a programmed fitness function the artistdoes the selecting ‘by hand’?

One reason for the Evo-artist’s choosing todo the selection by hand is in order to produceonly works in his/her own style. This is alsothe reason why the mutations that are allowedare usually very slight. For an artistic style is asustained pattern of activity, lasting over time.In Evo-art that allows radical mutations (andwhich does not ‘ration’ them to once every2,000th generation, for instance), no patterncan be sustained for long—not even if thehuman artist is trying to shape the results bymaking the ‘fitness’ selection at each stage. Onthe contrary, huge structural changes can occurin a single generation (cf. Sims 1991). Thisleads to fascinated amazement on the part of thegallery audience. Nevertheless, Evo-artists donot normally allow such mutations. They preferto explore a stylistic space which, despite manysurprising variations, remains recognisble as‘theirs’ to someone with an experienced eye.

There are some exceptions. The CG-artistPaul Brown (with others, including the authors ofthis paper) is involved in an ongoing Evo-artproject whose aim is to evolve robots that willmake aesthetically acceptable drawings which donot carry Brown’s general style, or ‘personalsignature’ (Bird et al. 2007; Boden forthcoming).Whether that can be done remains to be seen:Brown, after all, will be setting the fitness func-tions as the work proceeds.

This project raises questions also aboutthe relation between CG-art and embodiment.Many philosophers of mind discount

AI/A-Life in general (as models of mind orlife) for being concerned with virtual,body-less, systems. However, these R/Evo-artcreatures are robots moving in the real worldand are therefore subject to physical forces. It’sknown that truly fundamental changes—i.e.new types of sensory receptor—can evolve inrobots as a result of unsuspected physicalcontingencies (Bird and Layzell 2002).(Compare the biological evolution not of aprimitive eye into a better eye, but of alight-sensor where no such sensor existedbefore.) In principle, then, a fundamentally newstyle [sic] might develop in this way, whereas(arguably) that could not happen in a purelyvirtual, programmed, system.

Similar puzzles about authorial responsibil-ity arise in CI-art in general, of which ‘hand-selected’ Evo-art is a special case. Just where, inthe man-machine system concerned, is the trueauthor? That worry affects all I-art, of course–but is there any extra difficulty where CI-art isconcerned? (For present purposes, let us ignoreDuchamp’s suggestion that all art is multi-authored.) And what difference, if any, does itmake if—as sometimes happens–the audienceprovides feedback during the construction of theCI-work, so that its final form depends not onlyon the decisions of the artist but also on thereactions of the audience/s who encountered itin its prototype stage? Perhaps our distinctionbetween “decisions” and “reactions” is crucialhere, debarring the audience from earning any‘extra’ authorial credit in such cases?

To speak of a ‘worry’ here, however, isperhaps to counteract what CI-artists are tryingto do. Despite its sturdy roots in cyberneticsand computer technology, CI-art has attractedfavourable notice from post-modernists pre-cisely because of the ambiguity of authorshipinvolved. Ascott (2003), in particular, hasalways seen the value of CI-art as itsdemocratising ability to engage the viewer/participant as creator. In his words, “creativityis shared, authorship is distributed. . .” (1990,p. 238). If authorship is deliberately distributed,


39

Digita

lCreativity,

Vol.20,Nos.

1–2


then to worry about its locus (about ascribingthe status of author) is to miss the point.

(For all the heady talk of creativeparticipation, some CI-art is fairly limiting(Kahn 1996). That’s so, for instance, where thepossible changes in the artwork are explicitlypre-set by the artist, as opposed to theiremerging from the program’s “doing its ownthing”. The limitation is especially great wherethey are selected by the participant’s choosingfrom a Menu.)

Another way of putting questions aboutauthorial responsibility is to ask where thecreativity lies. But what, exactly, do we meanby creativity?

It certainly involves agency—which is whyconsiderations of autonomy and authorialresponsibility are inevitable. But what isagency? The interacting ‘arms’ and floatingbuoys identified above as examples of R-art aretypically described by the artists and tech-nicians concerned as agents—a word borrowedfrom AI/A-Life research on distributed cogni-tion. But does that research misuse the concept?Even if it does, does it include ‘agents’ ofinterestingly different types (Boden 2006,p. 13.iii.d-e), some of which are moredeserving of the name than others? If so, shouldwe at least reserve the term–and the ascriptionof creativity–for those cases of CG-art wherethe agents involved are of the more plausiblevariety? Again, such questions cannot beanswered without careful attention to thedetails of the programs and communicationsinvolved.

It’s commonly assumed that creativity—and art, too—involves unpredictability (Boden2004, ch. 9). But what is its source? Is it merelylack of predictive power on the part of humanminds? We have seen (in Section 3) that CG-art, like complex programs in general, is indeedunpredictable for that reason. But CI-art andEvo-art are unpredictable for other reasons aswell. CI-art, because the artist cannot predictthe sequence of interactions that will take place,even if he/she can predict what would happen

at a given moment if that audience-movementwere to occur; and Evo-art, because of themany random changes to the program andbecause of the choices made at successivegenerations by the artist. Does the unpredict-ability of traditional art have any deepersource? And if so, is this something whichcannot be ascribed to, or even simulated in,computers?

Another set of questions concerns ontology.How can we identify ‘the artwork’ when anartist’s computer program generates countlessunique images, or musical compositions, noneof which have been seen/heard by the artist? Iseach image/music produced by AARON orEmmy an artwork–or is the artwork theprogram which generates them? In Evo-art,does one and the same artwork exist at differinglevels of sophistication at different generations?Or does every generation produce a newartwork—or, perhaps, a new population of(sibling) artworks?

What counts as the artwork when theuniqueness is due not only to a richly genera-tive computer program but also to the contin-gent (and ephemeral) behaviour of aparticipatory human audience? Perhaps thefamiliar concept of artwork is well-suited onlyto the unchanging artefacts that form theoverwhelming majority of the cases insideMcCormack’s bubble? A traditional artist canfully comprehend the painting or sculpture thatthey have executed so carefully (althoughwhether this applies to the G-art dimension ofPollock’s paintings is questionable), butCI-artists cannot fully know the CI-artwork thatthey constructed with equal care. This is notmerely a matter of the unpredictability of detail:in sufficiently complex cases, it is not evenclear that they can recognise the generalpotential of their own work. With regard toCI-art, then, perhaps we should speak not of the‘artwork’ but of the ‘art system’—where thiscomprises the artist, the program, the techno-logical installation (and its observable results),and the behaviour of the human audience?

Boden and Edmonds

40

Digita


1–2


(And perhaps, if the concept of the ‘artwork’falls, then that of the ‘artist/author’ falls too?)

Or maybe we should think of each occur-rence of CI-art as a performance and theprogram/installation as the score? If so, the‘performance’ is more like a jazz improvisationthan the playing of classicalmusic, for it canvaryconsiderably fromone occasion to another. Evenif the form of each particular human-computerinteraction can be completely determined by theartist (which is not so, for instance, when thecomputer’s response can be modified by itsmemory of the history of previous interactions),the sequence of such events cannot.

Yet another problematic area concernsaesthetic evaluation. Are entirely novelaesthetic considerations relevant for CG-art ingeneral, or for some subclass of it? And aresome aesthetic criteria, normally regarded asessential, utterly out of place in CG-art:authenticity, for instance?

The devotees of CI-art, in fact, do not usethe familiar (inside-the-bubble) criteria to judgedifferent interactive installations. Or insofar asthey do, these are secondary to other con-siderations. The criteria they see as mostappropriate concern not the nature of theresulting ‘artwork’ (the beauty of the imageprojected on the wall, for example, or theharmoniousness of the accompanying sounds),but the nature of the interaction itself. There’sgeneral agreement on that point. But there’ssignificant disagreement on just what type ofinteraction is the most aesthetically valuable.

Some CI-artists, especially those engaged inVR-art, stress the disturbing sense of unrealityinvolved, and the participant’s new ‘take’ oneveryday experience that ensues. Many valuethe participant’s conscious control of theartwork; others aim to highlight their sense ofpersonal embodiment; while yet others stressthe audience’s disconcerting experience ofunpredictability triggered by their own actions.All of those criteria concern the participant’sexperience —but difficulties arise if one askshow that experience can be discerned, or

‘logged’, by anyone other than the individualparticipant. (As we saw in Section 3, if theobservers can never come to realise that theyare affecting what happens, then the ‘I’ in‘CI-art’ might better be thought of as the initialletter of ‘influence’, not of ‘interaction’.)

There are some especially jagged philoso-phical rocks lying in wait for VR-artists. Theconcept of virtual reality has been defined invarious ways (Steuer 1992). Most, like ourdefinition of VR-art, refer to the participant’sexperience of being immersed in a real worldand reacting accordingly. This notion seems tobe intuitively intelligible, especially if one hasactually encountered a VR-installation. But justwhat it means, whether in psychological orphilosophical terms, is very difficult to say. It isnot even clear that it is coherent. Several leadingphilosophers of mind have addressed thishornet’s nest of questions in writing about thefilm The Matrix (see especially the papers byHubert Dreyfus and Andy Clark on the WarnerBrothers website: http://whatisthematrix.warnerbros.com). That’s not to say that The Matrixcounts as VR-art, for it does not. Nevertheless, itraises some of the same questions that wouldattend highly plausible instances of VR-art.(Whether these would also be highly successfulinstances is another matter: we have seen thatVR in art typically highlights some unrealdimension of the experience.)

As for authenticity, this is a tricky concept.There are several reasons, of varying plausi-bility, why someone might argue that it is notapplicable to any instance of CG-art. And CG-artists have suffered as a result. For example,Cope (2006) has been continually disappointedby people’s failure to take his music seriously–not because they dislike it on hearing it(sometimes they even refuse to hear it), butsimply because it is computer-generated.Even when they do praise it, he has foundthat they typically see it less as ‘music’ thanas ‘computer output’—a classification whichcompromises its authenticity. For instance,even though each Emmy-composition is in fact


41

Digita

lCreativity,

Vol.20,Nos.

1–2


unique, people know that the program couldspew out indefinitely many more tomorrow.(The fact that human composers die, says Cope,has consequences for aesthetic valuation:someone’s oeuvre is valued in part because it isa unique set of works, now closed.) As a resultof this common reaction, Cope has recentlydestroyed the data-base of dead composers’music that he had built up over the last twenty-five years and used as a crucial source inEmmy’s functioning. (Emmy’s successor willcompose music only in Cope’s own style;whether audiences regard this is significantlymore authentic remains to be seen.)

Finally, what of the claims made by manyCG-artists to be exploring the nature of life? It’sclear from Rinaldo’s choice of the titlesAutopoiesis and The Flock (plus the rest of hisoeuvre—Whitelaw 2004, pp. 109–116), forinstance, that his R-art works are not intendedas mere fairground toys but as meditations onsignificant aspects of life. He is not alone inthis: many of the CG-artists who have beeninfluenced by A-Life see their work in thatway. Concepts such as emergence and self-organisation, and of course evolution, crop uprepeatedly in their writings and interviews—asdoes the key concept of life itself. One maywell agree that their work throws light on, oranyway reminds us of, important—and puz-zling—properties of life. But one need not alsoagree with the claim sometimes made by theseCG-artists, that purely virtual life (a.k.a. strongA-Life) is possible—and that their work, orsomething similar, might even create it.

6 Conclusion

We have mentioned a number of philosophicalquestions. But we have ignored what is perhapsthemost obvious one of all: “But is it art, really?”

Many people feel that computers are the veryantithesis of art. Indeed, some philosophersargue this position explicitly (e.g. O’Hear 1995).On their view, art involves the expression andcommunication of human experience, so that if

we did decide that it is the computer which isgenerating the ‘artwork’, then it cannot be an artwork after all. A closely related worry concernsemotion in particular: if computers are notemotional creatures then—on this view—theycannot generate anything that’s properly termed‘rt’ (Hofstadter 2001). Another common way ofdiscrediting computer art in general is to arguethat art involves creativity and that no compu-ter–irrespective of its observable perform-ance—can really be creative (for a discussion,see Boden 2004, ch. 11). And both authors ofthis paper have often observed someone’saesthetic approval of an artwork being instantlyrenounced on discovering that it is, in fact, aCG-artwork. Cope was so disturbed by thisreaction, as we have seen, that he destroyed thedata-base on which Emmy’s—or should werather say ‘his’?—CG-music rested.

It would not be appropriate to burden thisalready lengthy paper with a discussion of theslippery concept of art. But it is not necessary,either. For we have given many illustrations ofthe continuities between CG-art and non-computer art. Several of our analytic categoriesinclude examples drawn from both inside andoutside McCormack’s precious bubble–although, admittedly, most cases of traditionalart elude our definitions. And those categorieswhich apply only to CG-art cover manyindividual cases that are aesthetically related totraditional artworks.

Moreover, the art world itself–howeversuspicious it may be of computers in general,and however dismissive it may be of particularCG-art efforts–does sometimes award thesenew activities the coveted status of art. Some-times, this happens in a specialised corner ofthe art world: for instance, London’s Kineticagallery (opened in 2006), which is devoted tointeractive, robotic and kinetic art. But we havealso mentioned two examples (others couldhave been cited) where major ‘traditional’galleries clearly accept that traditional andCG-art are players in the same culturalballpark. These were the Tate’s one-man show

Boden and Edmonds

42

Digita


1–2


of Cohen’s AARON and the Washingtonexhibition featuring Edmonds’ work as adevelopment of that of the ColorField painters.The latter example is especially telling,precisely because it is not a show celebratingonly CG-art. On the contrary, the Washingtonexhibition is putting CG-art alongside theprecious bubble—or even inside it.

In response to the sceptic’s challenge “Butis it art, really?”, we therefore rest our case.

Acknowledgements

We are grateful for helpful comments given tous by Dustin Stokes, of the Centre forCognitive Science at the University of Sussex.This paper forms part of the research supportedby AHRC Grant no. B/RG/AN8285/APN19307: Computational Intelligence,Creativity, and Cognition: A MultidisciplinaryInvestigation. The work was also partly fundedby the Australasian CRC for InteractionDesign, which is established and supportedunder the Australian Government’s ResearchCentres Programme.

ReferencesAscott, R. (1966/1967) ‘Behaviourist art and thecybernetic vision’, Cybernetica, vol. 9 (1966), pp.247–264; and vol. 10 (1967), pp. 25–56. Rep-rinted as ch. 3 of Ascott, R. (2003), pp.109–156.

Ascott, R. (1990) ‘Is there love in the telematicembrace?’, Art Journal (College Arts Associationof America), vol. 49, no. 3, pp. 241–247. Rep-rinted as ch. 15 of Ascott, R. (2003), pp. 232–246.

Ascott, R. (1998/2003) ‘Technoetic aesthetics: 100terms and definitions for the post-biological era’ inAscott, R. (2003, pp. 375–382). First published inJapanese, trans. E. Fujihara in R. Ascott (1998) Artand telematics: toward the construction of a newaesthetics, NTT Publishing Co., Tokyo.

Ascott, R. (2003) Telematic embrace: visionarytheories of art, technology, and consciousness,University of California Press, London.

Bird, J., d’Inverno, M., and Prophet, J. (2007) ‘NetWork: an interactive artwork designed using an

interdisciplinary performative approach’, DigitalCreativity, vol. 18, no. 1, pp. 11–23.

Bird, J., and Layzell, P. (2002) ‘The evolved radioand its implications for modelling the evolutionof novel sensors’, Proceedings of Congresson Evolutionary Computation, CEC-2002,pp. 1836–1841.

Bird, J. and Stokes, D. (2006) Evolving minimallyCreative Robots. In S. Colton and A. Pease (eds.)Proceedings of the 3rd International Workshop onComputational Creativity (ECAI’06), pp. 1–5.

Bishop, C. (2006) Participation, MIT Press,Cambridge Mass.

Boden, M.A. (1999) ‘Is metabolism necessary?’,British Journal for the Philosophy of Science,vol. 50, no. 2, pp. 231–248.

Boden, M.A. (2004) The creative mind: myths andmechanisms, Routledge, London, 2nd edn.,expanded/revised.

Boden, M.A. (2006) Mind as machine: a history ofcognitive science, Clarendon/Oxford UniversityPress, Oxford.

Boden, M.A. (2007) ‘Authenticity and computerart’, Digital Creativity, vol. 18, no. 1, pp. 3–10.(Special Issue on Creativity, Computation, andCognition, ed. P. Brown.)

Boden, M.A. (in press, a) ‘Aesthetics and interactiveart’ in Art, body, embodiment, eds R. L. Chrisley,C. Makris, R. W. Clowes and M. A. Boden,Cambridge Scholars Press, Cambridge.

Boden, M.A. (in press, b), ‘Stillness as autonomy’,to appear in CADE-2007, Perth.

Boden, M.A. (in press, c) ‘Creativity and artificialevolution’ in Creativity: the mind, machines, andmathematics, eds J. Copeland and R. A. Brooks,MIT/ Templeton Press, Cambridge, Mass.

Boden, M.A. (forthcoming) ‘Personal signaturesand evolutionary art’, In M.A. Boden, ThreeRoads to Surprise: Essays on Creativity and Art.

Brown, P. (2003) ‘Generative computation and thearts’, Digital Creativity, vol. 14, no. 1, pp. 1–2.(Special issue on Generative Computation and theArts, ed. P. Brown.)

Brown, R. (2001) Biotica: art, emergence, andartificial life, RCA CRD Research, London.

Brown, R. (2006) ‘The electrochemical glass’,Strange Attractor Journal Three, available from


43

Digita

lCreativity,

Vol.20,Nos.

1–2


the Strange Attractor Shoppe, ,http://www.strangeattractor.co.uk.

Breton, A. (1969) Manifestoes of surrealism(University of Michigan Press, Ann Arbor),trans. R. Seaver and H.R. Lane. (Includes severalmanifestoes, the first published in 1924.)

Burnham, J. (1968) Beyond modern sculpture: theeffects of science and technology on the sculptureof this century, Allen Lane, London.

Candy, L. and Edmonds, E.A. (eds) 2002)Explorations in art and technology, Springer,London.

Cohen, H. (1995) ‘The further exploits of AARONpainter’ in Constructions of the mind: artificialintelligence and the humanities. special edn. ofStanford Humanities Review, eds S. Franchi andG. Guzeldere, vol. 4, no. 2, pp. 141–160.

Cohen, H. (2002) ‘A million millennial medicis’ inExplorations in art and technology, eds. L. Candyand E. A. Edmonds (2002), pp. 91–104.

Cope, D. (2001) Virtual music: computer synthesisof musical style, MIT Press, Cambridge, Mass.

Cope, D. (2006) Computer models of musical crea-tivity, MIT Press, Cambridge, Mass.

Cornock, S. and Edmonds, E.A. (1973) ‘The creativeprocesswhere the artist is amplifiedor supercededbythe computer’, Leonardo, vol. 6, no. 1, pp. 11–16.

Costello, B. and Edmonds, E.A. (2007) ‘Playfulpleasures: a user evaluation of three artworks’,University of Technology, Sydney: CCS Report2007–1.

Dawkins, R. (1986) The blind watchmaker,Longman, Harlow.

Dorin, A. and McCormack, J. (2001) ‘First iteration/generative systems (Guest Editors’ Introduction)’,Leonardo, vol. 34, no. 3, p. 335.

Duchamp, M. (1957) ‘The creative act’ in Theessential writings of Marcel Duchamp, edsM. Sanouillet and E. Peterson, Thames andHudson, London, 1975, pp. 138–140.

Duprat, H. and Besson, C. (1998) ‘The wonderfulcaddis worm: sculptural work in collaborationwith trichoptera’, Leonardo, vol. 31, no. 3,pp. 173–182.

Edmonds, E.A. (1969) ‘Independence of rose’saxioms for m-valued implication’, Journal ofSymbolic Logic, vol. 34, no. 2, pp. 283–284.

Edmonds, E.A. (1988) ‘Logic and time-basedart practice’, Leonardo, Electronic ArtSupplemental issue (Oxford: Pergamon Press),pp. 19–20.

Edmonds, E.A. (2003) ‘Logics for constructinggenerative art systems’, Digital Creativity, vol. 14,no. 1, pp. 23–38.

Edmonds, E.A. (2006) ‘New directions in interactiveart collaboration’, CoDesign, vol. 2, no. 4,pp. 19–194.

Edmonds, E.A. (2007) ‘Reflections on thenature of interaction’, CoDesign, vol. 3, no. 3,pp. 39–143.

Eno, B. (1996) Generative music: evolvingmetaphors, in my opinion, is what artists do.Available at: http://www.inmotionmagazine.com/eno1.html

Goodstein, R. L. (1951) Constructive formalism,essays on the foundations of mathematics, Uni-versity College Press, Leicester.

Grau, O. (2003)Virtual art: from illusion to immer-sion, MIT Press, Cambridge, Mass.

Hiller, L. and Isaacson, L. (1958) ‘Musicalcomposition with a high-speed digital computer,’Journal of the Audio-Engineering Society, vol. 6,no. 3, pp. 154–160.

Hofstadter, D.R. (2001) ‘Staring Emmy straight inthe eye - and doing my best not to flinch’, Cope(2001), pp. 33–82.

Hofstadter, D.R. (2007) I am a strange loop, BasicBooks, New York.

Jacob, C. and von Mammen (2007) ‘Swarmgrammars: growing dynamic structures in 3D agentspaces’, Digital Creativity: Special Issue onComputational Models of Creativity in the Arts,vol. 18, no. 1, pp. 54–64.

Johnson, B.S. (1969) The unfortunates, Panther,London.

Kahn, D. (1996) ‘What now the promise?’ inBurning the interface, exhibition catalogue oninternational artists’ CD-Rom, Museum ofContemporary Art, Sidney, eds M. Leggett andL. Michael, pp. 21–30.

Kowalski, T.L. and Levy, L.S. (1996) Rule-basedprogramming, Kluwer Academic, Amsterdam.

Krueger, M. W. (1991) Artificial reality, 2nd edn.,Addison-Wesley, Reading, Mass.

Boden and Edmonds

44

Digita


1–2


Laposky, B. (1969) ‘Oscillons: electronic abstrac-tions’, Leonardo, vol. 2, no. 4, pp. 345–354.

Leggett, M. (2000) ‘Thinking imaging software’,Photofile: Photomedia Journal, no. 60, pp. 28–31.

Leggett, M., and Michael, L. (eds.) (1996) Burningthe interface , international artists’ CD-Rom,Museum of Contemporary Art, Sidney.

LeWitt, S. (1967) ‘Paragraphs on conceptual art’,Artforum, vol. 5, no. 10, pp. 79–83. Reprinted inTheories and documents of contemporary art: asourcebook of artists’ writings, eds K. Stiles andP. Selz, University of California Press, London,pp. 822–826.

LeWitt, S. (1969) ‘Sentences on conceptual art’,Art-Language, vol. 1, pp. 11–13. Reprintedin Theories and documents of contemporaryart: a sourcebook of artists’ writings,eds L.R. Lippard, 1973, p. 75f, K. Stiles andP. Selz, University of California Press, London,pp. 826–827.

Lindenmayer, A. (1968) ‘Mathematical models forcellular interaction in development, Parts I and II’,Journal of Theoretical Biology, vol. 18, no. 3,pp. 280–315.

Lippard, L.R. (1973) Six years: the dematerializa-tion of the art object from 1966 to 1972, Praeger,New York. Page-references to the 1997 reprint,University of California Press, London.

McCorduck, P. (1991) AARON’s code: meta-art,artificial intelligence, and the work of HaroldCohen, W. H. Freeman, New York.

McCormack, J. (2003) ‘Art and the mirror of nature’,Digital Creativity, vol. 14, no. 1, pp. 3–22.(Special issue on Generative Computation and theArts, ed. Paul Brown.)

McCormack, J. (2004) Impossible nature: theaArt ofJohn McCormack, Australian Centre for theMoving Image, Melbourne.

McCormack, J., Dorin, A. and Innocent, T. (2004)‘Generative design: a paradigm for designresearch’ in Proceedings of Futureground, eds J.Redmond, D. Durling and A. de Bono, DesignResearch Society, Melbourne, pp. 156–164.

Mallen, G. (2005) ‘Reflections on Gordon Pask’sadaptive teaching concepts and their relevance tomodern knowledge systems’ in Creativity andCognition 2005. Proceedings of the Fifth Confer-ence on Creativity and Cognition, 12–15 April,

Goldsmiths College, ed. L. Candy, ACM Press,New York, pp. 86–91.

Marr, A. (forthcoming) ‘He thought it the deuil,whereas indeede it was a meere mathematicallinuention’: understanding automata in the lateRenaissance’ in The Sistine gap: essays in thehistory and philosophy of artificial life, ed. J.Riskin, Chicago University Press, Chicago.

Martin, K. (1951/54) ‘Abstract art’, Broadsheet No.1: Abstract Paintings, Sculptures, Mobiles (London:Lund Humphries, 1951), AIA exhibition catalogue;also in L. Alloway, Nine Abstract Artists, TheirWork and Their Theory: Robert Adams, Terry Frost,Adrian Heath, Anthony Hill, Roger Hilton, KennethMartin, Mary Martin, Victor Pasmore, William Scott(London: Tiranti, 1954).

Metzger, G. (1965) ‘Auto-destructive art’ (London:Destruction/Creation). Expanded version of a talkat the Architectural Association, updating themanifesto of 1959.

Muller, L., Edmonds, E.A. and Connell, M. (2006)‘Living laboratories for interactive art’, CoDesign,vol. 2, no. 4, pp. 195–207.

Nake, F. (1998) ‘Art in the time of the artificial’,Leonardo, vol. 31, no. 3, pp. 163–164.

Nake, F. (2005) ‘Computer art: a personal recollec-tion’ in Creativity and Cognition 2005. Proceed-ings of the Fifth Conference on Creativity andCognition, 12–15 April, Goldsmiths College, ed.L. Candy, ACM Press, New York, pp. 54–62.

Nees, G. (1969) Generative Computergraphik,Siemens AG, Berlin.

O’Hear, A. (1995) ‘Art and technology: an oldtension’ in Philosophy and Technology, ed. R.Fellows, Cambridge University Press, Cambridge,pp. 143–158.

Pask, G. (1971) ‘A comment, a case-history, and aplan’ in Cybernetics, art, and ideas, ed. J. Reich-ardt, Studio Vista, London, pp. 76–99.

Popper, F. (2007) From technological to virtual art,MIT Press, Cambridge, Mass.

Pinch, T. and Trocco, F. (2002) Analog days: theinvention and impact of the moog synthesizer,Harvard University Press, Cambridge, Mass.

Prusinkiewicz, P. (2004) ‘Art and science for life:designing and growing virtual plants withL-systems’ in Nursery Crops: Development,Evaluation, Production and Use. Proceedings of


45

Digita

lCreativity,

Vol.20,Nos.

1–2


the XXVI International Horticultural Congress,eds C.G. Davidson and T. Fernandez, ActaHorticulturae, 630, pp.15–28.

Prusinkiewicz, P. and Lindenmayer, A. (1990) Thealgorithmic beauty of plants, Springer-Verlag,New York,

Sims, K. (1991) ‘Artificial evolution for computergraphics’, Computer Graphics, vol. 25, no. 4,pp. 319–28.

Sims, K. (2007),Galapogos Interactive Exhibit,http://www.genarts.com/galapogos/index.html..

Smith, M. (ed.) (2005) Stelarc: the monograph, MITPress, Cambridge, Mass.

Steuer, J. (1992) ‘Defining virtual reality: dimen-sions determining telepresence’, Journal ofCommunication, vol. 42, no. 4, pp. 73–93.

Todd, S.C. and Latham, W. (1992) Evolutionary artand computers, Academic Press, London.

Tofts, D. (2003) ‘Avatars of the tortoise: life, long-evity, and simulation’, Digital Creativity, vol. 14,no. 1, pp. 54–63. (Special Issue on GenerativeComputation and the Arts, ed. Paul Brown.)

Tofts, D. (2005) Interzone: media arts in Australia,Craftsman House, Fisherman’s Bend, Victoria.

Webster, A. (2006-ongoing) Solar Stacking. (Anelectrochemical sculpture.) See ,http://www.andywebster.info/solar/.

Webster, A. and Bird, J. (2002) Tuning Pask’s Ear.Video (7 : :26) exhibited in several galleries.

Whitelaw, M. (2004), Metacreation: art andartificial life, MIT Press, London.

Wilson, S. (2002) Information arts: intersections of art,science, and technology,MITPress,Cambridge,Mass.

Woolf, S. (2004) ‘Expanded media: interactive andgenerative processes in new media art’. Unpub-lished DPhil thesis, Dept. Informatics, Universityof Sussex.

Xenakis, I (1971) Formalized music:thought andmathematics in composition, Indiana UniversityPress, Bloomington.

Zivanovic, A. (2005) ‘The development of a cyber-netic sculptor: Edward Ihnatowicz and the Senster’in Creativity and Cognition 2005.Proceedings ofthe Fifth Conference on Creativity and Cognition,12–15 April, Goldsmiths College, ed. L. Candy,ACM Press, New York, pp. 102–108.

Margaret A. Boden is Research Professor ofCognitive Science at the University of Sussex.She is a Fellow of the British Academy andof the American Association for ArtificialIntelligence (and its British and Europeanequivalents). In 2002 she was awarded an OBE“for services to cognitive science,” and (besidesher Cambridge ScD and Harvard PhD) she hashonorary doctorates from the Universities ofBristol and Sussex and the Open University.Her work is highly interdisciplinary and hasbeen translated into 20 languages. Her latestbooks are The creative mind (2nd edn.,expanded: Routledge 2004) and Mind asmachine: a history of cognitive science (Oxford2006).

Ernest A. Edmonds is Research Professor ofComputation and Creative Media at theUniversity of Technology, Sydney. He has aPhD in Logic from the University of Nottinghamand is a Fellow of the British Computer Societyand the Institution of Engineering andTechnology. As well as publishing widely in theareas of human-computer interaction andcreativity, he is a practicing artist, first using acomputer in that practice in 1968. He hasexhibited throughout the world, from Moscowto LA. He co-founded the ACM Creativity andCognition conference series and is Editor-in-Chief of the Leonardo journal’s Transactions.In 2005 Artists Bookworks (UK) published hislatest book On new constructs in art.

Boden and Edmonds

46

Digita


1–2


digital creativity what is generative art? - compart...

Documents