the case of the occasionally cheap computer pal, patra, nedevschi

49

Joyojeet Pal*[email protected] AssociateCenter for Information andSociety, U. of Washington4311 11th Avenue NE, Ste 400Seattle, WA 98195 USA(206)221-0558

Rabin Patra*[email protected]. CandidateDepartment of EECSU. of California, Berkeley2150 Shattuck Ave, PenthouseBerkeley, CA 94704 USA(510)717-2999

Sergiu NedevschiPostdoctoral ResearcherInternational Computer ScienceInstitute, Berkeley1947 Center St, Suite 600Berkeley, CA 94704 USA(510)282-4253

Madelaine PlaucheResearch AssociateHuman Language TechnologiesResearch Group, Meraka Inst.CSIR Site—Building 43, MeiringNaude RoadBrummeria, Pretoria 0001South Africa�27 128413028

Udai PawarAssistant ResearcherTechnologies for EmergingMarkets GroupMicrosoft Research IndiaScientia, 196/36 2nd MainSadashivnagarBangalore 560080India�91 80 66586000

The Case of the Occasionally Cheap Computer PAL, PATRA, NEDEVSCHI, PLAUCHE, PAWAR

Research Article

The Case of the OccasionallyCheap Computer:Low-cost Devices and Classroomsin the Developing WorldAbstract

The quest for the low-cost computer has been one of the most signiªcant pur-suits of ICTD since the 1990s. This article examines the experiences of low-costcomputing projects in developing regions and looks at some of the commonentrepreneurial and technical problems faced by past and current initiatives.Focusing speciªcally on the domain of education, we look at the quest forlow-cost devices and consider their economic and socio-cultural appropriate-ness to the typical classroom in the developing world. Using ªeld studies andinterviews conducted in rural Indian classrooms, we show that shared ratherthan single-user devices constitute a more realistic and sustainable approachfor low-cost computing projects targeting children’s education.

I. IntroductionThe past decade has seen a number of attempts to develop low-cost com-puters as a means of greater access to technology for underserved popu-lations. Such projects have employed a range of strategies at designinginterfaces and infrastructure differently to serve the unique needs of de-veloping regions and populations with limited technology and textual lit-eracy experience. At the same time, the devices themselves are designedas inexpensively as possible. However, most projects have fared poorly inmarkets, despite the apparent enthusiasm for many such initiatives. In thisresearch, we look at a number of projects in the low-cost computing do-main and examine them through the frame of how technology is de-signed and used in classrooms in the developing world. As our empiricalbase, we use research from our several years working with schools usingcomputer centers that indicated a salient point: the most striking short-coming, when considering technology in the classroom, has been the sin-gle-user model of computer use. As the starting point of our argument,we introduce evidence that computers tend to be shared by childrenthroughout the developing world and present further evidence that thissharing is not just one of economic need, but is strongly embedded in cul-tural approaches to learning and asset use.

Our argument here is divided broadly into three parts and looks pri-marily at computer use for young children. First, we discuss brieºy thereasons behind the recent boom in thought about technology and devel-opment and look speciªcally at major initiatives in the Computers for thePoor domain. We also take a detailed look at some of the practical rea-

*Corresponding Authors

© 2009 USC Annenberg School for Communication. Published under Creative Commons Attribution-NonCommercial-Share Alike 3.0 Unported

license. All rights not granted thereunder to the public are reserved to the publisher and may not be exercised without its express written permission.

Volume 5, Number 1, Spring 2009, 49–64

sons why some of these initiatives, despite their in-novation and technological sturdiness, have failedand at what newer initiatives may be doing differ-ently. Second, we look at sharing of computers, spe-ciªcally at prevalent trends in classrooms of thedeveloping world, especially in India where a bulk ofthis research has been done. Finally, after looking atthe economics of sharing versus single-device mod-els of technology, we turn to the mechanics of shar-ing to examine whether this offers any realisticlearning games worth pursuing as a serious, practi-cal option for computing.

II. BackgroundBefore we look at the speciªcs of why computers inschools for the developing world are at the centerstage of contemporary development discourse, weturn brieºy to the recent history of development,speciªcally at the perceived role of technology ingrowth. We believe that several of the key macro-and micro-economic propositions on the value oftechnology in engineering growth have contributedto a rise in interest from engineering, industry, andphilanthropy in developing low-cost computers.

A. IT and DevelopmentThe relationship between technology and develop-ment has been a consistent theme in social sciences.At a macro-economic level, the high-tech industryhas been cast as an important engine for regionalgrowth, while, at a micro-economic level, computershave been linked with human development. Theacademic community has been interested in themacro-economic aspect of this relationship sincethe 1960s with the emergence of a large body ofliterature that explored the knowledge economy(Machlup, 1962), especially in Japan, where work onKnowledge Society studies had emerged as a majorªeld around the time of its gaining competitivenessin the global economy (Masuda, 1981). By the late1970s and 1980s, this interest had taken on the hueof popular literature on post-Industrial economies(Bell, 1976; Gouldner, 1979) and a broader body ofknowledge on Information Society theory (Duff,2000). In the 1990s, the birth and rather rapid ex-plosion of the Internet renewed interest in the ideaof a changing world across scholarly and industrylines, in large part due to the interest of importantthinkers from all sides of the spectrum (Castells,1996; Negroponte, 1996), including, importantly, bya number of international development agencies

(Hanna, 1994; Wehn & Mansell, 1998; Nulens &Van Audenhove, 1999; Gaiha, 2001). In addition tothis idea of technology as a potential means of fur-thering development being fairly well represented,the work of management gurus like C. K. Prahaladon the bottom of the pyramid markets (Prahalad &Hammond, 2002) suggested that technology anddevelopment may not just be of interest to philan-thropy or government, but may indeed have proªt-making consequences for companies that choose todevelop products for an untapped market furtherdown the value chain.

While the late 1990s had already seen a few ini-tiatives in technology for low-income regions, we ar-gue that it was a burst of interest among engineersand scientists in issues of technology in the develop-ing world that created a signiªcant push and a sub-sequent slew of projects in this arena. This increasein interest started in the early 2000s with severalleading academics from key universities formingsmall research groups and publishing works in thegeneral area of technological innovation for theneeds of the developing world (Parikh, Ghosh et al.,2003; Bhagwat, Raman et al., 2004; Pentland,Fletcher et al., 2004; Brewer, Demmer et al., 2005).In addition, concrete international research partner-ships in technology and development emerged, anotable example being MIT’s Media Lab Asia, whichset up shop in India. Also around this time, researchwings in major technology groups, includingHewlett Packard Labs and Microsoft Research, setup establishments in the developing world, whichnaturally prompted thinking about application ideasrelevant to the needs of these markets, adding evenfurther momentum.

B. Inexpensive Computing DevicesHistorically, there have been numerous initiatives tar-geting the creation of “computers for the poor,”but the quest for such devices has been an elusiveone (see Table 1 for a selection). Arguably, the origi-nal “low-cost PC” was IBM’s PCjr, which waslaunched in 1983 with much fanfare, including amagazine devoted to coverage of it even before itsactual release. The product led to a wave of clones,some fairly successful, including the Tandy 1000,though it did not itself succeed in the market due todesign issues. This ªrst low-cost computer was notintended as a “computer for the poor,” but was in-stead an attempt to extend the range of peoplehaving access to computers—in this case, from busi-

50 Information Technologies and International Development

THE CASE OF THE OCCASIONALLY CHEAP COMPUTER

Volume 5, Number 1, Spring 2009 51

PAL, PATRA, NEDEVSCHI, PLAUCHE, PAWAR

Tab

le1

.Se

lect

edlo

w-c

ost

com

pu

tin

gin

itia

tive

s:C

om

par

iso

no

fco

sts

and

curr

ent

stat

us

of

imp

lem

enta

tio

n.

Init

iati

ve

Usa

ge

Mo

de

lE

nv

isio

ne

dC

ost

(US

$)

Act

ua

lC

ost

(US

$)

Ta

rge

ted

Use

rsB

uy

er

Sta

tus

in2

00

7D

esi

gn

Str

ate

gy

Sim

pu

ter

Sin

gle

use

r,sh

arab

le$

10

0$

20

0Lo

w-i

nco

me

us-

ers

Inst

itu

tio

ns,

In-

div

idu

als

Dis

con

td.

Co

mp

lete

red

esig

n,

PDA

form

fact

or

OLP

CSi

ng

leu

ser

$1

00

$1

35

($2

08

)1Lo

win

com

ech

ildre

nG

ovt

./

stat

eIn

pro

dn

.C

om

ple

tere

des

ign

,sm

all

lap

top

form

fact

or

Cla

ssm

ate

Sin

gle

use

r$

40

0$

20

0�

Ch

ildre

nIn

div

idu

als

Inp

rod

n.

Co

mp

lete

red

esig

n,

smal

lla

pto

pfo

rmfa

cto

r

Co

mp

uta

do

rPo

pu

lar

Shar

ed$

30

0—

Kio

sks

Ind

ivid

ual

s,su

b-si

diz

edN

ever

pro

du

ced

Stri

pp

ed-d

ow

nd

eskt

op

PC

NC

om

pu

tin

gSh

ared

—Th

inC

lien

t$

11

/ac

cess

term

inal

2$

66

/ac

cess

term

inal

2C

lass

roo

ms

Ki-

osk

sIn

div

idu

als

Inp

rod

n.

Serv

erw

/m

ult

iple

thin

cli-

ents

HP

44

1Sh

ared

$2

50

$2

50

Cla

ssro

om

sK

i-o

sks

Inst

itu

tio

ns

Dis

con

td.

Serv

erw

/m

ult

iple

du

mb

term

inal

s

1.

Wh

ileth

ees

tim

ated

pri

cefo

rfu

ture

sale

sis

aro

un

d$

13

5,

the

actu

alco

sto

fo

wn

ersh

ip(i

ncl

ud

ing

mai

nte

nan

ce)

agre

edu

po

nin

the

Mo

Ub

etw

een

OLP

Can

dLi

bya

was

$2

08

per

un

it.

2.

Excl

ud

esm

on

ito

r.

nesses to home users—by drastically reducing de-vice costs.

The second wave of low-cost PCs came in theearly days of the World Wide Web. Products such asthe Net PC were conceptualized (Russel, 1997), butnever made it to widespread production, becausethe 1990s saw such rapid decline in PC prices that alow enough threshold for a “computer for thepoor,” as then imagined, could be attained by themarket without any need to innovate. Back then, a$500 computer seemed like a quantum leap(Neugass, 1996), and most attention on the cost-minimizing side was directed toward optimizing thethin client architecture (Gaw, Marsh et al., 1998).This was, in some sense, understandable because ofthe dramatic drop in computer prices through mostof that period. Yet the computer industry remainedproªtable due, quite simply, to the increase in thenumber of families in the developed world that an-nually became new consumers of home computersthrough this period (Bresnahan & Malerba, 1999).We argue that, on the business side, it was partlythe normalization of demand through the developedworld that expanded the interest of major compa-nies in developing products for emerging markets.We also contend that a complementary argumentfor the rise in thinking around low-cost PCs tailoredfor the developing world came with the liberaliza-tion of several economies around the globe, wheremajor PC manufacturers found strong competitionfrom local non-branded PC assemblers (Dedrick,Kraemer et al., 2001).

This new wave of devices aimed to concurrentlydeal with what we see as three related, butsufªciently separate issues. The ªrst, and most em-phasized, was the reduction of the device cost. Thesecond was the creation of form factors andfunctionalities speciªc to usage in developing coun-tries, accounting for the lack of urbanization and in-frastructure. This second factor was, at times,equated with building robust machines that with-stood harsh weather, dust, and poor quality power,often gleaning inspiration from wearable computersfor combat situations (Zieniewicz, Johnson et al.,2002). The third factor was that of “usage appropri-ateness,” including issues related to literacy, culturalappropriateness, and social norms of resourcesharing.

The pioneer in this most recent wave was theSimputer project that originated in 1998. TheSimputer, or “Simple Inexpensive Multilingual Com-

puter,” (Chandru, Deshpande et al., 2001) aimed toaddress these three sets of issues. The device wassold at a considerably lower price point of US$200compared to the average computer cost ofUS$1,000 on the market at the time, even though itwas originally envisioned to cost as little as US$100.The Simputer attempted to work across the range ofissues in building for developing regions: it had adamage-resistant casing; a plastic cover for dustyand hot weather; large, sturdy buttons for roughuse; and an entirely new visual and input interface.The Simputer group put a signiªcant effort into de-veloping an intuitive UI with an OS interface de-signed with the needs of users new to technologyand textual interfaces in mind. The Simputer fea-tured icon-based screens and speech synthesis capa-bility and was intended to be easily shared, with anindividual ºash card for each user.

The Simputer did not do very well in the market,for reasons we discuss later, but another project,with a somewhat orthogonal strategy toward pro-viding low-cost computers, was taking shape inBrazil around the same time. Both projects camefrom academia in respective countries and werebuilt with a Linux backbone to reduce the cost ofthe OS. Unlike the Simputer, the Computador Popu-lar (CP) had very little device-level innovation. Infact, the CP was nothing more than a plain, stripped-down version of a PC running Linux, but the projectwas more important for a different reason: it wasthe ªrst project within the ICTD space to activelyseek government intervention to subsidize the costof personal computers through reduced taxes andloans. This device was to be priced at US$300.

By the turn of the millennium, there was a burstof projects in this same arena for a number of rea-sons (cited above) as reºected in the entry of bigtech companies into this space (Collins, 2007).Arguably, many of these tech companies departedslightly from their core businesses and competen-cies to try their hand at selling new devices innew markets, often with the unusual businessmodel of designing products meant for marketsthat could not buy the products themselves. Instead,they had to be sold through institutional buyers,such as governments, philanthropies, or interna-tional agencies. Oracle had a brief brush with the“New Internet Computer” (started around 2000and abandoned around 2003), which was pricedroughly at US$199. Chip manufacturer Via Technol-ogies designed a low-cost box-PC similar to the



AMD PIC1 at a price point of approximately US$250.Intel, AMD’s chief competitor, had its Community PCproject and Classmate,2 along with a collaborativeproject in China called the Beijing Rural PC. HP ex-perimented with the 441 device, with a changedLinux kernel to support four keyboards and screensfrom a single processor, and priced at approximatelyUS$1,200 for the entire unit. This attempt wasabandoned, along with its parent e-inclusion pro-gram, in 2005, although the technology has livedon in products such as the “Useful Desktop Multi-plier.” Recently, NComputing released the X300 thatuses low-cost access terminals connected overEthernet to share a single PC with up to seven us-ers, eventually hitting a price point of US$200 forthree users, excluding monitors and peripherals.A quick survey of ICTD projects3 shows that over50 projects in the past 10 years have attempted tocreate low-cost computers for developing regions,a large chunk of them small companies assem-bling PCs in the BRIC nations, featuring brandslike Fulong Mini-PC,4 and E-DUC,5 Sirius,6 andSofComp,7 as well as more rugged products such asthe SuperGenius Bharat PC, which, like the AMD-PIC, was built to withstand rough use. An unlikelyconstituent of the low-cost PC market was the NGOworld, with a number of experiments like the pedal-powered Jhai PC8 and social entrepreneurship ven-tures like Inveneo9 that again straddled the spacebetween being outright market products and exter-nal funding-dependent development projects.

Probably the most discussed project, and argu-ably the one with the largest expectations, is theOne Laptop Per Child (OLPC) initiative. Also origi-nally known as the $100 laptop, or more recentlythe XO-1, the current price of the device is aboutUS$208, but is expected to decrease with volume.This device, the brainchild of some of the leadingscientists of the MIT Media Labs, is an inexpensive,low-power laptop designed for harsh conditions indeveloping countries and intended for distributionto children around the developing world. In many

ways, the OLPC has come to exemplify the inexpen-sive computer space, but not just because of itscharismatic promoter, Nicholas Negroponte, and thehistory of the Media Lab behind it. Much of the me-dia attention focused on the project, plus its high-proªle approach (of negotiating sales primarily withheads of government), made it somewhat of an ex-emplar of a project within the ICTD space. TheOLPC project was in the news for impressive techno-logical innovation as much as for its approach ofrecommending individualized laptops as a means ofbetter learning for children in the poorest parts ofthe world, a position that was sometimes at oddswith some of the most inºuential commentators inthis area, including Bill Gates himself (Kraemer,Dedrick, et al., 2007). The idea behind the OLPC si-multaneously raised hopes and criticisms; partner or-ganizations frequently came and went (both Inteland Microsoft have, at different times, been sup-porters and opponents), and their participation fre-quently raised eyebrows.

C. The Occasionally Cheap ComputerWhile the complete fate of the computing-for-the-poor projects like OLPC remains to be seen, valuablelessons can be learned from the market experienceof other similar projects in the past, which, unfortu-nately, were either entirely ineffective or enjoyedvery limited success. Most such projects remainedoccasionally inexpensive at best, often movingswiftly from being inexpensive computers to aban-doned projects, or to computers that could barelybe considered cheap. The reasons for these out-comes were related to both the supply and the de-mand side of the market.

On the supply side, several of the companies pro-ducing these low-cost devices were either not typi-cal computer companies (e.g., the Simputer, CP), orthey were somewhat outside of their core compe-tencies (e.g., Intel, HP, AMD) in terms of productionfor the former category, and marketing for the lat-ter. In terms of production, these new devices couldnot promise large enough volumes at the produc-



1. http://50x15.amd.com/en-us/2. http://www.classmatepc.com/3. Low-cost computing devices and initiatives for the developing world http://www.infodev.org/en/Publication.107.html4. http://www.lemote.com/5. http://www.e-duc.com/6. http://www.ªveriverstech.com/sirius.htm7. http://www.ncoretech.com/mobilis/index.html8. http://www.jhai.org/jhai_remoteIT.htm9. http://www.inveneo.org/

tion end for their costs to be signiªcantly competi-tive. Cost considerations also prevented devicecustomization, and constrained manufacturers tobuilding a single version of a device, rather than asuite of products. For example, you can buy onlyone conªguration of the OLPC, two versions of theSimputer, one of the Classmate, and so on. Suchproducts are difªcult to sell to institutional buyerssuch as schools that are less inclined toward experi-mentation with untested products. On the market-ing side, a fairly common approach of projects inthe cheap computing space has been to emphasizethe social relevance of computers to the poor, thusmaking a case for the government to underwritethe cost of these projects—either directly by beingthe buyer or indirectly through tax concessions. Thishas been a risky strategy that has rarely worked welldue to a range of factors: state priorities in develop-ing countries tend to be focused toward more basicspending; political stability (the OLPC, in particular,has suffered the ºip-ºopping of states on purchasecommitments); equity (thus the problem of selectingthe recipient group of free computers, assuming“everyone” would rarely be provided for); and,ªnally, procurement processes. Finally, on the distri-bution side, working through the government isdetrimental in that it separates producers from themicro-environment within which technology salesand maintenance take place, such as training of lo-cal suppliers and support staff.

On the demand side of the market, the problemswere even more challenging. Some design decisions,while optimized for a certain group of users, couldend up defeating a product’s appeal for a generalaudience, especially when some such designs arehard-coded into a device. This was the case withfeatures of the Simputer and perhaps the keyboardsof the OLPC or Classmate, which are quite optimalfor young children, but a difªcult sell for adults orother potential buyers of the otherwise powerful de-vices. A major concern for these devices when theentire computing paradigm is redesigned has beenplanning for the creation of appropriate content andapplications. The Simputer, with its custom OS andinterface, was especially affected by this, as gettinga critical mass of developers working on creatingapplications for it depended on its widespreadusage—a chicken and egg conundrum. Likewise,while constructivist learning arguments guided de-sign arguments for a new OS and interface forOLPC, the approach exposes the device to the same

risks as Simputer, of not having enough developerscreating new applications for generic use on the de-vice. This led to a larger “lack of things to do” prob-lem for these devices, an issue ironically found to bea problem for another strand of ICTD projects—thecommunity computer kiosks (Keniston, 2002). De-signing applications for adults that do not have aconscious need for computers in their daily lives isnon-trivial, as it is difªcult to convince these adultsto incorporate technology in possibly disruptiveways into their livelihoods (Nedevschi, Patra et al.,2006).

From the free market perspective, the “computerfor the poor” faced its strongest challenge fromstandard low-end desktop computers. Below theUS$250 mark, there is a whole range of Linux-baseddesktop products available both in the U.S. (e.g.,Lindows Family PC), as well as several developingcountries (e.g., ApnaPC). However, the largest chal-lenger has been the neighborhood PC assembler,who puts together a computer with parts purchasedoff the shelf and shows up at the doorstep to ªxthings when stuff stops working. This segment of-fered machines with the same form factor and capa-bility as a typical branded PC. That most low-costPCs opted for Linux-based interfaces was an obviouscost-cutting approach that again turned out to be acompetitive disadvantage against smaller assemblersin the informal market that could afford to throw ina hot copy of the much-desired Windows OS for noadditional cost. In all, the real cost advantage of-fered by most low-cost computers was very small.Finally, an unanticipated demerit of the low-costcomputing machine may have precisely been itsbranding as a low-cost device. Research showedthat the association of ICTD products with low-income groups or low-attainment populations hada damaging brand impact, because the target mar-ket perceives purchasing “computers for the poor”as a climb-down of status (Ferraz, 2004). Some simi-lar ªndings have also been observed in user atti-tudes toward subsidized community computercenters (Kuriyan, Toyama, et al., 2006).

D. Computers and Children’s EducationWe turn now to the speciªc argument on many aninvestment plate for the use of low-cost computersby children in classrooms of the developing world.Frequently, this is the ªrst intended market of suchinitiatives, often engraved with brand names, as inthe cases of OLPC and Classmate. In developing re-



gions, we see two major reasons for the focus ofcomputer projects on children within the environ-ment of a broader developmental focus towardsustainability and human development investment.The ªrst reason has been the shortage of existingtrained teachers, thus shifting the emphasis towardwhat can be done independent of the systemicshortcomings. The relative importance of this is evi-dent in the structuring not just of the low-cost com-puter projects, but also in some of the more recentapplications of technology toward learning in theICTD space, such as language acquisition using tech-nology that has been designed to largely eliminatethe teacher and allow for greater independent inter-action between the child and the system (Kam,Ramachandran, et al., 2007). A second factor hasbeen an increase in interest in constructivist ap-proaches to learning that have inºuenced not justthe building of devices, as in the case of the OLPC,or for that matter, historically, in a number of pastprojects on children and learning (Papert, 1993), butalso in the structuring of technology projects forchildren more broadly (Baggio, 2000).

In addition to these factors speciªc to computersand children, the wave of increasing investment inICTD projects created a number of shared technol-ogy centers offering public access to computers, anumber of which were co-located in schools(Proenza & Montero, 2001; Colle, 2005). Someof the low-cost computing products such as theHP 441 were speciªcally designed for use by multi-ple users in school classrooms or telecenters.NComputing’s X300 was designed as a thin clientsolution for building school computer labs, and theVIA low-cost PC was also envisioned originally forclassrooms. The Simputer featured a number of chil-dren’s applications in its original design, but was in-tended to be shared, not in simultaneous sessions,but passed along from one child to another in turns.Though made for classrooms, the Classmate andOLPC were not intended as shared devices; instead,they were to be used as take-home computers withwhich children explored learning techniques. Nobroad estimates are published on the extent of ac-cess to computers in schools in the developingworld, but both computer centers and individualcomputer purchases are on the overall rise. In thecase of the OLPC, it has been reported that at least15 unbinding purchase discussions with countries,including Nigeria, Libya, and Rwanda, have oc-curred, but it is unclear how many actual purchases

have been made. The OLPC approach of selling inhuge quantities (orders for one million or morepieces) is good for cost reduction, but seems to bebetter suited for countries with a comparativelymore centralized decision-making process on largestate purchases. In India, for instance, the idea ofOLPCs in classrooms was rejected with the argu-ment that “it would be impossible to justify an ex-penditure of this scale on a debatable scheme whenpublic funds continue to be in inadequate supply forwell-established needs” (Mukul, 2006).

This leads to a larger question, given the thrustof spending on technology for classes: Can comput-ers actually complement teachers and make learningmore valuable? This question is actually far from set-tled as there is very little consensus on what thebeneªts of technology in classrooms are and evenless consensus on the comparative beneªts of com-puters versus other kinds of interventions. Whilestudies show that children’s access to computersyields clear gains in certain types of skill building(Attewell & Battle, 1999), especially when these arehome computers, there is an abundance of materialto suggest that the context within which the com-puters are used is critical to ensuring both educationefªciency and equity in education opportunity. Thereis strong evidence that investment in computers canbe highly inefªcient (Cuban, 2001), particularlywhen it is driven by an enthusiasm for technologyrather than the needs of the children. There is alsoevidence that the positive impact of access to anduse of computing facilities can be highly biased(Becker, 2000) due to cultural and cognitive factors.

III. Approaches to ClassroomComputing: Usage Models andEconomicsWe seek to address here the rather provocativequestion of how computers should be used in pri-mary school classrooms from a supply perspective.We consider the devices and the means of usingthem, not the larger issues of what speciªc learningend goals computers work stronger toward. Weturn thus to shared computing, which is central toour argument from this point on. Part of the goal oflooking at past initiatives targeting computers forchildren was not only to understand what they didwrong from a market perspective, but also to under-stand how well these projects ªt with thepracticalities of computer use in classrooms. Moving



on, we ªnd that the single biggest missing factorthat perhaps explains why computers in the class-room of the developing world have borne the bur-den of looking gimmicky has probably been the lackof attention to sharing. In our collective experiencefrom India, South Africa, Rwanda, and Brazil, theone constant feature of computer centers in poorprimary schools has been that computers wereshared among multiple users, anywhere betweentwo and 10 children per PC. We propose a fewshared- and single-use models of computer use tolook at some of the merits and disadvantages ofboth.

A. Usage ModelsWe distinguish here between three models of com-puter usage for child education: single ownership(Figure 1), single user per classroom computer/terminal (Figure 2), and multiple users per sharedclassroom computer (Figure 3). More models arepossible, and indeed, as we will note, some initia-tives share properties with each model. We ªndthese categories useful, however, for understandingthe design decisions of speciªc initiatives and fordemonstrating, in later sections, how the choice of

usage model must match existing contexts and prac-tices to have a long-term impact on establishingmore equitable education. Given the comparativelygreater portion of India-related research in this arti-cle, our sample has biases toward the speciªc condi-tions there, but our work in other countries leads usto believe that most of our ªndings resonate acrossconditions in other developing regions.

1. Single ownership: OLPC and Classmate: TheOLPC and Classmate are both designed for singleusers, and while the Classmate is oriented to indi-vidual and institutional purchases, the OLPC wasoriginally meant only to be purchased in quantitiesof one million-plus by governments. The recom-mended usage is individual child learning, parent-child collaboration, or teacher-child collaboration.The Simputer had a PDA form factor with a durablecasing, and a few buttons designed for dust andhead were also trying to target the same applicationdomain. This device was also intended for individualuse, but it was easily sharable, with pluggable ºashmemory cards for each user. Given these features,the Simputer could be passed from child to child, orfrom home to home, with each child or home re-sponsible for its individual ºash memory. This fact,as well as other design characteristics of the design(such as speech synthesis for illiterate users), pointsto some level of consideration for existing culturalpractices and economic realities of the targeted userbase. The OLPC probably presents a number of in-novations on the interface front, including a smaller,ºatter keyboard, a dual mode screen with low-power consumption, and white reºective sunlight toallow readability for open air conditions. However,



Figure 1. A single-child usage scenario for OLPCs inIraq. (© Wikimedia Commons)

Figure 2. A typical shared computer lab for middleschool children in India.

both the OLPC and the Classmate are designed astake-home devices, and their small screens makethem difªcult to share without an externally con-nected screen.

2. Single user per community-owned computer/terminal: A second usage model, which has beenthe standard in U.S. primary schools as well as someslightly better-off schools in the developing world, isthe computer lab: a classroom dedicated to com-puter usage where each child has access to a com-puter in the classroom environment. In this model,the responsibility of purchasing and paying for com-puter maintenance is shared by the community. Theteacher plays a primary role in educational settingsthat adopt this model, guiding and supervisingchild-computer interactions. The HP 441 was meantto share a single processor over multiple users, butonly have a single user per terminal. NComputing’sX300 was, likewise, designed to be shared by up tosix individual users, each at a terminal with its ownmonitor and keyboard. Devices that target the us-age model in which single users operate computersthat are shared by the community tend to featurelow-cost versions of standard PCs. The ownership bythe community allows for higher prices per unit, butin practice, the cost is comparable to devices thattarget a single ownership model. Community-owned computers operate in a group setting, wheregroup interactions external to the child-computer in-teraction play a central role in their overall use.

3. Multiple users per shared computer: Inpractice, there is no reason why the second mode ofdesktops in computer centers, as mentioned above,

cannot be shared further by groups of children, andindeed, in most cases, this is how the computers areactually used. Products such as the HP 441 or X300have the computing capability of a typical stand-alone PC. While not speciªcally marketed to multi-ple users per terminal, these devices could cater togroups of children huddled around each terminal asthey’d been designed. In fact, space and power re-strictions make it such that computers are mostlikely to be shared in simultaneous sessions, if theyever make it to a school in most parts of the devel-oping world. Given this, we ªnd that a striking miss-ing factor is the lack of innovation in the“computers for the poor” space on efªcient sharedcomputing. One initiative, started by some of theauthors of this article, looked at the use of multiplemice (on a single machine) to support up to ªve si-multaneous students; this would require addingminimal peripheral spending and tinkering with theapplications for better use across children (Pawar,Pal, et al., 2006). While this is clearly a restrictiveformat of interaction with a computer, we ªnd that,for the unique needs of primary schoolers using de-vices for computer-aided learning, a mouse alone isusually enough, given that the typical “narrative-interactive” loop of practically all of children’s learn-ing software relies mostly on multiple choice clickscreens. That being said, the potential of multiplekeyboard usage (Tse & Greenberg, 2004) and alter-nate text-entry, using mice or other devices, is botha possibility and an existing area of research. Fur-ther, we could also draw learnings in this area fromsome of the other projects, including the Simputer,to expand the use of ºash memory to give eachchild proªle individual memory for shared sessions.The intuitiveness of such usage models and screensharing has been shown (Moed, Otto, et al., 2009),and there is already much precedent for screen usewithin the realm of multi-player video games(Nintendo, Playstation, etc.) that allow for the samevariety of interaction modes (parallel, competitive,and collaborative).

Given what we know about the various modelsof technology use in classrooms and the prevalenceof the shared-use model in actual practice, we needa clearer idea of how this could work in practice andwhat the major beneªts or drawbacks of such a us-age model would be. We approach this question bycomparing shared-usage models with the two sin-gle-user scenarios, based on cost, educational effec-tiveness, and socio-cultural appropriateness of the



Figure 3. Shared computer using multiple mice. (Photocourtesy of Microsoft Research India)

shared model. For the two latter cases, we referenceexisting research on educational effectiveness ofmultiple-mice models and an interview-based studyof socio-cultural attitudes toward device sharing inrural India.

IV. Economic FeasibilityIn this section, we evaluate the economic feasibilityof various computing initiatives for schools and theirassociated usage models. Given the harsh ªnancialconstraints, cost is one of the most important con-siderations for any of these initiatives. We considerboth initial capital costs and running costs, includingreplacement, maintenance, and additional expendi-tures for developing appropriate content and train-ing of teachers.

An economic analysis of the “computers for thepoor” projects is necessary, because these projectshave typically targeted government buyers, project-ing the provision of computers to children as a stateresponsibility. The most important criticism of suchprojects, and often one very difªcult to address, isthe argument that money would be better spent onschool buildings, safe drinking water and toilets inthese schools, books, additional teachers and soon—all basic needs and with immediate returns oninvestment. In contrast, computers-in-education ini-tiatives address less stringent needs and can onlyshow a long-term return on investment. It is not ourgoal to claim here that the hierarchy of needs argu-ment applies absolutely, and that investment incomputers should be preceded by the solving of allother world problems. Governments all around theworld are introducing computers in schools in mod-eration, attempting to balance these expenditureswith more basic ones. However, it is essential to en-sure that already scarce resources are utilized in themost effective way. It is notable that the two coun-tries making the most progress in adopting OLPC-type programs are Libya and Nigeria, both nationswith limited political opposition to getting them ac-cepted at the highest levels.

A. Reduction of Computing CostMost low-cost computing initiatives are reducing thecapital cost of hardware by riding on the computerindustry’s exponential trends of increasing integra-tion and performance. They also downgrade or re-

move certain components from the ªnal device.Such initiatives often use low-end processors (OLPCuses the AMD Geode, Classmate uses low-powerIntel processors), replace hard disk with ºash mem-ory, and remove other capabilities like high-endgraphics, optical drives, and peripheral connectors.The cost of ownership can be reduced by loweringthe running cost of power. Power consumption canbe reduced by using lower-power displays andsmarter sleeping techniques. Interestingly, refur-bished computers have performed poorly due bothto maintenance and to disenchantment with “sec-ond rate” computers (Fonseca, 2004).

Though the overall cost of computing has gonedown, none of the low-cost devices have broken theoff-the-shelf US$200 mark (not including the cost ofmaintenance). OLPC’s XO-1 laptop expects toachieve this by selling the hardware in batches of atleast one million. However, according to the con-tract between OLPC and Libya, the cost per device,including maintenance, is US$208. Interestingly,Intel’s Classmate PCs were initially sold at a price ofUS$400, but their price is expected to fall to aroundUS$200 as well. The cost of NComputing’s X300 isalso expected to fall to US$11 per access terminal inlarge production volumes (Dukker, 2007).10

B. Cost ComparisonTo compare the economic viability of various deploy-ment models, we consider the capital and opera-tional cost of providing computers in the whole ofIndia. As discussed earlier in Section III, we considerthree scenarios:

1. Single ownership model.

2. Single user per community-owned com-puter/terminal: We assume a ratio of onecomputer per 10 children in the school. Dur-ing classes, each child gets his or her owncomputer.

3. Multiple users: We assume a ratio of onecomputer per 40 children. During computerclasses, three to ªve children share a com-puter. This can be done with or withoutmultiple input devices.

We use data about schoolchildren distribution inIndia from the Department of Education, Govt. ofIndia (2008). Approximately 1.04 million schools in



10. Dukker, S., & Bender, W. (2007). “Will Low-Cost Laptops Help Kids in Developing Countries?” The Wall Street Jour-nal, 5 September.

India serve about 165 million children between theages of 6 and 13. The schools vary in size, servingless than 30 students to more than 300 students.Approximately 91% of all schools in India, and over-whelmingly so in rural India, are government-aidedand/or managed schools (Ministry of Human Re-sources, Govt. of India, 2008). Infrastructure re-sources in these schools are very scarce: 59% of allIndian schools have no safe drinking water, 26% noblackboards, 89% no toilets, and less than a quarterof the schools have functional access to electricity(UNESCO, 2006).

Table 2 summarizes the cost parameters involvedin our comparison, based on latest ªgures fromOLPC and the Indian government. For scenarios(2 and 3) where desktop computers are used inclassrooms, we use cost ªgures reported from cur-rent deployments of computers in Indian schools (in-cluding expenditures for teacher salaries andmaintenance of equipment that the Governmentprovides) and recent market prices for desktops. Weassume a gradual deployment model where com-puters are introduced over ªve years. Assuming thatthe lifetime of the computers is also about ªveyears, the replacement capital cost for the hardwareevery year is one-ªfth of the total capital cost. Wealso assume that each school has only one com-puter teacher.

The cost comparison, showing the total cost peryear for all the three scenarios, is presented in

Table 3. As can be seen, the annual cost of provid-ing shared computers (one per 40 children) to all of165 million Indian students is only about US$1.06billion a year. In contrast, the annual expenditure ofproviding laptops to every child is about 12 timeshigher at US$12.42 billion. These are conservativeestimates that ignore the additional costs of runningsuch a large program, such as costs associated withrecurring power and other infrastructure needs inschools, with developing educational content in lo-cal languages, and with providing Internet connec-tivity to schools.

The total public expenditure for India on educa-tion was about US$22.9 billion (3.3% of GDP) in2004, of which 30%, or US$6.8 billion, was allo-cated for primary schooling. In these conditions, it isunrealistic to expect the Indian government tospend more than half of the education budget onbuying computers, especially without having anyguarantees on the educational beneªts of such aprogram.

V. Education EffectivenessAn important way to compare across usage modelsis by looking at the educational value provided bythose models. The question of how to best deªnevalue would be better settled in open debate by ed-ucation scientists. However, it is possible to compareacross usage models by relying on studies with easily



Table 2. Parameters for cost comparison (budgeted on state school computers in India).

Cost of desktop PC that is shared Rs 25000 or US$530

Cost of maintenance paid by Govt. for each PC (per year) Rs 1800 or US$38

Cost of teacher (per year) Rs 24000 or US$510

Cost of laptop (e.g., XO-1) Rs 9770 or US$208

Table 3. Comparison of costs for various scenarios for deployment of computers in all ruralschools of India (total of 165 million students in 1.04 million schools).

Metric Scenario 1 Scenario 2 Scenario 3

No. of computers (million) 149.40 15.19 4.00

Total initial cost ($billion/year) 31.06 8.08 2.15

Cost of replacement ($billion/year) 6.21 1.62 0.43

Cost of maintenance ($billion/year) 5.72 0.58 0.15

Cost of teachers ($billion/year) 0.48 0.48 0.48

Total cost ($billion/year) 12.42 2.68 1.06

quantiªable metrics based on objectively assessablelearning outcomes.

Few studies focus on learning outcomes in thesingle computer per student model. The OLPC proj-ect is starting ªeld studies with the XO-1 project in afew countries. Maine’s One Laptop per Child project(Muir, 2004) conducted some studies, but results didnot ªnd that a separate computer per child leads tostrong and measurable changes in learning outcomemetrics. However, for scenarios with computers in-side the classroom, we present preliminary resultsfrom India for various single- and multi-user scenar-ios, using single or multiple input devices. We ªndthat learning effectiveness with collaborative usageon multiple input computers can be as pronouncedas learning with single-user computers in class-rooms. This is a ªnding we attribute to the inherentsocial nature of learning provided in shared comput-ing scenarios.

A. Study Methodology in IndiaWe conducted observations and qualitative inter-views during May 2005 and August 2006 at 22schools in India, all catering to children from disad-vantaged backgrounds. Selection of schools wasbased on regional proªle and longevity of their com-puter-aided learning programs. These programswere state supported and contracted to Azim PremjiFoundation, an NGO that sets up computer-aidedlearning centers for children. A total of 179 inter-views were conducted with parents, teachers, vari-ous stakeholders in local education, and policymakers on their views about computers in schools,and the short- and long-term goals of their pro-grams. We observed ªfth and sixth grade children asthey sat in front of computers and used the applica-tions provided to them by the schools visited.

We tried two approaches to work toward thegoal of providing equity to all users around theshared resource (the PC with the learning applica-tion): ªrst, enforced resource sharing, and second,multiple input. Results published elsewhere (Pawar,Pal, et al., 2007) showed that, in the former case,there was often some degree of collaborationamong children, especially as the “alpha children”(i.e., the typical mouse controllers and usually thescholastic achievers within groups) tended to discusslearning material with others in a group, leading tosome impact on learning. However, in practice, thiswas a difªcult goal to realize as the alpha children

grew impatient with their role as surrogate teachersand wanted complete control for themselves.

Due to teacher shortages, children are often re-quired to learn and manage how to use computersthemselves with limited teacher intervention, mak-ing supervision-intensive tasks difªcult. As a result,children who established dominance among theirown small groups of colleagues tended to repeat-edly be “mouse controllers” who dictated the paceof computer-aided learning sessions. Observations ofeye contact with screens showed that mouse con-trollers were predominantly in command of the en-tire interaction and learning trickled down fromthem to other children. Thus, the main ªnding wasthat, with regard to computers in schools, evenwhere equitable access was available, the dynamicsof sharing among children often created new formsof power structures, generally to the detriment ofthe children who are most in need.

B. Experiments with Multiple InputDevices in IndiaFollowing our initial ªndings on device sharing, wetried using a single computer with multiple mice ontest applications to see if there was any difference inchildren’s learning in the new modality. Looking atthe educational applications being used in theschools, we tested a word learning application inSeptember 2006 with 238 children (11–12-year-olds) in various single- and multi-user scenarios, us-ing single or multiple input devices. The childrenwere shown a number of words that were new tothem and then asked to identify the words frommultiple choice options. The application used agame format and was tested in real classroom set-tings in rural India—two schools selected fromamong recently-instituted computer-aided learningprograms.

Children were tested for a list of English wordsbefore and immediately after the test, and thewords were included in the test application.Children were asked to play in four modes: (1) sin-gle-user, single input mode; (2) multi-user, single in-put mode; (3) multi-user, multi-input competitivemode; and (4) multi-user, multi-input collaborativemode. All multi-user modes had ªve users. The sim-ple single-user, single-child mode (mode 1) is themodel for which most applications are designed.Mode 2 is the closest to the typical usage scenarioobserved in India (and elsewhere) with many chil-dren at one computer, but only one controlling the



mouse. In mode 3, each child had a mouse, and theone who clicked the right answer ªrst earnedpoints. In mode 4, each child had a mouse, but theapplication moved to the next stage only if all chil-dren had identiªed the correct response.

During two rounds of experiments (Pawar, 2007)with a total of 238 children, we found that, for thespeciªc application of word learning, children wereable to consistently retain the most in mode 4 (seeresults in Table 4). The stark differences in resultscame when we looked at results separately by gen-der, and this emphasizes the social nature of learn-ing, as gender inºuences social behavior andengagement which, in turn, affects learning. Over-all, a key observation for this learning task was thatmultiple mice could offer the same beneªts. In thecompetitive mode however, learning was hampered,as competition lead to a decrease in collaboration.

Hence, we found that increasing access to inputwas not sufªcient in itself to make the learningmore effective; collaboration was an essential partto improve the quality of learning. More detailed re-sults from these tests are included in a work ofgreater exploration elsewhere (Pawar, 2007). Thetests were used only to establish short-term reten-tion. On the whole, they were useful in creating acase for collaborative learning over single displaygroupware (SDG), a case that has in the past alsobeen made of mathematics (Bricker, 1995; Inkpen,1995) and visual learning (Inkpen, 1999).

VI. Socio-Cultural SuitabilityA common criticism of “computers for the poor”devices has been that they have not been groundedin good design principles (i.e., that look at devicescontextually), but are designed in a lab-centricrather than need-oriented paradigm (Fonseca,2006). Two important factors that need to be evalu-ated are whether the computer usage model under

consideration is suitable in the actual social contextand whether it would ªt well with existing teachingmethods in developing countries. Our goal is towork within the limitations of the current deploy-ments that are already happening and workincrementally to increase access for all children.

Parents’ beliefs about computers and educa-tion: Our primary research showed that both par-ents and teachers controlled the amount of timethat children were allowed to work on computers inIndia, often very restrictively since the computer wasthe most expensive, or the only electronic gadget athome. As many as three years into having access tocomputers, teachers in some schools still let childrenuse the computers only under their supervision.Likewise, many parents did not allow their childrento use the television sets at home. The idea that par-ents, especially those in very poor families with nohousehold assets, will allow and encourage use ofcomputers in the same way as parents in developedcountries may be a gross overestimation.

To dig deeper, we conducted interviews with 165parents across four districts to determine parents’feelings about the use of computers in schools andhomes. The interviews sampled a wide range of par-ents whose ages ranged from 24 to 70 and whohad completed between 0 and 12 years of educa-tion. The interviews revealed an extremely importantrole played by parents in decisions about children’suse of computers in schools: When asked if comput-ers should be in schools, in the home, or both, par-ents overwhelmingly chose schools (Table 5). Inother words, they disfavored the single ownershipmodel adopted by OLPC and Intel Classmate. Mostof them cited the primary role of teachers as theirreason for preferring computers in schools.

Due to a system of education very centered onstructured learning coming from the teachers to thestudents, parents were not convinced that games



Table 4. Words learned during tests across single- and multiple-user shared-input modes(N �238, � � 0.5).

Mode Word Gain Engagement Response Error Decision makingDomination bysingle child

1 4.11 High, tails off Low Individual n/a

2 3.77 Low Very Low Collaborative Varied

3 3.60 Very high High Individual None

4 4.30 High Very Low Collaborative Varied

could have a positive role to play in children’s learn-ing. They were more concerned with ensuring chil-dren’s progression through curriculum. Computersare generally seen in a positive light and beneªcialto children, but mostly when working on them leadstoward immediate better learning. For example,when asked if the same amount of money shouldbe spent on teacher salaries or a new computer,60% of parents felt that additional teachers wouldbe a better investment for learning, whereas 40%thought that the one-time purchase of a computerwould positively impact student learning.

Compatibility with existing teaching methods:We also found that early exposure to computers indeveloping countries comes through curricular con-tent based on the dominant teaching methodolo-gies, which tend to be highly structured andinstructive rather than constructive. Almost univer-sally, we found that computer-based learning mate-rial for children reºected the classroom: it is createdin a “narrative-interactive” loop fashion, with theapplication feeding some content ªrst, and then fol-lows up with multiple choice-type questions thattest a child’s understanding of the material. Wefound this to be highly compatible with content onmulti-input, single-screen computers.

The central role played by teachers in makingsuch programs effective has been much discussed(Scott, 1992), as has the idea that class and cogni-tive issues impact the level and complexity of accessthat is available to children. Social class can alsohave an impact on computer learning: There is evi-dence that children from marginalized and under-served groups tend to do more drill-type activity,whereas children from afºuent backgrounds tend toget greater access to higher-level activities and cre-ative resources on computers (Means, 2001).

VII. ConclusionOur reference to the “Occasionally Cheap Com-puter” is meant to underscore the dilemma of low-cost computing in the developing regions. Many ini-tiatives in this space have attempted and succeededin bringing down the cost of computing devices,and in the process have often brought about re-markable new breakthroughs that will beneªt tech-nological advancement on the whole. Yet themarket experience of these projects has been pooroverall so far, and many of the projects don’t end uplooking so cheap after all once they get in shape fordistribution. Our argument here is that there is aneed to take a step back and look at what the mar-ket seems to be telling us. We propose that an in-cremental approach to bringing technology to theclassroom in the developing world is the direction totake ahead. Instead of building new devices that re-quire both expensive R&D and a greater threshold ofarguing for adoption, considering the institutionalbuying modes these projects operate in, it is far eas-ier to work within the realm of “off-the-shelf” com-puters, as we propose here. The shared-computeruse model that we propose here may not have thebest case beneªts of an ideal “laptops for all” sce-nario, but our argument is that it is not feasible tothink of the former, and besides the core economicarguments, we have also provided learning gainpotentials and parental attitudes toward shared-computer use to support our position. This approachto shared computing thus explores how we can bestmaximize the beneªts of computers for children thatare likely to be used in primary schools in the devel-oping world and in small computer centers wherechildren sit in groups. ■



Table 5. Parents’ beliefs about computer usage models.

ReasoningPreference for locationof computer

Fraction ofparents (%)

Cannot learn at home School 31

Only teachers can teach School 32

Children will learn better in collaboration School 24

Don’t want responsibility School 8

Lack of power, etc., at home School 4

Ease of access, device safety Home 3

AcknowledgmentsWe would like to acknowledge the valuable com-ments of six anonymous reviewers and the editors P.Balaji, Krithi Ramamritham, and Arlene Luck, as wellas the various people who have supported this proj-ect, including Eric Brewer, Meera Lakshmanan,Kentaro Toyama, and Rodrigo Fonseca. This materialis based upon work supported by the National Sci-ence Foundation under Grant No. 0326582.

ReferencesAttewell, P., & Battle, J. (1999). “Home Computers

and School Performance.” The Information Soci-ety 15(1): 1–10.

Baggio, R. (2000). “A Sociedade da Informação eaInfoexclusão.” Ciência da Informação 29(2): 16–21.

Becker, H. J. (2000). “Who’s wired and who’s not:Children’s access to and use of computer tech-nology.” The Future of Children 10(2): 44–75.

Bell, D. (1976). “The Coming of the Post-IndustrialSociety.” The Educational Forum 40(4): 574–579.

Bhagwat, P., Raman, B. et al. (2004). “Turning802.11 inside-out.” ACM SIGCOMM ComputerCommunication Review 34(1): 33–38.

Bresnahan, T., & Malerba, F. (1999). “Industrial Dy-namics and the Evolution of Firms’ and Nations’Competitive Capabilities in the World ComputerIndustry.” Sources of Industrial Leadership:Studies of Seven Industries: 79–132.

Brewer, E., Demmer, M., et al. (2005). “The case fortechnology in developing regions.” Computer38(6): 25–38.

Bricker, L., Inkpen, K., Stewart, J., Myers, B., et al.(1998). “Single Display Groupware: ExploringComputer Support for Co-Located Collabora-tion,” Panel submitted at ACM SIGCHI’99, Pitts-burgh.

Castells, M. (1996). Rise of the Network Society.Cambridge, MA, Oxford, UK: Blackwell Pub-lishers.

Chandru, V., Deshpande, V. L., et al. (2001). TheSimputer: Radical Simplicity for Universal Access.Development by Design: ThinkCycle Open Col-laborative Conference, Bangalore 2002.

Colle, R. D. (2005). “Building ICT4D capacity in andby African universities.’” International Journal ofEducation and Development using ICT 1(1).

Collins, L. (2007). “Cutting the Cost of Com-puting.” Engineering & Technology 2(2): 34–37.

Cuban, L. (2001). Oversold and Underused: Com-puters in the Classroom, Harvard University Press.

Dedrick, J., Kraemer, K. L., et al. (2001). “Impacts ofLiberalization and Economic Integration on Mex-ico’s Computer Sector.” INFORMATION SOCIETY17(2): 119–132.

Department of Education, Ministry of Human Re-source Development, Government of India.(2008). Statistics. Retrieved March 1, 2008,http://www.education.nic.in/Elementary/elementary.asp

Duff, A. S. (2000). Information Society Studies,Routledge.

Ferraz, C., Fonseca, R., et al. (2004). Computing forSocial Inclusion in Brazil: A Study of the CDI andOther Initiatives. First UCB-UNIDO Bridging theDivide Conference, 2004, Berkeley, CA.

Fonseca, R., & Pal, J. (2004). “Computing Devicesfor All: Creating and selling the low-cost com-puter.” 1st International Conference on ICT andDevelopment, Berkeley, CA. ACM IEEE Confer-ence Proceedings.

Gaiha, R. (2001). “Decentralization and Poverty Alle-viation in Developing Asia.” Economics and De-velopment Resource Center, Asian DevelopmentBank, Manila. Processed.

Gaw, D., Marsh, A., et al. (1998). Low-Cost Multi-Service Home Gateway Creates New BusinessOpportunities, Coactive Networks, http://www.coactive.com

Gouldner, A. W. (1979). The Future of Intellectualsand the Rise of the New Class: A Frame of Refer-ence, Theses, Conjectures, Arguments, and anHistorical Perspective on the Role of Intellectualsand Intelligentsia in the International Class Con-test of the Modern Era, Seabury.

Hanna, N. (1994). Exploiting information technologyfor development: a case study of India, TheWorld Bank.

Inkpen, K. M., et al. (1999). “This is fun! We’re all



best friends and we’re all playing: Supportingchildren’s synchronous collaboration.”International Society of the Learning Sciences.

Kam, M., Ramachandran, D., et al. (2007). “MILLEE:Mobile and Immersive Learning for Literacy inEmerging Economies.” TIER Research Group, Uni-versity of California at Berkeley.

Kraemer, K. L., Dedrick, J., et al. (2007). “One Lap-top Per Child (OLPC): An Education Project or aLaptop Project?”

Kuriyan, R., Toyama, K., et al. (2006). “IntegratingSocial Development and Financial Sustainability:The Challenges of Rural Computer Kiosks inKerala.” Information,Technologies and Interna-tional Development 2006. ICTD’06. InternationalConference: 121–130.

Machlup, F. (1962). The Production and Distributionof Knowledge in the United States. PrincetonUniversity Press.

Masuda, Y. (1981). The Information Society as Post-industrial Society, World Future Society.

Moed, A., Otto, O., et al. (2009). Reducing Domi-nance Behavior in Multiple-Mouse Learning Ac-tivities. 8th International Conference onComputer Supported Collaborative Learning(CSCL 2009)—Submitted, Rhodes, Greece.

Muir, M., Knezek, G., & Christensen, R. (2004).“The power of one-to-one: Early finding fromthe Maine Learning Technology Initiative.”Learning & Leading with Technology 32(3): 6–11.

Nedevschi, S., Patra, R., et al. (2006). “AMultidisciplinary Approach to Studying VillageInternet Kiosk Initiatives: The Case of Akshaya.”E-Learning 3(3): 291–316.

Negroponte, N. (1996). Being digital, RandomHouse Inc. New York.

Neugass, H. (1996). The $500 Internet Computer.13th TRON Project International Symposium,Washington, DC. IEEE Computer Society.

Nulens, G., & Van Audenhove, L. (1999). “An Infor-mation Society in Africa?: An Analysis of the In-formation Society Policy of the World Bank, ITUand ECA.” International Communication Gazette61(6): 451.

Papert, S. (1993). The children’s machine: rethinkingschool in the age of the computer, Basic Books,Inc. New York.

Parikh, T., Ghosh, K., et al. (2003). “Design studiesfor a ªnancial management system for micro-credit groups in rural India.” Proceedings of the2003 conference on Universal usability: 15–22

Pawar, U. S., Pal, J., et al. (2006). “Multiple Mice forComputers in Education in Developing Coun-tries.” International Conference on InformationTechnologies and Development, May 2006, IEEEConference Proceedings.

Pawar, U. S., Pal, J., et al. (2007). “Multiple mice forretention tasks in disadvantaged schools.” Pro-ceedings of the SIGCHI conference on Humanfactors in computing systems: 1581–1590.

Pawar, U., Pal, J., Gupta, R., & Toyama, K. (2007).“Multiple mice for retention tasks in disadvan-taged schools.” ACM SIGCHI’07 Conference Pro-ceedings, pp. 1581–1590.

Pentland, A. S., Fletcher, R., et al. (2004). “DakNet:Rethinking Connectivity in Developing Nations.”

Prahalad, C. K., & Hammond, A. (2002). “Servingthe World’s Poor, Proªtably.” Harvard BusinessReview 80(9): 48–57.

Proenza, F., & Montero, R. (2001). “Telecenters forsocioeconomic and rural development.” InterAmerican Development Bank, Working Papers.

Russel, J. (1997). “No End in Sight Over NC/NetPCDebate.” Communications Week.

Tse, E., & Greenberg, S. (2004). Rapidly prototypingSingle Display Groupware through theSDGToolkit, Australian Computer Society, Inc.Darlinghurst, Australia.

UNESCO Institute for Statistics (UIS). (2005).Children out of school: Measuring exclusion fromprimary education. Montreal: UIS.

Wehn, U., & Mansell, R. (1998). Knowledge socie-ties, Published for and on behalf of the UnitedNations by Oxford University Press, New York.

Zieniewicz, M. J., Johnson, D. C., et al. (2002). “TheEvolution of Army Wearable Computers.” IEEEPERVASIVE COMPUTING: 30–40.



the case of the occasionally cheap computer pal, patra, nedevschi

Documents