Marshall Van Alstyne and Erik Brynjolfsson†
Working with informationrequires time and attention. Awealth of information leads to apoverty of attention, creating aneed to allocate attentionefficiently. Not surprisingly,the online information explosionmakes explicit searching andfiltering not only possible butalso necessary. Automatedagents can scan the World WideWeb for focused informationbased on predeterminedpreferences while collaborativefilters master new preferences bycomparing hundreds of thousandsof user profiles to makestrikingly accurate suggestionson information of potentialinterest. Highly customizedonline journals, with asubscriber base of one, havebecome feasible.
Organizational structures are alsochanging. Plummeting costs ofinformation technology (IT)have changed the relativeefficiency of different structuresfor coordinating work in firmsand markets (1) and inuniversities (2). Increasedintegration and collaboration –as well as competition – mayfollow. Although most research(and press attention) has focusedon business restructuring, thestructure of scientific inquiry isnot immune to a changingtechnological environment. Inparticular, the unifying andintegrating benefits of access
†Nota Bene: This is an early
draft of an article appearing inScience 274(5292), Nov. 29. pp.1479-80. For either quote orcitation purposes, please refer tothe more recent publishededition.
MIT Sloan School
50 Memorial Drive E53-313Cambridge, MA 02142
technology should not be takenfor granted. Recent work showsthat greater access to specializedand remote resources can increasescientific productivity but thatscientific insularity might alsorise (3). Faced with a wealth ofresources and limited attention,researchers can use IT to focuson only those articles andcolleagues that really interestthem, regardless of location, andto the effective exclusion ofothers.
Figure 1.A Figure 1.BIT can reconstitute geographiccommunities (locations in 1.A)by research discipline (colors in 1.B).Benefits of ElectronicCollaboration
The proof of Fermat's lasttheorem offers an illustration.Discovery of a glitch in theoriginal version sparked anelectronic exchange of ideasamong internationally distributedalgebraic topologists. Longdistance dialogue then led theauthor, Andrew Wiles, tostrengthen his ideas and fix theproof. Focused interactionsenhanced productivity withinthis mathematical subspecialty –even as its members turned fromother tasks in their efforts to beamong the first to find newanswers.
1
Scientists who use IT appear tobe more productive – theyreportedly write more papers,earn greater peer recognition, andknow more colleagues (4).“Collaboratories” provide newways to coordinate large-scaleresearch projects and to accessremote data sets, researchspecialists, and equipment (5).This evidence supports thepromise of the World Wide Webto build broader, richer scientificcommunities. The particlephysicists group at CERN, afterall, designed the Web to enhancecollaboration.
Electronic BalkanizationYet, if IT helps an algebraictopologist in North Americaspend more time interacting withcolleagues in Europe, Asia, andelsewhere, what happens to hisor her interactions with thecomputer scientist, thebiologist, or the graduate studentwho works down the hall? Asquickly as IT collapses barriersbased on geography, it forces usto build new ones based oninterest or time. Ironically,global communication networkscan leave intact or even promotepartitions based on specialty,politics or perceived rank –divisions that can matter farmore than geography.
Thomas Kuhn identified thewidening gulf between scientificspecialists as a problem overthree decades ago (6). The powerof emerging IT to search,connect, screen, and select canexacerbate this problem. An ITtelescope that brings distantcolleagues into focus can
inadvertently produce tunnelvision and peripheral blindness.Geographic balkanization, whichmight have separated scientistsin physical space, can give wayto electronic balkanization,which separates them in “topicspace”.
Figure 1 shows how.
In figure 1, local connectionsare traded for long-distanceconnections with greaterrelevance by interest group.Because the Internet makes iteasier to find colleagues withsimilar interests, professionalintegration substitutes forgeographic integration. Theseinteractions can also be modeledformally. Let the topics beindexed by t ∈ {1, 2, ...T} andthe community members whoare aware of a given topic beM(t), then as one index of
11TT
M(t)∩M(s)2
≡1−TT−1t∈{1,2...∑T}∑s≠tM(t)M(s)
“balkanization” we can define:In the figure, balkanization risesfrom .35 to 1 as inter-groupinteraction declines. Moregenerally, models and numericalsimulations reveal that this andother metrics of balkanizationcan increase as technologyimproves searching, filtering andlong-distance collaboration.Preferences and Privateversus Social IncentivesAs access technology improves,individual preferences largelydetermine whether balkanizationincreases or decreases. Ifscientists prefer more focusedinteraction than is availablelocally, then to the extent IThelps scientists to satisfy thesepreferences, an increase in ITwill lead to more narrowscientific interactions. Becausethe Internet makes it easier tofind more interesting contacts,those less interesting contacts
near the threshold of attention
may be abandoned. Unlessscientists actively seek diversity,global access might thereforebalkanize interactions.
Thus marine hydrodynamics andcomputational fluid dynamicshave emerged among thousandsof new and distinct discussiongroups across the Internet. Thenarrower the focus, the greaterthe depth of interaction on agiven research topic. Focus is aresponse to the poverty ofattention. When dailyinteractions bombard us withirrelevant information, a strongdesire to focus might evolve as auseful heuristic for minimizingdistractions. By radicallyimproving filtering, however,advanced IT can lead this sameheuristic to inordinately favordepth at too great a cost inbreadth. Old strategies canbecome counterproductive innew environments.
Developing many new contactswithin a specialty also calls for asecond form of focus: anemphasis on quality. The WorldWide Web accumulates twohundred thousand new pagesdaily, along with thousands ofpostings to public discussiongroups – a wealth of informationthat creates a need to sort thediamonds from the dust. Topreserve the caliber ofinteraction, most prominentscientists retreat to small privatee-mail lists and inviteddiscussion groups (7, 8). Thisscreening process growsincreasingly efficient. Therecently announced Platform forInternet Content Selection willnot only enable the labeling ofmaterial in online journals (e.g.“seminal article”), but thecreators foresee “labels forUsenet authors according to thequality of the messages theypost: posts from those withpoor reputations could bescreened out” (9 p. 93).
2
Communications that oncedepended on geography,proximity, and serendipity arescreened and filtered for perceivedrelevance and reputation.Separation and stratification caneasily result.
Science advances not just frompublication, but from dialogue,apprenticeship, andcollaboration. Although wecommonly think of scientificknowledge as a public good –available at zero cost once it hasbeen produced – the specializedskills, education, andforeknowledge needed to use itimply that scientific informationis far from free. Incentives arenecessary to encourage creation,distribution and use.
The incentives faced byindividual scientists, however,do not automatically lead themto act in ways that maximizescientific progress.Overspecialization, for example,can be privately beneficial whilehaving the effect of erectingvirtual walls between scholarlycommunities. If intra-disciplinary interactionssubstitute for inter-disciplinaryinteractions, then the intellectualcross-pollination of ideas cansuffer. Consider that the Black-Scholes equation for pricingfinancial options is derived froman arbitrage model that leads tothe heat transfer equation (10).Conceivably, reducing thespillovers betweenthermodynamics and financecould have forestalled thedevelopment of options markets.Similarly, the Alvarez theorythat an extraterrestrial impact ledto the extinction of the dinosaursoriginated from the collaborationof a geologist and a physicist atUC, Berkeley. Crick’s trainingin physics and Watson’sbackground in zoology helpedthem develop their theories ofDNA (11). In general, theinsularity of sub-populations
negatively affects the speed atwhich new ideas percolatethrough an entire population(12).
Conclusions
The balkanization of science isby no means inevitable. Ascientist may use IT to selectdiverse contacts as easily asspecialized contacts. Whethertechnology contributes tointegration or fragmentationhinges on individual preferencesand factors such as whether thepressure to publish at thefrontier of one’s own disciplineis low enough to permit time forexploration in others.
New technologies give usoptions that we never had toconsider before. They enableboth the global village and thevirtual Balkans of scientificcollaboration. While no singlescenario is inevitable, certainoutcomes, once achieved, can bedifficult to reverse. At this earlystage of developing informationinfrastructure, we can, andshould, explicitly consider whatwe value as we shape the natureof our networks – with noillusions that a greater sense ofcommunity will inexorablyresult.References1.
T. W. Malone, J. F.Rockart, ScientificAmerican 265, 128-136(1991).
2.E. Noam, Science 270,247-249 (1995).
3.
M. V. Alstyne, & E.Brynjolfsson, ElectronicCommunities: GlobalVillage orCyberbalkanization?, Proc.of the InternationalConference on InformationSystems, Cleveland, OH(1996).
4.
B. Hesse, L. Sproull, S.Kiesler, J. Walsh,
Communications of theACM 36, 90-101 (1993).5.W. A. Wulf, Science 261,
854-855 (1993).
6.T. Kuhn, The Structure of
Scientific Revolutions(University of ChicagoPress, Chicago, ed. 2nd,1970).
7.J. Chao, “Internet Pioneers
Abandon World TheyCreated,” Wall StreetJournal, June 7 1995, pp.B8.
8.S. Steinberg, Technology
Review 97, p25-31 (1994).9.P. Resnick, J. Miller,
Communications of theACM 39, 87-93 (1996).10.F. Black, M. Scholes,
Journal of PoliticalEconomy 81, 637-654(1973).
11.A. S. Moffat, Science
(1993).
12.P. A. David, D. Foray, in
Global TelecommunicationsStrategies and TechnicalChange G. Pogorel, Ed.(Elsevier, Amsterdam,1993).
3
因篇幅问题不能全部显示，请点此查看更多更全内容