C OMPUTERS AND S CIENCE ing of ESTs that substantially alleviated, even if they and kindred U.S. legislation, see, e.g., J. H. Reichman, P. 37. The role of scientific organizations in facilitating did not totally resolve, this threat to science from Samuelson, Vanderbilt Law Rev. 50, 51 (1997). changes in U.S. policy is recounted in (38). overbroad patent rights. 36. See, e.g., National Research Council, Bits of Power: 38. P. Samuelson, Va. J. Intl. Law 37, 369 (1997). 34. Directive 96/9/EC of the European Parliament and of Issues in Global Access to Scientific Data (National 39. These efforts are recounted by J. H. Reichman and the Council of 11 March 1996 on the Legal Protection Academy of Sciences Press, Washington, DC, 1997) P. F. Uhlir [Berkeley Technol. Law J. 14, 793 (1999)]. of Databases, 1996 O.J (L 77) 20. (expressingconcernaboutEuropeanUnionÐstyledata- 40. I gratefully acknowledge research support from NSF 35. For a critical commentary on the EU database directive base protection). grant SEC-9979852. VIEWPOINT Computer Networks As Social Networks Barry Wellman Computer networks are inherently social networks, linking people, orga- first, computer scientists would be saying nizations, and knowledge. They are social institutions that should not be “netware” instead of “groupware” for sys- studied in isolation but as integrated into everyday lives. The proliferation tems that enable people to interact with each of computer networks has facilitated a deemphasis on group solidarities at other online. Often computer networks and work and in the community and afforded a turn to networked societies social networks work conjointly, with com- that are loosely bounded and sparsely knit. The Internet increases people’s puter networks linking people in social net- social capital, increasing contact with friends and relatives who live nearby works and with people bringing their offline and far away. New tools must be developed to help people navigate and situations to bear when they use computer find knowledge in complex, fragmented, networked societies. networks to interact. The intersection of computer networks Once upon a time, computers were not cial network of them all. Just one small por- with the emerging networked society has fos- social beings. Most stood alone, be they tion of the Internet—Usenet members—par- tered several exciting developments. I report mainframe, mini, or personal computer. ticipated in more than 80,000 topic-oriented here on two developing areas: (i) community Each person who used a computer sat alone collective discussion groups in 2000. 8.1 mil- networks on- and offline and (ii) knowledge in front of a keyboard and screen. To help lion unique participants posted 151 million access. people deal with their computers, the field messages (2–4). This is more than three times of human-computer interaction (HCI) de- the number identified on 27 January 1996 (5) Community Networks On- and Offline veloped, providing such things as more Computer scientists and developers have Community, like computers, has become net- accessible interfaces and user-friendly soft- come to realize that when computer systems worked. Although community was once syn- ware. But as the HCI name says, the model connect people and organizations, they are onymous with densely knit, bounded neigh- was person-computer. inherently social. They are also coming to borhood groups, it is now seen as a less Computers have increasingly reached out realize that the popular term “groupware” is bounded social network of relationships that to each other. Starting in the 1960s, people misleading, because computer networks prin- provide sociability support, information, and began piggybacking on machine-machine cipally support social networks, not groups. a sense of belonging. These communities are data transfers to send each other messages. A group is only one special type of a social partial (people cycle through interactions Communication soon spilled over organiza- network; one that is heavily interconnected with multiple sets of others) and ramify tional boundaries. The proliferation of elec- and clearly bounded. Much social organiza- through space [a low proportion of commu- tronic mail (e-mail) in the 1980s and its tion no longer fits the group model. Work, nity members in the developed world are expansion into the Internet in the 1990s community, and domestic life have largely neighbors (7)]. Where once people interacted (based on e-mail and the Web) have so tied moved from hierarchically arranged, densely door-to-door in villages (subject to public things together that to many, being at a com- knit, bounded groups to social networks. support and social control), they now interact puter is synonymous with being connected to In networked societies, boundaries are household-to-household and person-to-per- the Internet. more permeable, interactions are with diverse son (9). As a result, HCI has become socialized. others, linkages switch between multiple net- Although the support of collaborative Much of the discussion at current HCI con- works, and hierarchies are flatter and more work was the initial purpose of the Internet ferences is about how people use computers recursive (6–8). Hence, many people and or- (both e-mail and the Web), it is an excellent to relate to each other (1). Some participants ganizations communicate with others in ways medium for supporting far-flung, intermit- build “groupware” to support such interac- that ramify across group boundaries. Rather tent, networked communities. E-mail tran- tions; others do ethnographic, laboratory, and than relating to one group, they cycle through scends physical propinquity and mutual survey studies to ascertain how people actu- interactions with a variety of others, at work availability; e-mail lists enable broadcasts to ally relate to each other. This work has slowly or in the community. Their work and com- multiple community members; attachments moved from the lone computer user to deal- munity networks are diffuse and sparsely and Web sites allow documents, pictures, and ing with (i) how two people relate to each knit, with vague overlapping social and spa- videos to be passed along; buddy lists and other online, (ii) how small groups interact, tial boundaries. Their computer-mediated other awareness tools show who might be and (iii) how large unbounded systems oper- communication has become part of their ev- available for communication at any one time; ate—the ultimate being the worldwide Inter- eryday lives, rather than being a separate set and instant messaging means that simulta- net, the largest and most fully connected so- of relationships. neous communication can happen online as When computer-mediated communication well as face-to-face and by telephone. networks link people, institutions, and knowl- Systematic research on what people ac- Centre for Urban and Community Studies, University of Toronto, 455 Spadina Avenue, Toronto, Canada edge, they are computer-supported social net- tually do on the Internet has lagged behind M5S 2G8. E-mail: [email protected] works. Indeed, if Novell had not gotten there the Internet’s development. After a long www.sciencemag.org SCIENCE VOL 293 14 SEPTEMBER 2001 2031 C OMPUTERS AND S CIENCE period of pundit supposition, travelers’ tales, now used by a majority of North Ameri- tained; new ties are developed among people and laboratory studies of computer-mediated cans, although its growth rate is slowing sharing interests. It is not only that time and communication, survey-based and ethnograph- and may stabilize at about 60% of adults. space become less important in computer- ic research is now appearing. The digital divide is decreasing rapidly in mediated communication, but that it is easy These studies address a vigorous public North America, although socioeconomic to communicate with large groups of commu- debate about whether people can find com- status (education, occupation, and income) nity members (using lists) and to bring un- munity online. Critics wonder whether rela- remains an important differentiator (17– connected community members into direct tionships between people who never see, 20). The digital divide is much more sig- contact. The ease with which computer-me- smell, or hear each other can be the basis for nificant in two ways in less developed diated communication connects friends of true community [reviewed in (10); examples countries: (i) A much lower percentage of friends can also increase the density of inter- include (11–13)]. Other detractors make an the population use the Internet and (ii) the connections among clusters of network mem- opposite argument: The Internet may be so users are predominantly well-connected bers within communities. immersive that it lures people away from elites (21). In the developed world, the For one thing, as the newbies studied by other pursuits (14) and involves them in on- amount of time spent online is increasing, Kraut et al. (33) gained more experience with line interactions that only reinforce their ex- per capita as well as overall. For example, the Internet, their depression and alienation isting opinions. the average AOL user spent 31 min per day disappeared, and their social contact in- By contrast, enthusiasts see the Internet online in the first quarter of 1997; in 4 creased enough to have a positive impact on as extending and transforming community. years, this had more than doubled to 64 min their overall interactions with community John Perry Barlow asserts that “with the online in the first quarter of 2001 (22). Nor members. A comparative analysis found that development of the Internet...wearein does familiarity breed interpersonal con- social support obtained online helped people the middle of the most transforming tech- tempt: The more contact people have on- to deal with depression (34). nological event since the capture of fire” line, the greater the impression they make Other studies have found that the Internet (15). They point to the ability of the Inter- on each other (23).