Media scholars have become increasingly concerned with the possible negative social and civic impacts brought on by the diffusion of both traditional media like television and cable and new media such as videogames and the Internet. This concern is perhaps best known as the “bowling alone” hypothesis (Putnam, 2000), which suggests that media are displacing crucial civic and social institutions. According to Putnam, time spent with relatively passive and disengaging media has come at the expense of time spent on vital community-building activities. While few dispute Putnam’s richly detailed evidence of the general decline of civic and social life in America during the rise of television, some scholars have argued that online, Internet-based media are exceptions. The evidence to date is mixed (Smith & Kollock, 1999), with some scholars arguing that the Internet’s capacity for connecting people across time and space fosters the formation of social networks and personal communities (Wellman & Gullia, 1999) and bridges class and racial gaps (Mehra, Merkel, & Bishop, 2004), and other scholars arguing that the Internet functions as a displacer (Nie & Erbring, 2002; Nie & Hillygus, 2002) enabling little more than “pseudo communities” (Beniger, 1987; Postman, 1992).
A core problem on both sides of the debate is an underlying assumption that all Internet use is more or less equivalent (Bakardjieva, 2005). Online technologies enable a broad range of activities: searching information, visiting chat rooms, downloading music files, corresponding with friends and family by email, browsing political blogs, playing in 3-D virtual worlds, and others. It would be more plausible and empirically rigorous, then, to consider how specific forms of Internet activity impact civic and social engagement as a result of their particular underlying social architectures (Lessig, 1999)—their designed-in, code-based structures that afford some forms of social interaction and constrain others. In this way, we might determine what underlying variables are involved in each activity (Evelund, 2003) before drawing conclusions about the effects of online media as a whole.
In this article, we examine the effects on social engagement of one particular increasingly popular online activity: large, collaborative online videogames called “massively multiplayer online games” (MMOs). Our collaboration on this project is somewhat novel, combining conclusions from two different lines of MMO research conducted from two different perspectives—one from a media effects approach, the other from a sociocultural perspective on cognition and learning. Our joint product represents the culmination of these two lines of inquiry in terms of (a) the extent to which such spaces are structurally similar to “third places” (Oldenburg, 1999) for informal sociability (Bruckman & Resnick, 1995), and (b) their potential function in terms of social capital (Coleman, 1988). Despite differing theoretical and methodological vantage points, our conclusions are remarkably similar: By providing spaces for social interaction and relationships beyond the workplace and home, MMOs have the capacity to function as one form of a new “third place” for informal sociability much like the pubs, coffee shops, and other hangouts of old. Moreover, participation in such virtual “third places” appears particularly well suited to the formation of bridging social capital (Putnam, 2000), social relationships that, while not providing deep emotional support per se, typically function to expose the individual to a diversity of worldviews. In this article, we present our shared theoretical framework of third places and social capital, highlighting the consistent trends observed across two distinct sets of data gathered through two separate lines of inquiry.
MMOs are graphical two-dimensional (2-D) or three-dimensional (3-D) videogames played online, allowing individuals, through their self-created digital characters or “avatars,” to interact not only with the gaming software but with other players. Aesthetically, they are part of the long history of alternative worlds found in science fiction and fantasy literature (e.g., The Hobbit, Tolkien, 1938). Technically, they are the latest step in a progression of social games that originated with paper-and-pencil fantasy games (e.g., Dungeons and Dragons,Gygax & Arneson, 1973) and later migrated to computers, first as mainframe text-based multi-user dungeons (MUDs) (Trubshaw & Bartle, 1978) and later as the high-end 3-D digital worlds of today (Koster, 2002). The virtual worlds that today’s MMO players routinely plug in and inhabit are persistent social and material worlds, loosely structured by open-ended narratives, where players are largely free to do as they please, to slay ogres, siege castles, etc. They are known for their peculiar combination of designed “escapist fantasy” and emergent “social realism” (Kolbert, 2001): In a setting of wizards and elves, dwarfs and knights, people save for homes, create basket indices of the trading market, build relationships of status and solidarity, and worry about crime. The online gaming industry continues to prosper, with over nine million subscribers worldwide (Woodcock, 2006). MMOs are played heavily (average time spent in-game is 20 hours per week, Yee, 2002) and often with friends and relatives (Yee, 2006).
Divergent research, convergent findings
The goal of this article is to present a theoretical framework for understanding the social form and function of MMOs based on conclusions from two research projects: one an examination of the media effects of MMOs, the other an ethnographic study of cognition and culture in such contexts. Both projects used a mixed-methods approach; however, the former leaned toward quantitative data collection and analysis while the latter leaned toward qualitative. Although grounded in different theoretical perspectives and research traditions, the conclusions of both studies were remarkably aligned. Thus, we took it upon ourselves to collaborate in the development of a theoretical framework that might encompass and elucidate the findings of both. Despite the seeming novelty of such an enterprise, cross-disciplinary collaboration is frequently advocated in academic research generally (Lewis, 1997; Stake, 1995), and in games research specifically (Williams, 2005), under the assumption that the most fruitful advances are sometimes made when congruent findings are discovered through disparate means (Kuhn, 1961).
The first project was a traditional effects study of MMOs focused on issues of social capital and displacement that examined questions raised by Putnam, Nie, Wellman, and others that sought to discover whether the MMOs Asheron’s Call I and II added to or subtracted from players‘ overall social capital. The methods used were a combination of survey research and experimental design, with first-time MMO players given a free copy of the game or placed in a control group that did not play the game (Williams, in press). Seven hundred and fifty participants populated the two groups. The changes among the treatment group players were contrasted between the two groups and presented statistically to demonstrate the “effects” of game play vs. no game play.
This statistical evidence was buttressed by a complementary participant observation phase during which the investigator played the game and conducted 30 random interviews with other players over a period of one year. In semi-structured interviews, players were asked about their motivations for playing, their in-game social networks, and about their life outside the game. The investigator sampled players in central city locations and in remote wilderness areas across all of the games’ experience levels and kept field notes to inform the subsequent survey and experimental work. Unlike the survey data, these interviews were not intended to be representative, but rather to discover trends and norms that could be compared and contrasted with the quantitative data and that could help fashion appropriate survey questions. Nevertheless, these observations squared with the survey results and with the results of the second project detailed below.
The second project, a qualitative study of cognition and learning in MMOs (Steinkuehler, 2005), consisted of a two-year ethnography of the MMO Lineage (first I, then II) conducted from a sociocultural perspective (Gee, 1999; Lave & Wenger, 1991). The goal of this project was to explicate the kinds of social and intellectual activities in which gamers routinely participate, including individual and collaborative problem solving, identity construction, apprenticeship, and literacy practices. Cognitive ethnography (Hutchins, 1995) was chosen as the primary research methodology as a way to tease out what happens in the virtual setting of the game and how the people involved consider their own activities, the activities of others, and the contexts in which those activities takes place (Steinkuehler, Black, & Clinton, 2005). This “thick description” (Geertz, 1973) included 24 months of participant observation in the game (eight months averaging eight to 12 hours of gameplay daily, the remaining months averaging 20-40 hours per week); several thousand lines of recorded and transcribed observations of naturally occurring game play (digital screenshot images, video recordings, and fieldnotes); and collections of game-related player communications with informants including posts to official (NCSoft sponsored) and unofficial discussion boards (on guild and fan websites), chatroom transcripts, instant message conversations, and emails. Also considered were community documents from the Lineage fandom network (i.e., materials either linked directly to the corporate sponsored website or within the set of game-related sites that contained more internal links among them than external links beyond them; Barabási, 2003) including guild and fan websites, fan fictions, and community-written player manuals and guidebooks.
As part of the second project, repeated unstructured and semi-structured interviews were conducted with a snowball sample (Hammersley & Atkinson, 1986) of 16 key informants throughout the course of the investigation, with interviews lasting one hour on average, resulting in roughly 100 hours of recorded dialogue (either IM transcript or audio recording). Additional interviews were not recorded but used as a basis for fieldnotes, fact checking, triangulation of data, and verification of major themes. The data were organized using NUD*IST and NVivo qualitative data analysis software to identify major themes (e.g., learning through apprenticeship), and were coded inductively in the second year of the study in order to identify major patterns within those themes (e.g., scaffolding strategies). Discourse analysis (Gee, 1999) was used throughout the investigation as the fundamental basis for analysis in order to tease out how the underlying assumptions or “cultural models” (Holland & Quinn, 1987) participants held about the virtual social and material world were created, maintained, and transformed by specific individuals and social groups whose ways of being in the world underwrite them.
In combination, these two empirical studies provide a reasonable level of generalizability (random assignment to condition in the first study) and contextualization (ethnographic description of existing in-game social networks and practices in the second). The two franchises investigated, Lineage I and II and Asheron’s Call I and II, represent a fairly mainstream portion of the fantasy-based MMO market: All four titles involved players assuming the roles of archetypal medieval fighter types, social architectures which reward players for cooperation and the formation of long-term player groups or “guilds.” The Lineage series is a highly successful franchise in South Korea, with moderate U.S. success; the Asheron’s Call series was moderately successful in the U.S., but not abroad. The combined data corpus provided a range of in-game contexts for observation while exhibiting consistency in design features across all four.
Bridging and bonding in MMO third places
In the previous section, we examined the structural form of virtual worlds, demonstrating how MMOs satisfy Oldenburg’s (1999) eight defining characteristics of the third place. In this section, we turn to the function of such spaces. Even with Oldenburg’s eight criteria met, a fundamental question remains: Are virtual communities really communities, or is physical proximity necessary? Much scholarly work on the viability of online communities has been influenced by the work of Anderson (1991), who suggests that geographic proximity itself is neither a necessary nor sufficient condition for the emergence and preservation of “community.” As he points out, conglomerations such as “America” or “Iraq” are no more face-to-face than networked, online ones, yet we generally acknowledge them as large “communities” based on their internally coherent, shared sense of history and information—collective characteristics made possible by a shared national media (Feenburg & Bakardjieva, 2004). Rather than presume that a shared medium (here, an MMO) alone suffices to enable community, however, we take the functional characteristics of “community” that can be operationalized and observed and examine them in MMOs in order to see whether or not such virtual contexts serve the same ends as “real world” communities do. Toward this end, we turn to the concept of social capital and its component parts, bridging and bonding (Putnam, 2000).
Social capital (Coleman, 1988) works analogously to financial capital; it can be acquired and spent, but for social and personal gains rather than financial. For example, by comforting a friend, a person is then more readily able to seek comfort in his or her own time of need. Thus social capital operates cyclically within social networks because of their associated norms of reciprocity (Newton, 1997). Such patterns can occur online as well as offline (Resnick, 2001).
According to Putnam, bridging social capital is inclusive. It occurs when individuals from different backgrounds make connections between social networks, functioning as a kind of sociological lubricant. This form of social capital is marked by tentative relationships, yet what they lack in depth, they make up for in breadth. On the one hand, bridging social capital provides little in the way of emotional support; on the other hand, such relationships can broaden social horizons or worldviews, providing access to information and new resources. In contrast, bonding social capital is exclusive. It occurs when strongly tied individuals, such as family and close friends, provide emotional or substantive support for one another, functioning not as lubricant but more as a kind of social superglue. In contrast to bridging social capital, bonding social capital is marked by relationships with less diversity but stronger personal connections. It provides continued reciprocity among individuals who share strong emotional and substantive support. However, it can also result in insularity.
Granovetter’s (1973, 1974) early work on tie strength shows that bridging and bonding social capital are tied to different social contexts, given the network of relationships they enable. In the context of MMOs, the question is whether the online communities found within them tend to be large, weak networks (bridging) or small, strong ones (bonding). Evidence from studies of the Internet generally suggests that online social networks are characteristically broad, bridging-oriented networks (Constant, Sproull, & Kiesler, 1996; Pickering & King, 1995), although both weak and strong ties can be forged within them (Wellman, Salaff, Dimitrova, Garton, & Haythornthwaite, 1996). Our conclusion is that MMOs confirm this pattern: Virtual worlds appear to function best as bridging mechanisms rather than as bonding ones, although they do not entirely preclude social ties of the latter type.
Relationships and behaviors associated with bridging social capital were the most prevalent and noticeable across our MMO data corpus, especially in games with a vibrant social scene. Because Asheron’s Call II was plagued by low subscription rates, it was not a strong source of either form of social capital simply because there were not enough people present; its low population rates (see Figure 6) demonstrated that a truly unpopular third place can lead to poor outcomes of either form.
Asheron’s Call II players experienced the opposite outcomes from players in the other three titles: drops in the diversity of networks and drops in bringing social capital (Williams, 2006, in press). In this sense, it was the exception that proves the rule; a lonely game, much like an empty bar, produces effects opposite to those of a vibrant third place. All three other MMOs in our research generated positive and obvious bridging outcomes that one would associate with a real-world third place. Broad, weak social networks were common. Within such networks, individuals from a wide range of backgrounds mixed on the level playing fields that Oldenburg praises: Supervisor and supervisee, parent and child, classroom teacher and student generally left behind their out-of-game roles and participated as equals. Individuals with diverse worldviews found themselves interacting on a level playing field. For example, during the 2004 U.S. presidential election period, supporters of both presidential candidates cussed and discussed the televised debates while slaying monsters in game. This mixing extended to game play issues as well. In some cases, teenagers mentored adults twice their age and education in how to level their avatar, find the best territories for hunting, or lead a guild. However, while such bridging was frequent, it was more likely to lead to potential resources or new information within the longer-term groups such as “guilds” than within the temporary “pick up” groups that band together for short-term goals. Within the guild organizations in particular, players were able to establish enough of a relationship to exchange real-life information beyond the basic “a/s/l” (“age, sex, and location”) to include a greater diversity of viewpoints and experiences.
In contrast, bonding social capital was much rarer. Across our combined data corpus, there were only a handful of cases in which deeper, more substantive relationships formed. In most such cases, the bonds formed within long-term guilds and over a period of at least several months. For example, one informant met his fiancé when, after gaming online together for several years, they decided to meet face-to-face. In another case, a long-term international friendship was formed through online gaming, with one guild member from Singapore visiting the other in the U.S. In another instance, when a female guild member began living alone, a cadre of fellow guild members would call to check in on her when she was later than usual logging in. Such examples of the formation of bonding relationships are compelling, although for the most part (as in offline contexts) they are relatively infrequent. One could argue that, if the benchmark for bonding social capital is the ability to acquire emotional, practical, or substantive support, then MMOs are not well set up for the task: While deep affective relationships among players are possible, they are less likely to generate the same range of bonding benefits as real-world relationships because of players’ geographic dispersion and the nature of third places themselves. As to the former, it is difficult to provide one another rides to work or a literal shoulder to cry on when friends live in different states and time zones. As to the latter, the persistent “playful mood” (Oldenburg, 1999) of the third place of MMOs often stymied players’ attempts to engage in emotionally weighty conversation.