The "social factor" is often advanced to explain the popularity of MMOs: "it’s the people that are addictive, not the game." And indeed, social networks are essential components of the online gaming experience. Game designers build "virtual third places" to encourage interactions. Quality time in a MMO is time spent with others, be it raiding a dungeon in a group, socializing with bystanders in a cantina, or chatting with remote guildmates while exploring the wilderness on your own.
Or... is it? The data PlayOn has assembled from World of Warcraft challenges many of these assumptions. Could it be that a less, not more, social environment contributed to WoW’s success? Read on for more.
For the past 8 months, we have been collecting and analyzing data from WoW. We use "bots" on 5 different servers (covering PvP, PvE and RP realms) and take snapshots of the population every 10 minutes or so (more detail here), capturing data about each present character's level, location, grouped status, and guild affiliation (among other things). So far, we have observed about 150,000 unique characters. While this data is far from perfect (after all, we are only getting what Blizzard is willing to let transpire through their API), it can still yield a surprising amount of insight into the social dimensions of multiplayer games.
We have tried to assess the prevalence of social activities in WoW using a variety of metrics. The most obvious is the time players spend in groups. In the early stages of the game (level 40 and below), it hovers at around 30%. Even if we assume players spend a fair amount of time crafting, selling goods on the auction house, etc., it is hard to see these activities accounting for 70% of the time spent in-game. It is only at level 56 and above that players spend the majority of their time in groups (probably raiding high-end instances). Moreover, players favor "soloable" classes (warriors, hunters) that, by design, survive mob encounters better in solo play - the more social classes (e.g. priests) that require a group to work well are among the least favored. Compounding the problem, and despite features like WoW’s "group xp bonus", grouping is an inefficient way to level, which naturally steers the more "hardcore" players away from groups (at least, in the early stages of the game).
Overall, this data indicates that instrumental group play in the context of quests might not be an important source of social interactions in WoW. How about guilds? Excluding level 1 characters, 62% of WoW’s players are in guilds. Interestingly however, 17.5% of them belong to a "one-person guild" - in other words, nothing more than a custom tag below their name (which, incidentally, says volume about the trouble players are willing to go through to customize their avatars). That leaves only 44.5% of players in guilds with at least 2 members. Moreover WoW’s guilds are fairly small (median size: 9) and very few grow beyond 35 members (90th percentile) - a kind of Dunbar Number for guilds.
More importantly, guilds are also very sparse social networks. Our analyses indicate that the average guild member in WoW spends little time with a large majority of his/her guildmates. Over a month, the average guild member is online 20 minutes or less with 80% of his/her guildmates. The same average member collaborates (in quests, etc.) with only 11% of his/her guildmates for more than 10 minutes over the same month. In other words, most guilds in WoW are not highly social and/or collaborative environments.
Considering all this, it is easy to wonder why WoW’s players spend so much time (more than 10 hours per week) in this game instead of a single-player RPG! To me, WoW illustrates how a large number (the majority?) of online gamers enjoy being "alone together." The social factor at work here clearly has little to do with direct interactions and camaraderie in the context of quest groups or guilds. Instead, it looks as if other players are mostly:
- An audience (to showcase your latest "elite" gear and other accomplishments).
- A social presence (the constant flow of chat in the wide-reaching "general" channel, the movement of other avatars around you, make playing WoW somewhat analogous to reading a book in a densely populated café - while you may not necessarily choose to interact with the other patrons, the sense of being in a public social space can be attractive enough to conduct individual activities there).
- A source of spectacle and entertainment ("people watching" is fun in WoW, in part because Blizzard encouraged it by design with many humorous objects that can be appropriated by the players - the gnomes’ various trinkets with unpredictable effects, the infamous 20-lb catfish that can be used as a weapon, etc).
Does this mean that, to reach WoW's scale and attract gamers as-yet unfamiliar with the genre, future MMOs should focus less on collaborative questing and other traditional techniques to encourage interactions, favoring instead "looser," more indirect forms of social experience? A "society of the spectacle," to use Guy Debord's terminology? Or are there alternatives?
(P.S.: an expanded version of this analysis will appear in a CHI 2006 paper I'd be happy to send to interested readers).