We've seen WoW used for a lot of research, from epidemics to anthropological fieldwork, but this is probably one of the craziest and one of the most helpful (assuming it works) ways to use it. Psychologists at the University of Miami and the University of California, Irvine have been studying how guilds and groups form in World of Warcraft in the hopes that it'll help them figure out how gangs form in real life. It sounds like a wild idea, but following guilds and groups in World of Warcraft is much easier than trying to study spontaneous guilds in the real world, because you've got immediate access to data: when people joined and left and why. And the psychologists say putting data together like this will help, because it'll help answer questions about, for example, what happens when you decide to separate a group of people -- do they form their own groups again or do they stay separated?
They say there are other connections as well: though killing dragons is far less heinous than killing innocent bystanders, Warcraft guilds form, grow, stick together, and fall apart just like gangs and even other groups all over the world do. No matter what kind of group it is, the researchers say that "group ecology" is the same everywhere, so studying the way we work in endgame raids can lead to ideas about what we're doing elsewhere. Very interesting.
Unfortunately, they're full on potential but still pretty short on conclusions yet (listen, guys, all you have to do to break up gangs is ensure there's not enough loot to go around), but once again, Azeroth seems like a fertile ground for directly studying just how we players interact as humans.