< Game Theory


Friday, October 16, 2009

BOB GARFIELD: This is On the Media. I'm Bob Garfield.

BROOKE GLADSTONE: And I'm Brooke Gladstone.


ANNOUNCER: From all corners of the world heroes are being called to join the quest that will decide the fate of Norrath.

BROOKE GLADSTONE: EverQuest II is a massively multiplayer online roleplaying game. It is not the Gallup Poll. But some social scientists think it actually may offer a clearer picture of human behavior than standard surveys, which generally rely on respondents answering randomly selected phones. Dmitri Williams, assistant professor at the Annenberg School for Communication, is part of a group of researchers who believe that they can get better information about our public predilections than polls can provide, through these massively multiplayer online games, even in a realm of elves, dwarves, ogres and trolls. He argues that recording every single action by the human-controlled characters in EverQuest II is better than any study conducted offline.

DMITRI WILLIAMS: There’s a big difference between knowing what somebody did and asking them what they did. When we do standard survey research we always have to worry about whether or not the people are giving us accurate answers. They might give us inaccurate answers based on what they want us to hear or what they want to think about themselves, or they might simply get the information wrong.


DMITRI WILLIAMS: Well, the main reason that we used EverQuest II is because the gamemakers, Sony Online Entertainment, were nice enough to work with us. So why we picked that particular game wasn't because we really like gnomes or dragons. We simply got the data we could get. These online worlds record just about everything that anybody does, so you can imagine if your life was recorded second by second, everything that you did, every action, every interaction, every transaction -- you got in your car, you drove to work, you did this, you did that -- in a virtual space if you acquire some gold coins or, you know, slay a big spider or something, it wants to know that and it keeps a record of it. So it’s just like a very, very good journal-keeping of behaviors and actions.

BROOKE GLADSTONE: What is it you’re tracking?

DMITRI WILLIAMS: It’s hard to get your head around, you know, who’s slaying a dragon, but everybody understands buying and selling goods, and a big part of the action and the play of these games is people buying and selling items, making items, repairing items. That lets us track an entire economy and measure supply and demand. And we can do it with such a level of detail that it’s better than anything you could get offline. For example, when the United States wants to make an estimate about what’s going on with GDP, it is just that. It’s an estimate. What we have is every single transaction that ever happened, period.

BROOKE GLADSTONE: Not only are you dealing with a world that bears very little resemblance to the real world, you’re also dealing with demographics that are very different from the real world. These players are younger, they're wealthier, they're whiter; they're more male. I don't see how you can extract useful information.

DMITRI WILLIAMS: The phenomenon you’re talking about is something I call mapping, and that’s the sense that is the behavior in the virtual world valid outside. In other words, can we say that the things that happen inside do, in fact, happen outside? And the honest answer is, in some cases it is, definitely; in some cases it definitely is not. A couple years ago in the very popular online world, World of Warcraft, there was an incident called the Ironforge Plague. And so, what happened was you had an epidemiological event going on, where people were infecting each other from person-to-person around the city. And epidemiologists thought, wow, what a great way to study disease vectors. It would be, except that the behaviors in the virtual world are very different than the offline world, in this respect: The risks of death and the feeling of pain don't map from A to B. So you wouldn't want to look at people’s virtual behavior in that sense and say, oh, that tells us how people are going to behave in a disease outbreak. It wouldn't, because in a disease outbreak you don't run around cackling with glee and trying to kill your friends.

[BROOKE LAUGHS] But in the virtual world you would.

BROOKE GLADSTONE: If there are no consequences, maybe you'd be much more willing to go broke in a virtual world [LAUGHS] than you would in the real one.

DMITRI WILLIAMS: You could imagine that, you know, maybe the players wouldn't take it seriously and they would fritter away their money or they wouldn't make rational purchasing decisions, but it turns out they behave exactly like a real-world offline market. And so, it’s that kind of confirmation that was a nice first step.

BROOKE GLADSTONE: You could confirm that you could apply this method to track behavior in the market realm.

DMITRI WILLIAMS: You know, eventually what we'd like to do is be able to also then alter the world or do experiments, because we're not the game developers. We can't, you know, charge the price of something and then see what happens or change the structure. But you could see how powerful that method could be if you could basically do a live experiment, with control and treatment groups and hundreds of thousands of people. That’s something that no scientist has been able to do before.

BROOKE GLADSTONE: So how could studying EverQuest II have led you to possibly predict what happened in the real world, with regard to the recent economic collapse?

DMITRI WILLIAMS: You know, flashback five, six months ago when everybody’s arguing about what should we do to fix the economy. Do we bail out banks? Do we subsidize certain industries? Well, we essentially did a large-scale trillion-dollar experiment with live people. And the idea that you could do this in a virtual world, in a sense, a simulation but with real people, for substantially [LAUGHS] lower cost, for maybe, you know, 100,000 dollars, you could find out what would happen in an economy given certain policy choices. It’s the power of that idea that makes this potential so strong.

BROOKE GLADSTONE: Dr. Williams, thank you very much.


BROOKE GLADSTONE: Dmitri Williams is an assistant professor in the Annenberg School for Communication at the University of Southern California.