Virtual Behavior Labs Discover What Gamers Want

video game infographic
An infographic for the game "Red Dead Redemption" shows what players accomplished in the virtual Old West during the first two weeks of playtime. (Image credit: Credit: Rockstar/GameSpy Technology/Column Five Media)

Imagine the power to know every consumer purchase ever made, big or small, or the gory details of any crime ever committed. That's the new reality in the worlds of video games.

Tens of millions of gamers inhabit virtual worlds where behaviors or actions can be tracked and tallied, creating some astounding statistics. In the first two weeks of release for the cowboy-themed "Red Dead Redemption," for instance, 13,250,237 virtual U.S. soldiers were killed (or about the same number of actual German and Soviet military deaths combined during World War II). Players also committed a total of 131,904,068 counts of in-game murder and hunted down millions of virtual critters, including 55,813,649 wolves.

The rabbit hole goes deeper. Developers of the futuristic sci-fi game Mass Effect 2 found that 80 percent of the game's players used the face customization system to change their appearances, rather than use the default hero or heroine. Some games even track giggleworthy player behaviors; Mafia II records how long players spend staring at in-game Playboy centerfolds.

But companies and academic researchers aren't laughing; they're too busy turning green with envy.

"I think everyone would kill to have the data that games have," said Todd Northcutt, vice president of GameSpy Technology. "Maybe not kill. They'd love to have the data."

Game developers have slowly realized they're sitting on one of the greatest opportunities ever to study human behavior, as well as to turn a profit from it. They can see what choices players make in-game and even can figure out what gamers want before the gamers ask for it – useful knowledge for creating downloadable add-on content or designing that next best-selling game in the series.

Many developers don't have the tools to understand the social patterns of their player populations, but that's okay with a growing number of researchers, who are eager to help in exchange for a peek at the wealth of behavioral data. Some companies such as GameSpy Technologies even have built their business upon helping game companies track player behavior.

Drowning in the sea of possibility

Virtual world data may look ripe for the taking, but two huge problems stand in the way.

First, sifting through the sheer mass of raw game-play data requires some serious processing power and storage space. One group of researchers made that discovery when Sony Online Entertainment gave them access to anonymous user data from the online game Everquest II. The Virtual Worlds Exploratorium team received terabytes of data compressed as huge files on external hard drives, with each terabyte the equivalent of about 200,000 photos or MP3s.

"Without divulging trade secrets and violating a nondisclosure agreement, it is safe to say that these spaces generate terabytes of data per server per year of operation," said Dmitri Williams, a communications researcher at the University of Southern California.

Second, extracting useful information from the sea of data can prove tricky. Many of the first systems to collect game-play data were designed so that game developers could find and fix game bugs, not so they could analyze human behavior.

"You can't just flip a switch and take a debugging file and use it to study social behavior," Williams explained.

Companies would need to either design the proper data collection into their game code or hire the expertise of outsiders. But GameSpy Technologies' Northcutt admitted that tracking more metrics per player does not necessarily make it easier to figure out what the data mean.

In a recent game that was on several game consoles and PCs, "a pistol on this one platform was far more popular than on another platform," Northcutt said. "We had no idea why; the developer had no idea why."

Brave new world

The challenges have not discouraged outside researchers from taking a shot. One is Jamie Madigan, an industrial psychologist who updates his website (psychologyofgames.com) when he's not working for the federal government.

Madigan hypothesized that gamers with large friend lists on game social networking services such as the PC's Steam or Xbox Live would cheat less often than people with smaller friend lists. His thinking was that having a larger social network of gamers would put people on better behavior, because they would be more likely to want to guard their reputation among friends. He teamed up with GameSpy Technologies to get the game statistics, which, in fact, confirmed his hunch.

To fix that cheating bias, Madigan has suggested using the psychology notion of "priming" to influence gamers to act more cooperatively rather than selfishly. That might mean showing certain words, videos, or stories to gamers while they wait for online matches to begin.

"Even mentioning certain words can influence behavior," Madigan said. "It's not going to get someone to do a 180 [degree turn], but every little thing counts if you sort of nudge."

Game companies have also begun to hire psychologists, economists and communications researchers as in-house experts to sift through the game-play data.

Blizzard, the industry giant behind popular games such as World of Warcraft, has hired such experts for its business intelligence unit. PC game-maker Valve also has a full-time social scientist on staff, and it runs the popular online store Steam, which serves as a hub for buying and playing downloadable games, as well as for tracking player stats.

The game companies that are most attuned to patterns in player behavior are those behind the wildly popular games played on social networking sites, such as the Facebook game Zynga's FarmVille.

"People who are doing the most with metrics and game-play data are the social game folks," Northcutt told LiveScience. "They don't make a move within their game without testing; just a fractional change in player behavior can mean millions of dollars for those guys."

Sharing the wealth

The intense scrutiny of their actions does not seem to bother most gamers, in part because their anonymity is safe. But the experts say gamers themselves can find value in game statistics – whether it's to improve their own game-play strategies or merely to show off in front of friends for topping the public leader boards.

Game developers can build relationships with gamers by sharing their game data and analyses, Williams said. Such openness with the data also helps build up the community of players for each game, according to Northcutt and Madigan.

"When you track stats, you give players a chance to brag about it and create word-of-mouth evangelizing for the game," Northcutt said.

Many younger gamers have grown up accustomed to Facebook and Amazon tracking their personal information and purchase history, Williams pointed out. In that sense, tracking player behavior may not look all that different from the normal standards of online life.

"If you talk to younger consumers, they're much less bothered by giving data away, but they expect something in return," Williams said.

Williams and his academic colleagues have already formed their own company, Ninja Metrics, which will help to track populations within games and social networks. They hope that cooperation with game companies will someday give them permission to manipulate the virtual worlds to see player responses, rather than just watch the "natural experiments" unfold – just another step toward turning video games into true behavioral labs.

"I think we'd all like to get involved with some of the larger-scale social games, the Facebook stuff," Williams said. "That lets us learn more about general populations, as opposed to hard-core game players, and would allow us to push the boundaries of data analysis by sheer volume."

Jeremy Hsu
Jeremy has written for publications such as Popular Science, Scientific American Mind and Reader's Digest Asia. He obtained his masters degree in science journalism from New York University, and completed his undergraduate education in the history and sociology of science at the University of Pennsylvania.