The government may run a research project on public data in games. Yawn.

I was really hoping not to cover this, but the coverage I keep seeing is so over-the-top (e.g. TERRORISTS IN WORLD OF WARCRAFT!) that, having actually read the report in question, I’d be remiss not to at least briefly comment.

Here’s the real story of the Reynard Project (“Reynard” is a trickster fox from medieval European folklore and literature) from the report from the Office of the Director of National Intelligence to Congress (.pdf) that sparked the firestorm (report found via Wired’s Threat Level blog, emphasis is mine):

Reynard is a seedling effort to study the emerging phenomenon of social (particularly terrorist) dynamics in virtual worlds and large-scale online games and their implications for the Intelligence Community.

The cultural and behavioral norms of virtual worlds and gaming are generally unstudied. Therefore, Reynard will seek to identify the emerging social, behavioral and cultural norms in virtual worlds and gaming environments. The project would then apply the lessons learned to determine the feasibility of automatically detecting suspicious behavior and actions in the virtual world.

[...]

Reynard will conduct unclassified research in a public virtual world environment. The research will use publicly available data and will begin with observational studies to establish baseline normative behaviors.

So basically, they’re going to be standing on a virtual street corner noting how anonymous users interact and compiling data on that to try to establish a set of baselines from which they could, later, maybe spot deviations.

Commentary

This is simply not what everyone is making it out to be. The government is not investigating terrorists in World of Warcraft. They are not getting chat logs from providers. They are not secretly monitoring conversations. They’re just using cheap public data to see if they can spot patterns.

For another take on this, Juan Cole over at Salon debunks the idea of looking for terrorists in games pretty completely, but that article misses the much simpler point that that’s not even what they’re doing here. Not yet, anyway.

I think this is potentially pretty smart on the part of the analysts, and at least as written, it doesn’t raise privacy flags for me. They just appear to be hoping to mine the massive pile of conveniently anonymous publicly available data games produce regarding “social, behavioral, and cultural norms.” They hope this will, in the aggregate, eventually reveal patterns that they can apply later.

So, for example, they could learn that guild recruitment in a game typically follows a pattern of contact with a recruit that goes (a) leader, (b) third in command, (c) second in command, (d) leader. Not always, but often enough to be notable. They could then run that data over later contact patterns to try to spot apparent attempts at recruitment. They’re not going to be kicking in doors based on this kind of analysis, but it’s another data point, and collecting data points is what (typically boring) intelligence work is really all about.

There is sometimes good reason to suspect that a program like this will go beyond it’s intended scope, but this doesn’t have that feel to me. They’ve got an internal group advocating privacy issues, and the idea is to do all of this (at least at this stage) with publicly available data. If that’s all it is, I don’t really care. Sure, there’s a chance it’d go beyond the plan laid out here at some point, but that’s true of every program, and seems less true for this one than many. Privacy advocates have to pick their battles.

At bottom, it’s definitely a long shot, but they could get data they simply cannot get in the real world, at minimal cost and zero risk. It’s odd to think of analysts studying colonies of gamers like scientists study beehives, but if it stays at the macro and/or anonymous level and uses only public data, I sort of have to shrug. They’re not looking for terrorists in games, they’re just looking for social patterns that they can extrapolate from. That’s boring, and makes a lousy headline, but it’s really not all that bad — and not all that dumb.