David Plotz has a nice piece at Slate on Iraq: The Computer Game. Everyone’s heard of the Department of Defense using shoot-’em-ups like Doom to train soldiers, but another game genre is starting to take hold as well: massively multiplayer role-playing games. If you can create a virtual world, like the Sims or EverQuest, that matches real-world conditions, the outcome of the game can provide hints for how the real world will unfold.
This isn’t just blue-sky thinking, either. General Wesley Clark commissioned a didactic system called SENSE, in which each player played a stakeholder in postwar Bosnia. He had the new Bosnian government play different roles in the game, to show them the consequences of different policies. The game got so heated that the opposition leader had to go on television after one session and explain why the country fell apart while he was playing the role of President.
These games are also good for observing group behavior. This is where Iraq comes in. Edward Castronova from Cal State was approached by the DoD for suggestions on how to model the politics of post-war Iraq. He suggested that they update the War of the Roses strategy game Kingmaker. If you can accurately model the situation, and let the computer simulate each significant role, a few million simulations should give you a sense of how likely certain situations are: if the new Iraq doesn’t join NATO, Iran will invade 25% of the time. The utility of these simulations is not for prediction, but analysis: they can provide a list of outcomes that policymakers could apply to their pet theories.
The most intriguing idea is described by Plotz as NorthKorea.com: create a world that simulates the conditions of North Korea, and let thousands of gamers loose on it. Each player would act in their own interests, and the aggregate effect of their actions would provide an excellent insight on the internal politics of the country. The players would treat it as a game, of course, but observers could glean valuable intelligence from it.
The greater story here is in the use of actual humans to perform a simulation. It’s notoriously difficult to effectively model human behavior. The great insight here is that modelling human behavior is unnecessary: with a set of networked players, you can incorporate the genuine article.