CIA Takes a Gamble on Prediction Markets

In the last decade or so, prediction markets have forecasted everything from box office sales of movies to the outcomes of elections. Now, Verizon, Carnegie Mellon University, and In-Q-Tel (which develops technology for information agencies including the CIA) are working together to launch a prediction market for information security.

Markets that offer participants money or other valuable incentives for accurate predictions tend to be more reliable forecasters than traditional polls. According to “The Promise of Prediction Markets” published in Science, when there is something at stake for the predictors, “the potential for profit (and loss) creates strong incentives to search for better information.” And thus, the accuracy of the prediction goes up.  The Iowa Electronic Market, for example, is believed to be 75% more accurate than traditional polls in predicting the election results.

When bets are monetary, though, prediction market developers run up against state and federal online-gambling restrictions. In the past, prediction markets have gotten around these regulations by donating the winnings to charity or creating alternate currencies. Greg Shannon of Carnegie Mellon University’s Software Engineering Institute told Forbes, “In the long term, we want to have money involved. But getting through the legal issues isn’t trivial.”

The last time a government agency tried to implement a prediction market, a political storm erupted as soon as the program went public. In 2001, The Defense Advanced Projects Research Agency (DARPA) launched FutureMAP, a market that predicted strategies “for avoiding surprise and predicting future events” related to global terrorism and national security. In 2003, the program was shut down after senators Ron Wyden and Byron Dorgan called the predictions “terrorism betting parlors.”

An article on the CIA’s website defends FutureMAP, stating, “It was an experiment to see whether market-generated predictions could improve upon conventional approaches to forecasting.” It’s possible that if FutureMAP had attempted to forecast something less controversial than terrorism-related activity, Congress would not have reacted so strongly, but a prediction market for cyber threats and data security could be just as problematic – especially since information technology has permeated every commercial and governmental area of modern life.

Though the CIA’s prediction market sounds plausible, there are still a lot of questions developers will have to answer before launching. For example, who gets to participate? The more people allowed to participate, the more information collected. Presumably, this is a good thing; however, it could cause the accuracy of the overall prediction market to drop, making it little more than a federal betting pool. To become a respected governmental forecasting tool, prediction markets will have to distance themselves from gambling, both legal and in the eyes of the public. How can they do this while still providing participants with meaningful incentives to “win” a bet? What might such an incentive be?

Leave a Reply