Monday, November 5, 2012

What Can Intelligence Expect From Prediction Markets?

Opening screen of the prediction market
Prediction markets have long been touted as tools that have a wide variety of potential uses for intelligence professionals.  Far more accurate in many cases than expert judgement alone, these markets tend to incentivize good thinking and punish poor thinking in ways that, over time, produce quantifiably better results on topics like elections and sales forecasts.  Strong advocates of this method have even suggested that these markets might be able to replace traditional analysts entirely.

Naysayers have (and will likely continue...) to argue that the kinds of questions asked of intelligence professionals do not lend themselves to pat, numerical estimates.  Furthermore, they will say, even in the handful of cases where such answers would be of potential use, combining the estimates of people who know little to nothing about the details of a particular, narrow problem -- the kind that is usually of intelligence interest -- will only serve to create an estimate that is also of little to no use.  Finally, while these estimates are useless in forecasting the future (or so the naysayers will say), they will serve to anchor both intelligence professionals and policymakers alike, reducing their ability to see alternatives to the predicted outcome.

The purpose of this series of posts, then, is to explore both sides of this argument, to look at prediction markets from the point of view of the intelligence profession and in light of ongoing research and to come to some preliminary conclusions about the future of prediction markets as a tool for the working intelligence professional and the decisionmakers they support.

Informing this series will be the results of research done by the DAGGRE prediction market and the scientists involved in that effort.  DAGGRE is run by Dr. Charles Twardy of the C4I Center at George Mason University and is working in cooperation with a number of other universities and organizations (including Mercyhurst University and yours truly) to better understand prediction markets and their potential uses to the intelligence community.  

The DAGGRE project is one of five such projects funded by the Intelligence Advanced Research Projects Activity (IARPA) under their Aggregative Contingent Estimation (ACE) program.  Now in its second year, ACE has already produced a number of interesting results and promises to produce many more.  

In short, whatever your initial reaction is to the idea of prediction markets in intelligence, this series is designed to give the working intelligence professional inside access to some of the most interesting and intriguing results from research currently being done on prediction markets and intelligence questions.  My goal is to turn these results into “plain English” so that you can have an informed opinion about these unique tools.

Next Week:  What Is A Prediction Market?