The purpose of the site is to leverage knowledge that many individuals each have about one area into knowledge of a whole area or sequence of events.  DARPA says, “Prices and spreads signal probabilities and confidence.  Since markets provide incentives for good judgment and self-selection, the market will effectively aggregate information among knowledgeable participants.’’ It continued, “This approach has proven successful in predictions concerning elections, monetary policy decisions and movie box office receipts—DARPA is investigating its success in defense- related areas.”

The mechanism is to set up mutually opposing outcomes - each of which represents a futures contract that pays $1.  For example, the question, “Will terrorists attack Israel with bio-weapons in the next year?” has only two possible outcomes.  You can bet (purchase a contract on one side or the other. 

The interesting thing is that it doesn’t stop there.  There are derivatives and hedge contracts as well, that serve to combine information.  The example given on the PAM website is historical - two contracts, one on the likelihood of the collapse of the Jordanian monarchy, and the other on the likelihood of Saddam’s regime lasting a month once the US began hostilities.  In a matrix, there are four possible combinations of outcomes, and each of these represents a derivative.  The price on each is in effect an aggregate prediction.  Further there are hedge bets - for example, you could bet that if Saddam lasts longer than one month, then the Jordanian monarchy will collapse.  If the first part of the prediction doesn’t come true, you don’t lose money, but you would gain signifcantly if the whole thing did.

In all, this is very similar in concept to the idea of the Delphi boards in Brunner’s novel.  The boards were organized more on the model of racetrack betting, but they did involve speculation of future events.  And the underlying assumption was identical: if you ask enough people, even if they are unaware of the complete body of information regarding the question, the average of their answers will approximate the actual result.

If anything, the DARPA concept is potentially more powerful than the Delphi boards, because it seems to allow combinations of predictions in infinite variety.  In this manner, new questions conveniently packaged with answers can be discovered, rather than merely answers to questions explicitly asked.  Further, the price mechanism could allow more responsive and informative predictions than artificially moving the odds in response to betting on the part of participants.

Needlessly to say, I think this is very cool.  I am very upset that they got their first.  However, currently the site is only aimed at the middle east.  This could be expanded.  I am now on a crusade to convince those with the knowledge I need that they must help me.  Ross, you’re first on my list.

As a side note, some Democrats were upset in completely predictable ways.  Senator Byron Dorgan of North Dakota called it, “the most Byzantine thing I have ever seen proposed by a federal agency.’’ But more to the point, he was worried about people’s feelings:

How would you feel if you were the king of Jordan and learned that the U.S. Department of Defense was creating a futures market in whether you’re going to be overthrown?

Well, seeing as I am always going on about the power of markets, I kind of have to be behind this idea.  I am going to try to sign on when the site begins registration on Aug 1.  Should be interesting.