You bet your life - Futures markets won't solve a real intelligence problem.

04 August 2003
Financial Times
10

The Pentagon's rapidly cancelled scheme to sponsor a futures market in 
terrorism was an unusually clear example of "thinking the unthinkable". The 
idea was politically idiotic and incompetently marketed and, unsurprisingly, 
died quickly. But such markets will not go away. It is interesting to ask 
whether they will ever be useful.

The FutureMap programme was originally designed to connect analysts in the 
US government. It was supposed to solve two problems: first, that analysts 
find it hard to share information, especially between organisations; and 
second, that junior or unprepossessing informants are ignored. FutureMap 
would have invited them to bet on an event, such as a change of government 
in Saudi Arabia before 2005, and aggregated their opinions, weighted 
according to how much they were willing to stake. It was hoped that it would 
save intelligence agencies from cheap talk, political correctness and 
"group-think".

But FutureMap was opened beyond in-house communities of a few analysts; and 
because its proponents at the Defence Advanced Research Projects Agency 
(Darpa) have a political tin ear, they suggested that FutureMap be used to 
predict terrorist attacks as well as mundane political events. FutureMap 
became politically explosive and morally offensive. It was also optimistic 
economics. Markets aggregate some kinds of information very well but would 
not have helped Darpa.

Darpa had its eye on the Iowa exchange, which has done a better job of 
predicting elections than pundits or polls. This is no fluke. Such markets 
do well, if liquid - and if real money is at stake. The reason is clear: 
punters will take opportunities to make money by revealing private information.

The most obvious example is the stock market, where there is ample 
liquidity. But the view that the stock market is efficient has taken a knock 
over the past few years. Had Darpa followed the Nasdaq in the late 1990s, it 
would have acquired strange ideas. Tracking shares in Enron would have given 
it even stranger ones.

Markets that paid out on specific terrorist attacks would not be liquid, so 
would give little information, while markets for vague events would elicit 
vague information. If market participants were anonymous, terrorists could 
rig illiquid markets to mislead intelligence services. But if punters were 
traceable, noise traders would dominate. There is little public information 
about terrorist strikes to be aggregated, unlike markets for election results.

Credit is due to Darpa for recognising the need to cut through bureaucracy, 
combine sources and force analysts to say what they really believe and how 
confidently they believe it. The problem remains unsolved.

Market fundamentalists need not mourn, however. There are plenty of private 
markets available to bet on intelligence events, including, ironically, the 
chance that John Poindexter of Darpa may lose his job. Laissez faire, 
Admiral Poindexter.

US Edition 2.