Given our nation’s catastrophic intelligence failures of the past few years such as the failure to detect and stop the 9/11 hijackers and our false beliefs about Iraqi weapons of mass destruction, I was excited to learn that there was a new federal program that might do a bit better at predicting threats to the U.S.
Unfortunately I learned about the program in an article about its cancellation amidst loud denunciation by both Democratic and Republican politicians.
- “Defies common sense.”
“I think you ought to end the careers of whoever it was thought that up.”
“A sick idea.”
When that many politicians gather so quickly to condemn something, I start to think there might be something good in it.
And it turns out, maybe there is.
DARPA, the Defense Research Projects Agency (the folks who started experimenting with the Internet in the 60s), set up a betting market to identity vulnerabilities. In other words, they were going to calculate the odds of another terrorist attack with real bets just as we use bets to calculate football odds.
- The program is called the Policy Analysis Market. DARPA said it was part of a research effort “to investigate the broadest possible set of new ways to prevent terrorist attacks.”
Traders would have bought and sold futures contracts — just like energy traders do now in betting on the future price of oil. But the contracts in this case would have been based on what might happen in the Middle East in terms of economics, civil and military affairs or specific events, such as terrorist attacks.
Holders of a futures contract that came true would have collected the proceeds of traders who put money into the market but predicted wrong.
If you’re interested in how the idea works and why it turns out to work exceedingly well, you can check out this web page about idea futures. Oh, and it turns out that much of the underlying methodology is based on work by Vernon Smith, who won the Nobel Prize in economics this past October.
It turns out that when people place bets, their collective knowledge is brought together in ways that surveys and other forms of analysis simply can’t aggregate. And since real money is behind it, people internalize the cost of hunches — they discount sketchy information appropriately and place a premium on real likelihoods. And bureaucracy gets stripped away. And there’s no meely-mouthing and covering for potential mistakes. Betting markets have a tremendously accurate predictive value.
But it offends the sensibilities of Hillary Clinton, John Warner, Byron Dorgan, and Ron Wyden — so we’ll just stick to our CIA, thank you very much. (Thanks to Tyler Cowen of the Volokh Conspiracy for the pointer.)
Update: Tyler has more on this idea. For instance,
- Think of it as a substitute for government intelligence. If you want to know who will win the Super Bowl, check the Las Vegas odds (you wouldn’t ask the CIA, even if it were assigned to that task). Why should the best device for aggregating information be any different for terrorism or for that matter anything else?
Here are some criticisms from some prominent Senators.
Which criticisms make sense?
1. Would terrorists bet on attacks and then carry them out?
Maybe, but they can already do this with current stock markets, if they are at all clever.
2. Are we allowing some people to profit from the death of others?
Yes, but how does this differ from life insurance?
3. Would price/betting odds accurately reflect information?
Probably, provided you had a fair number of bettors. Return to the Super Bowl analogy. And the stock market figured out within minutes which contracting firm was responsible for the Challenger crash.
Update again: Brad DeLong is blogging about this too.