I’ve already shared Sam Savage’s take on the failure to keep the underpants bomber from getting on that plane to Detroit: that it’s just really really really hard to identify a vanishingly small segment of the population and keep them off airplanes without mistakenly preventing scads of harmless people from boarding. But there are other angles, one of which President Obama addressed Tuesday (and will presumably address again in more depth this afternoon)(update: he did):
[E]lements of our intelligence community knew that Umar Farouk Abdulmutallab had traveled to Yemen and joined up with extremists there. It now turns out that our intelligence community knew of other red flags — that al Qaeda in the Arabian Peninsula sought to strike not only American targets in Yemen, but the United States itself. And we had information that this group was working with an individual who was known — who we now know was in fact the individual involved in the Christmas attack.
The bottom line is this: The U.S. government had sufficient information to have uncovered this plot and potentially disrupt the Christmas Day attack. But our intelligence community failed to connect those dots, which would have placed the suspect on the “no fly” list.
My initial thought was that if all the various elements of the intelligence community had simply Tweeted their findings, the hive mind of the Internet (or, more specifically, some 14-year-old in his bedroom in Bakersfield) would have blown the whistle on Abdulmutallab weeks ago. In the Cold War there was one big, secretive, hierarchically organized foe. So it made sense to have a secretive, hierarchically organized intelligence agency trying to figure out what that foe was up to. When the foe is dispersed and decentralized, the advantages of that secretive hierarchy diminish rapidly. And these days the informational and technological advantages of large organizations are diminishing as well. Any halfway affluent individual can assemble a better set of communication devices and networks on her own than she’d ever get from the IT department of a large corporation (or large government agency).
Now I do get why we don’t want to put every last intelligence finding on Twitter. The bad guys can read Twitter too. But within the community of people with security clearances there obviously needs to be much more sharing and weighing of dispersed information. And what’s the best mechanism known for sharing and weighing dispersed information? A market. I know, I know: I wrote a whole book about how financial markets aren’t always rational, and on the big-picture, macro level, markets get things wrong all the time. But they are awfully good at finding and processing and weighing smaller pieces of information—and at motivating participants to dig for ever more.
Which is why, seven years ago, some creative folks at the Pentagon—inspired by the work of George Mason University economist Robin Hanson—proposed creating a futures market in which informed participants would bet on political developments in the Middle East. All sorts of political hell ensued, and the government shelved the project. This was a market for predicting major regional events, not whether a particular poor little rich kid from Nigeria posed a security threat. But still, such a market does seem like a mechanism that could encourage not just information-sharing but active seeking-out of the sort of information that could have kept Abdulmutallab off that plane.
And yet, in all the public discussion of what went wrong in the Abdulmutallab case, I have seen not a single mention of the Policy Analysis Market, as the Pentagon called its project, or the terrorism futures market, as everybody else called it. Hanson hasn’t even brought it up lately on his blog. So I figured it was time to rectify that. I’m not sure that a prediction market is the right answer here. But we ought to at least be talking about it.
What does Hanson think about all this? I e-mailed him, and here’s the meat of his response:
Having markets on each and every suspected person would be possible, but not likely or recommended as a first application. First you’d want to start out forecasting big aggregates and the effects of major policy choices. But of course even before that you’d have to have agencies at all willing to use prediction markets to aggregate info on anything. And it isn’t clear we really care enough about actually avoiding terrorism to be willing to bother to change how our agencies do things. It seems to me that we more want to signal concern by having agencies that “do something” conventional.