Payoffs For Prediction: Could Markets Help Identify Terrorism Risk?

By NPR Staff

Published November 28, 2015 at 8:00 AM EST

A flower is placed inside a bullet hole in the window of Le Carillon restaurant in tribute to the victims of the terror attacks on November 15 in Paris, France.

Trying to predict terrorism is a chancy business — and one that can have disastrous consequences.

But more than a decade ago, the Defense Department had an idea. What about a policy analysis market — not a think tank, not a board, not a center, but a market?

In a prediction market, people put real money on their ideas for what will happen in the future.

The idea was dropped because of public outrage. But Stephen Carter, a Yale law professor, novelist and columnist for Bloomberg View, thinks it might be wise to revive it.

He explains why to NPR's Scott Simon.

Interview Highlights

On how the markets would work, and why people objected

Prediction markets — markets like political prediction markets, where people risk money on the presidential election, things like that — have been remarkably successful in many different areas. So a little more than a decade ago, the Defense Department thought ... "Why don't we try to do a prediction market that would include all sorts of measures of international instability?" — where there'll be a coup, what government might be at risk, where the economy might fall. And it included questions about terrorism as something we might try to predict. ...

There was a public outcry. People complained [that] this would mean people are earning money from terror attacks. That is, if you bet right in the prediction market, obviously, you'd get a payoff. And so it was dropped.

On how putting money on the line improves people's predictions

One of the theories of prediction markets is that all of us have lots and lots of opinions. We rarely have anything at stake. And so not having anything at stake, we tend to make bad predictions or to stick to our ideological preferences and things like that.

There's a lot of evidence that when people actually have something at stake, when there's a payoff that comes as a result of being right, people's judgments over time tend to get better — especially if they're judging short-term events, but often when they're judging long-term events as well.

The policy analysis market, as it was planned back in the early 2000s, wasn't going to pay out large sums of money. It was going to give people small sums of money to see whether the market forces, the information held by different people being aggregated in the market, could serve as a kind of predictive tool to lay alongside all the other predictive tools that people use.

On who would participate in such a market

The idea would be, in the initial effort ... experts of various kinds. They didn't have to be necessarily experts in international relations but certainly experts in a variety of related fields, because one of the things that markets do very, very well is that they rapidly integrate information that's known only to a few people. ...

Think about the way, for example, the stock market works. If you and I own shares of a stock, and we notice the stock is dropping suddenly, we might not know why it's dropping. But that's certainly going to affect our judgment whether we should be holding that stock or not. And so it's the same idea that if there are different experts who have different views, especially if you can combine the views in different ways. For example, would a U.S. intervention in this country make a terror attack more likely in that country?

On whether terrorists could follow the markets and change their behavior

That would be an interesting question. Right now, we make a decision. We say we're going to harden particular targets to make them difficult for terrorists to attack. So yes, terrorists do act rationally and choose another target - they choose a less hardened target. The question is whether it could respond very rapidly to changing information.

Large terror attacks take a lot of work and a lot of planning. And if the market predicted a high probability of an attack in a particular place, it's not as though it's a trivial matter for a terror group to suddenly switch all of its resources to another place.

I'm not suggesting this would work. It would be a tool that would be used alongside other tools. And I'm just suggesting an experiment to see if it can improve the accuracy of intelligence forecast.

On the reaction from some that the idea is hideous

It is hideous in a sense. It is. There's no question. I share the revulsion of people who say, "But people shouldn't profit from this." On the other hand, if allowing people to profit from this, from making these predictions, improved our ability to predict, then it might be something worth trying because of the lives it would save on the other side.