Can Terrorist Strikes Be Predicted?

(Nov. 5) -- An elite group of scientists who advise the government on national security issues has issued a harsh critique of attempts to predict 9/11-type attacks. There is "no credible approach that has been documented to date" for predicting terrorist or weapons of mass destruction attacks," the group concluded in a study obtained by the Federation of American Scientists.

That statement may seem glaringly obvious to some, but it counters the Pentagon's predeliction for funding social scientists to come up with models that predict the future. The critical study was performed by JASON, a little-known group of academics who counsel the Pentagon and other agencies on science issues.

The report does not mention by name any of a number of ongoing Defense Department projects to build predictive models of various sorts. The Pentagon's Defense Advanced Research Projects Agency, for example, funds a program called the Integrated Crisis Warning System that "seeks to develop a comprehensive, integrated, automated, generalizable, and validated system to monitor, assess, and forecast national, sub-national, and international crises..." Another Pentagon office funds a program called "Human, Social, Cultural, & Behavioral Modeling," which also looks at ways to forecast behavior.


The JASON report focuses on models that purport to forecast specific, or "rare" events, such as the 9/11attacks. But its central conclusion can be generalized to other types of forecasting. "It is surprisingly difficult to evaluate models, more specifically quantitative models, reliably," the JASON note in the report. "The problem is that at any given time, it is only possible to evaluate a model against events that have happened."

In other words, predicting events after they happen is easy; it's predicting them ahead of time that's hard.

The JASON report underscores the difference between predicting events -- which it considers impossible -- and assessing risk. The group wrote that "another 9/11-scale event in the world is unlikely but not improbable in the next ten years." But what are the chances of another such event happening, according to the group?

They took a stab at that one at least: 7 percent.