Hugh Pickens writes writes: "Statisticians make predictions about all kinds of phenomena governed by power laws, everything from earthquakes, forest fires and avalanches to epidemics, the volume of email and even the spread of rumors. Now two statisticians — Aaron Clauset and Ryan Woodward — have put together a comprehensive global database of 13,274 terrorist events from 1968 to 2007 and created an algorithm that predicts the probability of a large-scale terrorist attack with similar or greater consequences happening again. Using the algorithm to test whether the size of the 9/11 terrorist events, which were nearly six times larger than the next largest event, could be an outlier, they estimated the historical probability of observing at least one 9/11-sized event somewhere in the world over the past 40 years to be 11–35%, depending on the particular choice of tail model used to fit the distribution’s upper tail. The results were robust to conditioning on global variations in economic development, domestic versus international events, and the type of weapon used. "That's important," says Technology Review. "It means that 9/11 itself was not at all unlikely given the pattern of terrorist activity leading up to it." Then looking to the future and assuming that the number of terrorists events per year remains roughly what it is now (2000 per year), they calculated that the likelihood of another large-scale terrorist attack anywhere in the world (with a death toll greater than or equivalent to 9/11) is between 20 and 50 percent over the next decade (PDF). A 50-50 chance, the top of that range, means another "catastrophic attack is as likely as not" in the next ten years."