Where the World's Views of America Come into Focus

March 5, 2005
By Stephane Foucart
Certain mathematicians are convinced of it: everything, absolutely everything yields to the laws of probability. And that includes terrorism, say Aaron Clauset and Maxwell Young, researchers at the University of New Mexico at Albuquerque. In a study made public in February on the ArXiv.com archive, the two mathematicians propose a model presumed able to describe the probability of terrorist attacks.
(http://arxiv.org/PS_cache/physics/pdf/0502/0502014.pdf)
At the end of their calculations, they regrettably conclude that attacks on the scale of September 11, 2001 have a strong probability of occurring again in the relatively near future, and that the number of deaths due to terrorism can only increase over the next few years.
In mathematical terms, their results could be summarized with this sentence: "the frequency of a terrorist act is inversely proportional to its" index of severity "raised to the power alpha."
The parameter alpha is, in this case, slightly lower than 2 ... In general: the more serious the event is, the less probable it is. There is nothing astonishing here.
But intuition, explain the authors, would hold that the most significant and spectacular terrorist acts are "outside the model," occurring in a completely random way, and statistically unspecifiable. This intuition is false. In fact, events of very large scale closely follow the general law. According to the model presented, a new hyperterrorist attack should take place  not necessarily in the United States  between now and 2012.
These quantified forecasts are, however, of doubtful validity. "The overall approach is reasonable", says a French statistician, in charge of research at France's National Scientific Research Center (CNRS). "But the quantitative forecast cannot be pushed too hard."