War & Health

How many Iraqi insurgent attacks will occur next year causing 1 casualty? 10 casualties? 200 casualties? How many FARC attacks will occur next year causing 30 casualties? How many Tamil Tiger attacks will occur next year causing 500 casualties? One very interesting research paper just released argues it knows the answers, and finding them requires a simple number: 2.5.

The paper by Neil Johnson, Mike Spagat, Jorge Restrepo, Oscar Becerra, Juan Bohorquez, Nicolas Suarez, Elvira Restrepo, and Roberto Zarama argues modern insurgent attacks all obey the same power law. In plain english: in all modern insurgent conflict, attacks causing many casualties will occur with the same relative frequency to attacks causing few casualties. Simply by knowing the average number of attacks per year, the authors argue it is possible to predict the number of future attacks causing a certain number casualties in a future year.

Want to calculate your own predictions? It is easy: First, take the average number of attacks per year for the conflict. Second, select a level of attack deadliness (number of casualties) you want to predict. Third, plug them into the formula below.

For example, imagine a conflict with an average of 1000 insurgent attacks per year and we want to calculate how many attacks will cause 10 casualties in a future year:

The model predicts slightly over 3 attacks causing 10 casualties will occur in a one year period.

Are the authors correct? I have no idea, but it is a very interesting approach to an important question. If you want to know more, check out this Physorg article containing a link to the original paper (might be gated).

Update: Check out Drew at Zero Intelligence Agents for a more informed discussion of the paper.