Ok, so I'm a bit behind on my reading, but Prof Johnson's group at OCCF came up with an intriguing study last year conjecturing that terrorist attacks follow Salpeter's Law.
see semi-technical article in PhysOrg or the pop article in the Grauniad
Richardson had found a similar law for the size of conventional warfare, but Johnson et al have now shown similar statistics apply to terrorist acts, with the frequency of attacks of a given size scaling roughly as t-2.3 -- -2.5
In a further refinement, instantly recognizable to all stellar theorists, the slope is flatter, but normalization lower in developed nations - big attacks are overall less frequent, but relatively more frequent in nations that present large targets. There is also some conjecture on the time evolution of the distribution of attacks, and how that could be used to understand the nature of evolving conflicts, with particular examples of Colombia and Iraq. Hey, maybe there is a quantitative measure to distinguish between mere "terrorism" and "civil war".
The analogy to star formation on poor stellar associations vs rich clusters is fascinating, probably meaningless, and yet may be a signature of underlying similar physics; namely self-organized hierarchical structuring of discrete clumps competing for fixed resources, whether that be the inter-stellar medium of a molecular cloud undergoing internal fragmentation, or the resources available for terrorists.
I guess the reassuring thing is that the integral converges in the high N limit...
So, we shouldn't be worried about being taken out by a few big attacks, it's the far more numerous small incidents that do the real harm!
Formally, yes. But since the measure is fatalities and the field is quantised, there is a natural lower limit cutoff that eliminatesd infrared divergences and keeps the whole integral convergent.