Quote Originally Posted by Rex Brynen View Post
What's the dataset coding definition for insurgency, Cavguy? Does it depend on battle deaths (and if so, is the threshold set in absolute numbers, or relative to population size)? How does it differentiate from terrorism?

I ask this because, arguably, the great success of democracies might be that groups never make the transition from protest movement > small terrorist group > full-blown insurgency, and that the "success" of democracies lies rather earlier than their ability to engage in full-scale COIN.

A critic might argue that it is rather like trying to measure the effectiveness of a bug-zapper atop Mount Everest (the body count isn't really getting at the issue of why there are so few mosquitos around...)
Agreed. The RAND Terrorist post-1945 dataset has over 600 listed entries. The Insurgency dataset has 89. The coding is based off of a number of factors, including size, goals, casualties, and other factors. Adding terrorist data actually reinforces the hypothesis, as they are often defeated and rarely successful.

Intuitively, I think you have hit on the reason - the structure of democracy makes it hard for terror groups to metastasize into full blown rebellions. There would have to be a huge reservoir of non-reconcilable discontent unresolvable through political means - which in turn suggests the contest is hardcore ideological (religious-like or separatist) in nature, rather than grievance based.