The logic of war

12 July 2009

I came upon this video on “The Best Defense,” a blog on Foreign Policy‘s website dealing with defense issues.

Briefly a New Zealand physicist (with a background in conflict studies, faintly bizarrely) was curious whether there were any statistical patterns in the Iraq war based upon available information on insurgent attacks.  He hit upon culling such information from various sources, including the news media, open source websites, etc. of both the number and lethality of each attack, and ran the data.  He discovered a very telling movement:  conflicts such as Iraq, Colombia, Afghanistan and Senegal (where I didn’t even know there was a war) all exhibited virtually the same pattern: an inverse relationship between the number of attacks and their size (or death toll).  All of the insurgencies studied possessed slopes which clustered around alpha=2.5 – that is for every 2 1/2 fewer attacks the death toll (or scope of the attack, if you like) increased by one.

From this the team inferred that an insurgency that’s successful as an insurgency sticks pretty close to the alpha score (or slope) of 2.5.  From this they extrapolated a score for the likelihood of an attack, which I don’t quite follow and which, in the event, doesn’t strike me.  What does is the application for the understand of insurgency generally.  The team theorizes that moving this alpha score away from 2.5 – in either direction – erodes the insurgency.  This means that as the ratio of number of attacks to deaths either increases or decreases (this being the effective scope of the insurgency), that it tends to erode. In the case of an increase it evinces an insurgency being split, fragmented, and devolving essentially into apolitical gangs: they might engage in a lot of violence but it will not be potent or organized.  In the case of a decrease it’s evidence of an insurgency coming out from underground: they will engage in relatively few attacks but these will be large and deadly.

This is why insurgencies will conclude if pushed away from their norm, and I think this is the real methodological innovation of the study.  (I haven’t found an actual working paper or anything yet – if you see one, let me know.)  Saying “attacking an insurgency erodes it” or “bigger insurgent groups mount bigger and deadlier operations” is hardly revolutionary.  But the operating assumption of many, especially in America after Vietnam, is that an insurgency is a robust operation and the onus is on an occupying force to “break” it.

This is not terribly well-founded. It is as difficult to run an insurgency as it is to fight it.  (Ask our own Founding Fathers – or for that matter Ho Chi Minh.)  Redefining a successful insurgency in terms of an effort to tack close to an established mean – and avoid either fragmenting or prematurely ballooning – puts it in proper perspective.  Insurgencies are not long lasting because they are so easy, or because they’ll keep cropping up because you’re an occupying power.  They won’t.  Rather it’s just exceedingly rare that a nation has the moral fiber, sound judgment and discipline necessary to put one down.

The group applied their model to the Iraq War, and found an interesting historical movement.  Through 2006 the insurgency was stable until the elections and the bombing of Samarra, at which point the insurgency fragmented.  (That is attacks became larger in number involving a more diverse set of groups but less deadly individually.)  The surge was an operation designed to reinforce this: their data suggests that it did not succeed, but rather shifted the momentum in the diametrically opposite direction.  (Fewer, more deadly attacks by fewer groups.)  This pushed past the equilibrium until the surge began to wind down, at which point fragmentation pushed the insurgency back above the mean into smaller, fragmented groups.

As the data was new the group had no explanations as yet.  I can’t speak with any authority, but if the data holds I’d say this: pre-surge, the insurgency was already in decline.  Its fragmentation and the downswing in attacks the research found support this.  My guess would be a combination of elections (though assholes like me always overvalue their peacebuilding value), the new agreement on the use of US forces in Iraq, the completion of training and deployment of the Iraqi Army and the consequent erosion of insurgent groups.  (Insurgent groups face the same manpower problems as anybody – after enough time in combat, experienced people die.)  The surge was supposed to reinforce this trend.

Instead, it reversed it.  The surge was successful in putting pressure on insurgent groups – so successful, in fact, that the smaller fragments either collapsed or were forced to merge into larger and more advanced units to survive.  A piece of research I read awhile back makes clear (along the lines of the fragility of insurgent groups) just how rare the technical skills required for IEDs are.  Lose your technician and the group loses its strongest player.  Along those lines, the imminent threat of the surge forced insurgents to shift their tactics, rearm and join forces.  This begs the question why they hadn’t before: perhaps they weren’t actually on the decline, but rather became complacent and morphed into street gangs when the Army wasn’t going after them and they discovered the government could be bought.  In the event the end of the surge has refragmented them as the lack of pressure allows power and ideological struggles within the groups to come to the fore.

Food for thought, anyway.  Feel free to throw in various denominations of change.