Concentrating Dynamically in Afghanistan

| Wed Nov. 18, 2009 6:34 PM EST

Matt Yglesias on how we should distribute additional troops in Afghanistan:

Afghanistan is a big country. So in addition to the question of how many resources should be sent to Afghanistan, there’s the question of where they should go. Recently, the tendency has been to throw additional resources at the parts of the country where things are worse. In his latest Carnegie Endowment report “Fixing a Failed Strategy in Afghanistan”, Gilles Dorronsoro argues that this would be a big mistake. The resources being contemplated, he argues, aren’t enough to win the war in the South. Sending them there would merely guarantee that we also lose the war in the North and the East, without making much progress in the South.

Instead, he prefers to adopt a more defensive posture in the South—securing main cities where the Taliban is disliked—and focus our attention on winning what he regards as the more winnable struggles in the North and East where the Taliban is making gains but isn’t deeply intertwined with local communities.

Hmmm.  This reminds me of the crime fighting strategy Mark Kleiman outlines in When Brute Force Fails.  It's too complicated to explain the whole thing here, but basically the idea is to concentrate overwhelming force on a small fraction of the population, which then shapes up because they have zero chance of getting away with anything.  As they become better behaved, resources then move to other areas, and eventually the whole population is well behaved.

In other words, it's pretty much the crime equivalent of clear and hold, which is a counterinsurgency staple.  It's also (very roughly) what the surge did in Iraq.  The overall increase in troops from the surge was only about 20%, which seemed plainly inadequate to the task, but most of those troops were concentrated in Baghdad, and it turned out that this was enough to clean up the city.

Now, cleaning up petty crime among drug probationers is not the same thing as stabilizing Afghanistan, but some of the principles are the same.  And as I recall, whether "dynamic concentration" works depends a lot on how widespread violence is to begin with; how good your monitoring and response is; whether your resource level is high enough in the initial target areas; and how much time you have.  Those would all be excellent things to stuff into a game theoretical model to see if an additional 40,000 troops can really make a difference in Afghanistan.  As I've said before, I'm less interested in the argument over the number of troops we send there (which tends to get sterile pretty quickly) than I am in what the detailed strategy is to deploy those troops.  Still waiting on that, though.