Study co-written by Princeton and West Point scholars:

Empirical evidence suggests that Great Powers and weak states alike are increasingly unable to wage successful counterinsurgency campaigns. We argue that this decline can be explained by rising levels of mechanization within state militaries. Unlike their earlier counterparts, modern militaries possess force structures that inhibit the creation of information-gathering networks among local populations. Mechanized militaries therefore struggle to wield their power discriminately, pushing fence-sitters into the insurgency. We test this claim using a new dataset of 238 insurgencies (1800-2000) and a microlevel comparison of two U.S. Army Divisions in Iraq (2003-04). We find that mechanization is associated with a decreasing probability of incumbent victory; that regime- and power-based explanations only account for nineteenth century outcomes; and that oft-cited factors such as terrain or ethnolinguistic divisions are largely unconnected to outcomes in counterinsurgency warfare.
http://www.princeton.edu/~jlyall/Rage2.3.pdf