They argue that as militaries mechanized it separated the "incumbent" away from the population, depriving themselves of much needed information. This leads to an "identification problem," the inability to separate insurgents from the population. The "information starvation" often fuels insurgencies as mechanized force attack both the insurgent and population. (Kalyvas makes this argument in the book: The Logic of Violence in Civil War ).
Thus all you need to "win" a counter insurgency is to create enough presence among the population to create a country wide "Zone 1," that is zone under the control of the incumbent with no influence from the insurgent, Kalyvas would argue. That is put enough soldiers in the country to create order and suppress the insurgency long enough to build institutions (like a military or police force) to make sure the insurgent never rears his head again.
This is interesting because maybe its not just mechanization that has led to the "losing" of insurgencies. But rather more normative and political issues like the way force is used and maintained or the fact that institution building is not necessarily a function of force but time and behavior. (The Utility of Force and Political Order in Changing Societies gives good insight in these topics.)
Update: To find the article "Rage against the Machines: Explaining Outcomes in Counterinsurgency Wars" google it, the pdf is first link. I apologize for the broken link.