I do not recall spotting this 2010 RAND paper appearing on SWC; the full title is 'Victory Has a Thousand Fathers: Sources of Success in Counterinsurgency' by Christopher Paul, Colin P. Clarke, Beth Grill

RAND's summary:
Insurgency has been the most prevalent form of armed conflict since at least 1949, as well as the subject of countless historical and contemporary studies. Contemporary discourse on the subject is voluminous and often contentious, but to date there has been a dearth of systematic evidence supporting the counterinsurgency (COIN) approaches, practices, and tenets that make for successful operations. Relying on a collection of the 30 most recent resolved insurgencies, along with a bank of factors that helped or hindered the COIN force in each case and in each phase of each case, several commonalities emerge. For instance, the data show that good COIN practices tend to “run in packs” and that the balance of selected good and bad practices perfectly predicts the outcome of a conflict. The importance of popular support is confirmed, but the ability to interdict tangible support (such as new personnel, materiel, and financing) is the single best predictor of COIN force success. Twenty distinct approaches to COIN are rigorously tested against the historical record, providing valuable lessons for U.S. engagement in and support for COIN operations. A companion volume, Victory Has a Thousand Fathers: Detailed Counterinsurgency Case Studies, presents in-depth profiles of each of the insurgencies.

I noticed a reference to the RAND paper in an Australian think tank, the Lowry Institute's emailing, with comments - opening with:
I extracted conflict duration from the dataset and analysed it. (My emphasis) The proposition that time favours the insurgent is totally wrong. The longer an insurgency lasts, the more likely the counterinsurgent is to win. If the duration of the conflict is broken into four year periods we find that there were 11 insurgencies that lasted four years or less and the insurgent won 10 of these. There were four insurgencies that lasted between 5-8 years and the insurgent won all four. There were eight insurgencies that lasted between 9-12 and the insurgent won five of these. Of the seven insurgencies that lasted longer than 12 years the counterinsurgent won four. The probability of success shifts in the counterinsurgent favour after the tenth year. Of the 30 most recently resolved insurgencies the counterinsurgent only won once in less than ten years. Once an insurgency enters the 11th year the counterinsurgent’s likelihood of winning increases to 55%. This rate increases as more time passes.


Followed by:http://www.lowyinterpreter.org/post/...nsurgency.aspx

Which I note refers to a paper by David Killcullen.