One method I used was to distinguish between the higher levels needed to pacify hostile Sunnis and the lower levels needed in either the more friendly Shia areas or the relatively calm Kurdish areas. This lowered the number of troops needed.
The second was to include all security forces rather than just American forces as many backers of the too-few troops side assumed. So Iraqi army, police, pipeline guards, and even militias (when they battled Sunni insurgents and terrorists) were put in the mix. I even tried to estimate private military contractors in this.
I also assumed that some fraction of our troops in Kuwait should be counted since they performed tasks that they'd otherwise perform while in Iraq if they didn't have Kuwait to operate from. If those troops were in Iraq, we'd have counted every one of them as being part of the troop-to-population ratio needed to win.
I even tried to estimate private military contractors in this. I mostly just looked for numbers on those directly involved in providing security, but I also mentioned that with support contractors doing jobs that in the past would have been done by troops in uniform--and thus counted toward the troop-to-people ratio--that some of those contractors should be counted.
Other than in the immediate post-major combat operations period before we could recruit Iraqis and in the immediate aftermath of the spring 2004 offensives when so many Iraqi security units collapsed, I judged we had the troops required to win.
In the end, we did win even though many critics of the surge said even that provided too few troops to make a difference. So we clearly had enough troops to win.
This analysis of post-war operations validates my feelings on counting contractors:
Assumptions on the number of soldiers required to provide post-conflict security were largely based on a study conducted by the Rand Corporation. It concluded that the security requirement was 20 soldiers for every 1,000 residents, or a ratio of 1 to 50, which concurs with the latest army counterinsurgency (COIN) manual. With nearly 25 million Iraqis, this put the troop level requirement at 500,000. In the period just after major combat operations, there were approximately 160,000 Coalition soldiers in Iraq. Not accounted for in the Rand study however, were the contributions of contractors, which in effect boosted troop strength. Although the exact number is not known, according to the Congressional Budget Office, there was approximately one contractor for every soldier, with 78 percent of them operating on Department of Defense (DoD) contracts as opposed to State Department or U.S. Agency for International Development (USAID) missions. This placed the number of contractors performing soldiers’ jobs at approximately 125,000, giving a total of approximately 285,000 personnel, of the 500,000 calculated by Rand to be required.
This goes farther than I am comfortable by assuming a 1:1 effectiveness of contractors compared to uniformed troops doing the same jobs. I assume it must be some fraction because even uniformed cooks, typists, radio operators, and truck drivers would be expected to pick up a rifle and sit in a bunker to defend a base. Civilian contractors would not have that reserve role. But the contractors do represent numbers that if in uniform would be fully counted as part of the occupation force.
As I always reminded readers, when judging whether we had enough troops (aside from judging how we used them) count all the security forces. This use of private security contractors is just a fact of life today and represents the norm over the long history of warfare.
And now that you ask, yes I do have a collection of essays from this blog on privatized warfare.