The crux of matter goes all the way back to the 2003 Rumsfeld "metrics memo" which castigated his staff for failing to come up with hard numbers for the insurgents, and if their ranks were shrinking or growing in response to U.S. operations. Without that crucial info it is impossible to know long-term trends for the conflict.
But it now appears there are as many 20,000 fighters hostile to the U.S presence in Iraq. The fighters flow in and out of the conflict, have various motivations and capabilities, and do not all answer to single command, but standard counterinsurgency theory suggests that the U.S. needs 200,000 troops in the field to deny the countryside to a guerilla force of that size and to swarm any concentration of fighters without losing control somewhere else. The U.S. presence in Iraq is several combat divisions short of that number now and is merely treading water until both the U.S. and Iraqi elections are over.
Iraqi fighters are also cheap and potentially plentiful while the magnificent units America deploys are expensive, incredibly lethal, and very finite.
Hence the choice Novak posits for the president early next year: more U.S. troops or get out. As Bush has already taken his "Mission Accomplished" victory lap for Iraq and Afghanistan, sending more troops to war would be cognitive dissonance of the first order. Declaring double-secret victory and bringing the troops home seems at least as plausible for a Bush II Administration.
I'd also add that a crucial, often overlooked factor this calculus is the continuing inability of the Iraqis to field any kind of fighting force that might take manpower demands off of the U.S.
You see, there is purpose behind those car bombs at Iraqi police stations and recruiting centers besides just murderous rage or "freedom hatred."
So long as the Iraqis are light in the manpower department, a Bush or Kerry White House must sooner or later confront the awful double-down or fold question.