Quite a few commentators have noted that the decrease of violence in Anbar province cannot be because of the surge since it clearly predated the surge. That's just basic logical reasoning. And the numbers seem to bear it out. But this blogger suggests a more precise theory: that it was the US midterm election that turned the tide.

Specifically, did the change in thinking among the Sunni tribal chiefs happen because (in response to the 2006 elections) they could see that the US military presence would be winding down?

Look at the graph on page seven of Petraeus's slides. November is where the shift happens. (One might also hypothesize a degree of intentional ramp-up of violence to affect the November election.)

I think this may be too neat and economical an explanation and may play on our own ability to project our worldview and circumstances on to an Iraqi one we don't understand very well. But I suspect that some version of this may actually play some role. Especially when you consider that the switch likely has more to do with preparing for a future confrontation with the Shia government than a rethinking of the tactical alliance with 'al Qaeda.'