Lancet

The basic point with this new Lancet survey is the same as the basic point about the last, which is that you can't beat something with nothing. The question for skeptics is "If that's the wrong number, what does the American government think is the right number?" The American government won't release any estimates. The only conceivable reason for that is that they believe that the true fugure is either much the same as or greater than the Lancet figure.
Imagine a court case in which there's a question about how many cattle you could run on a particular property. One side says "based on our statistical estimates, there are 4 to the acre or 4,000 in all." The side in posession says "Don't know. Haven't counted. Can't guess. Don't ask." The judge would be entitled to assume that the answer to the question wouldn't support the posessor's case.
Have to say, though, that in one way it's an impressive achievement. The Americans appear to have killed 183,000 Iraqis with the loss of only 2,700 Americans, or over sixty to one.
A little over one My Lai a week.