Tim F. at Balloon Juice compares the upcoming September report on the "surge" from Gen. Petraeus to the scientific method:

It goes without saying that even a country as disasterfied as Iraq must have something going well. That’s why respectable people establish criteria in advance that will determine whether or not an experiment worked. If all of us scientists could do our experiments and then carefully sift and manipulate our data to find the most appealing possible interpretation, huzzah! Every experiment would work. The Kagan experiment set out some fairly clear pass/fail criteria and then failed to meet them, but like a poor first-year grad student Bush can’t accept that his experiments don’t work so he’ll throw in a few numbers from column 2, switch the labels on a couple of figures and hope that nobody asks too many questions.