Old news now of course, but this was on xkcd a few weeks ago and Stoat pointed me to it in a comment:
I can't play all innocent because I've made similar jokes myself (eg here). However, I'm at least partly persuaded by Andrew Gelman (the comments thread is also interesting) that it's not an entirely fair criticism. Just because it's a valid frequentist calculation, doesn't make it a sensible one. On the other hand, it does highlight that even frequentists have to make sensible subjective decisions about what sort of analysis they are going to do (eg, which hypotheses to test). Which does knock a big hole in the claim you sometimes hear that frequentist methods are objective.