Just tweeted by Neuroskeptic; if this has implications for the recent issues with fMRI studies, I hope someone smarter than me will let me know.
Thought this might be interesting for people that know the No Free Lunch Theorems from classification/optimization too