This post by Andrew Gelman is the best blog post I have read in some time. It really captures something of what it feels like to be an empirical researcher who cares about getting the numbers right.
Not bad for someone who does not believe in utility functions, whatever that might mean. :)