Kevin Clarke and David Primo have a really nice piece in the NY Times this weekend. They boil down the essence of their recent book in arguing that:
1. Political scientists have a fetish for going about science as physicists do.
2. Physicists don't really do science this way.
The alternative they propose is to think of models -- both empirical and theoretical -- as maps. The consequence of this is that they are judged by how practical they are, given their stated purpose. It's a compelling argument. I've used various drafts of Clarke and Primo's book in teaching fourth year specialist students about the nature of political science. I've also recommended it widely to grad students in our program at Toronto. Most importantly, it's done a lot to shape my thinking around how to go about empirical political science. It's worth picking up.