Anyway, I've got more I want to say about the referee process, and maybe tell a few, but I'll save it for later. But I wanted to address something Jaffe says in the post linked above:
It’s a safe bet that most of us think that our papers are generally not improved in the process, but in the usual self-congratlulatory way, most of us probably think that we’re in the minority of good referees who actually make useful suggestions, or catch egregious errors. We can’t be right about both, not most of us anyway.
I do think that many of the changes suggested/mandated by a referee for my papers have, in fact, improved my papers. However, those changes have been fairly few in number. The problem, as I see it, is that a number of authors seem to do some research, assemble some text and graphs and submit. They then rely on the referees and the revision process to attain a better paper. I think that's an abuse of the system.
Finally, for more discussion on refereeing and such see the discussions linked to in the comments of Andrew's post.