Paul Graham is one of my favorite tech essayists. His essays on startup companies, Lisp, and — especially — American high schools are smart, funny, and insightful. On these topics, even when I think he's wrong, I learn something.
However, he also writes periodically about economics, about which he seems to know a lot less. His most recent foray is an article on unions. Boiled down, the argument is that unions were successful a few decades ago because the manufacturing industries they were organizing were growth industries, and because growth industries don't mind overpaying for stuff, since time to market is more important that expense. I know enough to know I don't know the economics behind why unions are not doing well — but this hypothesis is both naive and lacks predictive power. It attempts to explain why unions are not doing well by explaining why *manufacturing* unions are not doing well. What about all the other types of unions that are not doing well? Why aren't growth industries unionizing so the workers can get some of the wealth? I don't know the answers … but at least I know I don't know :).