One of his theses suggests that mathematical fireworks became more important than the utility of the models themselves. Paul is the perfect person to make this claim: his work is delightful in that it is amazingly lucid and insightful with sparing use of high-level math. Try and publish some of his same seminal works today and you would have trouble - "too simple," referees would inevitable say. And yes, this resonates with me as I strive for the same simplicity in my theoretical work but sometimes wonder whether I should dress it up will a bunch of useless math. And I can't tell you how many papers I have received to referee from good journals that have pages and pages of math and proofs all to make a simple point that could have been made with one simple algebraic equation. That said, the introduction of serious math into economics has been hugely beneficial - allowing for precision and insight that would have not been there in its absence. But like anything, there can be too much of a good thing, however, it is this pushing of the boundaries, however, that enhances the profession overall. I don't think that Paul would argue for less math, just more attention to what it buys you in terms of insight and a much lower reliance on the rule of thumb that more math means better economics. In fact, I think it is more likely that the opposite is true.
Also in the NY Times, a nice article about how stimulus spending is being dwarfed by state level cuts (in this case in education), something I have talked about before. Having just returned from dropping my child off at his second grade classroom that is overstuffed with little bodies, I can relate.