A little more on Roberts and Krugman

I'm pretty swamped right now, but I did want to share Russ's response to me, and Krugman's response to Russ. The Krugman post is really excellent. Russ's is quite good too, although (obviously) I disagree with a lot of it.

Regular readers know that I was (before starting school) a primarily empirical labor economist/policy evaluator by profession, and that one of my biggest interests in macroeconomic stabilization discussions is discussions of empirical estimates of the multiplier. That's one thing that Russ brings up here and he says the evidence is "discouraging". I don't think it's quite as discouraging, for some of the reasons I mentioned in the last post. What are your thoughts on that claim of Russ's?