If we're gonna throw mountains of taxpayer dough at the recession, in a desperate attempt to do something, it would at least be nice to learn from it. We're wasting a ton of money, and smart people know it. But some stimulus ideas are better than others. We just don't know with much certainty which ones those are. That's why I like Len Burman's unrealistic idea of running experiments:
The ideal “experiment” would assign different “treatments” to different states. States that begin with A-E: tax credits, F-K: infrastructure, and so on. We could even have a control group that would get nothing. Sorry, Wyoming.
OK, that's not going to happen.
Burman goes on to suggest giving grants to the states, letting them do whatever they want with it, but then very carefully gathering data about what happens when they do whatever they decide to do. The reason this is a relatively great idea is that it assumes that a lot of fiscal stimulus will fail, and that we don't know what we're doing. But policy that is remarkable for the honesty of its assumptions doesn't tend to be extremely popular.