Two guest writers on Olivia Judson's blog offer an interesting idea for spending stimulus money on research:

Instead of simply funding more grants, we suggest using some of the windfall to provide an opportunity for fresh college graduates to pursue two years of research in the nation’s service while the job market is bottoming out. Call it “Research for America.” Our proposal would put young Americans to work and support science — without setting off a later bust cycle in research support, as previous funding booms have done.

They make a good point about a boom and bust cycle in research funding - if we pump a lot of stimulus money into new grants, we're funding people who are going to need more money down the road, if they're going to stay in the business. A big, temporary research stimulus right now could not be sustained. (As a side note, we didn't have to have a bust after the NIH budget doubling boom from 1998-2003 - just letting the NIH budget keep up with inflation would have solved a lot of problems.)

The authors argue that new college graduates (most of whom presumably have no long-term aspirations for a science career) could dig right into these 'shovel-ready' (sick of that term yet?) projects:

Is there a better approach to funding research than this Malthusian boom-and-bust cycle? We think so. Modern research requires intelligence and drive, but people without long-term aspirations to become scientists could do much of the hands-on work. New college graduates could enter the laboratory workforce for just a few years, which would allow more research to be done under existing grant support without creating future obligations.

The problem is, I don't think this will work. Research is not only intellectually demanding; it takes physical skill - wet lab work is like gourment cooking, but with much, much smaller error margins. You don't bring an untrained intern in to do the work of a chef at a 5-star restaurant; you have them chop carrots and shell shrimp. But in many labs, you don't really have the lab equivalent of carrot chopping - much of the lab work takes training. And in general, fresh college graduates suck up a lot of time when they're being trained to work in the lab. That's a good thing when it comes to training future scientists; someone did it for me, and I'm happy to do it for others. It's a great system of training, but it's a horrible way of being productive.

Most of the research that gets published is done by individual graduate students and postdocs - people who are strongly motivated and well-trained, trained enough to, with some guidance, do the physical lab work and intellectually direct the project. Most research just can't be farmed out to interns - I achieve much more when I do my own work (or work with another experienced scientist) than I do when I'm directing someone inexperienced. Bringing in large numbers of untrained, short-term, fresh college graduates to do lab work will turn a professional research lab into summer camp.

If the goal is teaching and training, that's great, let's do it. But if the goal is productive research, this is the wrong way to go.