My new favorite commenter dhanson wrote in the thread on Chait, my emphasis:
And if we're talking about a large stimulus project like HSR or wind farms or new bridge projects, those initial phases employing high-valued people can last years, not weeks or months. By the time you're at the point where you can hire large quantities of lower-skilled labor, you'd hope that the recession will either be over or you're in a balance sheet recession or other restructuring recession that will last a long time, and for which Keynesian stimulus is probably not appropriate in the first place.
This is a terrific observation. All the talk about having "shovel-ready" stimulus projects overlooks the question of whether the human capital of the unemployed would match with the human capital required to get those projects going. Even if they were "ready for shovels" (which Obama has admitted they were not), the question of whether they were ready for the relevant, specific human capital was one that I don't recall ever being asked by the supporters of the stimulus. Again, if labor is not perfectly fungible, then it matters what sorts of human capital is required for stimulus jobs and at what point in the multi-period production process those complementary factors are required.
Human and physical capital are multi-specific and production processes take time: Austrian economics 101.
I can't emphasize enough how much an Austrian view of the multiple (though neither infinite nor perfect) specificity of human and physical capital matters for understanding the operation of real-world economies. Only in a world where both capital and labor are akin to shmoo (or play-dough) can one assume that stimulus projects are shovel-ready or skill-ready. Once you understand and take seriously the heterogeneity and specificity of human and physical capital, then you can grasp why stimulus spending was unlikely to do any good and perhaps do much harm.