Chemistry World has really touched a lot of nerves with this editorial by economics professor Paula Stephan. It starts off with a look back to the beginnings of the NIH and NSF, Vannevar Bush's "Endless Frontier":
. . .a goal of government and, indirectly, universities and medical schools, was to build research capacity by training new researchers. It was also to conduct research. However, it was never Bush’s vision that training be married to research. . .
. . .It did not take long, however, for this to change. Faculty quickly learned to include graduate students and postdocs on grant proposals, and by the late 1960s PhD training, at least in certain fields, had become less about capacity building and more about the need to staff labs.
Staff them we have, and as Prof. Stephen points out, the resemblence to a pyramid scheme is uncomfortable. The whole thing can keep going as long as enough jobs exist, but if that ever tightens up, well. . .have a look around. Why do chemists-in-training (and other scientists) put up with the state of affairs?
Are students blind or ignorant to what awaits them? Several factors allow the system to continue. First, there has, at least until recently, been a ready supply of funds to support graduate students as research assistants. Second, factors other than money play a role in determining who chooses to become a scientist, and one factor in particular is a taste for science, an interest in finding things out. So dangle stipends and the prospect of a research career in front of star students who enjoy solving puzzles and it is not surprising that some keep right on coming, discounting the all-too-muted signals that all is not well on the job front. Overconfidence also plays a role: students in science persistently see themselves as better than the average student in their program – something that is statistically impossible.
I don't think the job signals are particularly muted, myself. What we do have are a lot of people who are interested in scientific research, would like to make careers of it, and find themselves having to go through the system as it is because there's no other one to go through.
Stephan's biggest recommendation is to try to decouple research from training: the best training is to do research, but you can do research without training new people all the time. This would require more permanent staff, as opposed to a steady stream of new students, and that's a proposal that's come up before. But even if we decide that this is what's needed, where are the incentives to do it? You'd have to go back to the source of the money, naturally, and fund people differently. Until something's done at that level, I don't see much change coming, in any direction.