Given the potential for producing effective treatments in near all areas of medicine brought about by the ongoing revolution in biotechnology, a growing number of people are coming to see that the present established systems of funding, both public and private, are essentially broken. They are far too conservative, funding next to none of the most important early stage research. All of the most important and risky early stage research programs are funded by either administrative sleight of hand or by visionary philanthropy: established funding sources as a rule never offer grants unless the new science has already been discovered and mapped out with a fair degree of certainty. Thus they fund the process of fleshing out and developing a discovery, not the work needed to create and validate that discovery in the first place. If we want real progress and radical new directions in medicine, it is exactly the early stage and risky research that must be funded with greater confidence, however.
The difference between science and engineering is that scientific research starts without understanding and tries out various hypotheses until one seems to work; while an engineer works with a paradigm that she knows to be reliable enough to be a basis for results of her innovations in advance. A high failure rate is inseparable from good science. But the National Science Foundation (NSF) prefers to fund low-risk work, which is really engineering. One irony is that capitalism is pretty good at allocating funds for engineering. Once the science is well developed, the marketplace isn't a bad model for deciding where to invest engineering resources. We probably don't need NSF to fund the "D" half of "R&D". But the reason that we need NSF (and NIH and NIA) as public funding institutions is that the rewards of science are difficult to predict.
I venture to propose that the more unpredictable the result, the more important the experiment. The best prospects for future scientific breakthroughs lie in the direction of things that we already know but don't understand - things that don't make sense. Most of these will turn out to be mistakes in experimental technique or interpretation; but there are some that have such broad corroboration from diverse laboratories that this is unlikely. I could say that "professional scientist" is already a oxymoron. Scientists work best when they are driven by curiosity and a passion to find out, when they are doing what they love. How can that be consistent with centralized decision-making and bureaucratic control of research priorities? If we pay a scientist to do science, we should not make the payment contingent on studying anything in particular. No one in a government bureaucracy has the wisdom to predict next year's breakthroughs, or to single out the scientists most likely to achieve them.
In the late 1970s, when I was a low-level researcher at a government contract research house, we always worked one year ahead of our funding. By the time a proposal was written, we had worked out the science in sufficient detail that we knew the results. If the proposal was funded, we would use the proceeds to support us while we worked on next year's proposal. We may be outraged at 70% overhead rates for administration, and think of this as "slush money" that is ripe for abuse. I agree that bureaucrats receive too big a share of the pie, and scientists too little. But there is some portion of the overhead money that finds its way back through departments to the researchers themselves, and offers them some slack between contracts, their only real freedom to think and to innovate. I asked my collaborator at Prominent Midwestern U whether he had funding for the exploratory, groundbreaking work on population dynamics that he was doing with me, but I already knew the answer. He was doing it with soft funding for a follow-on to previously successful research. He had prudently kept the funders in the dark about this specific project. There's plenty of time to tell them about it if we succeed.