doing science properly

Last term I took a grant-writing/prelim-giving/paper-reading class, and one of the sessions was dedicated to writing a good grant. The guest speaker was a professor in the department who has never had to resubmit an NIH grant in his 30+ years of tenure as professor. He brought in his experiences as a grad student at MIT and how there were two professors there who did science very differently, but both ways were ideal, depending on your personal style.

The first way is to go off on crazy tangents from one root project. It's like saying, "mutant A has an obvious eye defect as heterozygotes, but homozygous mutants for A die." You could have someone in your lab working on what this eye defect is and how it relates to the mutated A, and another person working on the timing of death. Why do the homozygotes die? What stage? What went wrong?

The second way is to keep working on one central question, discovering things that relate to it along the way, but never fully answering the question. Going back to mutant A, you could ask "why do homozygotes die?" Then you could find out that they have defects in gene A, which is part of pathway B. But that doesn't mean pathway B is the ONLY one in the system that is making mutant A unhealthy. In other words, is there anything working with pathway B? Does it act alone? Is there a mutant A that doesn't have problems in pathway B, but has problems with receptor C instead?

The unfortunate part about getting grants is that the government expects neatly-laid out experiments, with not much wiggle room for failure (one of the main laments of one of the professors I talked to). But since NIH grants are stretched over a five year period, how is anyone supposed to have the insight to know that some project that looks promising now won't be a dead end? How do you account for pop-ups that are related to your project, but are completely unanticipated?

Science is already a competition for funds, but the restrictions on what you can write in your proposal and what you can't possibly foresee makes it more and more difficult. We can't possibly outsource research, but will the sheer difficulty of getting money drive down the population of scientists?