To a point.

There was a retrospective study looking at grant scores vs. outcomes (citations of the papers produced). The correlation was pretty weak. Reviewers can identify the obvious dreck, but beyond that was a bit of a toss-up.

If you have too much competition, people start rushing sexy, half-baked results out the door, regardless of whether they seem likely to hold up--or even make sense. This selects against the people doing the careful sort of research that we probably intend to encourage.

Yes, competition is good.

But as you say, the incentives are not aligned correctly. And yes, it's very difficult for reviewers to make the best decisions.

So instead, they reject and request preliminary data. Which requires money. But at least this is better than a sexy shot in the dark, which I think private institutions should be funding. Or things like MD Anderson's moonshot program.

Things aren't perfect, but I think we can both agree that there is no perfect solution to this problem, incremental improvement? Certainly.

Turn the competition knob to 0, everybody slacks. Turn it to 11, everybody cheats.

It's at 10. We're hearing more and more about large cheating scandals. I wonder, what could possibly be wrong with the system?

I don't think slacking in science is a bad thing. Less papers means less noise and if people are not sufficiently motivated by their ideas themselves may be it is best for everyone they are not publishing.

And beyond cheating we have a lot of bad science. * Reproducibility project.