[hugh3]

Are you speaking from experience, or from what you've heard?

I don't know much about NIH funding, but if you're looking for funding from (say) NSF, DOE or DARPA it doesn't have to be that "safe" except in the sense that they really want you to produce something publishable... not necessarily particularly useful. I figure that any worthwhile research project ought to be able to be massaged into a form where you're producing something worth publishing regardless of what happens.

NIH might be different, due to the larger and more expensive scale of these sorts of projects.