Grant Giving is a matter of allying yourself with your best bets.

From the point of view of applicants for research grants, they are competing to be one of the chosen. But from the point of view of the foundation or grants officer or investor, your problem is to choose to give your resources to:
1. Recipients who will deliver on what the promise. And what they promise is OK.
2. Long-shots that may pay off, or may not. 
3. People whose work has proved valuable and you are trying to have them be part of your stable of grantees. You want their work allied with your program or foundation. Not a matter of agreeing with it, so much as sharing in its prestige and influence.

Surely you want reliable recipients, and you may want to take some risks that you know are risky. But what you want as well is to support those who are likely to be very strong, and there you are competing with other funders--and it's not just a matter of dollars, it may be a matter of how you encourage them. Similar problems come up in venture capital, where #1 and #2 are nicely addressed by finance, but #3 is a very different sort.