Today's paper includes an article about grant money in cancer research and other research areas funded by the National Institutes of Health. It focuses on the process of awarding grant money, which is done based on the result of peer review of applications, in a process similar to that found in academic publication.
I'd bet that while every scientist you might talk to would agree that the system has flaws, they would probably also agree that it's difficult to imagine what a better system would look like. Highly innovative research is likely to also be risky, and peer review is bound to have a hard time with such proposals. On top of that, the NIH is charged with coming up with the "best" way of spending federal tax revenues, not exactly an uncontroversial activity. Politically speaking, private foundations can afford to take much bigger risks, even if from a functional point of view one might argue that the risk tolerance of the nation as a whole ought to be at least as great as across any subgroup.