The current system for allocating grant money favors senior investigators with established labs over junior faculty with smaller labs and shorter track records, and conservative projects that are almost guaranteed to work over high risk-high payoff projects that take a bigger step away from established knowledge. This isn't news; it's an issue that's been debated over and over in the biomedical research community. Richard Klausner and David Baltimore have laid out some solutions (yes, I know this is old - I'm getting caught up on back issues of Nature):

The National Institutes of Health (NIH) serves the US biomedical community by providing resources for experimentation, but it does so in ways that bias the enterprise towards short-range and unimaginative thinking... First, funding criteria will put more weight on judgements about the individual who is applying, not the details of the proposed project... The technical part of the review will shift from assessing the feasibility of the plan to the capabilities of the investigators.

I like this idea in principle, since its shifts the emphasis away from generating large amounts of preliminary data (and the practice of basically doing 50% of the proposed work before the funding comes in); the down side is that younger investigators with shorter track records will have less of an opportunity to get funding for something genuinely new, since, being in the early stages of a research career, their capabilities aren't so well established. Still, I'm in favor of trusting people who have proven that they can design and successfully execute research projects, instead of demanding that they get every new project up and running before it gets funded. Read the feed: