It's a very interesting idea, and I've read some of the coverage of that as well as a few studies of what has happened after that.
I think it's in the early days, so like everything else, we should be empirical. We should ensure that a pilot program is really tested and that we look carefully at the evidence with a clear eye.
On the other hand, we've also seen evidence that peer reviewers—and I'm talking here about grant reviewers, in other words, grant peer reviewers—often don't do any better if you were to measure impact later on, or even citations, as flawed a metric as that is. They don't do any better than a random selection. That, to me, is somewhat troubling. However, it would also argue for perhaps trying that as a system, at least for maybe some percentage. Maybe it's a pilot program.
I also want to recommend the work and the writing of someone named Stuart Buck. He's at the Good Science Project, and he has done a lot of really smart thinking about a lot of these issues, in particular about grant review—