It seems to me that the prize model really could work in areas where there's a lot of value snagged up in existing knowledge we already have and the prize-giver

It seems to me that the prize model really could work in areas where there's a lot of value snagged up in existing knowledge we already have and the prize-givers want to shake out that value in the form of good applications. I think this works for DARPA because DARPA these days seems to want deliverables that are in this vein. The prize model can also work well in a situation where the prize-giver wants the results to enter the public domain or be cheaply available to the public. I don't think this could work to fund blue sky research, though. It's asking for too much risk to be assumed by a researcher who probably doesn't have the resources to justify these kinds of attempts.