[dekhn]

I think the right way to think about science investment is the multi-armed bandit. There's a lot of uncertainty, but you also know that if you fund a portfolio with a good distribution, you'll have a chance of maximizing the total good science results that come out of it.

I've always been very skeptical of claims that heavy investment in the space sciences advacnced other technologies, more than if we had just invested that money directly in the other technology. I also see a lot of money wasted in space sciences running experiments on the ISS that aren't really that useful (like crystallography- sure, you can grow nice crystals in space and then bring them back down to earth, but for the money you spent on that, you could have funded 10 PIs, and also crystal structures aren't that useful for advancing science).

Interestingly, if you look at modern science funding, it does basically treat it as a multi-armed bandit and there is a portfolio of project funding that includes both large particle physics experiments and individual investigators.