Effective Research Funding: [...] it’s odd that the rules for NSF funding, which is the premier source of funding for basic science in the US, generally requires university participation on proposals. This restriction naturally makes it easier for researchers at universities to acquire grant money than researchers not at universities. I don’t understand why this restriction is desirable from the viewpoint of a government wanting to effectively subsidize research./cite> (Via Machine Learning (Theory).)
Hal comments that one of the NSF's criteria for funding is educational impact, which means funding of education institutions. I think that's a red herring, because good industrial research labs do a lot for education through their internship programs. Just to mention a few cases I know personally, Lillian Lee (Cornell), Andrew Ng (Stanford), Ben Taskar (Penn), Sanjoy Dasgupta (UCSD), David Yarwosky and Jason Eisner (JHU) are former Bell Labs/AT&T Labs interns.
My main objection to John's suggestion is that I saw it go badly wrong in European Union research funding, and I can't think of any good mechanism to fix those wrongs. The problems ranged from cost accounting manipulations to distortions of research agendas to fit government research objectives. Industry is much more complex and subject to more varied pressures than academia, making grant rules both more burdensome and also easier to game (these are not contradictory). Furthermore, Congress keeps authorizing R&D tax credits that are a more direct and less distorting form of support for industrial research.