There's a vicious cycle here: One must work on "important" problems in order to get funding that mostly goes to pay too many grad students whose work allows ne

There's a vicious cycle here: One must work on "important" problems in order to get funding that mostly goes to pay too many grad students whose work allows new grants applications to happen, which then reqires more grad students. As a result, departments deliberately take on many more grad students than they will reasonably be able to place in tenure-track jobs. Thus every incoming class of PhD students contains a large fraction (more than half) whose careers will never be served by holding their noses to an unpleasant grindstone. Much in the way humanities departments exploit PhD students for cheap teaching labor, science departments exploit PhD students to keep the stream of indirect funding coming in. A big grant that requires a lot of assistants provides a lot of indirect funding for hiring and paying administrators. (Also, I would argue that advancing the economy is not actually part of the compact for the NSF, which is really a jobs program for scientists based on the experience during WWII that having scientists around in a pinch was really useful.) Interestingly, not every field is on this economic treadmill. Mathematics (pure, not applied), seems to declare problems as important based on whether they're unsolved and subjectively interesting. And economics, unsurprisingly, doesn't consider it rational to run large PhD programs that flood the market with competitors and reduce the market value of an economics PhD.