This makes sense until you realize the people funding the grants are not the people approving the grants. It is really a lot easier to spend other people's money poorly than to spend your own poorly. Besides, if you want to know if a grant is worth funding who are you going to ask? Probably esteemed people in that field or something related, who all have the same poor incentives and institutional inertia to contend with.
And when politicians are funding long term science, they aren't funding the actual long term benefits, they are funding the APPEARANCE of producing long term benefits (and the funneling of money to those who can provide kickbacks). As long as the pretense can be maintained, anything can be funded, regardless of its actual utility.
The logic is not "if people buy it, it must be good", but more "if people buy it, there must not be something else that they can buy that gets them what they want for less".
I think that's within the working definition of "good", but even if it isn't, I think my concerns still apply. It might on average be true in certain narrow circumstances, but there are so many exceptions it conceals at least as much as it reveals.
Re:20 Years Later
by chrisweekly on Tuesday December 29, 2020 @06:37PM
(#25576434)
+1 insightful; 100% agreed; well-put!
Slashdot reference?
Re:20 Years Later
by chrisweekly on Wednesday December 30, 2020 @08:36PM
(#25588573)
Yep
This is very close to "There must be a pony in there somewhere". [1] If the question is whether the field is worth continuing to fund, the answer can't be, "Well since people are still funding it, it must be worth funding." By the same logic, we should just keep putting billions into WeWork.
Research is hard to value, and fundamental research especially so. The logic of "if people buy it, it must be good" only patchily applies to normal commerce, where results are relatively easy to measure and feedback loops are short. It's wholly inadequate for feedback loops on the scale of decades.
[1] https://www.google.com/search?channel=fs&client=ubuntu&q=the...