There was, as I remember, a curious observation, called goodhart's law or something similar stating that people will always try to affect outcome. When artifici

There was, as I remember, a curious observation, called goodhart's law or something similar stating that people will always try to affect outcome. When artificial rules of the funding game are known, people will game the rules and the just stop to be useful. For example, if we use "fund research at random", people will flood with research proposals. Or if HN upvotes will become measure of which research is funded, soon enough people will optimize research proposals for HN upvotes. I think, goodhart's law would not apply in certain situations, for example when signal is extremely costly to fake. In research world, some unquestionable proof of personal time of a researcher spent would suffice, I think. I.e three years in a row researcher comes in the same grant proposal and his colleagues say in writing, that yes, he is working only on it and nothing else for three years, no funny stuff.