>>11670173I don't think the idea of reproducibility quite applies to math, but there's a similar issue of work being published that's not sufficiently verified
Wild guess I'd estimate somewhere between 2-5% of papers contain results that are outright false or proofs that just don't work, and a much larger percentage is at least a little bit broken but salvageable.
The issue in math is that through a combination of many details being left out (don't want to insult your reader's intelligence by providing them with full arguments), theory getting so obscure and complicated that even the experts only have a shaky grasp of what's going on, and computations (also left out of the paper of course, because computations are trivialities) being so ugly and laborious that the referee can't be fucked to re-do them to check, the rate of errors that slip through the referee process is quite high.
