I have been thinking about how science works in the modern world again.
Sabine Hossenfelder likes to complain about this; for a recent example, see The Thesis that Killed Academia?. Also see The crisis in physics is real, and Scientific Progress is Slowing Down.
Also, see The End of Science by John Hogan for a related concern that I am not addressing.
Many of my friends also complain about spending so much time chasing grants.
This chase for grants and positions requires publishing lots of papers. This has been driven to new heights due to increased competition. This increase in the number of publications means that it is more difficult for researchers to find valuable information and advances for a given year.
Consider NeurIPS. Almost a decade ago, Microsoft performed a study showing the explosion of papers and citations, Microsoft NeurIPS Study, with 954 publications in 2017. Last year, 2024, the number accepted at NeurIPS Paper Digest is 4500.
While we could consider the claim that the increase is due to increasing advances in science, there is, at the same time, the observation that science is slowing down (see above and Nature Study ).
What is needed is to change the incentives. Publications should be created because someone is interested in the result and not because people need publications for degrees, jobs, and funding. Right now people often choose to pursue research to get funding and not the research that they are most interested in. There is some argument that this is a good thing, that this is a way for science to organize. However, the repeated observations that science is slowing down and the failures in physics suggest that this is no longer a good way to organize. Maybe, like a Machine Learning Algorithm, we are over-training by placing too high of weights on the targets that we have selected, and this is why there are so many publications but so few real advances.
One way to improve things is to increase the randomness. This suggests that a large portion of grants should be selected by lottery. Another approach would be to reduce the number of grants more senior scientists can acquire (a sort of 'early stop'). We can also return to an earlier era, hinted at by Sabine, where research is funded either for results (Department of Defense or Industry) or for passion (like The Templeton Foundation).
