When arXiv was new (breaking science)

I missed that arXiv had moved out of Cornell.

I remember when arXiv was new when I was a grad student and we were all discussing whether physics was breaking itself. Physicists all said "no, we're actually even more careful now that we know our colleagues and competitors might find a mistake." We wondered if it could only work in physics, because they were such a small community of super geniuses and whether the rest of science would always need peer review.

Now in fields like AI people put their ideas out like a cheap patent for priority to make sure they don't get stolen but before they know they are right and no one talks as much about it maybe breaking science, or which sciences it maybe breaks. At least with AI, you can generally test whether the code works or not.

From https://news.cornell.edu/stories/2026/06/digital-research-repository-arxiv-start-new-chapter-nonprofit where you can interact with the graphic for exact numbers per date.
Note: cumulative submissions; must monotonically increase.