The first reply from the thread's OP got me thinking about something I'd never really noticed: > I used to think it was because as a quasiscientific community

The first reply from the thread's OP got me thinking about something I'd never really noticed: > I used to think it was because as a quasiscientific community we highly value peer review but ML people are totally fine with citing papers on arxiv that haven't been accepted anywhere as long as you can replicate the results - I think perceived prestige nails it[1] Peer reviewed citations are excellent, and unstable citations like Wikipedia are obviously concerning. (At bare minimum, you'd want to link to the page history instead of the page.) But the willingness to cite arXiv content without hesitation definitely suggests that peer review is only part of what's going on there. [1]https://twitter.com/adversariel/status/1054836640307392512 https://twitter.com/adversariel/status/1054836640307392512