Arxiv, and it's various follow-ons - I think I was one of the first submissions to medRxiv, are excellent as far as they go, but if we take the COVID-19 pandemic as a stress test, they still have a long way to go.

Beyond that, their curation is...lackluster at best, even if we're not talking about gatekeeping, filtering, etc. but just "What if I want to do something other than key word search?"

I find unexpectedly interesting articles all the time reading journals. I have yet to do so for a preprint server unless someone sends it my way, or it ends up on my Twitter feed.

> I find unexpectedly interesting articles all the time reading journals.

Interesting. In math, I don't think anyone "reads journals", but lots of people check the arXiv frequently to see what has been posted.

Math is probably the furthest along this particular pathway - which is both a good sign for this being able to work, and also not generalizable.

Journals can still do curation of the stuff that's on preprint archives. Some already do. It's a lot less friction than the current model.