Comment by grues-dinner

> clearly a lot of knowledge held in silos.

I don't think it's quite that, since the eventual goal is to publish, not only publicly, but as publicly as possible. More like it seems like everyone tends to hold their cards quite close to their chest until the moment of pre-print publication. Which means you can be working on something that someone could have told you months or years ago you have a problem.

The scientific equivalent of polishing a branch before making a pull request, only to be told "this has a huge memory leak and moreover what you want already works if you use this other API".

I'm not really sure there's a human-scale solution: the research landscape is so vast that you can't connect everyone to everyone else and have everyone in need of valuable input get it, and have everyone able to give it not be inundated with half-baked rubbish. Even if you assume everyone from the top to bottom has pure motivations and incentives for doing the research in the first place (in the pull request analogy CVE spammers, for example).

Perhaps not having the universities themselves so keen for PR that they'll slap a press release together about anything that looks clickable without due diligence would at least prevent making a public spectacle outside of the academic circle now and then, but it wouldn't solve the fundamental issues.