Hacker News Reader

Comment by anotherhue | Hacker News Reader

Is there an argument for keeping publishers at this stage? More and more they seem to be net negatives.

Significant sub-areas of computer science do their own publishing nowadays. Peer review is already volunteer, and authors are already expected to produce their own publication-ready PDFs (mostly via LaTeX, sometimes Word), so it’s not a huge additional step to arrange the resulting PDFs into a website with a table of contents, and grab an ISBN for it. NeurIPS is a prominent example of this model: https://papers.nips.cc/

Yeah, version control seems like a better approach in every way. Some of the articles I've been asked to review are clearly written by a harried grad student writing English as a second language, as though the senior author didn't even bother proof-reading it. I see no reason for me to expend so many hours reviewing to figure out what in the hell they're talking about, effectively editing and proofing as I go so I know it's at least structured correctly. I think I'm effectively being co-oped by the senior author to teach their students. Why not just post to GitHub and see if it gets traction? If it does, cool. If not, let it die quietly under the fallen leaves.

I feel Arxiv is kind of this.

Back when I was doing research I had a couple of "papers" that I think had interesting ideas but I just hated the submit/publish dance (part of why I left academia after I years). So I just uploaded the "unedited" PDFs into Arxiv, thinking that maybe someone would find them useful.

Fast forward 10 years, I was surprised one of those PDFs was cited by more than 10 published articles. I'm happy I had my cake and ate it too: I could help someone with my ideas and didn't need to do all the bureaucracy.

As I think Churchill said: " everyone wants to be famous, but nobody wants to be useful" (sorry I I completely botched the quote) .

People nowadays care so much for their image .

Reply on news.ycombinator.com