We are in the midst of the academic slop-pocalypse. It seems like people aren’t even reading what the machine churns out before they share it. A friend of mine who was reviewing for a top CS conference recently told me
It’s so bad dude. So so bad. Reading two paragraphs of claudeslop is bad enough, let alone eight pages of it.
I’ve omitted some profanity. This isn’t unique; I’ve heard similar things from people in academia across a number of roles and institutions. There’s a lot of concern that it’s now hard to find good pre-prints, and that even sorting through conference proceedings and journals is no guarantee of quality.
So I built papers.club. One thing about really awful papers is that no one is going to cite them, and no one is going to recommend that others read them. If a lot of people who you trust are citing a paper or reading paper that’s a pretty good signal that it could be worth checking out. This happens informally already on a lot of applications, like twitter and bluesky, where people share papers as posts and you can see papers from the people you follow. If you want to see what it looks like, you can find me at https://papers.club/~brice.green.
The difference here is that this is a citation manager first and social media third. The citation manager exports to bibtex, and makes it easy to import bibliographies, tag and search papers, and filter for things you’re looking for. You can tag things as you’d like to make them searchable and add things to your reading lists, checking them off as you go. You can add to your papers straight from a feed. Once we get enough users I’ll add simple social-graph based recommendations. In a couple days you’ll be able to create (possibly dynamic) lists that contain specific papers or papers which match topics / are posted by certain people / posted by people in a certain field.
There’s no way to comment on other’s activity, you can’t see how many followers you have, and you can’t see who liked your posts. It’s kind of the opposite of an attempt at dopamine hacking, while still being a repository of useful, searchable information. The only real way to communicate with others is through an AIM-style status, which I included because I was born in 1993.
Anyway this is all in super-duper alpha, but the idea is simple: the social graph has the information we need to filter out the slop and focus on the helpful papers. So if that sounds cool, check it out! You can log in with your atproto account (e.g. your bluesky) or make an account with me using invite code et-al if that sounds interesting.