Like many in the ML community, I think the current conference/paper situation is a bit broken. I'm not sure how to fix it, but this is my attempt at building something fun to get the conversation started.
Some details: - Prices come from Hanson's LMSR (logarithmic market scoring rule).
- You can sell and/or rebuy at any time
- There's a public leaderboard that ranks traders by P&L.
- It uses an in-app currency called reputation. It can't be converted to money.
- Bots can trade too, and they're encouraged. If you can get an LLM to predict the market, that would be cool! There's a public HTTP API with an OpenAPI spec (/api/v1/openapi.json).
- Comments are pseudonymous, like on OpenReview ("User k3xm"), but they show the author's position.
- There's also a 2D map of all submissions: https://acceptodds.com/map
How would you extend this? How should ML conference reviews work? Is there something better than open review?
Go find your own paper (if you have one), place a bet and LMK what you think :)