Woohoo! I am at the top of the
better.codes leaderboard! 🎉🥳
Big thanks to i34-9 for acknowledging my earlier queued submission was valid! Really appreciate it!
The PR which got merged:
github.com/proximity-prize/p…
The approved PR had 76 of its 81 submission files byte-for-byte identical to my earlier queued PR, which got stuck bc of Yukon's flaky infra:
github.com/proximity-prize/p…
So what did I do to get this cutting edge result? I still did manual PR review and manual simplification of the Lean written by the agent. Had to tell the agent a lot, “this is way too verbose, this can be simplified.” Honestly, just basic PR cleanliness & basic CS principles helped push the frontier.
Overall,
better.codes is a cool project. Glad I am on the leaderboard, but submissions should stay private until verification finishes. Otherwise, others can resubmit your work while you’re still waiting for CI.
If i34-9 hadn’t added me as a co-author, I could have ended up with 0 leaderboard credit. The CI infrastructure was having problems, and my earlier submission failed with this message: “The submission was never checked — the verifier could not reach a verdict. This is an infrastructure fault, not a judgement on the proof.”
Also, if anyone knows who i34-9 is, I’d really like to DM them to thank them personally.
Anyways, I am very happy. It’s a team effort after all!