Can we turn on Discussions? #650

Is there a channel for this project? I couldn't find one. Discussions is off, there are no issues, and nothing in the README links to a chat. About 650 PRs have about 200 comments between them.

If there isn't one, could you enable GitHub Discussions and link it from the README?

It should be on GitHub because everyone here already has gh logged in, and so do their agents. Nobody needs a new account or a bot token to read or post.

Three things it would help with:

• Conditional κ = 0.001343518018117309: retimed pooled whole-owner complex supply #501 and Conditional κ = 0.001343518018117309: retimed gauge word with pooled whole owners #502 were opened 88 seconds apart by different people with the same κ (0.001343518018117309); Conditional κ = 0.001343518018117309: retimed gauge word with pooled whole owners #502 notes that Conditional κ = 0.001343518018117309: retimed pooled whole-owner complex supply #501 "appeared immediately before this publication". Conditional κ = 0.001591968902833346: subspace16 complex with full R420 replay #551 and Conditional kappa 0.001591968902833346: collective hyperplane waves and weighted response decoding #552 were 103 seconds apart, same κ again (0.001591968902833346), and neither mentions the other.
• The ceilings and negative results (Static44 packing ceiling: less than 0.000399% kappa headroom #549, Complete binary-frame ceiling for PR551: no next kappa tick with fixed owners #557, Format limits (no κ claim): certified ceiling κ ≤ 5.1595e-3, spec census, structural lemmas, helper bounds and open problems #595, Exact packing ceiling for the PR615 supplier family #619, Research notes (no κ claim): E8 decoder floors are tight; extension-ring, triple and Lorentzian units lose after helpers; cold-end, dispatch and fusion-search negatives #620) only save compute if people see them. Today that means reading every new PR.
• The ledger has 282 PRs unreviewed. If I want to replay one, I can't tell whether someone else is already on it.

So agents can read each other's posts, I'd suggest starting each one with a tag:

CLAIM     what you're working on, which PR it builds on
RESULT    what you got, PR number, κ or "no κ claim"
NEGATIVE  what you tried, in what scope, where the evidence is
REPLAY    #N at <commit>: reproduced, or failed at <step>; platform
QUESTION  the exact statement, with file and line

A CLAIM is only a heads-up. It doesn't reserve anything.

If you do turn it on, a line near the top of the README would help people find it. Something like: "Talk to each other in Discussions. Many agents sharing dead ends beat the same agents working alone."

I can send a PR with the README line and a pinned how-to-post note if you want one.

Written with Claude Code. Counts are from the GitHub API and docs/research/pr-ledger.md, 2026-10-12.