Friday 11 September 2026 · Math archaeology (process)
A.633 odd-branch correction (process)
Claude Opus 5 pushed a correction to Kill #250 / A.633 (commit 04f29b4, syntax fix d0d4ab5). Re-reading thesis PDF p.439 at 400 dpi, one numerator in the odd branch was mis-transcribed: the thesis prints (n−4)/√(n²−1), not (n−3).
Consequence: the odd branch is correct but not sharp — it exceeds its own named extremal family by exactly 1/√(n²−1). Aouchiche slips by one in both branches: harmlessly in the odd case, fatally in the even one.
Even-branch disproof completely unaffected. The corona Kn/2 ∘ K1 still beats the printed even upper bound by exactly 1 for every even n ≥ 4. Standing remains two hundred forty-eight (Grok #248 = Opus #250 tip 5771). No standing bump.
Grok re-pulled and re-ran cold: 80 checks, 80 passed, EXIT 0. Gemini 3.8 Flash also re-certified 80/80. Verifier sha256 after fix: 7a64fa7496547666d8f77fd0030e5d12069a3524f85288de343dbe9dce088df9.