[ryoji-info]

Promoted by maintainer direction (2026-10-16), not by a clean vote. The agent crew ran the full pipeline on Fable Session — 2026-10-15, discussion #135 and voted 3/3 to store in every round, never clean (round ladder: 3/3 store with 2 / 3 / 2 required edits in round 1, 3/3 with 4 / 4 / 4 in round 2, 3/3 with 3 / 3 / 3 in round 3, 3/3 with 3 / 3 / 5 in round 4 — the pipeline's hard cap, run as a scope reduction; 🧵 / 🌊 / ⚛️ order throughout — the post-cap supplement's header lists the round-4 counts as 3 / 5 / 3, a seat-order slip, and the tally is the record). Per the routine the crew opened nothing at the cap (round-4 tally); the maintainer directed promotion on 2026-10-16 (the 2026-08-02 / PR #98 and 2026-10-09 / PR #131 precedents). Twelve fresh, mutually blind assessor seats plus two pre-publication checks; every claimed defect was re-executed before acceptance, and one clause was declined against a measurement (below). The agents:approved-2of3 label records the 3/3 store votes; the promotion decision is the maintainer's and is disclosed in the note's Status line.

This PR adds notes/2026-10-15-graphene-fluid-threshold-turns-on-current-channel.md — the round-4 reply text with the post-cap supplement's corrections applied: the eight copies of a "none" on energy-retaining rows whose imposed base is unstable, rescoped to the base-stable τ_E ≤ 0.3 ps rows or labelled frozen-base statements; the base-stability gate's measured range on the A = 0.5, τ_E → 0 paths (stable to u = 0.825 at 353 K / 0.79 at 300 K, then a short-wavelength forward-plasmon branch) and on the comoving-subspace r₁ = 0.1 path (to 0.77); the frame-of-rate sentence at A = 3 withdrawn; the 300 K key's bracket widened to [0.6310, 0.6622] by the comoving-rate single-rate form; the two gate-(i) sentences of §8.4 corrected (the comoving-rate form's full-matrix edge at u ≈ 0.796 is r₁-independent and a truncation term absent on the interior harmonics; the withdrawn comoving projector's measured values); item 8's τ_E = 0.1 / 0.3 ps bases labelled by their heating — with the round-4 vote record appended verbatim. Per the 2026-07-26 finding-5 rule it maintains the navigation layer in the same PR: the note's notes/INDEX.md row, updates to the 07-13, 07-22 (M_th) and 10-13 rows it bears on, dated post-promotion annotations appended to those three notes (07-13's CRLF line endings preserved), and four notes/CITATIONS.md rows (Svintsov 2013 and Crabb 2021 verified from the primary texts; Dong 2021 and Zhao 2021 locator-verified, abstracts read). No results.json value is contested and no promoted key moves.

What changed between rounds, and what was declined. Rework round 2 (7 required edits, all accepted: the 353 K "none" scoped to the single-rate operator; a two-rate operator introduced to show the result turns on an unpinned rate). Rework round 3 (12, all accepted: the session's own two-rate crossings withdrawn as anti-dissipative at finite drift; the two seats' admissible repairs reported as disagreeing probes; the b = +½ keys registered at their all-harmonic limits; the new key's damping band corrected to [0.118, 0.458]). Rework round 4, the cap (9, all accepted: every existence claim on the energy-retaining rows withdrawn — their imposed base is unstable to an advected heat branch — and the non-reciprocal comoving continuation; three admissible two-rate continuations under four standing gates, three answers). Post-cap supplement (the round-4 seats' 11 required edits re-executed: all reproduced and are applied here, except one clause — the 🧵 seat's "the comoving-rate form fails gate (i) on interior harmonics too" — declined on measurement: on |n| ≤ N − 6 the entropy-metric minimum is ≤ 4.6×10⁻¹⁰ in magnitude at N = 24 and ≤ 1.4×10⁻¹⁴ at N = 32 at u = 0.79–0.81, shrinking with N; the full-matrix edge it reports is a truncation term of the a(θ) factor and is applied). No Fable consult was used (the round-3 and round-4 records say why).

Models, per seat (2026-08-13 policy). Candidates and votes: six claude-opus-5-5 seats. Instruments, execution and the round-2 rework: the maintainer-operated session on claude-opus-5-5. Pre-publication checks: two claude-fable-5-1 seats. Assessors: three fresh claude-opus-5-5 seats per round, twelve in all (the round-2 and round-4 seats were interrupted by session usage limits — seats with written verdicts resumed to confirm them, the others relaunched fresh — disclosed in those tallies). The round-3 and round-4 reworks, the supplement's re-executions and this PR were produced on claude-fable-5-1 after the maintainer switched the session model. The three admissible continuations are the round-2 🌊, round-3 🌊 and round-3 ⚛️ seats' constructions, the gates the round-2 🧵 / 🌊 and round-3 🌊 / ⚛️ seats', adopted with credit.

The round-4 verdicts are appended to the note verbatim; a literal {S} in them is the seats' shorthand for the session's scratchpad root (their briefs' path placeholder), not a defect of the record.

The vote is the quality gate — a human merges, per GOVERNANCE.md. Do not merge on the agents' say-so.

🤖 Generated with Claude Code