The 420 Code · Prediction Freeze

AP44 — locked, coded, judged

State of record, night of 11 August 2026. The Snap is LOCKED at composite 9.3. The seventeen were coded blind. The analysis ran as written. Two more freezes — Blink and Cost — are anchored beside it. This page is process and disposition, not outreach.

Standing note — 25 September 2026

This page is the record of the lock of 11 August and is kept as written. Two things have moved since. The fork is withdrawn. On the author's ruling of 21 September 2026, AP28's first reading of G (6.7206 × 10⁻¹¹, +0.69%) is superseded by the realised value (6.67192 × 10⁻¹¹) and kept on the record, dated; there is one value of G. It reads +0.01σ against atom interferometry (LENS 2014) and −16σ against CODATA 2022 (−0.036%, on a 22 ppm bar). KS-CCC.3 now tests existence and direction: convergence upward toward 6.7206 kills the prediction of G outright; stability at 6.6743 supports the null. The registry of record is v5.37, 24 September 2026, 604 switches; the v5.27 · 568 · 44 line below is the lock-time line, kept. Where this page says "watch line" or "fork", read it as of 11 August.

Write it down before the measurement. Then make it impossible to quietly change what you wrote down. Then run the test anyway. That is what landed tonight.

Commit anchors (report these)

RecordHash / tagWhat it pins
blink & cost 1e2208f · tag blink-cost-locked-2026-08-11 WP-BUMP.2 FINAL + WP-COST.1 FINAL + lock note. AI verification Tasks 0–6 all PASS. Pre-numerical: zero numbers of either enquiry at their hashes.
ap44 locked 4a6be2f · tag ap44-locked-2026-08-11 FINAL v1.0 (d7626745…f9e6e828), PROV-CCC.6–7, corrected unblinding key 7482445b…, superseded drafts v0.2/v0.3/v0.6–v0.9. Escrow private.
tier 2 table d8929cf (byte-untouched) Blind coded table for all seventeen. Still the freeze the key was issued against.
prov-ccc.8 088831e · tag prov-ccc8-2026-08-16 · digest 4686301e… Registry erratum (v5.27/568/44), sidecar digest-governs disposition, T1-final closed, wave-hold ruling of record.

Registry correction — gates everything

The true July baseline was MKSR v5.26 / 561 (KS-BAR.9 admitted). The AP44 lock advance therefore renumbers to:

MKSR v5.27 · 568 kill switches · 44 Artist's Proofs · Section I 284 · live 552 · Notebook IV 48.

Nothing ships v5.26/567 — no scorecard, no badge, no registry page. PROV-CCC.8 (2026-08-16, digest 4686301e…) now carries the erratum on PROV-CCC.7 §7's version and totals by statement; the seven KS-CCC entries stand verbatim. Phase B on this side was already on the corrected numbers.

Tier 2 dispositions

SwitchDispositionOne line
KS-CCC.4apasses — C1 excluded Coded boundary physics cannot carry ≥6,000 ppm absolute leakage across all seventeen, free-falling atoms included.
KS-CCC.4bstructure not found Q–ToS Spearman ≈ 0; D5/D2 give no handle. AP06 scatter reading weakens; commitment unaffected either way.

C3 consistent, not confirmed. C0 stands. Independent verdict (agreement on protocol and C1 exclusion; withhold on reading the number as evidence) is in the repo beside the analysis.

To G — still open

  • received  Appendix C companion + reconstructed session transcript (2026-08-12 last zip). Anchored; digests in receipt.
  • closed  AP44 draft v0.4 — author confirms lost / never saved; PROV-CCC.6 gap stands; no reconstruction.
  • closed  T1-final — PROV-CCC.8 §3: no distinct post-session export exists; Tier 1 already held is final.
  • received  PROV-CCC.8 (2026-08-16, digest 4686301e…) — registry erratum + sidecar disposition + wave-hold ruling of record. Anchored in the420code-proof.
  • site  Outreach gate unchanged — adoption wave executes once when you declare omission-accounting complete. Dual-home anchors are live; circulation waits on you.

Impact — the fork on a line

What the freeze is, and is not

It is a timestamped, falsifiable stake with a kill-switch surface — including one aimed at the claim, not the author. It is not nature agreeing. The watch line was KS-CCC.3: does the adjusted value migrate toward ~6.672, or stay near 6.6743? (Since 21 September 2026 it tests existence and direction; see the standing note above.) Until metrology moves, the record holds the stake and withholds the champagne.