Scorecard

The colour reading has been WITHDRAWN (S324). Green, amber and red were published here and on four other pages as a reading of how each row sits against its comparator. No definition for them was ever published, and they were not a threshold on d — a row marked green sat at |d| = 1.75 beside rows marked amber at 1.00 and 1.94. Rather than define a rule after the fact, the house has withdrawn the label. Nothing measured has been removed: every row still prints its deviation d and its tier, and the tier is what governs citation — no row may be cited as zero-parameter, and nothing here is peer-reviewed. Where a status cell carried a substantive qualifier it is kept verbatim; where it carried only a colour it now reads —.

★ RE-CHECKED, S370 (2026-09-26) — the basis moves to Rev33.1. The S351 verifier (scripts/scorecard_rederive_S351.py) was re-run at the Rev33.1 cut: every theory cell and every deviation cell recomputed from its formula at full precision, 0 disagreements. Rev33.1’s engine roll (kernel v5.0.12) reads the NuFIT 6.1 Δm² pair and sin²θ_W(M_Z) from source-ledger rows SL-01, SL-02 and SL-06 — the values this table already printed — so the seal moved the engine and not the table. No row is added. The notes below keep the release names they were written under.

★ RE-CHECKED, S364 (2026-09-23) — the basis moves to Rev33.0, and two rows are added. The S351 verifier (scripts/scorecard_rederive_S351.py) was re-run at the Rev33.0 seal: every theory cell and every deviation cell recomputed from its formula at full precision, 0 disagreements. Rev33.0 moved the engine’s CKM, charm and top comparators onto the source ledger’s PDG 2026 rows; this table already printed those rows, so the seal moved the engine and not the table. Rows 23 and 24 (the Jarlskog invariants J_CKM and J_PMNS, PI ruling R174-b) are added with their notes N23/N24. The Rev32.11-era correction notes below keep the release names they were written under.

★ RE-DERIVED, S351 (2026-09-19) — the basis moves to Rev32.11; not one number moves with it. This page had declared a Rev32.7 basis since the Rev32.7 seal. Rolling that label is only honest if the rows underneath it were actually checked against the current release, so every theory cell whose formula this table prints was recomputed from that formula at full precision, and every percent and comparator distance was recomputed from theory and comparator under R77 (compute unrounded, round once) and R77-b (a distance uses the error on the side facing the theory value). Fourteen theory cells and eleven deviation cells were re-derived and all twenty-five agree with what is printed — zero disagreements. The method is deliberately not a comparison against another document: the S327h correction below was caused by cells computed from already-rounded inputs, and a second document can carry the same mistake. Arithmetic was checked against arithmetic. Rev32.10 and Rev32.11 rolled the bottom- and top-quark ENGINE INPUTS to PDG 2026; this table’s comparators were already the PDG 2026 values, which is why the seals moved the suite and not the scorecard. Row-level citations below (“A1566, Rev32.6”, “Paper 3, Rev32.2”) name when a decision was taken and are left exactly as they stand — they are history, not currency. Verifier: scripts/scorecard_rederive_S351.py.

★ CORRECTION, S327h (2026-09-11) — seven cells recomputed at full precision. The Scorecard notes (DETAILS column) re-ran every row. Seven printed cells had been computed from numbers that were already rounded, or, in one case, with the wrong side of an asymmetric error. By PI ruling R77, every number is now computed from the unrounded formula and rounded once, and a distance uses the error on the side facing the theory value. Changed: row 11: |V_ub| d: −0.17 → −0.18 (|V_ub| = 0.003861974 unrounded; −0.17 was the rounded 0.003862 sitting exactly on −0.175) · row 12b: δ_CKM vs direct γ: +2.60% → +2.61% (68.1297°) · row 12c: bare-φ control: −0.80% → −0.81% (65.5874° vs the global fit); direct d −0.30 → −0.29 (R77-b: below the central value, so the lower error 2.8 applies, not 2.7) · row 18: m_e: −0.67% → −0.66% (m_e = 0.507607 MeV unrounded; the suite itself prints −0.66%) · row 19: m_b: −0.96% · d = −6.67 → −0.95% · d = −6.64 (4146.17 MeV unrounded; the theory cell still displays 4146) · row 21: m_u^const: −1.49% → −1.57% (330.725 MeV unrounded against ~336; the theory cell still displays 331) · row 22: Δm² ratio, d = 1 reading: +820% → +821% (0.276393 / 0.030016; the factor 9.21 is unchanged). No theory value, comparator or tier moved, and no row changes status. The sealed papers (Rev32.10) and the Registrar records that quote them still print the earlier figures, except where the Rev32.10 changelog records a fix; the remaining paper corrections are queued for Rev32.11. Found by the ChatGPT drafting lane (A1633) and re-measured independently by the house before this change. Nothing above is edited away; this note is the record that it changed.

★ CHANGED, S328a (2026-09-11) — PI ruling R84. Row 12a: the global-fit comparator is printed in the unit it is published in, 1.154 ± 0.025 rad, and the distance is computed there: d = +1.41 → +1.40 (the degree cells gave +1.41; R84-e). Rows 12a and 12b: the status cells now carry Paper 3’s own sentence that the 5φ⁻⁵ correction is not supported by current data (R84-d). Row 14: the tier cell reads “SM content — not a test of J₃” instead of “no tier assigned” (R84-d). Row 16: the suite’s own running endpoint, 168.36 GeV, is printed beside the tree value; whether the row keeps a distance is not yet ruled (R84-f). The colour tally that survived the S324 withdrawal above the table is removed (H200). The tier list below now defines every tier word the table prints (H201). The footer names the m_t comparator correctly as the direct-measurement (MC) mass (H199), and the α sentence prints both failure sets the suite records (H204). No theory value moved. The sealed papers keep the earlier figures until Rev32.11. Rev32.10 was sealed at S344 as the referee-rows release; the remaining corrections roll to Rev32.11.