Phase 1 Option G+ audit COMPLETE — methodological lessons + outcome summary

§1 — Outcome (numbers)

Phase 1 audit period: Sunday 26 April 2026 (D-9 / D-10 / D-11 closed) + Monday 27 April 2026 (D-12 through D-18 closed in single session).

9 decisions adjudicated (D-9 through D-18); 9-of-9 Rich-confirms:

DecisionOutcomeAmendmentCommitEffect
D-9 Apache scope expansionDEFER(none)e6e7a96Sovereignty discipline reinforced
D-10 Wills↔Probate seam visibilityAUTHOR(visibility doc)ecc0d95Wills↔Probate seam locked
D-11 Q3+Q4 RE-VERIFYCONFIRM-BOTH(none)f5970fcModule structure stable
D-12 Probate Alloy pull-forwardPULL-FORWARD-PHASE-1-5A-1183224911Probate-Alloy layered formal verification
D-13 OwnershipBasisPROMOTEA-119 + scheme #2097d1b25OwnershipBasis #22
D-14 IHTReferenceStatePROMOTEA-120 + scheme #211b4d377IHTReferenceState #23
D-15 EstateAccountsPROMOTEA-121 + scheme #22bae36d0EstateAccounts #24
D-16 AssentEventPROMOTEA-122 + scheme #238eb8785AssentEvent #25 (2nd 3-module-bridge paralleling A-98)
D-17 IAS Tier-3 architectural-asymmetryHYBRIDA-123 (methodology)49d5471IAS forward-looking Tier-3-primary A-numbers
D-18 Phase-1.5 stress-test reorderingOPTION-CA-124 (sequencing)(this commit)Module-by-module + per-module 3-rule MVP

Net outcome:

  • 7 substantive A-numbered amendments registered (A-118 + A-119 + A-120 + A-121 + A-122 + A-123 + A-124)
  • 4 cross-module primitives added (#22 OwnershipBasis + #23 IHTReferenceState + #24 EstateAccounts + #25 AssentEvent); cross-module primitive count 21 → 25 (+4)
  • 4 SKOS classifier schemes added (#20 OwnershipBasisScheme + #21 TaxReferenceTypeScheme + #22 EstateAccountSectionScheme + #23 AssentTypeScheme); SKOS classifier count 19 → 23 (+4)
  • 3 visibility/methodology updates: Wills↔Probate seam articulation v1.0 → v1.2; IAS state v1.1 → v1.2; ARCHITECTURE-STATE-INDEX v1.0 → v1.1
  • Tier 1 arch-state: v2.71 → v2.78 (7 minor-bumps); amendment_registry_version v2.24 → v2.31
  • Phase-1.5 cell count: 189 → ~215-236 (+26-47; ~14-25% increase); still within original 4-6 weeks Phase-1.5 budget
  • Module count unchanged at 9; A-43 VC distribution unchanged at 10; A-21 CI gate count unchanged at 19 (gate 20 spec’d for A-118 + A-NEW-22 pending Phase-1.5 wire-in)
  • 1 NEW locked discipline: feedback_display_numbers_in_scorecards (locked Monday 27 April 2026; sharpening of feedback_always_display_full_scorecard)

§2 — Methodological lessons (for future audit phases + future scorecard work)

Lesson 1: Numerical scoring matrix discipline surfaces stalls + ties that verbal reasoning hides

Trigger: Rich locked feedback_display_numbers_in_scorecards Monday 27 April 2026 same-turn as D-13 close, after D-12 + D-13 docs lacked numerical matrices. Applied to D-14 onwards.

Concrete lesson: D-14 IHTReferenceState scored 0pp tie A-vs-C at 80% in numerical matrix; without the matrix, the tie would have been invisible and Option A would have been recommended on verbal-reasoning-only basis. Numerical matrix forced a feedback_reframe_beats_reweight reframe via Probate-Alloy + A-115 precedent + discoverability. Reframe carried A.

Discipline value: every forthcoming scorecard MUST include explicit numerical scoring matrix (per-option × per-criterion 0-5; weighted sum; margin to second-best; sensitivity perturbations). Verbal reasoning + trade-off tables alone are NOT sufficient.

Lesson 2: Recalculation post each Rich-confirm strengthens subsequent decisions (precedent compounds)

Concrete lesson: D-13 OwnershipBasis recalc post-D-12 strengthened by Phase-1.5 cell-count expansion (+7-14 cells from A-118 created headroom for OwnershipBasis 8-jurisdiction stress-test cells). D-14 recalc post-D-13 strengthened by D-13’s promotion-pattern precedent (5-module reach passed bar). D-15 recalc post-D-14 strengthened by D-13 + D-14 just-locked dependencies (capital section splits + tax section IHTReferenceState reference). D-16 recalc post-D-15 STRENGTHENED MOST (4-Barlow-primitives convergence + A-98 parallel + Probate-Alloy distribute_bequest clean EFFECT — strongest A-win 92.5%).

Discipline value: per feedback_decision_analysis_one_at_a_time, recalculate from scratch based on Rich’s actual answers (not the projected answers). Recalculation isn’t ceremonial — it MATERIALLY strengthens subsequent decisions because each Rich-confirm becomes evidence/precedent for the next.

Lesson 3: Reframe per feedback_reframe_beats_reweight works at narrow-margin AND wide-margin

Concrete lesson: D-14 reframe (narrow-margin 0pp tie A-vs-C → reframe via Probate-Alloy precedent + A-115 + discoverability tie-broke A). D-18 reframe (wide-margin: original “3-rule MVP vs 10-rule corpus” framing recast inside feedback_first_vertical_slice_symmetric_2026_04_24 discipline → Option C “Module-by-module + per-module 3-rule MVP” surfaced as genuinely better path; 10pp margin C-vs-B; sensitivity 7-of-7).

Discipline value: reframe isn’t only a tie-break tool. When the original framing is structurally incomplete OR violates a binding discipline, reframe surfaces a better option even at wide-margin. Test: would a reframed option score materially better on the criteria? If yes, surface the reframe.

Lesson 4: Margin honesty needed for ties AND clean wins

Concrete lesson: D-14 0pp tie reframed (margin honesty surfaced the reframe need); D-15 10pp clean win (margin honesty noted “this is the cleanest A-win”); D-16 15pp strongest win (margin honesty noted “strongest A-win of the 4 Barlow primitives”); D-17 15pp clean win (margin honesty noted “C HYBRID wins decisively”); D-18 10pp clean win (margin honesty noted “no reframe needed; C IS the reframe option”).

Discipline value: per feedback_weak_criteria_to_avoid_in_scorecards 5-test checklist (scarcity / clean-break / double-counting / temporal-mismatch / effect-size), margin honesty applies regardless of size. Document margin tightness OR margin clarity in every scorecard.

Lesson 5: Discipline-priority sequencing — methodology-last anti-pattern

Concrete lesson: D-17 IAS Tier-3 architectural-asymmetry is methodology-only; it correctly came AFTER D-13/14/15/16 substantive primitive promotions per feedback_scorecards_one_at_a_time_optimal_sequence anti-pattern: “methodology-last — decisions about methodology/process typically come last so they don’t contaminate current adjudication.”

Discipline value: when sequencing decisions in a multi-decision audit, methodology-only decisions go LAST. Substantive primitive/schema decisions go first. D-17 (IAS A-number convention) correctly came after D-13/14/15/16 (4 cross-module primitive promotions). D-18 (sequencing convention) correctly came after all primitive promotions (depends on cell-count picture).

Lesson 6: Tier 2 InheritKit precedent unlocks Tier 3 IAS HYBRID convention

Concrete lesson: D-17 reframed via Tier 2 InheritKit precedent — Tier 2 already uses same A-N sequence flagged “T2-primary” per Tier 2 state §10 (“Total Tier 2 entries: 18 (10 primary; 8 cross-tier-referenced from Tier 1)”). HYBRID convention generalises Tier 2’s pattern to Tier 3 IAS for structural symmetry.

Discipline value: when an architectural-asymmetry surfaces, look for an existing precedent in a sibling tier/area. Often the precedent is the answer. Future-extensibility test: does the convention scale to 4th tier? (Yes for HYBRID — same pattern applies to consumer-facing brand state files when they emerge.)

Lesson 7: Phase-1.5 cell count grows naturally with primitive promotions

Concrete lesson: 4 Barlow primitives + 1 Probate-Alloy added 26-47 Phase-1.5 cells (+14-25% from 189 baseline) → ~215-236 cells. Each primitive promotion adds proportional Phase-1.5 stress-test cells (8-jurisdiction or 8-jurisdiction × scheme-leaves). Phase-1.5 budget (4-6 weeks) holds because per-module phasing per Option C distributes cleanly (3-7 days/module × 9 modules = 4-9 weeks; lean-end fits).

Discipline value: Phase-1.5 cell count is an emergent metric — track it after each substantive amendment. Per feedback_loc_not_significant_metric, framed as Gate 1/2/3 structural-completeness uplift not LOC budget. The cell count tells you whether Phase-1.5 budget is at risk.

§3 — Phase 2 audit kickoff (next)

After D-18 lock, Phase 2 of the Option G+ audit begins:

Phase 2 task: re-score Option G+ vs B/C/D/F using the sharpened 17-criterion frame from criteria-frame-sharpened.md v1.0, applying the locked discipline feedback_display_numbers_in_scorecards.

Predicted Phase 2 outcome per criteria-frame-sharpened.md §4 prediction: G+ overtakes F by ~9pp in the 17-criterion academic frame (G+ moves from 65/80 81% → ~75/85 88%; F drops 67/80 84% → 67/85 79%; commercial-scorecard lead G+ 84% vs F 50% likely widens further).

Phase 2 will produce actual numbers under the locked numerical-scoring discipline. Predicted outcome is guidance, not pre-locked answer.

§4 — Memory cross-references

  • feedback_decision_analysis_one_at_a_time — one-at-a-time discipline operated correctly across all 9 decisions
  • feedback_scorecards_one_at_a_time_optimal_sequence — methodology-last sequencing operated correctly
  • feedback_display_numbers_in_scorecards — NEW (Monday 27 April 2026; sharpening of feedback_always_display_full_scorecard)
  • feedback_actively_use_t_files_in_scorecard_authoring — verbatim quotes used in every decision doc
  • feedback_reframe_beats_reweight — reframe applied at D-14 (narrow-margin) AND D-18 (wide-margin)
  • feedback_weak_criteria_to_avoid_in_scorecards — 5-test checklist applied; margin honesty in every doc
  • feedback_introduction_module_rule — Core fallback for cross-cutting concepts; D-13/14/15/16 all promoted via this rule
  • feedback_first_vertical_slice_symmetric_2026_04_24 — binding discipline ruled out E&W-first sequencing in D-18
  • feedback_v2_ik_ias_build_plan_sovereign — boundary test applied to every decision
  • feedback_loc_not_significant_metric — framed Phase-1.5 expansion as Gate 1/2/3 structural-completeness uplift
  • feedback_architecture_state_file_discipline — refined by D-17 HYBRID convention
  • project_law_vs_practice_split — invoked by D-15 EstateAccounts (statutory skeleton goes Apache; practitioner tooling goes InheritKit)
  • project_round_5_gold_2026_04_25 — IAS £25-40m UK Y3 ARR opportunity (referenced as IAS commercial-moat-affecting criterion in D-17)