project_full_phase_2_3_reversion_2026_05_01

Trigger: Rich audit directive 2026-05-01T20:35 BST: “i want to go back to the beginning of phase 2 and use our prompt to ask all the questions. i feel the enhanced clarity from the prompt, and the improved research, will give us a huge advantage over time, as the 16 [now 17] questions we have answered, are so significant.”

Scope: 17 questions reverted:

• ζ-Q1 (criteria-frame) → A-126 discarded
• ζ-Q2 (architecture options G+) → A-127 discarded
• ζ-Q3 (asset taxonomy ε) → A-128 discarded
• ζ-Q4 (upper ontology ω.γ) → A-129 discarded
• MQ-001 R2 (ω.γ→ω.α reversal) → A-131 discarded
• ζ-Q5 (ψ.γ SSSOM-canonical) → A-130 discarded
• ζ-Q6 (φ.α’ R2-reframed t-box) → A-132 discarded
• ζ-Q7 (χ.γ vi-6+ Hybrid) → A-133 discarded
• ζ-Q8 (Λ.β multi-anchor SSSOM) → A-134 discarded
• MQ-002 (Phase-2 closure) → A-135 discarded
• ζ-Q9 (Ξ.β verified-with-4-gaps) → A-136 discarded
• ζ-Q10 (Φ.α Wills MVP brief) → A-137 discarded
• ζ-Q11 (π.β BillingConfigurator) → A-138 discarded
• ζ-Q12 (Θ.β Organisation.ratings) → A-139 discarded
• ζ-Q13 (Δ.β registrations[]) → A-140 discarded
• ζ-Q14 (Ψ.γ verificationTier) → A-141 discarded
• ζ-Q15 (✕.ζ ★ Wills.Will class schema) → A-142 (re-issued) discarded

State file reversions:

• inherit-v2-architecture-state.md v3.28 → v3.10 (logical equivalent post-MQ-000)
• inheritkit-architecture-state.md v1.2 → v1.1
• ias-architecture-state.md v1.3 → v1.2
• 2026-04-29-multi-phase-audit/INDEX.md v2.18 → v2.1
• inherit-v2-risk-register.md v1.2 (no change; pin_kind tagging preserved; Phase-2-onwards risks may need re-validation)
• inherit-v2-standards-engagement-state.md v1.0 (no change)
• richard-tasks-tt.md v1.16 → v1.17 (annotation only — tasks 189-202 marked PRE-REFINED-PROMPT-archived; #203 stays)

Cascade Q-files moved to forensic-record archives (NOT deleted):

• 15 ζ-Q files → answered-questions/_pre-refined-prompt/<filename>-PRE-REFINED-PROMPT.md
• 2 MQ files (MQ-001 R2 + MQ-002) → meta-questions/_pre-refined-prompt/<filename>-PRE-REFINED-PROMPT.md
• MQ-000 STAYS (defines Phase 2 scope; not subject to re-ask)

What survives unchanged:

• MQ-000 (ζ.2 scope amendment)
• Pre-Phase-2 architecture state (arch-state v3.10 baseline; A-1..A-125)
• Pin_kind tagging convention
• 2026-05-01-inheritkitbench-explainer.md
• Master plan v1.10 + per-repo BUILD-PLAN.md v1.4-1.6
• 2026-04-30-architecture-rederivation-design.md v1.1
• Library indexed books + T-files + spike files + WR-bequest-class-2026-05-01 file (research-only-deferred or research-complete)
• brand-architecture v1.16
• Refined prompt itself
• richard-task #203 (test-suite Phase-5 deferred — independent)

Re-ask sequence (~10-11 hours over 8 sessions):

• S-1: ζ-Q1 (criteria-frame) — 60 min
• S-2: ζ-Q2 (architecture options G+) — 60 min
• S-3: ζ-Q3 (asset taxonomy) — 45 min
• S-4: ζ-Q4 (upper ontology) — 60 min
• S-5: ζ-Q5..ζ-Q8 (Shape X angles) — 120 min
• S-6: MQ-001 R2 + MQ-002 (Phase-2 closure) — 30 min
• S-7: ζ-Q9..ζ-Q14 (Phase 3 derivative) — 180 min
• S-8: ζ-Q15 (Wills.Will class schema) — 45 min

Methodological observations:

1. Largest reversal in audit history (17 Qs vs ζ-Q15..Q18 reversal of 4 Qs)
2. Anchoring-bias mitigation — full reversion forces fresh re-derivation
3. Off-github → git move is the enabling condition for first-class WR-file version-control
4. Reversal-readiness preserved
5. Foundations compound — Rich-judgement that load-bearing Qs benefit most from refined-prompt discipline
6. Cost ~10-11 hours bounded
7. Phase 3 module-authoring (ζ-Q16 Bequest framing was in flight; ✕.ε ★ recommended but not yet locked) pauses until re-ask completes

NOT a methodology pivot (staying with ζ.2 + refined prompt). NOT abandoning the audit (continuing under refined-prompt foundation). NOT distrust of prior locks (validated under sensitivity perturbations + Rich-confirmed). NOT permanent — locks may come out same or different.

NEXT: Rich opens new session for ζ-Q1 (criteria-frame 14-criterion) re-ask under refined prompt — the FIRST re-asked Q under the full refined-prompt discipline.

Phase 2 status post-revert: 0 ζ-Qs answered (re-ask in progress); MQ-000 stands; Phase-2 budget 18-30 Qs (was at 11 done pre-revert; now 0 + 1 MQ-000 = 1 Q in scope already-locked).