experiment: test live no-confirmation closure guard

docs/current-handoff.md

@@ -3467,3 +3467,28 @@ Read-only structural audit completed. `apply-proposal.js` is 4,480 lines with on
 **Decision:** RETURN TO REASONING WORK FIRST — no implementation directive this session. Next extraction should happen when schema changes naturally touch the validation layer, avoiding forced refactoring without external pressure to ensure correctness.
 
 Medium-term target for apply-proposal.js: ~4,250-4,300 lines (clear orchestration file, not tiny wrapper-only).
+
+
+---
+
+### Experiment 60B.71 — No-Confirmation Premature-Closure Guard (PASS)
+
+**Date:** 2026-08-14
+**Branch:** `feature/decision-sufficiency-module-v0.44`
+**Experiment commit:** a5b71ad experiment: map remaining apply proposal boundaries
+**Result:** A + E — CONSERVATIVE GUARD CONFIRMED; NEW MATERIAL FACTOR IDENTIFIED
+
+Inverses 60B.66: the enterprise-customer-signing factor resolves, but raw user answer lacks explicit "no other material uncertainty" confirmation.
+
+Key outcomes:
+- Decision `n_product_launch_decision` **remains unknown** (KEPT OPEN) — neither model proposal nor deterministic closure fired
+- Customer `n_enterprise_customer_signing` resolves to resolved with negative meaning preserved in reason text
+- `finalActiveUnknownNodeId = "n_product_launch_decision"` (decision itself becomes the genuine unresolved target)
+- `finalSelectedQuestion` = non-null, decision_threshold strategy — "What outcome would demonstrate enough value to justify launching?"
+- Model introduces new observation node `n_revised_launch_year_revenue` (£500k/year) and edge e-customer-confirmation-to-revenue — quantified financial context, not spurious uncertainty
+- Decision NOT in proposal updatedNodes, NOT in resolvedUnknownNodeIds — zero closure signals from any source
+
+The no-confirmation guard works live post-refactor. Parent decision stays open when raw confirmation is absent, matching the expected conservative behavior of experiment 60B.64's gate.
+
+
+---