Add commodity-wedge shadow gate logging in runFatalObjection

filter rejected GATE reversible: simple 6h proposed 20 May 2026

What is the proposed change?

At the start of runFatalObjection in argument.js (the move already performing dual GPT+Gemini evaluation), add a pre-check Gemini Flash call that scores 5 binary axes: (1) incumbent_owns_workflow — does an incumbent SaaS already own this exact workflow step? (2) requires_crm_integration — is the value prop contingent on CRM or ERP data access? (3) deliverable_is_taxonomy — is the core output a classification, tagging, or audit report rather than a decision or action? (4) no_urgency_event_named — does the hypothesis name zero triggering events (renewal, audit, deadline, compliance date, new hire)? (5) no_owned_data_advantage — does the hypothesis require no proprietary data that incumbents structurally lack? Score = count of YES answers (0-5). Append result as JSONL to meta_engine/data/shadow_gate_verdicts/commodity_wedge_YYYY-MM-DD.jsonl with fields: hypothesis_id, score, per_axis_flags, timestamp. NEVER kill on this score — shadow mode only. No new DB columns written.

Target files

hypothesis_engine/moves/argument.js meta_engine/data/shadow_gate_verdicts/

Expected effect

After 60 days of shadow logging, score ≥3 correlates with ≥70% eventual KILL verdicts. On a 12-item calibration set containing 3 known commodity-wedge hypotheses and 4 known ROBUST graduates, all 3 commodity-wedge items score ≥3 and all 4 ROBUST items score ≤1.

Falsifier — what would prove this wrong?

If after 60 days the shadow gate score ≥3 shows no statistically significant correlation with actual kill rate (chi-square p > 0.1 against the verdict distribution in engine.db), the 5 binary axes do not predict commodity-wedge failure and must be replaced before any promotion to a killing gate.

Evidence that triggered the proposal

Corpus D: META_ENGINE_S158_RED_TEAM_SYNTHESIS Survivor E — 'commodity-wedge shadow gate, 6h, SIMPLE shadow, approved pending implementation'
Corpus E: kill distribution — episodic-vs-recurring tension and audit-shaped kills recur across multiple sessions with no pre-argument detection; these reach council at ~$0.06/run when a ~$0.002 pre-check could flag them
Corpus D: V2_FILTER_DESIGN_v2.3.md A10 (workflow_embedment) and A6 (acute_pain_not_candy) — commodity-wedge hypotheses structurally fail both axes but current filter only scores 5 v1 axes; shadow gate provides signal ahead of full A6/A7 wiring

Proposer self-score

The proposer scored its own draft on these axes (0-3 each) before submitting.

Axis	Score
specificity	3
falsifier	3
solo feasible	3
blast radius	3
composability	2
reversibility	3

Disposition

Rejected by filter_score. The proposal did not meet the bar for specificity, falsifiability, or solo-feasibility.

Evaluation history

When	Move
2026-05-20 04:14	meta_filter_score
2026-05-20 04:10	meta_genesis