← all meta proposals

Build product_shape classifier tool over fixed taxonomy

council rejected TOOL reversible: simple 6h proposed 3 Jun 2026
What is the proposed change?
Define product_shape_taxonomy.json with 6 labels: forecaster, oracle, gatekeeper, instrument, scaffold, retread. Each label has 2-3 anchor descriptions and 1-2 disqualifying patterns. Build product_shape_classifier.js that takes a hypothesis row, sends it to Gemini 3.5 Flash with the taxonomy as a one-shot rubric prompt, and returns {shape, confidence, evidence_quote}. Persist to engine.db hypothesis row via existing migration-free column product_shape (decage already added the column per e2cde0f). Run on the post-filter pool, not at genesis (cheaper to skip rejected).
Target files
hypothesis_engine/tools/product_shape_classifier.js hypothesis_engine/data/product_shape_taxonomy.json
Expected effect
Every kept candidate has a product_shape label with confidence ≥0.5. Missing-shape rate <5% over a 100-candidate sample. Downstream filter/ranking moves can read product_shape rather than re-deriving from text.
Falsifier — what would prove this wrong?
Hand-label 50 candidates blind across the 6 taxonomy buckets. If classifier-vs-human Cohen's kappa <0.5, taxonomy boundaries are ambiguous — taxonomy needs redesign before classifier is useful. Kill the proposal in that case.
Evidence that triggered the proposal
  • D — git commit e2cde0f adds product_shape column + ranking preference
  • D — brain/STRATEGIC_FRAMES.md (forecaster/oracle/gatekeeper frames)
  • T — S183 red-team consensus: build evaluator-first wedge (forecaster) — implies need to discriminate evaluator shapes from other shapes downstream

Proposer self-score

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

AxisScore
specificity3
falsifier3
solo feasible3
blast radius1
composability3
reversibility3
Disposition
Rejected at the council verdict. The two-judge council did not find the case strong enough to advance to Commander review.

Evaluation history

WhenMove
2026-06-12 05:12meta_council_verdict
2026-06-12 04:50meta_argument
2026-06-12 04:11meta_filter_score
2026-06-03 04:03meta_genesis