Add pre-filter source-diversity gate (G_DIV)

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

What is the proposed change?

Insert a new pre-filter gate G_DIV between genesis and filter_score. Reject any single-cycle batch where >60% of proposals trace back to the same corpus source_id (computed from proposal.evidence.source_corpus + source). On reject, signal genesis to re-draw with the over-represented source temporarily down-weighted by 0.5 for that cycle only. Log to engine.db gate_events table with reason='source_skew'.

Target files

hypothesis_engine/gates/source_diversity.js hypothesis_engine/pipeline.js

Expected effect

S200 cross-memo dedup runs post-genesis; G_DIV catches the upstream skew before any filter compute is spent. Expect ~10-15% of cycles to trigger re-draw initially, dropping to <5% as corpus rotation matures.

Falsifier — what would prove this wrong?

Run G_DIV in shadow mode (log but do not reject) for 7 cycles. If shadow rejection rate is <2% or if rejected batches' downstream graduation rate is ≥ the kept batches', gate adds no signal and should be removed.

Evidence that triggered the proposal

D — e336df0 S200 diversity guard: cross-memo semantic dedup (post-genesis only — leaves upstream skew unaddressed)
D — S198_ENGINE_REAIM (rotating pain corpus implies need for cycle-level source balance)

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	2
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
2026-06-20 04:05	meta_filter_score
2026-06-20 04:03	meta_genesis