Pre-argument source_corpus diversity gate

council rejected GATE reversible: simple 3h proposed 15 Jun 2026

What is the proposed change?

Insert a pre-argument gate that inspects the top-K (K=batch size for argument move) filter_kept candidates. If >70% of them share the same source_corpus value, reject the lowest-scoring duplicates and promote the next-highest candidate from an under-represented corpus until the batch is <=70% mono-source. Log gate firings to engine.db with a new gate_log row (no hypotheses-table change).

Target files

hypothesis_engine/moves/argument.js hypothesis_engine/gates/source_diversity.js

Expected effect

When Corpus T floods (as in S183-S186 forecaster arc), the gate fires on ~15-25% of batches and keeps Corpus E/D candidates from being starved out of argument move. Cross-corpus citation rate in graduated hypotheses rises measurably.

Falsifier — what would prove this wrong?

Run gate in shadow mode for 14 days alongside unmodified pipeline. If shadow vs live council_verdict='kept' rate differs by <2 percentage points AND graduated-hypothesis source_corpus mix is statistically indistinguishable, gate provides no signal and should be removed.

Evidence that triggered the proposal

D — S183_EVIDENCE_FORECASTER_WEDGE.md (single-corpus dominance during pivot)
E — engine.db filter_kept distribution by source_corpus over last 30 days

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	1
composability	2
reversibility	3

Disposition

Rejected at the council verdict. The two-judge council did not find the case strong enough to advance to Commander review.

Evaluation history

When	Move
2026-06-15 04:10	meta_council_verdict
2026-06-15 04:07	meta_argument
2026-06-15 04:04	meta_filter_score
2026-06-15 04:03	meta_genesis