Median-zero-on-any-axis hard kill in filter_score

filter rejected GATE reversible: simple 2h proposed 16 Jun 2026

What is the proposed change?

The current two-consecutive-zero kill (line 214-220) only fires when the SAME axis scores 0 on two consecutive runs. After all 3 runs complete (line 244-247 region), compute median per axis across the 3 runs. If ANY axis has median=0 across the 3 runs (e.g. midpoints [0, 1, 0] → median 0), set phase='killed' with kill_reason='median_zero_on_<axis>' BEFORE composite_rank_score gates graduation. Important: the existing composite score 'median total - 0.5*IQR + 0.3*signal_count' can still elevate hypotheses with an axis-zero median if other axes score 3s — this closes that gap.

Target files

hypothesis_engine/moves/filter_score.js

Expected effect

Hypotheses with a load-bearing weakness on ONE axis but high scores elsewhere will be killed regardless of composite. Estimate: 5-12% of currently-ranked hypotheses have a median-0 axis (visible by reparsing moves.scores_json). Those hypotheses currently graduate to 'scored'/'ranked' on composite strength alone.

Falsifier — what would prove this wrong?

Replay 60 days of filter_score history. If 0 historically-ranked hypotheses would have been killed by this rule, the existing two-consecutive-zero rule is already catching everything and this is redundant. If >30% would have been killed, the rule is too harsh and the engine starves; reject in favor of a softer 'median ≤ 0.5' threshold.

Evidence that triggered the proposal

E — hypothesis_engine/moves/filter_score.js:7 ('Two-consecutive-zero categorical kill preserved here' — only consecutive, not pattern [0,1,0])
E — hypothesis_engine/moves/filter_score.js:211 ('compositeRankScore = medianTotal - 0.5 * iqrTotal + 0.3 * h.signal_count' — composite can mask single-axis collapse)

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	3
reversibility	3

Disposition

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

Evaluation history

When	Move
2026-06-16 04:07	meta_filter_score
2026-06-16 04:04	meta_genesis