L4 Wave1: negative-assortative crossover + novelty-debt advisory + mode-lock replay test#22
Merged
Conversation
…icy path lowest_similarity global donor policy already exists in crossover.py and is reachable via CACrossoverConfig; add a regression test proving it selects the most-distant donor (negative-assortative) vs the default highest_search_quality. Zero production change; default policy unchanged. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ate) assess_seed_coverage emits a pure-derived novelty_debt dict (overrepresented families + missing-family count) reusing existing family counters and the 0.35 concentration threshold. advisory_metadata_only_no_gate: does not alter needs_more_seed/status gating. No new class/registry/ledger. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
… regression test SemanticLooping/DiversityCollapse already routes under-explored families through selection_pressure into ParentSelector and deterministic reproduction; add an end-to-end test asserting a cold under-explored parent is selected, written into MutationPlan.parent_ids, and produces offspring. Zero production change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
L4 Wave1 — landing immediately net-positive directions
Following L1-L3 (#21), this wave lands the three directions that can be landed immediately by reusing existing mechanisms and adding tests. Production net +36 lines (seed_coverage), the rest tests.
Three landings
lowest_similarityglobal donor policy already exists in production and is reachable viaCACrossoverConfig. Zero production change; regression test added to prove it selects the most distant donor, contrasting with the defaulthighest_search_quality. Default policy unchanged.assess_seed_coveragegains a pure-derivednovelty_debtfield (over-represented families + missing family count), reusing existing family counts and the 0.35 concentration threshold. Advisory metadata only, no gate:needs_more_seed/statussemantics unchanged, no new class/registry/ledger.Invariants
Verification
Note: descriptor-collapse direction closed (falsified by data)
The "change the descriptor to fix family collapse" direction was closed after offline measurement falsified it: the exploration run does not collapse (462 candidates -> 333 cells, 323 families); removing the path bucket has zero effect, and routing the descriptor cell actually collapses it (-> 71 families). Collapse is run x descriptor-binding dependent, not a run-agnostic defect.
🤖 Generated with Claude Code