Pre-registered ▤ Framework

Retiring never-surfaced frameworks does not reduce coverage

hypothesis + method published before the result.

Hypothesis

The 6 frameworks that never surface on benchmark v1 (SS-FW-005/006/007/009/018/023) can be retired or merged without reducing recommendation coverage or quality.

Method

For each candidate: check retrieval reachability; if unreachable, remove from the active set and re-run evaluate.js; confirm pattern coverage and answer recall do not fall vs baseline. Structural merge-candidates evaluated separately.

Dataset

benchmark-v1 + knowledge graph (25 frameworks, 30 patterns).

Metrics

pattern coverage	≥ 29/30
answer recall	≥ baseline

Result

Pending — pre-registered; the result will be published here.

Decision

REGISTERED. Usage signal is live (framework-evolution.js flags 6 never-surfaced); the retire decision awaits the coverage re-run.

Lessons learned

— (recorded when the experiment completes)

Confidence

N/A

Evidence

evolution report — bench/history/evolution-*.json

← all experiments