Define the entity-search relevance model: which signals rank results (citation strength, Fact consensus, freshness, completeness) versus which are forbidden (raw engagement/watch-time), plus the anti-gaming rules and change governance.
Why this matters
G-010 covers recommendation and G-055 the external AEO strategy, but neither defines on-platform search ranking — and there is no search on the site today. The signal choice is load-bearing for both quality and values. Open decisions:
- Permitted signals — citation strength (Entry 020), Fact consensus level (G-060), freshness/recency (a stated edge over Fandom, Entry 030), graph completeness, contributor standing (G-051).
- Forbidden signals — raw engagement, click-through, watch-time; the explicit line that prevents recreating engagement-optimized ranking.
- Anti-gaming — defenses specific to search: citation-padding, consensus-gaming, edit-spam to boost freshness (ties to anti-flood, Entry 024, and data quality, G-026).
- Change governance — whether ranking changes ship unilaterally, are published, or are constitutional-tier; and how they're A/B-evaluated (G-064) with quality guardrails (G-059).
- Infrastructure — relationship to the OpenSearch/discovery layer named in the storage architecture (Entry 023).
Related
- Entry 032 — the bridge strategy that surfaces this gap
- Entry 030 — discoverability posture (freshness as an edge)
- G-055 — discoverability moat & migration (the strategic parent)
- G-010 — discoverability & recommendation (the sibling rec surface)
- G-026 — Fact-graph data quality (the signal substrate)
- G-059 — gamified contribution (quality guardrails for ranking A/Bs)