Metrics
What the knowledge base currently holds, and how evidence, belief-gaps, and hype are distributed across it. How these are defined →
Coverage
| interventions | 33 | |
| outcomes | 20 | |
| claims | 123 | |
| beliefs | 24 | |
| interactions | 14 | |
| folkTerms | 10 |
Evidence grades
| A | 5 | |
| B | 33 | |
| C | 63 | |
| D | 9 | |
| gradeless | 13 |
Claim states
| well supported | 26 | |
| mixed | 53 | |
| weak | 31 | |
| null result | 12 | |
| untested | 0 | |
| safety flagged | 1 |
Belief-vs-evidence gaps
| belief ahead | 20 | |
| belief behind | 1 | |
| belief contradicts | 1 | |
| belief matches | 2 |
Hype positions
| emerging | 3 | |
| surging | 6 | |
| evidence catching up | 1 | |
| settled | 7 | |
| declining | 2 | |
| debunked | 2 |
Interaction severities
| contraindicated | 1 | |
| serious | 3 | |
| moderate | 5 | |
| uncertain | 2 | |
| minor | 3 |
Answer-shell performance (last 30 days)
Production LCP for answered Ask pages, collected only in fixed anonymous duration buckets. Target: at least 75% at or below 2.5 seconds, with at least 25 publishable observations.
Insufficient production sample (0/25 observations).
Freshness
123 claims carry a review date — 123 fresh, 0 past the review cadence (flagged for re-review).
Change feed
81 recorded changes · 74 in the last 90 days.
Trust KPIs
The numbers that make us look bad publish here too — see independence.
| SLA compliance (claims within re-review window) | 100% |
| Credential verifiability (reviewers verifiable against a primary registry) | 50% (1/2) |
| Corrections published | 0 — see corrections log |
Embedded evidence (last 30 days)
EMBEDDABLE-EVIDENCE §2 — co-citation surface, not referring domains. Every embed is self-updating from the live claim, so correction propagation is definitional: there is no stale copy to catch up.
| Distinct claims embedded | 1 |
| Total embed views | 1 |
Branded search (last 12 weeks)
ACQUISITION-ANALYTICS §3 Tier 4 — the zero-click-proof signal. Imported from Google Search Console weekly.
No GSC data imported yet — POST to /api/v1/admin/branded-search to seed.
Trigger → publish latency
FRESHNESS-ENGINE §5 target — median ≤ 14 days (general), ≤ 72 hours (safety). Derived from the change ledger.
No dated triggers in the ledger yet.