420 Trust protocol¶
420 Trust is the canonical domain-scoped evidence and reputation-signal protocol for 420 Integrated. It records authenticated facts and measurements about subjects without creating a universal social, credit, political, financial or behavioral score.
Authority boundary¶
420 Identity answers who or what an entity is and which credentials it holds. 420 Trust records verifiable protocol history about a referenced subject. Neither protocol may silently absorb the other's authority. Trust signals do not create Wallet authority, custody, governance weight, validator weight, bridge authority, oracle-routing authority or Identity credentials.
Canonical components¶
TrustIssuerRegistry420owns issuer identity, operator, active state and epoch.TrustPolicyRegistry420owns domains, metrics, units, revisions and exact issuer-to-metric authorization.TrustSignalRegistry420owns immutable signals, corrections, revocations and replay protection.TrustAggregator420exposes deterministic per-subject/per-metric active sum and count.ITrust420is the stable consumer read boundary.
A consumer should bind to the exact metric/domain it needs. There is no protocol-wide score endpoint.
Signal lifecycle¶
A signal is accepted only for an active issuer whose current operator is authorized for the exact active metric. Signal IDs are single-use and the issuer/subject/metric/evidence tuple is replay-protected. Every signal pins the metric revision and issuer epoch applicable when recorded.
Corrections are append-only: the predecessor remains readable and points to its replacement. Revocation removes an active signal from aggregation exactly once without deleting history. Disabling an issuer blocks future mutations without erasing previously valid evidence.
Privacy and security¶
Canonical Trust state contains typed identifiers, values, commitments/references and timestamps needed for verification. Sensitive plaintext such as KYC records, addresses, health data, private review text or unrelated personal information belongs off-chain.
Security-critical invariants include exact issuer/metric authorization, replay protection, append-only correction, aggregate conservation, domain separation, Identity separation, prospective issuer disable and historical reconstructability.
Integration rules¶
Applications must never convert Trust output into hidden universal ranking semantics. A policy using Trust evidence must identify the exact metric, domain, revision and interpretation it applies. Derived UX scores remain application policy, not canonical Trust truth.
Indexer/Search/Analytics views are rebuildable projections. When authoritative interpretation matters, read canonical Trust state or ITrust420 and preserve source/version context.
Source model¶
The frozen implementation model remains docs/420-TRUST-V1-MODEL.md. This governed page owns 420Docs architecture placement and integration boundaries; it does not supersede protocol contracts or generated reference material.