Model reputation as counterparty assurance gradient across four tiers

Add research on the journey from gamable opinion signals (reviews, ratings)
through observed metrics (PAYDEX, SLA stats), financial commitments (bonds,
escrow), and adjudicated outcomes (arbitration, courts). Resolve OpenQuestions:
no Reputation entity; use Reputation Signal, Performance Evidence, Commercial
Commitment, and Adjudication Outcome with Counterparty Assurance Gradient pattern.

OpenQuestions.md

@@ -201,11 +201,27 @@ reporting volatility vs. enduring CDD collection obligations.
 
 ### Reputation as canon concept
 
-**Status:** Open — leaning toward Evidence Source aggregation.
+**Status:** Resolved — tiered Evidence Source pattern; no Reputation entity.
 
-Credit scores (PAYDEX), performance history, and repeat-play trust may not need
-a separate Reputation entity if modeled as Evidence Source + Trust Relationship
-with temporal scope.
+**Decision:** Model reputation as a **Counterparty Assurance Gradient** across
+four tiers — opinion (Reputation Signal), observed (Performance Evidence),
+committed (Commercial Commitment), adjudicated (Adjudication Outcome). **Trust
+Relationship** cites `assurance_basis`; do not equate star ratings with bonds or
+court judgments.
+
+**Rationale:** Star ratings are gamable and scope-local; PAYDEX and SLA metrics
+are observed evidence; bonds and escrow are Commercial Commitments; arbitration
+and courts produce Adjudication Outcomes. A single "reputation" root would collapse
+enforceability and attribution differences.
+
+**Citations:**
+
+- `research/commercial-identity/reputation-assurance-gradient.md`
+- `research/commercial-identity/commercial-trust-binding-theory.md`
+- `research/commercial-identity/duns-commercial-credit-identity.md`
+
+**Remaining nuance:** Whether segregated platform escrow is `committed` without a
+separate contract artifact; cross-platform reputation portability via Synonymity.
 
 ### Registry identifier subtype