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.
2026-06-21 22:57:28 +02:00
parent 0741a2e3a5
commit bd272151af
10 changed files with 383 additions and 11 deletions
--- a/DownstreamRecommendations.md
+++ b/DownstreamRecommendations.md
@@ -126,4 +126,9 @@ later explicit package is extracted.
 - Link registry identifiers for the same entity via Synonymity Assertion when
  multiple registries describe one Organization/Legal Entity.
 - Separate CRM Account and Stripe Customer as Commercial Records; never merge with login Account.
- Use qualified credentials (eIDAS seal, VC) as Evidence for Commercial Commitment where applicable.
+- Use qualified credentials (eIDAS seal, VC) as Evidence for Commercial Commitment where applicable.
+- Map reviews and star ratings to Reputation Signal (opinion tier); never merge with credit scores or legal outcomes.
+- Map PAYDEX, SLA metrics, and credit bureau data to Performance Evidence (observed tier).
+- Map bonds, escrow, and signed SLAs to Commercial Commitment (committed tier).
+- Map arbitration awards and court judgments to Adjudication Outcome (adjudicated tier).
+- Trust Relationship projections must cite assurance_basis tier; weight opinion weak by default.
--- a/OpenQuestions.md
+++ b/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

--- a/canon/CanonicalGlossary.md
+++ b/canon/CanonicalGlossary.md
@@ -327,6 +327,11 @@ relationships, policies, or configuration in a scope.
 A relationship where one actor, issuer, verifier, system, or scope relies on
 another for claims, identifiers, credentials, or decisions.

+For commercial counterparty reliance, cite an **assurance_basis**: which tier of
+the **Counterparty Assurance Gradient** supports the trust (opinion signal,
+observed metric, commercial commitment, or adjudication outcome). Do not treat
+weak opinion-tier evidence as equivalent to committed or adjudicated assurance.
+
 ## Synonymity Assertion

 A scoped, evidenced assertion that two or more identifiers, records, accounts,
@@ -349,6 +354,67 @@ schema.org sameAs (weak by default).
 A source, document, event, issuer, import, observation, or verification process
 supporting a claim, relationship, or synonymity assertion.

+Recommended commercial field: `assurance_tier` — `opinion` | `observed` |
+`committed` | `adjudicated` (see **Counterparty Assurance Gradient**).
+
+## Counterparty Assurance Gradient
+
+A four-tier pattern for how counterparties earn reliance, from weak social proof
+to enforceable outcomes. Higher tiers do not erase lower tiers but constrain how
+much weight lower tiers may carry.
+
+| Tier | Label | Examples | Canon anchor |
+| --- | --- | --- | --- |
+| 1 | Opinion | Star ratings, reviews, karma, badges | Reputation Signal |
+| 2 | Observed | PAYDEX, SLA stats, chargeback rate, KYC pass | Performance Evidence |
+| 3 | Committed | Contract, bond, escrow, SLA penalties, payment mandate | Commercial Commitment |
+| 4 | Adjudicated | Arbitration award, court judgment, regulator enforcement | Adjudication Outcome |
+
+Identity attribution strengthens up the gradient: opinion may attach to Persona
+or platform Profile; adjudication attaches to Legal Entity with Registry Identifier.
+
+## Reputation Signal
+
+An **Evidence Source** with `assurance_tier: opinion` — crowd-sourced or
+platform-computed social proof attributed to an actor, profile, or commercial record.
+
+Examples: star ratings, customer reviews, marketplace feedback scores, community karma.
+
+Default trust strength: weak. Subject to gaming (fake reviews, Sybil accounts,
+review bombing). Scope-local to the issuing platform unless portable via explicit
+Synonymity Assertion. Not a **Commercial Commitment** and not interchangeable with
+credit metrics or legal outcomes.
+
+## Performance Evidence
+
+An **Evidence Source** with `assurance_tier: observed` — metrics or attestations
+grounded in observable transactions, registry facts, or third-party measurement.
+
+Examples: PAYDEX, business credit score, on-time delivery rate, chargeback ratio,
+sanctions-screen result, LEI renewal status.
+
+Usually links to **Commercial Record**, **Organization**, or **Registry Identifier**.
+Supports medium-to-strong **Trust Relationship** when issuer is authoritative.
+
+## Adjudication Outcome
+
+An **Evidence Source** with `assurance_tier: adjudicated` — formal dispute or
+enforcement result from arbitration, courts, regulators, or binding platform
+resolution processes.
+
+Examples: arbitration award, court judgment, regulatory consent order, binding
+chargeback ruling, bankruptcy filing.
+
+May trigger **Commercial Commitment** lifecycle change (breached, fulfilled) and
+**Trust Relationship** revocation. Attaches to **Legal Entity** / **Organization**
+actors, not merely to profiles or opinion aggregates.
+
+## Non-Canonical Convenience Term: Reputation
+
+`Reputation` is overloaded (reviews, credit, brand, legal standing). Do not use
+as a canonical root. Resolve to **Counterparty Assurance Gradient** tier and the
+specific Evidence Source or Commercial Commitment before modeling.
+
 ## Lifecycle State

 The current state of a record, account, relationship, credential, claim, or
--- a/model/ConceptualModel.md
+++ b/model/ConceptualModel.md
@@ -226,6 +226,29 @@ Identity representations vary in persistence based on commercial stake:
 Commercial binding does not merge layers. It increases assurance requirements and
 lifecycle rigor on the records and relationships already in the model.

+## Counterparty Assurance Gradient
+
+Counterparty reliance escalates through four evidence tiers. Model each tier
+explicitly; do not collapse into a single "reputation score."
+
+| Tier | Assurance | Typical evidence | Canon elements |
+| --- | --- | --- | --- |
+| 1 — Opinion | Weak; gamable | Star ratings, reviews, karma | Reputation Signal (Evidence Source) |
+| 2 — Observed | Evidence-based | PAYDEX, SLA metrics, KYC outcome | Performance Evidence (Evidence Source) |
+| 3 — Committed | Financial / contractual | Bond, escrow, signed SLA, mandate | Commercial Commitment + Evidence |
+| 4 — Adjudicated | Legal / binding dispute | Arbitration, judgment, enforcement | Adjudication Outcome (Evidence Source) |
+
+**Trust Relationship** should cite `assurance_basis` (tier + evidence references).
+Escalation path: dispute on committed terms → automated platform resolution →
+contractual ADR → courts. De-escalation via supersession lifecycle, not silent delete.
+
+**Attribution rule:** opinion may bind to Persona/Profile in a platform Scope;
+observed and committed tiers prefer Commercial Record + Registry Identifier;
+adjudicated tiers require Legal Entity / Organization actors.
+
+No standalone Reputation entity — aggregate downstream if needed; preserve tier
+provenance in canon.
+
 ## Scenario Gaps

 No scenario requires glossary or principle changes that the current model
--- a/research/CorpusIndex.md
+++ b/research/CorpusIndex.md
@@ -66,6 +66,7 @@ The repository is focused on research and terminology. The corpus should collect
 - `salesforce-crm-commercial-record.md`
 - `beneficial-ownership-kyc-boi.md`
 - `registry-identifier-subtypes.md`
+- `reputation-assurance-gradient.md`

 ## Source Note Template

--- a/research/commercial-identity/commercial-identity-synthesis.md
+++ b/research/commercial-identity/commercial-identity-synthesis.md
@@ -53,7 +53,9 @@ From commercial theory and practice:
 1. **Attribution**: counterparties know **who** bears liability (legal person, BO, agent).
 2. **Commitment**: contracts, subscriptions, and payment authorizations create **costly exit**.
 3. **Evidence**: KYC, LEI, registry credentials, and credit files provide **verifiable history**.
-4. **Reputation**: PAYDEX, performance history, and repeat play increase **trust without re-verification**.
+4. **Reputation / assurance**: tiered reliance from opinion signals (reviews) through
+   observed metrics (PAYDEX) to committed stakes (bonds) and adjudicated outcomes
+   (courts) — see **Counterparty Assurance Gradient**.
 5. **Enforcement**: law of agency and contract makes promises **actionable** beyond platform ToS.

 Trust Relationship in canon should often be **justified by** Commercial Relationship +
@@ -87,6 +89,8 @@ Commercial Commitment + Evidence, not declared ad hoc.
  Organization/Legal Entity for KYC/CDD (not Ownership subtype).
 - **Registry Identifier** and **Proxy Commercial Identifier** — Reference layer
  subtypes with authority class, ICD scheme, and renewal lifecycle.
+- **Counterparty Assurance Gradient** — opinion → observed → committed →
+  adjudicated; Reputation Signal, Performance Evidence, Adjudication Outcome.

 ### Unchanged roots

@@ -116,7 +120,8 @@ Model as lifecycle events, not silent merges:
 - Payment Credential vs. authentication Credential boundary in PCI contexts.
 - Smart contracts and automated Commercial Commitment lifecycle.
 - Synonymity strength bands for LEI ↔ DUNS ↔ company reg crosswalks.
- Reputation as first-class canon concept vs. Evidence Source aggregation.
+- Cross-platform reputation portability (Synonymity between Reputation Signals).
+- Smart-contract / oracle outcomes — observed vs. adjudicated tier placement.
 - Standard `control_basis` enum for Beneficial Ownership across jurisdictions.

 ## Source Notes in This Stack
@@ -130,6 +135,7 @@ Model as lifecycle events, not silent merges:
 - `salesforce-crm-commercial-record.md`
 - `beneficial-ownership-kyc-boi.md`
 - `registry-identifier-subtypes.md`
+- `reputation-assurance-gradient.md`
 - `../commercial-subscription/b2b-saas-subscriber-tenancy.md`
 - `../commercial-subscription/stripe-customer-billing.md`

--- a/research/commercial-identity/commercial-trust-binding-theory.md
+++ b/research/commercial-identity/commercial-trust-binding-theory.md
@@ -95,7 +95,8 @@ sharing, and enforceable promises.
 | --- | --- |
 | Fluid identity | Persona / Scoped Identifier without Commercial Commitment |
 | Commercial binding | Commercial Commitment on Commercial Relationship |
-| Reputation capital | Evidence Source history + Trust Relationship |
+| Reputation capital | Performance Evidence history + Trust Relationship (assurance_basis) |
+| Star ratings / reviews | Reputation Signal (opinion tier) |
 | Counterparty identification | Commercial Record + Legal Entity + Identifiers |
 | Contractual promise | Commercial Commitment (contract subtype) |
 | Assurance mechanism | Assurance Level + Evidence Source |
@@ -107,7 +108,7 @@ sharing, and enforceable promises.
 - Should Commercial Commitment be a Relationship subclass or metadata on
  Commercial Relationship?
 - How should fluid-to-bound transitions be modeled (trial → paid, anonymous → KYC)?
- Does reputation warrant a canonical concept or remain Evidence Source aggregation?
+- Resolved: tiered Evidence Source pattern — see `reputation-assurance-gradient.md`.

 ## References

--- a/research/commercial-identity/duns-commercial-credit-identity.md
+++ b/research/commercial-identity/duns-commercial-credit-identity.md
@@ -77,7 +77,11 @@ interest among counterparties.

 ## Open Questions

- Should credit scores be canonical metadata or strictly downstream risk signals?
+- Numeric score vs. assurance_tier band mapping in downstream risk engines.
+
+## Resolved (see reputation-assurance-gradient.md)
+
+- PAYDEX and credit scores → **Performance Evidence** (observed tier), not opinion-tier Reputation Signal.

 ## References

--- a/research/commercial-identity/reputation-assurance-gradient.md
+++ b/research/commercial-identity/reputation-assurance-gradient.md
@@ -0,0 +1,241 @@
+# Reputation and Counterparty Assurance Gradient
+
+## Source Type
+
+Cross-domain synthesis. Online reputation systems, credit reporting, contract
+bonding theory, payment dispute automation, and alternative dispute resolution
+(ADR) / litigation practice.
+
+## Domain
+
+How counterparties move from weak social proof to enforceable commercial reliance
+— and how identity-canon should model that journey without collapsing tiers.
+
+## Why This Source Matters
+
+"Reputation" is overloaded: a five-star Yelp review, a D&B PAYDEX score, a
+performance bond, and a court judgment all influence whether a counterparty is
+trusted — but they differ radically in **evidence quality**, **gaming risk**,
+**attribution strength**, and **enforceability**. Software often stores them in
+one "rating" field. Canon must preserve the gradient so downstream systems do not
+treat gamable opinion as legal fact or ignore contractual stakes already modeled
+elsewhere.
+
+## The Assurance Gradient (Journey)
+
+Counterparty assurance typically escalates through four tiers. Higher tiers do
+not replace lower ones; they **constrain** how much weight lower tiers may carry
+for a given decision.
+
+```text
+Tier 1  OPINION        Star ratings, reviews, karma, badges
+        (weak/gamable) Low cost to fake; Sybil-prone; scope-local
+
+Tier 2  OBSERVED       PAYDEX, on-time %, chargeback rate, audit logs,
+        (evidence)     verified transaction history, KYC outcome
+
+Tier 3  COMMITTED      Contract, bond, escrow, guarantee, insurance,
+        (financial)    SLA penalties, payment mandate, subscription lock-in
+
+Tier 4  ADJUDICATED    Arbitration award, court judgment, regulator action,
+        (legal)        enforced settlement, lien, bankruptcy filing
+```
+
+### Tier 1 — Opinion signals (weak, gamable)
+
+**Examples:** Amazon/Yelp star ratings, eBay feedback scores, Stack Overflow
+reputation, Uber driver rating, Trustpilot reviews, Airbnb host score.
+
+**Properties:**
+
+- **Low cost of manipulation** — fake reviews, review bombing, sock puppets,
+  Sybil accounts (Jøsang reputation attack taxonomy).
+- **Scope-local** — reputation on eBay does not transfer to Etsy without
+  explicit portability (reputation bank problem).
+- **Voluntary participation bias** — satisfied and angry customers over-represent;
+  silent majority absent.
+- **Identity attribution weak** — reviewer may be unverified persona; linkage to
+  Natural Person or Organization often absent.
+- **Economic effect real but bounded** — eBay seller ratings correlate with price
+  premium, but platforms add escrow and buyer protection because ratings alone
+  insufficient.
+
+**Canon mapping:** **Reputation Signal** — an **Evidence Source** with
+`assurance_tier: opinion`. Attach to **Profile**, **Commercial Record**, or
+**Actor** with explicit **Scope** (platform namespace). Default synonymity and
+trust strength: **weak**. Do not promote to Commercial Commitment.
+
+**Gaming defenses (downstream):** verified-purchase flags, rate limits, graph
+analysis, moderation — model as separate Evidence Source metadata, not as tier
+upgrade by itself.
+
+### Tier 2 — Observed metrics (evidence-based)
+
+**Examples:** D&B PAYDEX, business credit scores, platform completion rate,
+on-time delivery statistics, SLA attainment dashboards, chargeback ratio,
+sanctions-screen clear result, KYC pass, LEI renewal status.
+
+**Properties:**
+
+- **Grounded in observable events** — payment dates, shipment scans, registry
+  lookups, transaction logs.
+- **Stronger attribution** — usually tied to **Registry Identifier**, **Commercial
+  Record**, or verified **Account** history.
+- **Third-party or platform issuer** — D&B, credit bureaus, marketplace operator,
+  KYC vendor acts as **Evidence Source** issuer.
+- **Still revisable** — metrics update; disputes may correct; not legally
+  conclusive.
+- **Monitoring lifecycle** — ongoing CDD and PAYDEX refresh mirror **Lifecycle
+  State** on evidence, not one-time truth.
+
+**Canon mapping:** **Performance Evidence** — **Evidence Source** with
+`assurance_tier: observed`. Link to **Commercial Record** / **Organization** via
+**Registry Identifier** or **Commercial Relationship**. Supports **Trust
+Relationship** with medium-to-strong confidence when issuer is authoritative.
+
+### Tier 3 — Committed stakes (contractual / financial)
+
+**Examples:** Performance bonds, surety bonds, letters of credit, escrow deposits,
+service-level agreements with liquidated damages, signed MSAs, active subscription
+with payment mandate, insurance certificates, qualified electronic seals on
+contracts (eIDAS).
+
+**Properties:**
+
+- **Costly to breach** — Klein-Leffler bonding: quality assurance through
+  market forces when reputation alone insufficient; hostages and penalties.
+- **Explicit parties** — **Legal Person** / **Organization** actors bound via
+  **Commercial Commitment** and **Representation** chains.
+- **Automated enforcement partial** — smart-contract escrow, Stripe retention,
+  auto-renewal billing, SLA breach triggers — automation executes **committed**
+  rules without yet reaching court.
+- **Identity stakes rise** — counterparties need stable **Registry Identifier**,
+  **Commercial Record**, and often **Beneficial Ownership Relationship** because
+  liability is real.
+
+**Canon mapping:** **Commercial Commitment** (contract, subscription, payment
+mandate, bond) with **Evidence Source** attesting execution. Assurance tier:
+`committed`. **Trust Relationship** here should cite the commitment ID, not
+opinion aggregates.
+
+**Distinction:** A five-star rating is not a bond. A bond is not a review.
+Model separately; combine only in downstream risk engines with explicit weighting.
+
+### Tier 4 — Adjudicated outcomes (automated dispute → legal resolution)
+
+**Escalation path:**
+
+1. **Platform automation** — chargeback dispute rules, marketplace arbitration
+   (eBay Money Back Guarantee), payment processor outcome.
+2. **Contractual ADR** — mandatory arbitration clause (AAA, ICC, JAMS); neutral
+   award binding per contract and statute.
+3. **Courts** — breach of contract, fraud, collections, judgment lien, bankruptcy.
+
+**Properties:**
+
+- **Third-party or state authority** — arbitrator, court, regulator issues outcome.
+- **High attribution** — parties identified in proceeding; ties to **Legal Entity**.
+- **Enforceable beyond platform** — judgments attach to legal persons; credit
+  reporting may follow.
+- **Lifecycle durable** — satisfied, appealed, vacated, enforced — explicit
+  **Lifecycle State**.
+
+**Canon mapping:** **Adjudication Outcome** — **Evidence Source** with
+`assurance_tier: adjudicated`. May trigger **Commercial Commitment** state change
+(breached, fulfilled), **Trust Relationship** revocation, or **Lifecycle State**
+on **Commercial Record**. Do not model as "bad review."
+
+## Cross-Tier Dynamics
+
+| Transition | What changes | Canon event |
+| --- | --- | --- |
+| Opinion → Observed | Platform verifies purchase; metric computed from logs | New Performance Evidence; optional Synonymity link reviewer Account to transaction |
+| Observed → Committed | Parties sign contract / post bond | Commercial Commitment created; Trust Relationship cites commitment |
+| Committed → Adjudicated | Breach → ADR/court | Adjudication Outcome Evidence; commitment lifecycle update |
+| Adjudicated → Observed | Judgment paid; credit file updated | Performance Evidence refresh (credit bureau) |
+
+**De-escalation:** Adjudicated fraud finding may **invalidate** opinion signals
+(moderation) but should not silently delete Evidence — supersede with lifecycle.
+
+**Identity coupling:** Higher tiers require stronger **actor attribution**.
+Opinion may attach to **Persona**; adjudication attaches to **Legal Entity** +
+**Registry Identifier**.
+
+## Relationship to Existing Canon
+
+| Concept | Role in assurance gradient |
+| --- | --- |
+| Evidence Source | Carrier for all tiers; use `assurance_tier` metadata |
+| Trust Relationship | Counterparty reliance; must cite tier basis |
+| Commercial Commitment | Tier 3 anchor |
+| Commercial Relationship | Scope for which assurance applies |
+| Registry Identifier | Attribution for tiers 2–4 |
+| Beneficial Ownership Relationship | Liability chain for tier 3–4 entity customers |
+| Assurance Level (NIST) | Orthogonal — identity/auth proofing, not commercial performance |
+| Synonymity Assertion | Link platform persona to legal entity when tiers mix |
+
+## Reputation Systems Literature (Practical)
+
+Jøsang survey and Resnick criteria for effective reputation systems:
+
+1. Long-lived entities with predictable future interaction.
+2. Capture and distribute feedback from prior interactions.
+3. Use feedback to guide trust.
+
+**Implication for canon:** Tier 1 only works when **Scope** is stable and
+interaction history is modeled as Evidence with temporal bounds. Reputation
+**capital** (economic value of good history) is aggregate **Performance Evidence**
+over time — not a separate ontological root.
+
+**Attacks:** self-promotion, Sybil, slandering, whitewashing — map to
+`integrity_risk` metadata on opinion-tier Evidence; downstream concern, but canon
+should flag tier-1 default weakness.
+
+## Candidate Canonical Mappings
+
+| Source artifact | Canonical mapping |
+| --- | --- |
+| Star rating / review | Reputation Signal (Evidence Source, tier: opinion) |
+| Verified purchase review | Reputation Signal + Performance Evidence link |
+| PAYDEX / credit score | Performance Evidence (tier: observed) |
+| SLA dashboard | Performance Evidence on Commercial Relationship |
+| Signed MSA / bond | Commercial Commitment + Evidence Source (tier: committed) |
+| Escrow release | Commercial Commitment lifecycle event |
+| Arbitration award | Adjudication Outcome (tier: adjudicated) |
+| Court judgment | Adjudication Outcome + may affect Legal Entity lifecycle |
+| "Trust score" UI | Downstream projection — not canonical root |
+
+## Resolved Canon Question
+
+**Do not add Reputation as a first-class entity.**
+
+Instead:
+
+1. **Counterparty Assurance Gradient** — modeling pattern (four tiers).
+2. **Evidence Source** specializations by tier: **Reputation Signal** (opinion),
+   **Performance Evidence** (observed), **Adjudication Outcome** (adjudicated);
+   tier 3 uses existing **Commercial Commitment**.
+3. **Trust Relationship** carries `assurance_basis` referencing tier + evidence IDs.
+
+**Convenience term only:** "Reputation" in prose — resolve to specific tier and
+Evidence Source before modeling.
+
+## Open Questions
+
+- Standard `assurance_tier` enum vs. numeric score with tier mapping.
+- Whether platform escrow counts as `committed` or `observed` when no separate
+  contract artifact exists (lean: committed when funds legally segregated).
+- Cross-platform reputation portability — Synonymity between Reputation Signals
+  across scopes?
+- Smart-contract automated adjudication (on-chain oracle outcome) — observed or
+  adjudicated?
+
+## References
+
+- Josang, "A survey of trust and reputation systems for online service provision" — https://doi.org/10.1016/j.dss.2005.05.019
+- Hoffman et al., "A survey of attack and defense techniques for reputation systems" — ACM Computing Surveys
+- Klein and Leffler (1981), quality assurance through bonding / price premiums
+- RFC 7070, An Architecture for Reputation Reporting — https://www.rfc-editor.org/rfc/rfc7070
+- Wikipedia, Reputation system — https://en.wikipedia.org/wiki/Reputation_system
+- Internal: `commercial-trust-binding-theory.md`, `duns-commercial-credit-identity.md`,
+  `legal-person-agency-contract.md`, `kyc-aml-commercial-identity-binding.md`
--- a/terminology/TerminologyInventory.md
+++ b/terminology/TerminologyInventory.md
@@ -51,7 +51,16 @@ has incompatible meanings across source families.
 | alei / ibrn | Registry Identifier (government_registry) | ISO 8000-116 | Authoritative legal entity identifier from government register. |
 | iso 6523 / icd | Registry Identifier scheme | ISO/IEC 6523, PEPPOL | ICD + organization identifier encoding. |
 | legal person | Legal Person | eIDAS, civil law, agency | Natural or juridical person under law. |
-| paydex | Evidence Source | D&B | Credit/payment performance history. |
+| paydex | Performance Evidence | D&B | Observed-tier payment performance metric. |
+| reputation | Resolve by assurance tier | marketplaces, credit | Not canonical — see Counterparty Assurance Gradient. |
+| star rating / review | Reputation Signal | Yelp, Amazon, App Store | Opinion-tier Evidence Source; weak, gamable. |
+| feedback score | Reputation Signal | eBay, Uber | Platform-local opinion tier. |
+| credit score | Performance Evidence | bureaus, D&B | Observed-tier counterparty metric. |
+| performance bond / surety | Commercial Commitment | construction, procurement | Committed-tier financial assurance. |
+| escrow | Commercial Commitment | marketplaces, Stripe | Committed-tier funds segregation. |
+| arbitration award | Adjudication Outcome | AAA, ICC, JAMS | Adjudicated-tier dispute result. |
+| court judgment | Adjudication Outcome | courts | Adjudicated-tier enforcement outcome. |
+| assurance gradient | Counterparty Assurance Gradient | commercial identity | Four-tier reliance model (opinion → adjudicated). |
 | kyc / cip | Evidence Source + Assurance | FinCEN, FATF | Regulated commercial identity onboarding. |
 | crm account | Commercial Record | Salesforce | Company/household commercial record. |
 | fluid identity | Persona / weak binding | theory | Low commercial stake; intentional mutability. |