Canon 0.4.0: joinable consumer proof and EvidenceBasis

§10.3 now accepts a consumer record that validates against the live
catalog; resource-control's backup restatement is that proof. ITC-GOV
gains EvidenceBasis in tiers (invoiced and measured are peers) and
ITC-CAP requires it on every consumption row (CAP-R10).

infospace/models/governance/InfoTechCanonGovernanceModel.md

@@ -1066,6 +1066,53 @@ meeting decision
 audit sample
 ```
 
+A quantity offered as evidence MUST declare an **EvidenceBasis** (§11.34a)
+so a decision can be graded by the weakest thing it actually rests on.
+
+---
+
+## 11.34a EvidenceBasis
+
+An **EvidenceBasis** is how a quantity was obtained. A counted object and an
+assumed hourly rate are both numbers; they are not both knowledge.
+
+The closed vocabulary is held in `models/governance/evidence-basis.yaml` and
+is not restated as a second list here. Strength is a **tier**, not a total
+order: `invoiced` and `measured` are peers. `invoiced`, `measured`, and
+`quoted` are **evidenced**. Everything below that tier is inference.
+
+Normative rules:
+
+- A derived value is only as strong as its weakest input, compared by
+  tier. Members of the same tier do not rank against each other.
+- Basis `unknown` carries no quantity and MUST name the gap and its owner.
+  Any other basis MUST carry a quantity. Zero means measured-and-none;
+  unknown means not measured.
+- A derived value MUST record its inputs. Only a derived value may.
+- `invoiced` names a booked financial fact owned by the financial counterpart
+  (`fin-hub`). This model does not originate booked costs.
+
+Optional qualifiers: `proxy_for` (the value measures a different quantity
+than the one being reasoned about) and `observed_at` (the measurement has an
+age).
+
+A set of load-bearing values has a **decision grade**: `evidenced`,
+`projected`, `indicative`, or `insufficient`. The grade is no stronger than
+the weakest resolved basis.
+
+EvidenceBasis is not the intended/declared/applied/observed/assessed state
+qualifier (assimilation OQ-3). Those name *which world* a claim is about.
+This names *how a number* in that world was produced.
+
+Imported by ITC-CAP `CapabilityConsumption` (CAP-R10). Reusable on any
+measured or modelled quantity — actuals, thresholds, forecasts — without
+those records becoming capability artifacts.
+
+**Provenance.** Adapted from consumer demand `demand/EvidenceBasis.md`
+(resource-control `docs/evidence-basis.md` / `tools/basis.py`), 2026-08-15.
+The first application graded a provider comparison `indicative` that had
+been stated to the cent.
+
 ---
 
 ## 11.35 Assertion

infospace/models/governance/evidence-basis.yaml

@@ -0,0 +1,110 @@
+# InfoTechCanon Evidence Basis Catalog
+#
+# How a quantity was obtained. Owned by ITC-GOV; imported by CapabilityConsumption
+# (CAP-R10) and reusable on any measured or modelled value.
+#
+# Adapted from demand/EvidenceBasis.md (resource-control). Strength is a tier,
+# not a total order: invoiced and measured are peers. The consumer's first
+# implementation used a strict list rank and corrected it.
+canon:
+  id: itc-gov-evidence-basis
+  name: InfoTechCanon Evidence Basis
+  artifact_id: catalog/evidence-basis
+  version: 0.1.0
+  status: draft
+  canon_version: 0.4.0
+  owner: model/governance
+  purpose: Closed scale for how a quantity was obtained, with a propagation rule
+    and decision grades. Tiers, not a spurious total order.
+  normative_document: models/governance/InfoTechCanonGovernanceModel.md
+  provenance:
+    demand: demand/EvidenceBasis.md
+    consumer: resource-control
+    consumer_docs:
+      - docs/evidence-basis.md
+      - tools/basis.py
+bases:
+  - id: invoiced
+    tier: evidenced
+    description: Taken from an authoritative financial record. This canon names
+      the basis; it does not originate booked costs, tax, or currency semantics.
+  - id: measured
+    tier: evidenced
+    description: Directly observed from the authoritative system.
+  - id: quoted
+    tier: evidenced
+    description: Stated by a provider or counterparty in a citable source.
+  - id: derived
+    tier: null
+    description: Computed from other values by a stated rule. Not a resolved
+      tier — effective basis is the weakest input.
+  - id: projected
+    tier: projected
+    description: Interpolated between, or extrapolated beyond, observations.
+  - id: estimated
+    tier: judgement
+    description: Human judgement, neither observed nor computed.
+  - id: assumed
+    tier: judgement
+    description: A modelling constant chosen by the author.
+  - id: unknown
+    tier: unknown
+    description: No value exists. MUST NOT carry a quantity. MUST name the gap
+      and its owner.
+# Strongest to weakest. Members of one tier are peers: weakest() among only
+# invoiced and measured MUST NOT imply a ranking.
+tiers:
+  - id: evidenced
+    rank: 0
+    members: [invoiced, measured, quoted]
+    note: invoiced and measured are peers. quoted is evidenced for grading but
+      does not outrank or sit below a measurement of a different quantity.
+  - id: projected
+    rank: 1
+    members: [projected]
+  - id: judgement
+    rank: 2
+    members: [estimated, assumed]
+    note: both yield decision grade indicative; they are not ranked against
+      each other.
+  - id: unknown
+    rank: 3
+    members: [unknown]
+rules:
+  - id: weakest-input-wins
+    statement: A derived value is only as strong as its weakest input, compared
+      by tier. Members of the same tier do not rank against each other.
+  - id: unknown-is-not-zero
+    statement: Basis unknown carries no quantity. Any other basis MUST carry a
+      quantity. Zero means measured-and-none; unknown means not measured.
+  - id: derived-names-inputs
+    statement: A derived value MUST record derived_from. Only a derived value
+      MAY record derived_from.
+  - id: invoiced-not-originated
+    statement: invoiced names a booked fact owned elsewhere. This catalog does
+      not define ledger semantics.
+  - id: tiers-not-total-order
+    statement: Strength is a tier, not a total order. weakest([invoiced,
+      measured]) MUST NOT imply that one outranks the other.
+qualifiers:
+  - id: proxy_for
+    description: The value measures a different quantity than the one being
+      reasoned about. It does not weaken the measurement; it weakens the
+      inference.
+  - id: observed_at
+    description: When a measurement was taken. Age and sample size are not
+      basis. A value may be measured and still be a single noisy sample.
+decision_grades:
+  - id: evidenced
+    when: Every load-bearing value resolves to the evidenced tier.
+  - id: projected
+    when: The weakest resolved tier is projected.
+  - id: indicative
+    when: The weakest resolved tier is judgement (estimated or assumed).
+  - id: insufficient
+    when: At least one load-bearing value is unknown.
+note: >
+  EvidenceBasis is how a quantity was obtained. It is not the
+  intended/declared/applied/observed/assessed state qualifier (OQ-3). Those
+  name which world a claim is about; this names how a number in that world
+  was produced.