From 455abfae1a28e84619f42f008aa34e90fb16abb4 Mon Sep 17 00:00:00 2001 From: vp Date: Sun, 1 Mar 2026 23:01:35 +0300 Subject: [PATCH 1/4] fix: align ClaimScope typing with USM scope object semantics --- FPF-Spec.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/FPF-Spec.md b/FPF-Spec.md index 78f8b91..cee951a 100644 --- a/FPF-Spec.md +++ b/FPF-Spec.md @@ -4027,7 +4027,7 @@ Cross‑context transfer maps a **set** `Scope` via a **Bridge** and penalizes * What teams often want to “compress into a G number” is actually (a) the **quality of expression** and (b) the **completeness** of the declared scope. The first is an **F threshold** (e.g., require **`U.Formality ≥ F4`** so the scope is predicate‑like and addressable); the second is handled by explicit **ESG guards**: “**Scope covers TargetSlice**,” “**`Γ_time` is specified**,” and “**freshness window holds**” (R‑lane). A ladder for G adds confusion but no additional control. **Normative directive.** -`U.ClaimScope (G)` **SHALL** remain a **set‑valued** characteristic; **no ordinal or numeric ladder SHALL be defined** for G. Authoring and gating **SHOULD** use **F thresholds** (C.2.3) and **explicit guard predicates** (A.2.6) rather than pseudo‑levels of G. +`U.ClaimScope (G)` **SHALL** remain a **set‑valued USM scope object**; **no ordinal or numeric ladder SHALL be defined** for G. Authoring and gating **SHOULD** use **F thresholds** (C.2.3) and **explicit guard predicates** (A.2.6) rather than pseudo‑levels of G. #### A.2.6:7.5 - Translation across Contexts (Bridge & CL) From c0e587ec6c7d9a70218c2e964677faf2f9084e3b Mon Sep 17 00:00:00 2001 From: vp Date: Sun, 1 Mar 2026 23:24:40 +0300 Subject: [PATCH 2/4] fix: align B.3 triad typing and add CoverageMetric(G) guardrail --- FPF-Spec.md | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/FPF-Spec.md b/FPF-Spec.md index cee951a..fb676ff 100644 --- a/FPF-Spec.md +++ b/FPF-Spec.md @@ -4027,7 +4027,9 @@ Cross‑context transfer maps a **set** `Scope` via a **Bridge** and penalizes * What teams often want to “compress into a G number” is actually (a) the **quality of expression** and (b) the **completeness** of the declared scope. The first is an **F threshold** (e.g., require **`U.Formality ≥ F4`** so the scope is predicate‑like and addressable); the second is handled by explicit **ESG guards**: “**Scope covers TargetSlice**,” “**`Γ_time` is specified**,” and “**freshness window holds**” (R‑lane). A ladder for G adds confusion but no additional control. **Normative directive.** -`U.ClaimScope (G)` **SHALL** remain a **set‑valued USM scope object**; **no ordinal or numeric ladder SHALL be defined** for G. Authoring and gating **SHOULD** use **F thresholds** (C.2.3) and **explicit guard predicates** (A.2.6) rather than pseudo‑levels of G. +`U.ClaimScope (G)` **SHALL** remain a **set‑valued scope object**; **no ordinal or numeric ladder SHALL be defined** for G. Authoring and gating **SHOULD** use **F thresholds** (C.2.3) and **explicit guard predicates** (A.2.6) rather than pseudo‑levels of G. + +**Report-only proxy (optional).** `CoverageMetric(G)` MAY be emitted for dashboards/reports as a derived scalar projection of `G` (with declared domain/measure references). It **MUST** be marked report-only and **MUST NOT** substitute `G` in norms, gate predicates, or conformance logic. #### A.2.6:7.5 - Translation across Contexts (Bridge & CL) @@ -23557,7 +23559,7 @@ Every non‑trivial result in FPF—*a composed system is safe*, *a model is cre To make such claims comparable and auditable across domains, B.3 introduces a **Trust & Assurance Calculus** that: -* uses a **small set of characteristics** (F–G–R) governed by CHR principles (these are **not** a state space), +* uses **two characteristics** (**F**, **R**) plus one **scope object** (**G**); only **F/R** are CHR characteristics (this is **not** a state space), * accounts for **integration quality** via **Congruence Level (CL)** along the edges of a `DependencyGraph` (B.1.1, A.14), * and composes these values with **Γ‑flavours** while respecting the **Invariant Quintet** (IDEM, COMM/LOC or their replacements, WLNK, MONO). @@ -23592,9 +23594,9 @@ Without a disciplined calculus, four chronic failures appear: B.3 defines **what** to measure, **how** those measures live on nodes and edges of the dependency graph, and the **shape** of the aggregation that any Γ‑flavour must honor when producing an *assurance result*. -#### B.3:4.1 - The F–G–R characteristics (CHR‑compliant) +#### B.3:4.1 - The F–G–R assurance components (two CHR characteristics + one scope object) -We standardize three characteristics on **nodes (holons)** plus one **edge** characteristic: +We standardize two **node characteristics** (**F**, **R**), one **node scope object** (**G**), plus one **edge** characteristic (**CL**): 1. **Formality (F)** — *how constrained the reasoning is by explicit, proof‑grade structure.* @@ -23605,15 +23607,17 @@ We standardize three characteristics on **nodes (holons)** plus one **edge** cha 2. **ClaimScope (G)** — *how broadly the result applies in the relevant domain space.* - * **Scale kind:** **coverage / span** (set‑ or measure‑based; domain‑specific). - * **Monotone direction:** larger, but only when **correctly supported** (see WLNK and CL below). + * **Kind:** **set‑valued scope object** over `U.ContextSlice` (USM set algebra; not a CHR scale). + * **Order semantics:** set inclusion / coverage predicates only; “larger is better” is **not** a universal invariant under bridge translation. + * **Optional report-only projection:** `CoverageMetric(G)` MAY be published as a scalar proxy for reporting, but it MUST be explicitly derived from `G` with declared domain/measure references. + * **Normative guardrail:** `CoverageMetric(G)` **MUST NOT** replace `G` in normative clauses, gate conditions, or conformance checks. 3. **Reliability (R)** — *how likely the claim/behavior holds under stated conditions.* * **Scale kind:** **ratio** in `[0,1]` (or a conservative ordinal proxy when numeric modeling is unavailable). * **Monotone direction:** higher is better. -2. **Congruence Level (CL)** — *edge property: how well two parts fit* (semantic alignment, calibration, interface Standard). +4. **Congruence Level (CL)** — *edge property: how well two parts fit* (semantic alignment, calibration, interface Standard). * **Scale kind:** **ordinal** with a **monotone penalty function** `Φ(CL)` where `Φ` decreases as CL increases. * **Canonical levels (example):** From ff076c018f0ba1acd972cf1d232b0c5088d97977 Mon Sep 17 00:00:00 2001 From: vp Date: Sun, 1 Mar 2026 23:25:37 +0300 Subject: [PATCH 3/4] docs: use explicit B.3 formula two characteristics plus scope object --- FPF-Spec.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/FPF-Spec.md b/FPF-Spec.md index fb676ff..652b22d 100644 --- a/FPF-Spec.md +++ b/FPF-Spec.md @@ -23559,7 +23559,7 @@ Every non‑trivial result in FPF—*a composed system is safe*, *a model is cre To make such claims comparable and auditable across domains, B.3 introduces a **Trust & Assurance Calculus** that: -* uses **two characteristics** (**F**, **R**) plus one **scope object** (**G**); only **F/R** are CHR characteristics (this is **not** a state space), +* uses **two characteristics (F,R) + one scope object (G)**; only **F/R** are CHR characteristics (this is **not** a state space), * accounts for **integration quality** via **Congruence Level (CL)** along the edges of a `DependencyGraph` (B.1.1, A.14), * and composes these values with **Γ‑flavours** while respecting the **Invariant Quintet** (IDEM, COMM/LOC or their replacements, WLNK, MONO). From 86f39884cdcabcb512e1a7dd38e5df8adf3a7af9 Mon Sep 17 00:00:00 2001 From: vp Date: Sun, 1 Mar 2026 23:36:33 +0300 Subject: [PATCH 4/4] docs: replace residual F-G-R characteristics wording in E.4 --- FPF-Spec.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/FPF-Spec.md b/FPF-Spec.md index 652b22d..4c51b6e 100644 --- a/FPF-Spec.md +++ b/FPF-Spec.md @@ -32619,7 +32619,7 @@ The FPF ecosystem is formally stratified into three canonical **artefact familie * **Pedagogical Companion:** Provides a tutorial on how to model a water pump using that profile. * **For an `U.Episteme`:** - * **Conceptual Core:** Defines `U.Episteme` and the F-G-R characteristics. + * **Conceptual Core:** Defines `U.Episteme` and the F-G-R components. * **Tooling Reference:** Provides the reference linting tool to automatically score epistemes. * **Pedagogical Companion:** Provides a case study on how a scientific theory's R-score evolves over time.