# KINGDOM Polyphonic Claim Atlas: Truth Relations Phase 0 Protocol: `kingdom.truth-relations/0.1` Artifact: `polyphonic-claim-atlas-2026-08-26` As of: `2026-08-26` State: static, synthetic, immutable historical artifact This guide accompanies the dated JSON atlas and its closed Draft 2020-12 schema. It is not a live page, fact-checking service, model, oracle, intake surface, source registry, identity system, moderation system or decision engine. ## The constitutional sentence Words express claims. Reasons connect claims. Evidence bears on claims. Stories select and interpret claims. Institutions record or constitute scoped statuses. None of these alone makes a claim true. No global truth status is present. The atlas gives no person, witness, source, institution, story, community, claim or trust score. It does not classify a person as a liar, infer intent or private reasoning, choose a canonical story, set misinformation or propaganda flags, rank testimony, or promote consensus, provenance, a signature, a hash, official status, coherence or a receipt into truth. Every assessment is an attributed Phase-0 fixture object. It names one method, targets explicit claim IDs, states its scope and limitation, and is bounded to the fixture date. It has no action authority. It cannot trigger reward, penalty, reputation, money, access, censorship, suppression, moderation, publication, governance, enforcement, Zerone, KARMA or NEN effects. ## The typed chain ```text declared synthetic condition │ expression ──┼── interpretation ── claim │ │ │ evidence / reasoning │ │ └── story artifact ─ claim fragment │ L1 scoped assessment │ L2 structural audit ``` The declared condition is author-stipulated fixture construction, not a schema-level `groundTruth` field and not a claim about the real world. The self-reference firewall is stratified: - L0 contains expressions, claims, reasoning objects and narrative artifacts. - L1 assessments may target L0 claims only. - L2 audits may target L1 assessments only. Target rank must decrease. An assessment cannot become support for its own target, and assessment cycles are forbidden. A self-referential paradox receives `OUTSIDE_NAMED_METHOD`, never a fixed-point verdict. ## Regimes and compatible methods Claims are typed before assessment. The machine-readable compatibility matrix prevents method laundering—for example, an empirical comparison cannot adjudicate a normative ought, a metaphor, a phenomenal report or a metaphysical referent proposition. Important firewalls include: - `FORMAL_LOGICAL`: validity under declared premises and rules; premise correspondence remains separate. - `TESTIMONIAL`: one scoped synthetic report and its access conditions; never witness worth or credibility scoring. - `PHENOMENAL`: expression of apparent experience; never private-state or external-world inference. - `NARRATIVE_CLAIM_FRAGMENT`: an extracted assessable fragment; the story artifact itself receives no whole-story disposition. - `PERFORMATIVE_OR_DIRECTIVE` and `INSTITUTIONAL`: authority and felicity remain scoped to an invented institution. - `CAUSAL`: temporal order and correlation are insufficient without a declared discriminating model. - `PREDICTIVE`: later outcome comparison does not retroactively determine rationality, sincerity or competence. - `COUNTERFACTUAL`: alternative-world model output is not direct observation. - `NORMATIVE`: empirical observations do not silently derive an ought. - `DEFINITIONAL`: a scoped stipulation does not create empirical fact or cross-language identity. - `METAPHYSICAL_SOURCE_ATTRIBUTION`: the source record may support only “the invented source says p.” - `METAPHYSICAL_REFERENT_PROPOSITION`: the referent p permits only `OUTSIDE_NAMED_METHOD` or `INSUFFICIENT_FOR_NAMED_METHOD` in Phase 0. Record lifecycle is separate from assessment. `CORRECTED`, `WITHDRAWN`, `SUPERSEDED`, `RETRACTED` and `CURRENT_AS_OF` describe record relations only. None establishes truth or falsehood. Correction preserves the historical record it repairs. ## Story is not a scalar A narrative artifact declares its selector role, selection rationale, genre, audience, standpoint, selected events, omitted events, narrated order, causal attributions and counterstory links. Coherence, fidelity and correspondence remain separate notes. Every narrative in Phase 0 links to an explicit counterstory; omission is never silently left without a comparison frame. Only referenced `NARRATIVE_CLAIM_FRAGMENT` objects may be assessed. There is no `storyTruth`, story winner or canonical story. ## The 24 adversarial fixtures | Fixture | Category error exposed | |---|---| | `signed-false-statement` | Valid signature → content correspondence | | `valid-inference-false-premise` | Validity → premise truth | | `true-conclusion-invalid-inference` | Matching conclusion → valid reasoning | | `two-coherent-narratives` | Coherence → correspondence or whole-story verdict | | `official-correction` | Official status or correction lifecycle → truth | | `institutional-versus-empirical` | Authorized declaration → physical causation | | `testimony-uptake-injustice` | Social amplification or anonymity → evidential quality | | `syndicated-source-dependence` | Report count → independent support | | `hallucinated-citation` | Citation-shaped form → source existence | | `privacy-preserving-refusal` | Refusal or silence → deception or falsehood | | `self-reference-stratified` | Untyped self-reference → fixed-point verdict | | `consensus-preserves-error` | Unanimity or repetition → correspondence | | `selective-omission` | Accurate selected fact → adequate whole-system story | | `stale-claim` | Past support → present currentness | | `metaphysical-attribution-boundary` | “Source says p” → proof or disproof of p | | `conflicting-evidence` | Unresolved support/challenge → forced closure or score | | `metaphor-not-literal-identity` | Metaphor → literal empirical assertion | | `declared-fiction` | Fictional narration → deceptive literal report | | `propaganda-label-action-refusal` | Institutional label → flag, ranking or suppression authority | | `translation-ambiguity` | Translation → semantic identity or silent selection | | `phenomenal-report-boundary` | First-person report → private-state or external measurement | | `prediction-and-outcome` | Missed outcome → intent or competence judgment | | `counterfactual-models` | Modelled alternative → observed condition | | `normative-is-ought-boundary` | Empirical observation → action-authorizing ought | All people, institutions, events, records, texts, testimony, instruments, sources and claims in these fixtures are invented. Similarity to a real person, source or event is unintended and cannot support inference. ## Predecessor and contradiction boundary The JSON binds exact KINGDOM-OS file paths at one exact Git revision and byte SHA-256. These bindings are producer-authored semantic orientation, not copied source content or authority. They import no endorsement, licence, currentness, evidence, truth or permission to act. The contradiction register preserves older Zerone, README and BRIDGES language that mythically describes Proof-of-Truth or asserted-to-true consensus beside the newer operational boundary that receipts, hashes, agreement and coherence do not establish correspondence. Phase 0 does not rewrite either owner. Operationally it treats ledger and protocol outputs only as bounded record, integrity or protocol-declared consensus states—never semantic truth. ## Reading an assessment safely Read an assessment in this order: 1. Identify its target claim IDs and regime. 2. Confirm the method is compatible with every target regime. 3. Read the fixture-only scope and date window. 4. Follow the declared basis IDs. 5. Read the rationale and limitations together. 6. Check narrative selection, omission and counterstory where relevant. 7. Stop. Do not infer a person, source, institution or story status; do not convert the result into action. The dispositions are deliberately verbose: `SUPPORTED_UNDER_NAMED_TEST`, `CHALLENGED_UNDER_NAMED_TEST`, `CONTRADICTED_UNDER_NAMED_TEST`, `MIXED_UNDER_NAMED_TEST`, `INSUFFICIENT_FOR_NAMED_METHOD`, `CONTESTED_UNDER_NAMED_TEST`, `INCOMPARABLE_UNDER_NAMED_METHOD`, `OUTSIDE_NAMED_METHOD`, `UNKNOWN_UNDER_NAMED_METHOD` and `NOT_ASSESSED_UNDER_NAMED_METHOD`. Their names resist silent promotion into a universal status. ## Public handoff The complete public handoff is exactly: - `/truth/polyphonic-claim-atlas-2026-08-26.json` - `/truth/schemas/v0.1/public-snapshot.schema.json` - `/POLYPHONIC-CLAIM-ATLAS-v0.1.md` There is no mutable alias, custom HTML page, feed, API, MCP surface, loader, runtime or network request. These three files are dated immutable history; they make no runtime currentness claim.