MathGraph Resolve · permanent version 1

MathGraph Resolve — When Mathematical Evidence Changes

A historical demonstration of scoped mathematical evidence and changing documentary support. It preserves what changed, what did not, and what the evidence still cannot establish.

Six independent axes

No single badge can describe this record.

Historic support resolution, catalogue state, research maturity, current applicability, freshness, and authorization answer different questions. They are kept separate here and in the machine-readable index.

Historical epistemic resultSCOPED_MIXED

Mixed, exact per-claim outcomes in the pinned snapshot.

Catalogue lifecycleCURRENT

The catalogue entry is current; the evidence object remains historic.

Research maturityBOUNDED_QUALIFIED

Bounded qualification, not universal resolution.

Current applicabilityUNKNOWN

An offline replay cannot establish absence of newer changes.

FreshnessNOT_CHECKED

No authenticated live lifecycle check was performed.

Action policyREAD_ONLY_NO_AUTOMATIC_AUTHORIZATION

This record authorizes no automatic external action.

Historic resolution

Withdrawal changes support, not truth by decree.

The publisher withdrew three original arguments. Their prior documentary support paths are stale; mathematical truth remains UNKNOWN.

Documentary original-support routes were modeled as active before publisher withdrawal; this does not prove their correctness.

Publication history is a source notice, not a formal theorem verifier.

Read-only evidence explorer

Inspect a resolution without rerunning it.

The historical default below is server-rendered from the qualified snapshot. Optional controls select only precomputed entries from the exact published scenarios artifact; this browser does not resolve claims.

Selective consequence

What changed — and what did not

Three exact objects. No inferred edge between them.

PUBLISHED_AFTERPublished after publisher withdrawal

The publisher withdrew three original arguments. Their prior documentary support paths are stale; mathematical truth remains UNKNOWN.

Changed · documentary support

The documentary support path changes from SOURCE_DOCUMENTED to STALE.

openai.original_support.1
BeforeSOURCE_DOCUMENTEDAfterSTALE

Documentary support status is distinct from mathematical truth. This view does not establish that the mathematical proposition is false.

Unchanged · mathematical proposition

The mathematical proposition remains UNKNOWN.

openai.theorem.1
BeforeUNKNOWNAfterUNKNOWN

It is not REFUTED. No qualified documentary-to-theorem proof or refutation relation is admitted.

Unchanged · independently bounded observation

The independently bounded observation remains WARRANTED_BOUNDED.

jpl.l4yaml.finite_parser_acceptance
BeforeWARRANTED_BOUNDEDAfterWARRANTED_BOUNDED

nasa-jpl/L4YAML@62bf7077910e888a0bc8adfc8e08a5f500ff3ca3:yaml/yaml-test-suite@da267a5c4782e7361e82889e76c0dc7df0e1e870:48-cases

Published withdrawal receipts: openai.original_publication_receipt.1openai.original_publication_receipt.2openai.original_publication_receipt.3

Invariant boundary: these exact objects remain distinct. Selecting a view does not infer a relation between them or change any object's declared scope.

Precomputed selection only. This is a simulation/counterfactual interface; the browser neither resolves, qualifies, nor mutates historical evidence.

Display-only. Controls unlock only after the same-origin artifact passes its pinned byte digest and identity checks. Every hypothetical view is a simulation and never mutates the immutable historic snapshot.

Published record views

The modeled baseline is not proof of correctness. Withdrawal controls show precomputed consequences over that model, not additional historical events.

Synthetic conformance simulations

Synthetic kernel tests, not historical facts about OpenAI or JPL.

These small kernel tests are structurally separate from the OpenAI/JPL historical graph and establish no new fact about it.

Static qualified default: Published after publisher withdrawal.

If an interaction failure occurs after activation, the qualified default is restored directly from its trusted server-rendered clone; the browser renderer is not used as a fallback.

Qualified historic snapshot

Published after publisher withdrawal

PUBLISHED_AFTER

The publisher withdrew three original arguments. Their prior documentary support paths are stale; mathematical truth remains UNKNOWN.

Before and after support resolution for the selected precomputed scenario
Claim at exact scopeBefore supportAfter supportSelected scenario state
jpl.l4yaml.finite_parser_acceptanceWARRANTED_BOUNDEDWARRANTED_BOUNDEDWARRANTED_BOUNDED
openai.original_support.1Algebraicity of Weil classes on split abelian eightfoldsSOURCE_DOCUMENTEDSTALEUNKNOWN
openai.original_support.2Algebraicity of Kuga–Satake Correspondences for K3 SurfacesSOURCE_DOCUMENTEDSTALEUNKNOWN
openai.original_support.3The rational Hodge conjecture for products of K3 surfacesSOURCE_DOCUMENTEDSTALEUNKNOWN
openai.theorem.1Algebraicity of Weil classes on split abelian eightfoldsUNKNOWNUNKNOWNUNKNOWN
openai.theorem.2Algebraicity of Kuga–Satake Correspondences for K3 SurfacesUNKNOWNUNKNOWNUNKNOWN
openai.theorem.3The rational Hodge conjecture for products of K3 surfacesUNKNOWNUNKNOWNUNKNOWN

Support transitions come from resolutions. Raw selected-scenario claim states come from states. STALE means support was withdrawn; it is not a refutation.

Declared evidence receipts

  • mathgraph.l4yaml.48-case-qualificationformal · nasa-jpl/L4YAML@62bf7077910e888a0bc8adfc8e08a5f500ff3ca3:yaml/yaml-test-suite@da267a5c4782e7361e82889e76c0dc7df0e1e870:48-cases
  • openai.original_publication_receipt.1documentary · openai/math:Algebraicity of Weil classes on split abelian eightfolds:original-withdrawn-edition
  • openai.original_publication_receipt.2documentary · openai/math:Algebraicity of Kuga–Satake Correspondences for K3 Surfaces:original-withdrawn-edition
  • openai.original_publication_receipt.3documentary · openai/math:The rational Hodge conjecture for products of K3 surfaces:original-withdrawn-edition
Exact withdrawn_evidence field
  • openai.original_publication_receipt.1Listed in withdrawn_evidence
  • openai.original_publication_receipt.2Listed in withdrawn_evidence
  • openai.original_publication_receipt.3Listed in withdrawn_evidence

Declared graph routes

  • mathgraph.l4yaml.48-case-qualification jpl.l4yaml.finite_parser_acceptanceDeclared in the selected scenario graph
  • openai.original_publication_receipt.1 openai.original_support.1Declared in the selected scenario graph
  • openai.original_publication_receipt.2 openai.original_support.2Declared in the selected scenario graph
  • openai.original_publication_receipt.3 openai.original_support.3Declared in the selected scenario graph
Exact active_routes field
  • mathgraph.l4yaml.48-case-qualification jpl.l4yaml.finite_parser_acceptanceListed in active_routes

Claim states

  • jpl.l4yaml.finite_parser_acceptanceWARRANTED_BOUNDEDnasa-jpl/L4YAML@62bf7077910e888a0bc8adfc8e08a5f500ff3ca3:yaml/yaml-test-suite@da267a5c4782e7361e82889e76c0dc7df0e1e870:48-cases
  • openai.original_support.1UNKNOWNopenai/math:Algebraicity of Weil classes on split abelian eightfolds:original-withdrawn-edition
  • openai.original_support.2UNKNOWNopenai/math:Algebraicity of Kuga–Satake Correspondences for K3 Surfaces:original-withdrawn-edition
  • openai.original_support.3UNKNOWNopenai/math:The rational Hodge conjecture for products of K3 surfaces:original-withdrawn-edition
  • openai.theorem.1UNKNOWNNo qualified support route in this snapshot
  • openai.theorem.2UNKNOWNNo qualified support route in this snapshot
  • openai.theorem.3UNKNOWNNo qualified support route in this snapshot

Historic graph boundary — unchanged by simulations. No documentary-to-theorem edge is admitted. The withdrawn publication paths and the three mathematical propositions are distinct objects. Each proposition remains UNKNOWN.

Declared receipts in the immutable historical graph
ReceiptKindExact scope
openai.original_publication_receipt.1documentaryopenai/math:Algebraicity of Weil classes on split abelian eightfolds:original-withdrawn-edition
openai.original_publication_receipt.2documentaryopenai/math:Algebraicity of Kuga–Satake Correspondences for K3 Surfaces:original-withdrawn-edition
openai.original_publication_receipt.3documentaryopenai/math:The rational Hodge conjecture for products of K3 surfaces:original-withdrawn-edition
mathgraph.l4yaml.48-case-qualificationformalnasa-jpl/L4YAML@62bf7077910e888a0bc8adfc8e08a5f500ff3ca3:yaml/yaml-test-suite@da267a5c4782e7361e82889e76c0dc7df0e1e870:48-cases

Exact record withdrawn_evidence field

  • openai.original_publication_receipt.1 — listed in withdrawn_evidence
  • openai.original_publication_receipt.2 — listed in withdrawn_evidence
  • openai.original_publication_receipt.3 — listed in withdrawn_evidence

What this record does—and does not—establish

Three propositions remain UNKNOWN.

Stale documentary support is not a counterexample, refutation, or proof that a proposition is false. No unqualified implication from documentary publication to mathematical truth is present in the graph.

  • An independently qualified JPL 48-case record survives without new verification or false promotion.
  • The retained JPL observation is only the exact 48-case source-bound result; it is not universal YAML correctness.
  • Pinned hashes detect byte substitution; self-hashes do not authenticate publishers or qualify new evidence.
  • Git commit pins are provenance references under a separate repository and key trust policy.
  • The snapshot is not continuous monitoring: current applicability is UNKNOWN and freshness is NOT_CHECKED.

Explicit unknowns

  • current_applicabilitycurrent applicability
  • freshnessfreshness
  • mathematical_truth_of_withdrawn_argumentsmathematical truth of withdrawn arguments
  • new_jpl_lean_replaynew jpl lean replay
  • independently_owned_adoptionindependently owned adoption

Independent replay boundary

The RealityGraph consumer is a same-owner independent implementation. It does not rerun the original JPL Lean proof; this published replay did not rerun the original JPL Lean proof and is not outside-organization adoption.

The later P0 replay runs from three isolated inputs without MathGraph.org, producer imports, hosted services, an LLM, or network access. Its trust-policy digest is supplied separately from the bundle.

A policy downloaded from this host does not authenticate itself. Obtain its expected SHA-256 through a separately trusted channel: 09e8d5ba97bca126a81dfe0fcbd5d57f3915ffe71402a497ece4d5e6117f2b4a. Hashes supplied by the same compromised host cannot establish publisher authority.

The Node example is a local HTTP client, not an offline isolation proof. Neither client checks current revocation or supersession; that axis remains NOT_CHECKED.

Durable machine-readable resources

Immutable snapshot
/evidence/mg-living-warrant-openai-jpl-20261010/record.json
0a850b112cce200f2fa97f0d861db712f3752712a6326e62f4c0d56996a95d87 transport SHA-256
Precomputed scenarios
/evidence/mg-living-warrant-openai-jpl-20261010/scenarios.json
ed458ccbee85e3aa3f49d9d38e7ad5380515d00d707bac09c584cc4955b88181 transport SHA-256
Presentation metadata
/records/mg-living-warrant-openai-jpl-20261010/v1/metadata.json
Non-authoritative sidecar; it cannot change a warrant.

Metadata updated 2026-10-11T00:00:00Z: Metadata and presentation changed without changing evidence; the historical snapshot remains unchanged.

Inspect this record’s canonical object projection → · All objects and declared relations