Public explanation, private executable evidence.
The publication layer is intentionally downstream of the research runtime. It explains the ideas and exposes small first-party instruments without publishing the private source, raw reports, owner material, or Matter runtime.
The chronological evidence now has a consolidated architecture and maturity boundary.
Research lineage
The public story is derived from two private research tracks and one candidate Library extraction.
Representation, transport, topology, patching, equivalence, coherence.
Mathematical experiments establish the representation-aware machinery and expose the importance of ordered composition.
Relation kinds, law, applicability, version, negative knowledge, semantic time, supersession, change impact, non-canonical assembly, perspective drift/authority, and bounded partial-trust disclosure.
Cross-domain falsification work determines which geometric ideas generalize and which must remain domain-specific.
Reusable API above frozen Core.
The extracted Library surface provides witnessed relations, explicit relation-law profiles, ordered coherence, transport receipts, and historical replay.
Related architecture, separate product.
Matter preserves consequential work, reliance, reconsideration, and authority. Semantic Transport may eventually support it, but this publication does not connect to or mutate Matter.
What is deliberately not published
The boundary is a feature of the research program, not an omission.
External reference points
These were useful public standards in the semantic research. Links are provided for readers who want the authoritative source material.
SKOS Reference ↗ — used as an external case for relation-specific semantic behavior.
FHIR specification ↗ — used as an external case for versioned scope, contextual applicability, and mapping semantics.
The private executable experiment tests semantic-law retirement after use. It preserves historical validity, blocks new use of the predecessor after cutover, and shows that live dependents need explicit transition policy. This public page does not expose the private fixtures or promote supersession into a Library API.
The private executable experiment tests graph-level change impact. It deliberately includes reachable but non-causal controls, preserves them as unaffected, propagates only through declared causal edge semantics, returns unknown-impact when causal law is missing, and rejects cycles until a fixed-point rule is justified. No Change Impact API is promoted here.
The private executable experiment tests whether three intact local semantic perspectives can jointly support a global assembly witness without a canonical master. Required overlap roles must connect, contradictory observations fail, and alternate translation laws must agree even outside the currently observed sample. No Local-to-Global Assembly API is promoted here.
The private executable experiment tests asynchronous local revision drift and role-scoped authority. A stale non-authoritative projection can remain visible while the authoritative role resolves current meaning; newer revision numbers do not win automatically, authority conflict/gaps fail closed, and contestation triggers review without becoming automatic override. No Drift/Authority API is promoted here.
The private executable experiment tests selective semantic disclosure under partial trust. The assembler consumes bounded packets rather than raw local views; undisclosed, unknown and unauthorized-to-disclose remain distinct; true-but-untrusted and false-but-trusted attestations both fail closed. The experiment deliberately makes no cryptographic privacy claim and promotes no new API.