124-Retire unique operational semantics as the form criterion

Type: reference/types/adr.md · Status: accepted

Status: accepted Date: 2026-11-02

Context

The phrase “unique operational semantics” conflates a programming-language term with a criterion for classifying content. The associated conjecture—that an artifact's language admits a formal semantics—does not distinguish an implementation's behavior from the semantics its users hold it to. A compiler and a frozen model both admit formal descriptions of their computation. Such a description includes implementation defects; it does not by itself supply a governing reference for C or English.

The practical distinction is whether users hold consumers to a reference that settles what the artifact permits. Inspection cost and success at the intended task are further questions, not membership tests.

Decision

Retire unique operational semantics from current Commonplace vocabulary. Use the criterion in representational form: localized content is symbolic under a consumption path when its consequences are fixed by the reference of a language that has a formal semantics or is close to one. Here “close” means users hold consumers to a standard or implementation from which a formal semantics could be derived. Short glosses may say “held to a semantic reference” when the definition supplies that qualification.

Codification names the move into symbolic form, not a separate criterion. The move may rewrite the artifact or adopt a consumer as the reference for an operation. Mere pinning or repeatability is insufficient. Adopting a model package does not make it the reference for English, establish task adequacy, or make its rules affordable to inspect.

Ask localization first and assigned consequences only for localized content. Keep opacity relative to a consumption path, behavior or change, and inspection budget. Preserve the distinctions between specified behavior, implementation conformance, and requirement adequacy.

Considered alternatives

Retain the phrase with a narrower gloss. Rejected because the programming-language sense remains distinct and the formalizability conjecture alone fails to select a governing reference. Renaming only the phrase would leave that conceptual error intact.

Exclude prompts or opaque references from symbolic form. Rejected because surface syntax and inspection difficulty do not decide whether a system has adopted a mechanically specified reference. Opacity changes the available checks, not membership.

Treat every fixed consumer as a reference. Rejected because a formal account of an implementation's behavior does not establish its authority over a language. Users must adopt it or hold it to another governing reference.

Consequences

Authors and reviewers consume the revised definitions through the root vocabulary links, theory arguments, tag navigation and the vocabulary collision ledger. Those natural-language instructions constrain classification and writing; no validator computes semantic membership. The ledger retains the external programming-language and measurement senses without using them as the Commonplace criterion.

Conformance tests can reveal disagreements against a reference but do not prove its adequacy or correctness outside tested cases. An adopted model can conform exactly while failing its task. Independent review remains necessary after substantive definition changes; previous freshness records do not attest to the new text.

This decision governs Commonplace's current technical vocabulary. It does not rename the external discipline's operational semantics, alter quoted sources or frozen historical evidence, assert that all natural language is unformalizable, or prescribe when a project should adopt a model as its reference.