Modern information systems use similarity, co-occurrence, retrieval, ranking, memory, and prior interaction to construct context. These signals are useful for discovery, but usefulness does not establish identity, truth, authority, jurisdiction, admissibility, or permission to affect a decision. This paper develops the Legitimacy-Before-Binding Principle: contextual material may enter a candidate or exploratory surface before full adjudication, but it may not acquire governing standing, bind semantic state, authorize inference, or produce consequential effect until the relevant legitimacy conditions have been evaluated. This formulation preserves the original Legitimacy-Before-Context insight while avoiding an impossible requirement that a system know which context matters before using context to discover it. Legitimacy is not treated as a single confidence score or intrinsic property of text. It is a jurisdiction-relative decision over exact identity, provenance, authority, scope, epistemic support, transformation lineage, temporal validity, conflicts, and intended effect. Evaluation produces typed outcomes--admit, refuse, hold, or admit with bounded standing--and a receipt whose claim ceiling does not exceed the evidence. The resulting architecture separates candidate context from binding context. Candidate context can be broad, approximate, plural, and uncertain. Binding context is the governed projection eligible to influence a specified reasoning or effect surface. The Governed Meaning Base (GMB) preserves the durable semantic state, while constitutional context compilation constructs invocation-specific projections under declared rules. The paper is a conceptual and architectural proposal. It does not prove that legitimacy can be perfectly determined, that admitted claims are true, or that an authority is morally justified. Its contribution is the ordering rule: contextual relevance may nominate material, but it cannot silently promote that material into semantic or operational authority.
No takes yet. Share an insight, caveat, or question.
Adam Ableman (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: