UNKNOWN × REM: Toward an Evidence-Bound Minimal State Language for Reproducible Representation

This paper introduces UNKNOWN × REM as an evidence-bound minimal state language for reproducible representation. The work examines how states that cannot be established from available evidence should be represented without inference, forced resolution, or implicit completion. It distinguishes UNKNOWN from REM and defines their relationship as a minimal representational structure for preserving epistemic boundaries across transformation, compression, and subsequent reuse. The proposed language is designed to support evidence-bound representation, provenance preservation, and reproducibility by making unresolved or intentionally inactive states explicit rather than silently converting them into asserted knowledge. The paper focuses on the representational and governance implications of this distinction. It does not claim that UNKNOWN × REM constitutes an established external standard. Rather, it presents a minimal conceptual framework and examines the conditions under which such state representations can remain faithful to available evidence during composition, transformation, and reconstruction. Keywords: UNKNOWN, REM, evidence-bound representation, reproducibility, provenance preservation, state language, epistemic boundaries, institutional memory, AI governance.