Constitutional Ordering of the Epoch-Governed Intelligence Papers

This project presents a sequence of architectural papers proposing a constitutional framework for artificial intelligence grounded in explicit authority, sealed epochs, bounded learning, and non-authoritative treatment of time and context. Rather than optimizing models or benchmarks, the work focuses on architectural failure modes observed in contemporary AI systems—such as temporal hallucination, narrative drift, authority escalation via confidence, and unbounded memory—and argues that these failures are structural rather than ethical or data-driven in origin. The papers were released incrementally and are intended to be read as a single, ordered architectural system. Authority flows strictly downward: foundational papers define non-negotiable invariants, while later papers assume—but do not reinterpret or modify—those constraints. An explicit constitutional ordering is provided to prevent misreading of individual papers in isolation. This project does not propose new learning algorithms, optimization techniques, or empirical benchmarks. Its contribution is architectural: defining what intelligent systems must never be allowed to do, and specifying the structural boundaries required for auditable, bounded, and fail-closed intelligence.