The Record-Closure Problem — What Makes a World Internally Complete Enough to Be Lived?

What makes a world a world rather than a fragment, hallucination, dream, simulation-frame, Boltzmann-brain episode, or unstable configuration? This paper introduces the record-closure problem: the problem of specifying what degree of internal record coherence, stability, depth, channel-connectivity, and update-admissibility is required for a domain to function as a livable world for finite agents. A world need not be metaphysically complete in an absolute sense. It must, however, be sufficiently closed under record-based continuation: memories must connect to admissible histories; perceptions must connect to stable environments; actions must generate coherent updates; errors must be detectable and correctable; and law-like regularities must support expectation. The paper develops record-closure as a criterion of worldhood within Record-Based Ontology (RBO) and as a companion notion to Bit-Collapse Theory (BCT). If BCT asks when possibilities become different experienced worlds for a finite observer, record-closure asks when a domain is internally coherent enough to count as a world at all. The framework is applied to dreams, hallucinations, Boltzmann brains, simulations, Everettian observer-worlds, historical reconstruction, and ordinary reality. The central thesis is that a world is not merely a substrate, an object-set, or a formal state. It is a record-closed domain of admissible continuation.