Entropy as an Epistemic Artifact Under Deterministic Causation

Abstract This paper clarifies the ontological status of entropy under a single explicit axiom: deterministic evolution on full physical state. Under this assumption, probability cannot be fundamental and must be epistemic, reflecting incomplete knowledge rather than physical indeterminacy. Because all standard formulations of entropy—Shannon, Boltzmann, and Gibbs—presuppose probability distributions or coarse-grained state descriptions, entropy cannot be ontological once full-state causation is fixed. The paper reclassifies the Second Law of Thermodynamics as an inference law over macrostates rather than a fundamental law of motion, without disputing its empirical validity or practical indispensability. Structural consequences are derived for physics, information theory, and artificial intelligence, including the result that stochasticity is not required for exploration, creativity, or intelligence. Deterministic systems can generate novelty through structured traversal of state space under constraint. The contribution is not a new physical theory but a necessary clarification. It establishes that entropy, while real and measurable, is an epistemic artifact under deterministic causation and therefore cannot serve as a foundational ontological primitive.