Precedence under Irreversibility

This work presents a structural result on the limits of reflexive control in dynamical systems under irreversibility. It formalizes the insight that no purely reflexive mechanism—based on feedback, learning, anticipation, or gain modulation—can robustly prevent irreversible transitions when such states remain reachable. The main contribution is the Structural Precedence Theorem, which establishes that safety in the presence of irreversible states must be enforced by architectural constraints that restrict reachability, rather than by improving decision-making or learning mechanisms. Reflexive control is shown to be reliable only within forward-invariant safe subsets of the state space. This document consolidates and formalizes conclusions that emerged from earlier theoretical, exploratory, and empirical analyses, including toy models and negative results reported separately. The present work provides a unifying structural explanation for those findings and clarifies the necessary ordering between architecture, decision, and learning in irreversible systems. Provided for reproducibility and conceptual clarity.