Execution Governed by Admissible State: A Deterministic Model for Real-Time Reliance Control in Safety-Critical Systems

This paper defines a deterministic model for governing execution in safety-critical systems based on admissible state rather than assumed reliability. Existing assurance models rely on historic certification, periodic inspection, and inferred operational continuity. These approaches do not provide a mechanism to determine whether a system is in a condition suitable for reliance at the moment it is required. This work introduces a structured separation between evidenced system condition (State), the determination of suitability for reliance (Admissible State), and the permission to act (Execution). It establishes admissibility as a formal decision boundary that governs whether execution is permitted. Execution is therefore controlled by the presence of valid, current, and independently verifiable evidence. Where admissibility cannot be established, execution must not proceed as a matter of control, not discretion. The model produces enforceable outcome states in which execution is allowed, escalated, or blocked based on evidence sufficiency at the moment of reliance. This paper contributes a governance-aligned framework for real-time reliance control and forms part of a broader body of work addressing the structural gap between historic assurance and present-state evidence in safety-critical environments.