Search for a command to run...
Following the establishment of substrate-agnostic unpredictability in MMG-TR-001, this paper presents a formal computational model for internal human agency. We argue that "Free Will" is neither a mystical intervention nor a retrospective illusion, but a technical property of Self-Referential Modeling Systems. We propose its replacement with Functional Agency: a technical property of systems that possess an Incomputability Firewall against external reduction. Utilizing the Halting Problem (Turing) as a formal template for self-reference limits and the Principle of Computational Irreducibility (Wolfram), we demonstrate a Reflexive Prediction Barrier that prevents any resource-bounded predictor from generating intervention-stable predictions of a complex agent without either equivalent computational cost or a violation of the agent’s deliberative integrity. We further expand the metric of Agency Depth ($D_A$), first introduced in TR-001, defining it as a scalar function of Temporal Horizon, Counterfactual Width, and Historical Integration. We define Effective Agency ($A_e$) to incorporate Model Fidelity as a calibration multiplier, representing the functional capacity of an agent to act as a Salient Cause. This model demonstrates that sovereignty is not a metaphysical constant but a variable resource subject to environmental and cognitive constraints, providing the rigorous, non-dualistic foundation required for the Meaningfulness Protocol and the systemic defense of human dignity in an age of hyper-resolution predictive modeling. Keywords: Functional Agency, Agency Depth, Computational Irreducibility, Halting Problem, Active Inference, Sovereignty Turing Test, Disclosure-Stable Prediction, Process Sovereignty, Cognitive Science, AI Safety, Systems Theory, Self-Referential Systems, Predictive Modeling Limits. Deep Dive into this paper on Youtube: [LINK] Metadata & Implementation The official human-readable LaTeX code for this paper is maintained at the public repository:https://github.com/MeaningfulnessMediaGroup/MMG-Technical-Reports Protocol ID: MMG-TR-002License: CC-BY-4.0 Authorship & IdentityWriting as Djeff Bee, Dominic Branchaud (ORCID: 0009-0008-2046-4378) applies 25 years of systems engineering methods to philosophical modeling.