Effort Across Stochastic, Controlled, and Adaptive Worlds: Epistemic License, Changed Agents, Interaction Separation, Learning-Bearing Retry, and Rational Exit (Standalone theory draft v0.3)

Abstract

What this is. A standalone theory draft (v0.3, 7 September 2026) in the Human–AI Readout Programme on effort and retry under stochastic, other-controlled, and adaptive worlds. It keeps Effort Under Stochastic and Other-Controlled Worlds v0.1 as its anchor and v0.2 as its mathematical scaffold; v0.3 is an adversarial elevation, not a replacement. It adds four closures forced by hostile review: a distinction between conditional stochastic modeling and strong stochastic attribution; a witness-preserving cross-domain objection license; a positive interaction-separation condition with an explicit falsifier; and a local-theorem containment principle. Central claim (the author's own words). A theory of retry is incomplete merely because it specifies an outcome distribution. It must state whether stochasticity is assumed, supported, or strongly attributed; what the focal agent can read; what is retained; whether the retained difference becomes certified learning; whether the effective agent changes in a distinction to which the next interaction is actually sensitive; who else controls the outcome; and whether another attempt remains viable under finite cost and risk. A local theorem may fully defeat a local claim but earns no automatic right to erase epistemic, phenomenological, or agent-world transformation that its state/readout bridge does not preserve. Equation provenance. Mathematical provenance follows the Toledo source policy: existing equations are only those registered in Toledo, the programme's equation library (concept DOI 10.5281/zenodo.22537318), and are cited by their Toledo codes; every new equation is explicitly labelled and is registered in Toledo as a coded reading with this record as its origin. Ontology is interpreted through Readout Genesis and epistemology through Readout Universe. Status. Standalone theory draft, K0. Checked under the glosa methodology before deposit (assessment recorded in the programme's research journal). An AI assistant assisted under the author's direction; no AI system or vendor is an author or contributor.