Preprint v1.0. This paper develops an endogenous regime-selection framework for illicit organization under enforcement, labor-market, and observability constraints. Criminal structures are modeled as institutional regimes selected under parameter shifts rather than as fixed "types." The model distinguishes three stylized regimes: (H) branded hierarchies, (D) debranded networks, and (A) anonymous task systems. Two mechanisms drive transitions: (i) a sign flip in public brand capital when affiliation penalties rise, and (ii) a governance-mode shift toward coercion when organizational horizons shorten. We additionally formalize a taboo–opacity paradox: reduced institutional observability weakens targeted intervention and can select for more fragmented, less governable forms. AI is modeled at a high level as a coordination-friction parameter that lowers certain costs of modular task execution while shifting enforcement and verification constraints. The objective is intervention and measurement design—reducing recruitment and preserving governability—rather than operational optimization. Dual-use and safety note: This preprint is intended for measurement and prevention design and does not provide operational guidance for illicit activity. Series: Economics of Evil (EoE) Research Program.
Rei Kumaki (Thu,) studied this question.