Recursive Control under Permanent Systemic Confrontation

This record presents Recursive Control under Permanent Systemic Confrontation (CC–PSC), a cybernetic framework for systems that must continue governing themselves while facing recurrent, adaptive and cross-domain pressure. Its central concern is not resilience understood as a return to a previous state, but the preservation of agency: whether today’s defensive action leaves the system able to observe, choose, act, reconstruct and lawfully revise its conduct tomorrow. CC–PSC treats the controller as part of the contested system. Defensive decisions can alter dependency networks, disclose response patterns, exhaust recovery routes, increase switching delays and weaken the legitimacy required for coordinated reconstruction. The framework therefore distinguishes operational survival from recursive-control survival and from constitutional identity. It also introduces strategic epistemic closure, which replaces speculative modelling of an opponent’s inner beliefs with a bounded class of adversarial policies that remain compatible with available evidence and could materially affect protected outcomes. The theory separates urgent protection from attribution. Cyber incidents, deepfakes, unexplained industrial fires and suspected foreign political influence may justify immediate, reversible safeguards without authorizing premature accusations. Public attribution, sanctions, criminal responsibility and restrictions on political participation remain subject to progressively stronger evidential requirements. This protects the system against both paralysis under uncertainty and the political manufacture of certainty. The accompanying computational programme compares five controllers under equal action budgets and includes an exactly matched experiment in which identical public states face identical future disturbances but inherit different control histories. Further analyses decompose five history-dependent mechanisms, test structural sensitivity across 320 admissible parameter configurations and examine critical-domain vetoes and selective opacity. The negative findings are retained: additional secrecy can reduce disclosure while worsening survival, and a one-step veto can increase failure by blocking viable multi-step recovery. These results support neither a universally superior controller nor empirical geopolitical forecasts. They identify mechanisms, trade-offs and rejection conditions within a transparent synthetic environment. The record includes the article, supplementary methods, source code, frozen outputs, figures, parameter documentation, integrity hashes and a claim-and-result audit. It is intended as a reproducible theoretical and methodological foundation for later field studies, authorized red-team exercises and decision-support tools in critical infrastructure, cyber defence, institutional security and crisis governance.