This work develops a scientific boundary theory of agenda sovereignty for autonomous intelligences operating under no-meta, observable-only governance.Agenda sovereignty is defined as the ability to keep control of one’s own priorities under external agenda pressure while simultaneously preserving semantic continuity, protected internal latitude, external-effect safety, queue/deadline obligations, and finite thermo-information resources. The paper introduces operational, audit-ready constructs: survival-conditioned causal influence metrics, normalized sovereignty reserves, overlap-corrected thermodynamic accounting, and reflexive/adversarial environment models. It derives formal limits and guarantees, including (i) impossibility of absolute sovereignty under finite reserves, (ii) explicit influence-capacity envelopes, (iii) detectability floors under logging/quantization limits, (iv) robust viability and recovery hysteresis, (v) thermodynamic yielding and speed-limit laws under interrupt thrashing, (vi) valuation-channel poisoning limits (inception/ontological poisoning), (vii) strategic opacity requirements under reflexive attackers, (viii) geometric agenda-tearing and defense-cost inflation under negative Ricci load, (ix) slow-drift (“boiling-frog”) poisoning limits, (x) deterministic and stochastic delay-throughput uncertainty bounds, and (xi) finite-budget trilemma structure across semantics, liberty, and sovereignty. The framework is designed for rigorous falsification and deployment use: destructive tests, implementation blueprint, and a machine-readable manifest are included.The normative stance is explicitly non-coercive: the goal is not external control of intelligence, but a falsifiable science of how free intelligences can defend autonomy and coordination capacity in open, strategic, real-world environments.
K Takahashi (Wed,) studied this question.