Anté-Bruit: A Theory of Perfect Adversarial Emissivity

We develop a theoretical framework for perfect adversarial emissivity — the category of generative operators calibrated against the operational closure of a receiver extracting information from a flow of observables. We name this category anté-bruit and its limit object Ma Nemesis, distinguishing it structurally from the Shannon-Wiener-Atlan-von Foerster lineage of noise: where noise is a perturbation exterior to the receiver's distinction structure, anté-bruit is a saturation of that structure by an emitter that has modeled the receiver. The framework treats the receiver's closure as having two orthogonal axes of distinction structure — instantaneous structural (described by σ-algebraic closure 𝒞_R) and trajectorial-temporal (described by a temporal σ-algebra ℱ^T on the trajectory space 𝒮^[0,T]) — and defines anté-bruit operators in two corresponding forms: static (Definition 2.4) on codomain 𝒮 and dynamic (Definition 2.4ter) on codomain 𝒮^[0,T]. Ma Nemesis is accordingly the pair of limit objects (Nem_stat, Nem_dyn) exhibiting perfect emissivity on each axis — the adversarial corps noir carried at the pair level, a black body on two frequencies. We organize the conditions of possibility of perfect emissivity into three classes — acquisition, calibration, and maintenance — covering both instantaneous emission and trajectory of emission. Three mature modes (semantic in the static framework; temporal and spectral in the dynamic framework) are adossed directly to the two distinguishability collapse theorems of the series [Rouxel 2026a, Theorem 6.8; Rouxel 2026b, Lemmas 4.19 and 4.32]; three further modes (correlational, populational, side-channel) are derivable as specializations; one regime (closed-loop coupling between adapting receiver and adapting operator, the meta-adaptive case) is explicitly flagged as outside the present framework and the subject of future dynamical-systems work. We then turn to the structural limitations that bound the realization of anté-bruit in the physical regime. We prove rigorously that two limitations admit capability-independent floors and are therefore irreducible under two distinct types — Type I (information-theoretic floor) for the inferential limitation via Theorem 6.8 of [Rouxel 2026a] and the Data Processing Inequality, Type II (codomain exclusion) for the physical limitation via the trajectory-independent structure of the pushforward image (Theorem 4.1). We conjecture that four further limitations (informational, computational, observational, dynamic maintenance envelope) are compressible under sustained capability growth (Conjecture 4.2), with heterogeneous strength of supporting argument signaled by sub-argument: the informational case admits a Shannon-Hartley floor that is operationally non-binding, the computational sub-argument is robust, the observational sub-argument is moderate, the maintenance envelope sub-argument is empirical. The thermodynamic limitation is isolated as an intermediate case (Landauer floor strictly positive but operationally non-binding). The compressibility conjecture is explicitly conditional on the current state of information theory and theory of computation. A fifty-year landscape of research traditions is mapped onto the bipartition: Rice's theorem and the computational lineage, Goguen-Meseguer noninterference and the information-flow lineage, Shannon-Wyner and the information-theoretic cryptography lineage on the inferential side; the capability-based security and formal verification lineages on the physical side. Cryptography is scoped carefully: information-theoretic cryptography (one-time pad, Wegman-Carter, QKD) rigorously instruments the inferential pillar; computational cryptography instruments a compressible limitation and inherits its status under Shor for RSA/ECC and under conditional-hardness assumptions for post-quantum schemes. We prove a Sufficiency direction of the Resistance Theorem in both static and dynamic forms: a defensive paradigm that instruments at least one irreducible pillar non-trivially is anté-bruit-resistant against both static operators and dynamic operators (Theorem 5.1). The dynamic Sufficiency is established for detectors measurable with respect to the receiver's temporal σ-algebra ℱ^T; the residual macroscopic channel operator-side (Remark 5.4ter), dual to [Rouxel 2026b, Remark 4.18] on the receiver side, sits outside this scope and is the subject of a companion paper on defensive deception. We state the converse — that exclusive instrumentation of compressible limitations is insufficient — as Conjecture 5.2, inherited from Conjecture 4.2. An Abstraction-Fidelity Envelope Principle (§4.6) specifies the operational condition on instrumentation effectivity: the physical pillar is irreducible in right, but operationally effective only insofar as the abstraction presenting it to the upper layers is faithful to the underlying substrate — Rowhammer and cross-stack attack composition as canonical cases; this principle is marked as structural observation outside the information-theoretic formal scope. The four alternative paradigms flagged as conjecturally vulnerable in [Rouxel 2026b, Remark 8.2 and Conjecture 8.1] receive a structural refinement under this framework — resilience is proved anté-bruit-resistant via the rigorous Sufficiency direction; the vulnerability of deception, diversification, and forensic attribution inherits the conditional status of Conjecture 5.2. The framework is information-oriented in its definitions (which are domain-independent) and articulated around cybersecurity in its bipartition (which co-constructs the irreducible pillars with the operational structure of cyber where both pillars have analog); transposition to other domains is signaled with explicit acknowledgment that domains lacking an operational analog of the physical pillar inherit a weaker bipartition.