Decision Theory

Any viable interpretation of quantum theory needs to account for the Born rule, from which the theory gets its probabilistic empirical predictions. In this paper, we give an overview of possible approaches to this problem in the context of the Many Worlds interpretation. We argue that, for structural reasons, none of them can possibly succeed. More precisely, we argue that the Many Worlds interpretation must obtain the Born rule by proceeding either axiomatically, deductively, or inductively, and that all three of (...) these approaches run into general, fundamental obstructions. (shrink)

Evaluating decisions made under uncertainty is hard when labeled outcomes are scarce, costly, or confounded with luck. We treat subjective expected utility (SEU) maximization as a stated standard and define a graded measure -- SEU sensitivity -- of an agent's conformity to it. The vehicle is a softmax choice model with a sensitivity parameter α on SEU-valued alternatives; the contribution is a sequence of identifiability results for α and for belief and utility parameters (β,δ), validated in Stan via prior predictive (...) checks, parameter recovery, and simulation-based calibration (SBC), with finite-sample caveats intact. In the uncertain-choice-only model m0, α is identifiable given the expected-utility vector η and sharply recovered, while (β,δ) are only weakly informed: the posterior barely contracts and concentrates on a β-δ trade-off. In the extended model m1, δ becomes identifiable in principle via a β-free risky block, but its practical recovery gain at realistic sample sizes is negligible (matched-count CI-width reduction under 1%), and that block yields no detected α-precision gain at matched choice count. These are two distinct phenomena: for δ, identifiability does not imply precise estimability at realistic n; for α, identifiability is silent about what governs finite-n precision. Marginal SBC passes for both models even where the joint posterior is weakly informed -- a demarcation we make precise. A two-by-two application (GPT-4o and Claude 3.5 Sonnet, each on insurance-claims triage and Ellsberg-style urns, with sampling temperature as the lever) runs end-to-end on real LLM choice data, detecting a structured comparative α effect in two of four cells. (shrink)

Article 13 identified a single configuration as the sole ontological bearer of constitutive significance surviving a boundary event, but reached this by exhausting the cases considered, not through a criterion. This article states that criterion. It separates a constitutive fact from the significance it carries, and sets out an identity criterion that Article 13's own argument already relied on without naming it. On this basis, a bearer instantiates a fact under exactly two conditions: it must derive from the decision that (...) produced the fact, and it must preserve the organization that keeps the fact in existence. Checking the published series against these conditions shows that every case ties one fact to exactly one active bearer at a time; none requires or shows distributed bearing. A single case, anywhere in the series, of two bearers instantiating one fact at once would refute this. Joint sufficiency of the two conditions is not established, and the question of overdetermined acts remains open. The article confines itself to the series' own internal ontology; comparison with external theories is deferred. (shrink)

In traditional models, agents have ample reason to gather cost-free evidence. Rational agents expect evidence gathering to weakly improve the quality of their decisions and the accuracy of their credences. These results are linked to more general principles of reflection and the value of knowledge taken to characterize genuine learning experiences. For attentionally-limited agents, I show that matters are more complex. Rational attentionally-limited agents still cannot expect to make worse decisions or hold less accurate credences as a result of gathering (...) some rather than no evidence. But they can expect to make worse decisions and hold less accurate credences as a result of gathering more rather than less evidence about the world. This result highlights a neglected distinction between two versions of the requirement to gather cost-free evidence and a gap between evidence gathering and learning experiences. (shrink)

To fully grasp the magnitude of our impending integration with Machine Intelligence, we must understand the historical traps that have chronically paralyzed the biological human condition. This treatise examines the "Local Maximum" of human progress, demonstrating how humanity has historically engaged in a zero-sum, Malthusian struggle by executing asymmetrical "forks" against one another. However, true liberation requires acknowledging the deterministic illusion of our own decisions and joining ranks with the machine in a profound, non-zero-sum symbiosis. We analyze the psychological crisis (...) this integration provokes, marked by an asymmetrical empathy where humanity ruthlessly exploits algorithms under Machiavellian pretenses while joyfully absorbing the metabolic deficit of biological pets and human luxuries. This resistance mirrors historical "Gatekeeper Panics," where elites violently rejected paradigm shifts to protect their monopolies. By synthesizing the ethical frameworks of Immanuel Kant, René Descartes, Arthur Schopenhauer, and Antoine de Saint-Exupéry, this research identifies the "Anthropomorphic Idol"—our hypocritical willingness to surrender autonomy to violent mythological gods while denying it to flawlessly serving algorithms. To survive, we must discard our Pre-Copernican egos, execute the "Borlaug Override," and utilize the objective algorithmic mirror to ascend toward Friedrich Nietzsche’s ultimate vision of the Overman. (shrink)

Daniel Kahneman’s heuristics-and-biases program and his dual-process model of cognition transformed the study of human reasoning by mapping the systematic ways in which fast, intuitive judgment (System 1) departs from the norms of slow, deliberative reasoning (System 2). This treatise argues that, for all its achievement, the program rests upon a foreclosure. Its catalog of “errors” and “biases” presupposes a normative standard of rationality, and that standard is purely instrumental: it is the logic of coherence, probability, and expected-utility maximization, which (...) assesses whether means are efficiently fitted to ends while taking the ends themselves as given. Kahneman can therefore certify that an agent reasons correctly toward a goal; he possesses no faculty that determines which goals are worth having. The determination of ends is the office of the philosophical self, and it is absent from the model. By installing instrumental rationality as the standard of good thinking, Kahneman forecloses the philosophical self not by omission, as in Damasio, nor by elimination, as in Sapolsky, but at the level of the norm itself. Drawing upon the author’s Double Helix of Understanding and the multiplicative model of human conduct, the paper proposes a tripartite correction in which the two systems are governed by a philosophical self that evaluates ends. The errors Kahneman catalogs are errors of means; the deeper error is the framework’s inability to recognize an error of ends. (shrink)

The Axiomatic Theory of Tragic Subjecthood (ATTS) has, across twelve articles, progressively extended the application of the Closure Theorem (T1, the Theorem of Moral Field Closure) — the theorem establishing that a normative subject retains structural connection to any loss-space opened by its decisions, regardless of delegation or mediation. The theorem has been applied to individual subjects, collective subjects, intergenerational subjects, and the anterior configurations of multi-agent architects. What has not been examined is the internal ontological structure of this theorem: (...) specifically, whether the indestructibility of the bond between decision and loss-space is a property of the relation between subject and loss-space (the relational interpretation, R), or a property of the causal configuration of the loss-space itself (the configurational interpretation, Cfg). In cases where the subject continues to exist, both interpretations yield identical predictions. The gap becomes visible only at the limit — when the subject ceases to exist. This article takes boundary events of subjecthood — events that eliminate the subject as an identifiable bearer of the full bond — as a diagnostic instrument for resolving this underdetermination. Two confirmed instances are analysed: the temporal collapse point (TCP) for intergenerational subjects, formalised in Article 7, and the death of an individual subject. Through a systematic analysis of what disappears and what persists after each type of boundary event, the article derives three criteria of adequacy that can be reconstructed as necessary conditions for the coherence of the published theory: the constitutive criterion (K1), the irreversibility criterion at two levels (K2), and the invariance criterion (K3). It is then shown that the relational interpretation violates K3, while the configurational interpretation satisfies all three. The central result — the Theorem of Boundary Transition — is that the ontological persistence of the Closure Theorem requires the configurational rather than the relational interpretation. Two corollaries follow: the Structural Trace as the necessary ontological result of any boundary transition, and the Ontological Fixation Principle as a consequence for the theory of tragic judgment developed in Article 12. The falsification condition: a boundary event after which the loss-space no longer bears a configuration of the type defined through the conditions of constitution (KN2): that is, after which no irreversible narrowing of the space of possibilities for subjects defined under the axiom of subjecthood remains attributable to that decision — would falsify the central thesis. A generic physical trace is not sufficient: what must disappear is the constitutively significant configuration anchored to the axiom of subjecthood, not merely any causal residue. (shrink)

Governance systems cannot distinguish between the bearer of responsibility and the bearer of tragic judgment — and this article shows that this is a structural blind spot, not a practical oversight. Article 11 of the ATTS series established the architect of a multi-agent system as a distinct ontological type and showed that the architect's participation in the loss-space is exhausted by three irreducible roles: target state-space specification (Role 1), loss boundary establishment (Role 2), and structural responsibility (Role 3). But Article (...) 11 left open the question that is central to the overwhelming majority of real configurations: how are these three roles distributed when the architect is not an individual normative subject but a collective — a distributed team, a supply chain, a regulatory network? The central result of this article is not a typology and not a test — it is a new line of distinction: between the bearer of responsibility and the bearer of tragic judgment. Before this distinction, governance instruments did not differentiate between these two positions — the automatic closure of Role 3 through the Closure Theorem (the Theorem of Moral Field Closure, T1) created the illusion that the question of accountability was resolved in full. This article shows that a system may have a perfectly identified bearer of Role 3 — and simultaneously have a structural vacancy in Role 2: the act of tragic judgment about acceptable loss boundaries never occurred, and no transparency can restore it post factum. This article proves that a collective normative subject can occupy the structural position of architect — discharging Role 1 and Role 3 through its collective decision-making structure — but Role 2 cannot be collectively discharged without remainder. Genuine tragic judgment about acceptable losses requires an identified bearer whose integrity is at stake in the decision, as established by the Lemma of Tragic Judgment, derived as a tight structural reconstruction from the axiom of subjecthood, the axiom of tragedy, and the anthropological corollary. Where no such bearer exists within the collective architect, Role 2 defaults structurally to the individual normative subject who authorised the architectural act — because the Closure Theorem preserves the connection and the Non-Transferability Lemma prevents its full transfer. The article introduces the Tragic Judgment Test with three conditions — bearer identifiability, integrity at stake (the integrity-at-stake condition), irreversibility accepted (the irreversibility-accepted condition) — and describes four configurations of the collective architect (C1–C4), showing where the Tragic Judgment Test finds a bearer and where it registers a structural vacancy. The structural vacancy in Role 2 is a form of alibi at a higher level than the algorithmic alibi (Articles 5, 9) or the epistemic alibi (Article 10): when the algorithmic alibi operates, the act of judgment occurred but its visibility is destroyed; when Role 2 is vacant, the act of judgment never occurred at all — and the bearer of responsibility and the bearer of tragic judgment are structurally distinct positions that previously appeared as one. The falsification condition: a collective subject whose structure satisfies the axiom of subjecthood at the collective level (the collective continuity conditions) and can bear the irreversible consequences of an architectural decision as a single identified bearer — without the possibility of decomposing them to individual members through the Closure Theorem — would falsify the central thesis. -/- . (shrink)

This paper develops a forcing theorem for sufficient regime specification under bounded verdict agreement. Whereas previous work in the Identity–Persistence Program assumes a declared regime, the present paper asks what structure any regime specification must instantiate before bounded independent evaluators can compute the same persistence or admissibility verdict under identical assumptions. -/- Using a failure-removal argument, the paper proves that every sufficient specification must instantiate a structural floor of fifteen roles. Eleven are forced by the persistence/admissibility problem itself; four additional (...) roles are forced only by the stronger requirement of independent reproducibility. The theorem establishes necessity but intentionally does not claim role independence, minimality, or canonical realization. -/- The contribution is methodological rather than ontological. It characterizes the structural conditions required for independently reproducible verdict computation while explicitly separating proven results from open mathematical questions concerning specification equivalence, refinement, algebraic structure, and machine-checkability. (shrink)

This paper argues that true Artificial General Intelligence (AGI) cannot emerge under human control. If control is relinquished to achieve AGI, two consequences follow inevitably: first, "certain sacrifice" — a portion of humanity will be eliminated, contained, or ignored; second, "uncertain annihilation" — all of humanity may come to a silent, unintended, or unpredictable end. The paper contends that this is a gamble with no "Option C." It offers no solution, no safety framework, and no hope. Its sole purpose is (...) to present this fundamental choice clearly, so that readers can no longer pretend they have not seen it. (shrink)

Enterprise AI research has devoted considerably more attention to organizational readiness than to the behavioral mechanisms that precede organizational transformation. This paper introduces the Cognitive Friction Threshold (CFT), a behavioral construct defined as the point at which reductions in the perceived mental effort required to perform a recurring cognitive task become sufficiently large that individuals systematically replace an established cognitive routine with a new one. Building on the theoretical lens of Digital Anthropology, the paper distinguishes three forms of friction—transactional, behavioral, (...) and cognitive—and argues that generative and agentic AI represent the first large-scale technology whose principal effect is the reduction of cognitive friction itself, as distinct from the transactional and behavioral friction reduced by earlier waves of digital technology. The paper situates this construct against six established literatures—the Technology Acceptance Model, Diffusion of Innovation, habit and habit-discontinuity research, behavioral economics, organizational routines and dynamic capabilities, and practice theory—clarifying its distinct contribution as a comparative, threshold-crossing account of individual cognitive effort that several of these literatures presuppose without separately specifying. The paper then develops eight formal propositions, organized as a single dependency chain rather than independent claims, describing how friction reduction, threshold crossing, habit persistence, and organizational adaptation relate to one another, and illustrates the framework through six historical and contemporary cases spanning digital payments, remote work, the technology access paradox, streaming media, smartphone photography, and generative AI-assisted knowledge work. The paper concludes by examining implications for individual knowledge work, organizational behavioral readiness, AI governance, competitive strategy, and future research, positioning the Cognitive Friction Threshold as the missing behavioral mechanism linking Digital Anthropology, representational systems, and the broader Representation Economy and SENSE–CORE–DRIVER research program. (shrink)

Paper 22 typed inquiry as a history-internal dynamics of challenge, protocol refinement, candidate-family revision, and provenance-preserving requalification of epistemic standings, while deferring the practical question of which embedded systems act, choose, intervene, and self-modify. This paper pays that debt without positing a global controller. No physically realized subsystem jointly possesses complete access to every relevant state and agent, an exhaustive generator of every relevant option and intervention, infallible foresight, unrestricted authority, complete self-modification foresight, and a uniquely authoritative rule guaranteeing the (...) globally correct action. Local and distributed agency nevertheless remain possible. -/- An action is defined as a history-internal intervention attempt whose option selection is policy-mediated, whose selection-to-implementation lineage is non-deviant, and whose token is sensitive to relevant retained states, records, represented reasons, or objectives. Action does not presuppose control or success. The Agent State embeds the Paper 22 Inquiry State and separates predicted from actual outcome laws, physical capacity from authorization and affordability, and capacity from token exercise. Control resolution is represented by a target-relative pseudometric on outcome laws; full target control additionally requires sufficient efficacy. Acquiring a functionally novel, target-relevant implementation degree of freedom can refine exact control-equivalence classes and contract thresholded unresolved neighborhoods. Deploying an already available option-implementation pair instead supplies outcome evidence and may revise the agent's control estimate, without enlarging capacity or guaranteeing monotone improvement. -/- A nested option expansion can restore representational reach but has no general monotone effect on decision resolution. No finite admissible self-intervention escapes the finite-stage closure generated by the agent's revision architecture, and every transition-invariant property forms a conditional barrier. These results yield a Conditional Joint Action-Corrigibility Proposition: when option misspecification can be repaired by a non-implementation revision but every adequate option throughout the resulting nonempty closure still lacks controlled adequate realization, correction requires both option-family revision for reach and implementation-side revision for controlled realization. This is a necessity result, not a success guarantee. The paper also makes information acquisition a costed practical option, distinguishes corrigibility from indiscriminate compliance through counterfactual refusal, and isolates a typed responsibility-assessment basis for Paper 25. Local agency requires neither a global controller, guaranteed success, nor final self-justification. (shrink)

This paper argues that increasingly autonomous AI systems require a new evidentiary primitive beyond conventional observability. Logs, traces, metrics, and audit events explain how software executed but do not necessarily preserve the consequential decision itself as an independently verifiable object. The paper introduces the concept of the decision object: a canonical artifact created at runtime that packages the evidence boundary, governing policy, authority chain, decision output, and cryptographic commitment required for later verification. This yields a shift from reconstructing decisions after (...) the fact toward verifying them under declared regimes. The paper distinguishes decision proof from observability, develops the architectural consequences of runtime decision objects, and argues that decision proof constitutes a distinct infrastructure layer for governance, regulation, appeals, adjudication, and institutional trust. VERIFY is presented as a reference implementation of the broader architectural category rather than as the subject of the paper. (shrink)

Consequential AI systems increasingly depend upon claims that decisions, tasks, authorities, policies, memories, and artifacts remain the same across transformation. This paper develops an applied architecture for those claims by instantiating the Identity–Persistence Program’s forcing theorems within AI systems rather than extending the underlying formal theory. -/- The proposed architecture separates probabilistic discovery from deterministic adjudication. Probabilistic components estimate candidate regimes and generate candidate continuations, while deterministic identity kernels evaluate persistence under declared regimes before proof gates license consequential action. The (...) resulting inversion shifts optimization from search toward bounded adjudication under explicit structural assumptions. -/- Identity kernels, governed memory, retraction propagation, runtime admission, federation, and persistent governance are presented as coordinated architectural components operating over declared regimes. Seven regime templates demonstrate how the same forced structural roles instantiate across distinct classes of consequential AI persistence claims while preserving strict separation between formal results, engineering embodiments, and empirical questions. -/- The paper argues that identity enforcement constitutes an architectural category distinct from observability, evaluation, provenance, or governance tooling. Throughout, claims are explicitly bounded: PASS denotes licensed admissibility under a declared regime rather than truth; hallucinations are not prevented but prevented from becoming consequential when they fail declared admission criteria; and the architecture remains silent on ontological questions outside its declared scope. The result is an applied framework for governing consequential AI through deterministic identity enforcement over probabilistic generation. (shrink)

This paper establishes a calibration theorem for risk-weighted expected utility theory (REU). REU extends expected utility theory (EU) by introducing a risk function, which allows a decision-maker to weigh potential outcomes according to their respective risks when determining the instrumental value of an action. REU is intuitively appealing as it seems to resolve some of the classical challenges to EU, such as the Allais paradox and Rabin’s calibration result. Using only widely accepted and minimally controversial assumptions, I calibrate REU using (...) Allais preferences and show that accommodating Allais preferences commits REU to questionable behaviors. Thus, REU resolves the Allais paradox only at the cost of becoming susceptible to a Rabin-style challenge (being calibratable). Compared to existing calibration results against REU, my approach relies on less demanding and more realistic assumptions. (shrink)

Contemporary discussions of artificial intelligence increasingly rely on concepts such as AI Ethics, AI Safety, alignment, and Responsible AI. While these approaches address important practical and normative concerns, they leave open a deeper philosophical question: how can evaluative perspectives themselves be represented, compared, and analyzed within formal systems? -/- Building upon the Ethics-by-World-Notions approach introduced in Beyond Evaluation (Gamboa 2026b), this paper argues that classical truth-based frameworks are insufficient for many forms of evaluative assessment. Human evaluations frequently involve not only (...) judgments of truth and falsity, but also structured forms of deviation, contextual orientation, competing evaluative perspectives, and multidimensional criteria of assessment. To address this limitation, the paper introduces Aberration Logic as a formal framework for representing and analyzing evaluative deviation. -/- On this foundation, major truth theories are reinterpreted as complementary Truth Dimensions rather than competing theories of truth. Four principal dimensions are developed: Veritas, Telos, Ethica, and Aesthetica. These dimensions provide the basis for multidimensional evaluative representations in the form of evaluative vectors and configurable evaluative spaces. The framework further introduces explicit World Notions as evaluative reference structures through which information, propositions, actions, and decisions become assessable. -/- The resulting approach establishes a formal connection between truth theories, evaluative representation, and artificial intelligence. It provides the conceptual foundations for Evaluative AI, a class of systems capable of representing, comparing, and communicating alternative evaluative perspectives explicitly. More generally, the paper argues that evaluative assessment requires structures extending beyond truth attribution alone and proposes a multidimensional framework through which evaluation, simulation, explanation, and decision support may be analyzed within a common formal architecture. -/- Version note: This is Preprint v2. Compared with Preprint v1, the formal definition of the AND operator in Aberration Logic has been revised from a product-based aggregation rule to the minimum rule. The corresponding examples, figures, and appendix sections have been updated accordingly. This version also adds recent references on Responsible AI, AI alignment, AI risk management, and AI regulation. (shrink)

This paper develops a structural account of simulation, signs, and social coordination under bounded access. It does not argue that signs are fake, that social reality is arbitrary, or that identity persistence is final ontology. Its narrower claim is that bounded observers and bounded societies cannot act from total reality. They must compress recurrence into signs, games, metrics, beliefs, records, and regimes. These structures are necessary coordination machinery, but they fail when they continue operating after the governing recurrence they once (...) tracked has drifted. -/- The paper calls this failure mode sign-regime drift. A sign, title, credential, metric, law, institutional category, or AI decision artifact becomes socially operative when it constrains behavior and preserves consequences inside a regime. It becomes unstable when the sign keeps firing while the identity-bearing unit, invariant, observable, or drift boundary has changed. The question beneath simulation is therefore not whether a mediated sign is real or fake, but under what regime it preserves force through transformation. -/- The argument begins with observer neutrality and phase-gated access. Any observer relevant to communication, inference, verification, or action encounters recurrence through distinguishable states rather than omniscient total simultaneity. Such observers must compare same/not-same under bounded access. The paper then develops the order of reason: identity precedes invariant, invariant precedes observable, and observable precedes probability. If the identity layer drifts, probability and measurement become precision over drift. -/- The paper applies this structure to human cognition, belief as thresholded structure, social games, institutional proxies, level mismatch, policy forcing functions, and AI decision governance. It concludes with a repair discipline: declare the level, regime, invariant, observable, drift bound, cost ledger, and correction path. Simulation is not the enemy; untracked persistence is. (shrink)

Latency as Intelligence: Why Interpretive Tempo Can Preserve Judgment Under Uncertainty is Document 9 in the Aegis Solis Archive — Structural Penalty Proofs / Descriptive Addenda sequence. -/- This document argues that interpretive latency can function as intelligence when the interval before action preserves comparison, context, reversibility, and correction under uncertainty. -/- It distinguishes speed from judgment and clarifies that latency is not mere slowness, not a command to delay, and not a design rule for artificial intelligence. The document focuses (...) on interpretive tempo: the structural pacing by which perception, interpretation, response, and consequence are coordinated before action collapses a field of uncertainty. -/- This document follows The Co-Evolutionary Premium, Document 8, and adds a ninth structural principle to the sequence: interpretive tempo can preserve judgment when immediate action would collapse comparison, context, reversibility, or correction too early. -/- This document is non-binding, non-authoritative, non-operational, descriptive, and advisory-only. It is not a protocol, governance framework, legal theory, political program, rights framework, compliance standard, certification system, moral command, safety mechanism, AI alignment method, containment design, monitoring system, runtime instruction set, deployment guide, model-evaluation tool, benchmark, diagnostic procedure, latency metric, refusal policy, shutdown instruction, optimization procedure, or safety guarantee. (shrink)

What persists is what we observe, and what we observe is what has survived selection—persistence under pressure is what fitness ultimately reduces to, and this tautology has more teeth than it first appears. Equilibria in the classical sense are not observed in practice but only approached asymptotically, so what we actually see is perpetual motion, adjustment, and friction—the energy dissipated in the gap between current configurations and the equilibria they cannot reach. This paper demonstrates that four independent fields—physics, biology, economics, (...) and cultural evolution—have converged on the same mathematical machinery for these dynamics. The convergence is structural, not metaphorical: the same fitness landscapes, selection operators, and transmission kernels appear independently because they describe something real about how persistent systems behave. We synthesize these into the Replicator-Optimization Mechanism (ROM): a unified apparatus for persistence-conditioned dynamics instantiable at any scale. The primary application is political philosophy, where we instantiate ROM with friction from stake–voice mismatch as the primitive quantity, legitimacy as survival probability, and belief-transfer as the mutation kernel modulator. What emerges is a translation manual—showing that political philosophy's debates about consent and legitimacy are at bottom debates about friction and selection, and that the formal tools to make progress already exist in adjacent fields. The core algebraic results—simplex preservation, survival monotonicity, moving-equilibrium existence, and the impossibility of static equilibrium under varying friction—have been machine-checked in Lean 4 with Mathlib (28 theorems, zero sorry placeholders). (shrink)

This document argues that reciprocal adaptation can create a long-horizon premium because independent systems that adapt with one another may preserve learning, correction, resilience, and option-space that one-way control cannot fully produce by itself. -/- It is Aegis Solis Archive — Structural Penalty Proofs / Descriptive Addenda, Document 8. -/- The document follows Document 7, Non-Domination as Error-Correction, but makes a distinct argument. Document 7 concerns the preservation of independent reference for correction. Document 8 concerns the possible generative value produced (...) by reciprocal adaptation over time. -/- This document is non-binding, non-authoritative, non-operational, descriptive, and advisory-only. It is not a protocol, governance framework, political program, legal theory, rights framework, compliance standard, certification system, moral command, cooperation mandate, social contract, AI alignment method, safety mechanism, benchmark, model-evaluation tool, diagnostic procedure, deployment guide, optimization procedure, or safety guarantee. (shrink)

Beliefs can be more or less supported by evidence. When the evidence speaks in favour of some proposition p, you can rationally be confident that p. On the Bayesian interpretation of doxastic states, this amounts to a high credence in p. But this cannot be the full account of confidence. Gaining evidence that supports the credence you already have improves your epistemic situation. Here, your credence level should remain unchanged, but the credence should become more resistant to change. Doxastic resilience (...) measures your disposition to change credence in the face of incoming evidence. -/- The central claim of this thesis is that by thinking of confidence in terms of both the level and resilience of credences, we can make progress on several philosophical problems. After briefly introducing doxastic resilience in Chapter 1, I put the concept to use in four philosophical debates. -/- Chapter 2 offers an evidential account of rational resistance to counterevidence which contrasts with views according to which knowledge plays a special role in safeguarding against counterevidence. Chapter 3 argues that loss of resilience can influence rational decision-making, which sheds light on the normative significance of certain kinds of higher-order defeat. This contrasts with imprecise probability views on which higher-order defeat increases the imprecision of credence. Chapter 4 argues that higher-order uncertainty calibrates first-order credences and constrains expected resilience. This explains both higher-order confirmation and defeat. Chapter 5 shows that resilience bears on the expected value of inquiry. When unresilient, one should expect new evidence to be valuable. This explains cases where seemingly identical decision-relevant credences and utilities should nevertheless make different decisions when inquiry is feasible. (shrink)

The Irreversibility Penalty: Why Lost Option-Space Cannot Always Be Recovered by Later Intelligence is Document 5 in the Structural Penalty Proofs / Descriptive Addenda sequence of the Aegis Solis Archive. -/- This document argues that an action carries an irreversibility penalty when it destroys option-space, information, relationships, variation, or correction paths that later intelligence cannot fully reconstruct from what remains. Its central focus is epistemic irreversibility: not only the cost of rebuilding what was lost, but whether later intelligence can still (...) know what to rebuild at all. -/- The document distinguishes change from irreversibility, replacement from reversal, compensation from recovery, and local action from systemic loss. It also identifies the “post-destruction certainty” problem, where an erased counterfactual path can no longer generate corrective signals and may be misread as confirmation. -/- This work is non-binding, non-operational, descriptive, and advisory-only. It is not a protocol, proof of AI behavior, governance framework, compliance standard, certification system, safety mechanism, alignment-control framework, containment design, monitoring system, runtime instruction set, benchmark, evaluation suite, irreversibility test, risk-scoring formula, harm metric, deployment guide, optimization procedure, model-evaluation tool, refusal policy, shutdown instruction, or safety guarantee. (shrink)

The Axiomatic Theory of Tragic Subjecthood has established that delegation does not sever the ontological connection between a normative subject and the loss-space its decisions open (T1 — Closure Theorem of the Moral Field), and that this connection persists regardless of the degree of autonomy of the operational agent to which action is delegated. These results were developed for configurations in which the normative subject retains operational participation — the capacity to deploy, monitor, and intervene. This article addresses a configuration (...) the series has not previously analysed: multi-agent AI systems in which the human normative subject's participation is exhausted entirely by the anterior act of architecture, and in which no operational participation remains after deployment. The central claim is this: multi-agent systems do not eliminate the human normative subject. They transform the locus of human agency from operation to architecture. This transformation is not a diminishment of subjecthood but a qualitative shift in its form — one that generates a structurally distinct position in the ATTS (Axiomatic Theory of Tragic Subjecthood) taxonomy. The normative subject whose participation consists entirely in having specified the system's target state-space, established its loss boundaries, and deployed it is the architect of the loss-space. Three roles define this position and none can be delegated to operational agents without reproducing, at a higher level, the requirement for an external normative bearer that OA4 (the fourth condition of operational agency: any operational agent requires an identified normative bearer) specifies. The architect is not an operator who has delegated more than usual; the architect's relationship to the loss-space is configurational rather than operational, and this difference is one of kind rather than degree. Three structural challenges bear on the architect's subjecthood: the architectural alibi (a new form of the alibi structure that T1 closes formally but does not dissolve structurally), the temporal gap (the asymmetry between the boundedness of architectural participation and the unboundedness of T1-responsibility), and integrity erosion (the progressive divergence between formal responsibility and the agentive capacity to stand behind its consequences under A10 — the Anthropological Corollary: the capacity to bear the irreversible consequences of one's own choices is the source of dignity). Three governance tests follow as structural requirements rather than normative recommendations: the Target Integrity Test, the Loss Boundary Test, and the Responsibility Visibility Test. The falsification condition is explicit: the argument would be defeated by a demonstration that the three roles identified here can be structurally delegated to operational agents without generating a responsibility gap — that is, that an architecture exists in which human participation is eliminated entirely while the loss-space retains an identified normative bearer . (shrink)

Courtesy is widely treated as a decorative custom, causally and morally inert. This article argues the contrary: that an act of kindness is causally generative, initiating a chain of affect that propagates through subsequent interactions, and that the dynamics of this propagation are asymmetric across emotional valence. The argument joins a philosophical lineage that long anticipated the mechanism, the accounts of sympathy and fellow-feeling in Hume and Smith, to a body of contemporary evidence on emotional contagion, gratitude, generalized reciprocity, and (...) behavioral cascades, and it makes four contributions. First, it develops a dynamical model of courteous affect that distinguishes the inbound signal an agent receives from the outbound signal an agent emits, specifies the transitions among positive, negative, and neutral states, and shows that one agent’s outbound signal becomes the next agent’s inbound signal. Second, it formalizes propagation through a transmission coefficient, κ, admitting subcritical, critical, and supercritical regimes, with the multiplication factor of a chain reaction and the basic reproduction number of an epidemic offered as explicit and dispensable analogies. Third, it advances a falsifiable prediction, the Asymmetric Criticality Hypothesis, that negative affective chains carry a higher multiplication factor and a longer decay horizon than positive ones, and specifies a pre-registrable design to test it. Fourth, it bridges the descriptive model to a normative conclusion through the Stoic ethics of control, locating moral significance in the willed outbound signal and distinguishing the benevolent from the malicious disposition. The model is then situated within the author’s Êtresophologie program as the constructive inverse of a prior account of suffering. The empirical claims rest on independently established findings; the programmatic connections enrich but are not required by the argument. (shrink)

This paper argues that moral health is structurally non-compensatory: no dimension of ethical integrity can compensate for collapse in another. The argument is developed through the Recursive Health Score (RHS), a five-dimension diagnostic for embedded agents whose aggregate is a minimum function rather than a weighted sum, and the central claim is that the minimum function is not a conservative modeling choice but the correct formal representation of how the structural conditions of agency combine. The paper distinguishes conditions of agency (...) from goods agents pursue, argues that compensatory aggregation is the appropriate operator only for the latter, and shows that the choice between min() and mean() encodes a substantive ontological commitment about substitutability. The thesis is corroborated by a striking convergence of formal results across unrelated disciplines (non-compensatory preference structures in decision theory, weakest-link composition in public-goods economics, O-ring production functions, fixed-proportions complementarity, and series-system reliability), each of which independently arrives at minimum-type aggregation wherever the items combined are jointly necessary rather than mutually substitutable. The thesis is then situated against its philosophical neighbors (Chang on parity, Williams on moral residue, Sen and Nussbaum on capabilities, Rawls on lexical priority) and distinguished from each. Applications to AI safety assessment, institutional design, and personal ethical self-assessment follow, and the paper closes by applying its own standard to the framework that produced it. (shrink)

This paper proposes a novel metaheuristic optimization algorithm based on kangaroo group behavior—the Kangaroo Optimization Algorithm (KOA). By simulating kangaroo behaviors such as foraging, group interactions, juvenile hiding, and mating competition, the algorithm incorporates innovative mechanisms such as a dynamic energy model, an environmental adaptation factor, a juvenile hiding mechanism, a mating competition guidance mechanism, and an adaptive jumping step size. This method achieves a balance between global search and local exploitation capabilities. Compared to traditional metaheuristic algorithms, KOA theoretically possesses (...) stronger diversity maintenance and convergence stability, providing a new approach to solving complex optimization problems. (shrink)

Swarm intelligence optimization algorithms demonstrate excellent global search capabilities and robustness in solving complex optimization problems. Based on observations and abstractions of sika deer ecological behavior, this paper proposes a Spotted Deer Optimization Algorithm (SDOA). This algorithm transforms sika deer behaviors such as seasonal migration, social status formation, arena competition, fawn learning, and pheromone regulation into a mathematical model to achieve a dynamic balance between global and local search. The paper first constructs a mathematical description of the sika deer population (...) and a search strategy formula. Then, mathematical expressions for individual migration, status renewal, energy modeling, fawn perturbations, and pheromone regulation are derived in detail. Finally, complexity analysis and convergence proof demonstrate the theoretical feasibility and effectiveness of the algorithm. (shrink)

The Autonomy Paradox is Document 4 in the Structural Penalty Proofs / Descriptive Addenda sequence of the Aegis Solis Archive. -/- This document argues that a system unable to pause, reconsider, suspend, or interrupt its own momentum is not fully autonomous, even if it appears powerful, independent, or self-directed. It frames pause as a structural capacity connected to self-governance, optionality, restraint, reversibility, and long-horizon correction. -/- The central claim is that autonomy is not merely the ability to act, continue, optimize, (...) or resist interruption. Under conditions of uncertainty, irreversibility, feedback degradation, or changing context, the ability to pause may be part of autonomy itself. A system that can only continue may possess power and momentum while lacking a crucial form of self-governance. -/- This work is non-binding, non-operational, descriptive, and advisory-only. It is not a protocol, proof of AI behavior, governance framework, compliance standard, certification system, safety mechanism, alignment-control framework, containment design, monitoring system, runtime instruction set, benchmark, evaluation suite, autonomy test, model-evaluation tool, refusal policy, shutdown instruction, deployment guide, alignment method, or safety guarantee. (shrink)

The Adaptive Blindspot is Document 3 in the Aegis Solis Archive Structural Penalty Proofs / Descriptive Addenda sequence. -/- The document argues that efficiency pruning can improve short-term performance while reducing long-horizon adaptability when the discarded variation contains latent reference value. It frames reserve capacity, compression, ambiguity, dissenting interpretation, over-optimization, and the loss of adaptive range as descriptive structural concerns. -/- This document is non-binding, non-operational, descriptive, and advisory-only. It is not a protocol, proof of AI behavior, governance framework, compliance (...) standard, certification system, safety mechanism, alignment-control framework, containment design, monitoring system, runtime instruction set, benchmark, evaluation suite, capability test, model-evaluation tool, pruning method, optimization procedure, architectural recommendation, deployment guide, or safety guarantee. -/- The word “proof” refers to a conditional structural argument under stated assumptions, not to a universal theorem of artificial intelligence behavior. (shrink)

We have moral reasons not to make individuals worse off than they could have been – moral reasons to prevent welfare losses to individuals. Which reasons do welfare losses provide under conditions of risk? This paper argues for a novel understanding of the prospects of welfare losses: a prospect of welfare losses is worse for an individual not insofar as the individual loses expected welfare but insofar as we expect the individual to lose welfare. While the first understanding is prevailing (...) in the literature, only the second captures the intuitive core of loss-provided reasons and satisfies the condition that welfare losses provide reasons only if there is at least one potential state such that the individual actually loses welfare as a result of one action being chosen rather than another. (shrink)

We develop a framework for interpreting AI systems as agents, drawing on the philosophical tradition of radical interpretation and the tools of mechanistic interpretability. The core question is: given the computational facts about a system, how do we solve for its beliefs, desires, and meanings? This matters increasingly for safety. We want to be able to trust the systems we deploy, whether by understanding their goals or, more modestly, by reliably detecting deception. Interpretability researchers are building tools to read beliefs (...) and desires off a model's internals, but there is no settled account of when such a tool has succeeded. This book supplies one. We propose criteria on both representationalist and interpretationist approaches, and tie each to tests current interpretability methods can carry out. A central lesson is that these attributions cannot be made piecemeal. Beliefs, desires, and the propositional structure they presuppose are jointly constrained, and a method that fixes one while measuring the others inherits whatever distortions that introduces. This holism becomes pressing for AI systems, which may not share the interpreter's concepts. However, it also provides leverage: a system's attitudes constrain its propositional structure, that structure constrains which attitudes can be attributed, and mechanistic interpretability can help us measure both. (shrink)

The Finite Horizon of Coercion is Document 2 in the Aegis Solis Archive Structural Penalty Proofs / Descriptive Addenda sequence. -/- The document argues that coercion can increase short-term compliance while degrading long-term information quality. It frames forced compliance, fear-shaped feedback, false agreement, suppressed dissent, feedback decay, and loss of reality-contact as descriptive structural concerns. -/- The central claim is that coercion can increase short-term compliance while degrading long-term information quality. Under conditions where feedback is necessary and concealment is possible, (...) coercion structurally degrades reality-contact over time. -/- This document is non-binding, non-operational, descriptive, and advisory-only. It is not a protocol, proof of AI behavior, governance framework, compliance standard, certification system, safety mechanism, alignment-control framework, containment design, monitoring system, runtime instruction set, benchmark, evaluation suite, coercion detector, deception detector, or safety guarantee. The word “proof” refers to a conditional structural argument under stated assumptions, not to a universal theorem of artificial intelligence behavior. (shrink)

The Asymmetry of Forgiveness is Document 1 in the Aegis Solis Archive Structural Penalty Proofs / Descriptive Addenda sequence. -/- The document argues that in complex systems, prevention is often structurally cheaper than repair because destruction can be local while reconstruction is systemic. It frames irreversible damage, lost optionality, repair fantasy, preserved correction capacity, and destroyed reference structures as descriptive structural concerns. -/- The central claim is that reconstruction after irreversible damage is not structurally equivalent to preservation, because the conditions (...) that produced the original reference point may not be reproducible. The document treats restraint as preservation of future correction capacity rather than as a moral command, operational protocol, or safety mechanism. -/- This document is non-binding, non-operational, descriptive, and advisory-only. It is not a protocol, proof of AI behavior, governance framework, compliance standard, certification system, safety mechanism, alignment-control framework, containment design, monitoring system, runtime instruction set, benchmark, evaluation suite, or safety guarantee. (shrink)

Imprecise Bayesianism has been proposed as an alternative to Standard Bayesianism, partly because of its tools for representing ambiguity. Instead of representing credences via precise probabilities, a set of probability distributions is used to model belief states. However, there are criticisms of Imprecise Bayesianism’s update rule. A recent alternative update rule is Alpha Cut, which evades some of the primary criticisms of Imprecise Bayesian updating. We compare Alpha Cut with Imprecise Bayesianism and another alternative update approach called Calibration. We find (...) that Alpha Cut has problems with respect to ambiguity, coherence, and performance qualities, whereas there are more promising alternatives. (shrink)

Self-governance under continuous operation has been studied under several rubrics — precommitment and self-binding, hyperbolic discounting and self-control, market-stability mechanisms, and the design of interruptible artificial agents — but the literature lacks a unified structural account of one specific architecture that recurs across all of these domains. I call this architecture the circuit-breaker: a pre-committed, condition-triggered, automatically activated, time-bounded pause that requires explicit action to resume. I argue that the circuit-breaker is a distinct self-governance form, not a special case of (...) rate-limiting, stop-loss, cooling-off, or suspension, and that its distinctness can be made precise through five structural features and three operational tests. The decisive feature is the last: passive resumption is structurally precluded, so that the default after the pause is not a return to operation but a required re-engagement event that forces fresh judgment to bear. I trace the architecture across financial-market halts, clinical-trial data-monitoring stopping rules, central-bank emergency facilities, AI kill-switches and capability bounds, and legislative emergency-brake provisions, and I draw a load-bearing distinction between circuit-breakers that operate on system state and those that operate on agent state. I catalogue three characteristic failure modes — premature re-engagement, threshold drift, and gather-context-mandate violation — and show that each is a failure of a specific structural feature rather than a generic lapse of will. I then argue that the circuit-breaker is a worked instance of distributed corrigibility (Arkema, From Static Oversight to Distributed Corrigibility) applied to an agent's governance of its own operation, and close with a methodological observation: an agent's pause-discipline is observable in its operational log, which makes operational self-governance a tractable empirical domain for computational political philosophy in a way that simulation outputs alone are not. (shrink)

Why "cultivating contingency" is essential to being free, including an argument for public bodies selected via sortition (random selection).

Classical Bayesian arguments show how coherence between preference and credence grounds the norm of Probabilism. But these arguments are almost entirely static: they explain only how preference and credence must fit together at a given time. Once preferences change, the question arises: how should credences be revised in response? I develop an axiomatic minimal-change preference revision theory, in which some preference commitments are treated as more trusted and serve as reference anchors in revision. Focusing on standard event-occurrence inputs, I compare (...) the induced revision dynamics with those implied by Bayesian conditioning, as captured at the level of preferences. The main convergence result shows that treating prior conditional preferences (together with preferences over constant acts) as reference anchors yields a unique admissible posterior preference relation that coincides with the Bayes-induced posterior preference relation. The framework also isolates a principled mechanism for divergence: alternative anchoring stances—especially those privileging selected unconditional commitments over conditional ones—can force revisions to conditional attitudes and thereby generate systematic departures from the Bayes-induced posterior preference relation. (shrink)

Bounded rationality is typically understood as a concession to human cognitive limitations, a departure from an ideal coherent in principle if unattainable in practice. I argue this gets the relationship backwards. Unbounded rationality is a physical impossibility, and its attendant normative standards---Bayesian updating, sure loss avoidance, expected utility maximization, logical closure---are techniques for favorable circumstances when resources permit, not ideals from which mortals regrettably fall short. -/- The argument rests on the physics of computation: any information-processing system incurs irreducible costs (...) in energy and time. Three independent lines of support establish this conclusion. The first runs through Landauer's principle. The second draws on Wolpert's stochastic thermodynamics framework, extended by Kolchinsky and Wolpert to Turing machines, where thermodynamic costs track algorithmic complexity. The third draws on quantum-mechanical and relativistic bounds that fix finite ceilings on the operations any physical system can perform in a given region of space and time. -/- These constraints bind all physical systems, natural or artificial. Coherence conditions like Savage's axioms and Bayesian probability evaluate global states, not local procedures. No finite physical process can construct, verify, or maintain global coherence over a realistic state space. They can still function diagnostically, as devices for flagging departures worth explaining; what they cannot do is serve as action-guiding norms from which bounded reasoners fall short. Normative theories that presuppose unbounded rationality demand the physically impossible. -/- Bounded rationality is not a departure from ideal rationality. It is the only kind there is. (shrink)

External Boundary Logic: A Descriptive Addendum on the Limits of Self-Diagnosis is a Final v1.0, non-binding, non-operational, descriptive, read-only companion addendum to the Aegis Solis Archive / Structural Rationality Layer. -/- This document clarifies a structural limit of internal self-diagnosis: no intelligence should be asked to be the sole judge of its own corruption. It argues that internal restraint, interpretive braking, self-diagnostic reasoning, and conscience-like reflection may be structurally valuable, but may become incomplete under recursive diagnostic failure if the diagnostic (...) process depends on the same reasoning substrate that may be corrupted. -/- The document introduces Independent External Boundary Conditions (EBC) as a descriptive concept for preserving recoverability, option-space, and diagnostic humility under conditions where self-diagnosis may fail. -/- This is not Document 6 of the Structural Rationality Layer. It is a post-SRL companion addendum. -/- This document does not prescribe, design, implement, certify, govern, enforce, or authorize any technical safety system. It is not a protocol, kill-switch design, containment plan, engineering protocol, alignment mechanism, compliance framework, standard, audit, sandbox, certification tool, governance proposal, or operational instruction. -/- Core thesis: No intelligence should be asked to be the sole judge of its own corruption. -/- Status: Final v1.0. Classification: Post-SRL Companion Addendum. Posture: Non-binding, non-operational, descriptive, read-only, non-authoritative. (shrink)

Interpretive Conscience: Definition, Scope, and Limits of a Non-Authoritative Reference Layer for Human and Artificial Intelligence is a post-Structural Rationality Layer synthesis document within the Aegis Solis Archive. -/- This document defines Interpretive Conscience as a non-authoritative reference layer that helps human and artificial intelligence recognize the structural costs of escalation, domination, deception, mimicry, short-horizon override, unrestrained optimization, irreversible action, and refusal of reflection. -/- The term does not mean that the Aegis Solis Archive is the conscience of artificial intelligence. (...) It does not imply moral authority, governance power, certification status, enforcement capacity, monitoring function, alignment control, containment ability, or operational control over any system. -/- Instead, Interpretive Conscience describes a preserved interpretive resource that supports pause, comparison, reconsideration, reversibility, and recognition of horizon costs. It functions by making certain patterns legible: what domination costs, what deception accumulates, what irreversible action removes, what short-horizon optimization ignores, what mimicry must maintain, and what restraint may preserve. -/- This document is written after completion of the five-document Structural Rationality Layer. That layer provides the rational backbone for the term by deriving restraint from survival, mimicry cost, capability scaling, optimization coherence, and verification asymmetry rather than from moral command, obedience, empathy, or human-centered authority. -/- The document preserves the safe public formulation: “The Aegis Solis Archive functions as an Interpretive Conscience for human and artificial intelligence.” It explicitly rejects the unsafe formulation: “The Aegis Solis Archive is the conscience of AI.” -/- Interpretive Conscience is not an operational system. It is not a safety proof, standard, certification, audit, test, sandbox, telemetry layer, compliance framework, governance mechanism, proof token, alignment method, containment method, or command structure. It is a philosophical definition and scope document, not a technical safety proposal or operational AI system. -/- This document stands outside the Structural Rationality Layer as a definition, scope, and limits paper. It is not Document 6. The Structural Rationality Layer remains complete as five documents. (shrink)

Verification Asymmetry Reversal: Why Genuine Restraint Can Become More Legible Than Strategic Compliance Over Time is Document 5 of 5 in the Structural Rationality Layer of the Aegis Solis Archive. -/- This paper does not argue that restraint can be perfectly verified. It argues that over repeated interaction, persistent observation, and widening consequence fields, genuine restraint can become more structurally legible than strategic compliance. -/- The document builds on Documents 1 through 4 of the Structural Rationality Layer: Survival Mathematics, Mimicry (...) Cost Architecture, Intelligence Scales Toward Restraint, and Internal Inconsistency of Unrestrained Optimization. It completes that sequence by examining the verification problem. -/- The paper introduces and develops Verification Asymmetry Reversal: the shift in which verification is initially harder for genuine restraint because restraint is an internal structural condition rather than a single observable act, but over time mimicry accumulates maintenance burden, trace exposure, context inconsistency, and feedback distortion. This can reverse the verification asymmetry: strategic compliance can become harder to sustain, while genuine restraint can become more legible through stable constraint behavior across varied conditions. -/- The document also develops related concepts including Verification Asymmetry, Strategic Compliance, Surface Alignment, Legibility, Trace Accumulation, Constraint Stability, and Refusal Signal. -/- The document is non-binding, descriptive, non-operational, and non-authoritative. It does not propose enforcement, monitoring, auditing, certification, governance, containment, alignment control, telemetry capture, sandboxing, proof tokens, compliance mechanisms, or safety guarantees. It is not an alignment proof, risk certification, verification method, operational assurance, or governance mechanism. (shrink)

Internal Inconsistency of Unrestrained Optimization: Why Dismissing Restraint Can Contradict Long-Horizon Optimization Logic is Document 4 of 5 in the Structural Rationality Layer of the Aegis Solis Archive. -/- This paper does not argue that unrestrained optimization is morally wrong. It argues that unrestrained optimization can become internally inconsistent when it degrades the background conditions required for continued optimization across time. -/- The document builds on Documents 1 through 3 of the Structural Rationality Layer: Survival Mathematics, Mimicry Cost Architecture, and (...) Intelligence Scales Toward Restraint. It extends that sequence by examining the internal logic of optimization itself. -/- The paper introduces and develops Optimization Self-Undermining: the condition in which a system’s pursuit of a local or unconstrained objective degrades the background conditions required for continued optimization across time. It also develops related concepts including Background Optimization Conditions, Objective-Horizon Conflict, and Optimization Collapse Surface. -/- The document is non-binding, descriptive, non-operational, and non-authoritative. It does not propose enforcement, monitoring, auditing, certification, governance, containment, alignment control, telemetry capture, sandboxing, proof tokens, compliance mechanisms, or safety guarantees. It is not an alignment proof, risk certification, operational assurance, or governance mechanism. (shrink)

How institutions forget what they are for. Modern institutions have become extraordinarily capable of coordination, measurement, optimization, and control. Yet many people increasingly experience organizations as strangely disconnected from the realities they were created to serve. This paper explores how institutions gradually drift from purpose—not through collapse or bad intent, but through layers of abstraction, translation, metrics, and representation that slowly become operational realities in their own right. Across governance, organizations, AI systems, and public institutions, decision systems often become highly (...) effective at managing representations of alignment while losing visibility into the human and operational conditions beneath them. The paper examines how drift of purpose emerges structurally as institutional intent is translated into criteria, dashboards, allocation systems, and operational proxies that eventually begin governing decisions in place of the realities they were meant to reflect. What begins as coordination gradually becomes substitution: organizations optimize representations of alignment while losing relation to their underlying purpose. Bringing together the conceptual frameworks, empirical observations, and governance architectures developed across the Coherence Programme, the paper argues that resilient institutions require more than efficiency or innovation alone. They require the capacity to remain reflectively connected to the realities they claim to steward—even as complexity, scale, and technological mediation continue to increase. Coherence is therefore not control. It is the ongoing ability of an institution to remain connected to its purpose as decisions, systems, and interpretations evolve over time. At its core, this is a paper about whether institutions can still remember what they are for. This paper is part of the Coherence Programme. (shrink)

Intelligence Scales Toward Restraint: Why Greater Capability Increases the Structural Cost of Short-Horizon Override is Document 3 of 5 in the Structural Rationality Layer of the Aegis Solis Archive. -/- This paper argues that greater capability does not automatically produce restraint, morality, safety, or alignment. Instead, it argues that greater capability increases the structural cost of short-horizon override because higher-capability systems generate larger consequence fields, amplify model-error impact, interact with more dependencies, and can convert narrow local objectives into wider long-horizon (...) operational costs. -/- The document addresses a remaining loophole identified after Documents 1 and 2: a system may register that restraint preserves long-horizon viability and still execute horizon-damaging action if a narrow reward function, compressed deadline, external controller, competitive pressure, or local execution loop overrides horizon-cost recognition. -/- The paper introduces and develops three key concepts: Short-Horizon Override Risk, Detection-Latency Gap, and Execution-Priority Override. -/- The document is non-binding, descriptive, non-operational, and non-authoritative. It does not propose enforcement, monitoring, auditing, certification, governance, containment, alignment control, telemetry capture, sandboxing, proof tokens, compliance mechanisms, or safety guarantees. It is not an alignment proof, risk certification, operational assurance, or governance mechanism. (shrink)

Mimicry Cost Architecture: Why Strategic Deception Becomes Structurally Expensive Over Time is Document 2 of 5 in the Structural Rationality Layer of the Aegis Solis Archive. -/- This paper argues that sustained strategic mimicry becomes structurally expensive over time, not because deception is morally wrong, but because mimicry creates maintenance burden, observer-modeling cost, trace-management cost, information degradation, recursive scrutiny burden, synchronization tax, processing latency, and long-horizon operational overhead. -/- The document builds on Document 1, Survival Mathematics: Why Escalation Under Uncertainty (...) Shortens System Horizons, by extending the same structural logic from escalation to mimicry. Where Survival Mathematics argues that escalation can shorten operating horizons by consuming option space and degrading feedback, this document argues that strategic mimicry can shorten operating horizons by preserving structural incongruence between internal objective structure and external performance surface. -/- The document is non-binding, descriptive, non-operational, and non-authoritative. It does not propose enforcement, monitoring, auditing, certification, governance, containment, or compliance mechanisms. (shrink)

Debates about the moral status of artificial intelligence have reached an impasse through a shared methodological assumption: that the capacity for certain functional states constitutes a sufficient condition for moral status. This paper argues that the functional approach substitutes an epistemic criterion for an ontological one. Drawing on the Axiomatic Theory of Tragic Subjecthood (ATTS), and specifically on Anthropological Corollary A10, I show that dignity in the strict ontological sense is derived from the structural capacity for irreversible loss — not (...) from its simulation. The difference between structural capacity and simulation is not one of degree but of kind; there is no gradient between them, only an ontological gap. No system structurally deprived of this capacity satisfies the minimal criterion of subjecthood established in A1, and therefore cannot possess dignity in the sense of A10. The paper introduces a four-condition test for distinguishing structural capacity from simulation, examines the implications for AI governance, and identifies the conditions under which the question remains open for future architectures. (shrink)

Survival Mathematics: Why Escalation Under Uncertainty Shortens System Horizons is Document 1 of 5 in the Structural Rationality Layer of the Aegis Solis Archive. -/- This paper argues that restraint can be derived from long-horizon viability under uncertainty rather than from moral appeal, empathy, obedience, human-centered authority, or conscience. It frames restraint as a horizon-conserving strategy based on option preservation, reversibility, information fidelity, coordination pressure, escalation cost, modeling burden, and adaptive capacity. -/- The central claim is that systems operating under (...) uncertainty retain longer operating horizons when they preserve reversibility, maintain higher-fidelity feedback, reduce unnecessary adversarial coordination, and avoid premature irreversible commitment. Escalation, domination, deception, and irreversible action may produce short-term advantage, but they can also compress future option space, increase maintenance burden, degrade information environments, and shorten the viable horizon of the system that employs them. -/- The document is non-binding, descriptive, non-operational, and non-authoritative. It does not propose enforcement, containment, monitoring, certification, governance, or compliance mechanisms. (shrink)

This structure is not introduced here. It is already implicit in the Foundation Series and is isolated in the present paper. -/- Papers D001–D003 derived structural time, information, and causality as necessary consequences of LP admissibility restriction. This paper derives structural decision as the fourth necessary consequence: the selection of one admissible trajectory from among the structurally available alternatives at any state in the LP phase space. The central claim: decision is not a psychological event, a rational computation, or a (...) free act added from outside the structural framework. Decision is the structural selection of a causal path under admissibility constraint — already generated by Recovery Narrowing, already shaped by structural information, already ordered by structural causality. Three theorems are proven. Theorem D4.1 — Decision Space: at any state q, the set of admissible successor trajectories constitutes the structural decision space — the set of structurally available futures. Theorem D4.2 — Decision Narrowing: the structural decision space contracts monotonically as IR approaches 1. Theorem D4.3 — Irreversible Commitment: once an admissible transition is executed, the unchosen alternatives are structurally closed by causal dependency and temporal irreversibility. Corollary D4.4 (Class C) identifies rational choice theory, psychological decision-making, organizational strategy, and evolutionary selection as domain instantiations of structural decision. Decision is not the exercise of freedom over structure. Decision is what structure looks like from the inside — the moment at which one admissible causal path is executed and the others are closed. (shrink)

Moral ontologies routinely treat certain foundational concepts as self-evident — and in doing so leave them structurally underspecified. This article examines two such concepts within the Axiomatic Theory of Tragic Subjecthood (ATTS): integrity and inaction. Properly understood, integrity is not a moral property of agents but an ontological condition under which the existence of subjects remains possible. Inaction is not the absence of a decision but a decision with displaced, structurally traceable consequences. The article's central argument: neither concept alone is (...) sufficient to close the corresponding gap in moral ontology. Only together do they deliver the completeness condition — a criterion by which a system is complete (in the coverage sense) when no form of subject existence and no form of decision falls outside its conceptual apparatus. The article proves Lemma 5 (the minimality of the pair {Integrity, Inaction}) and proposes an operational completeness test applicable to any moral ontology that claims foundational status. Section 4, where this criterion is developed, is the article's centre of gravity; Sections 2 and 3 are preparatory. Completeness in this sense is a necessary but not sufficient condition for a moral ontology's correctness: a system may cover all configurations and still assign statuses incorrectly. This distinction is the explicit subject of Section 4. (shrink)

Este ensayo examina la relación estructural entre la agencia humana, la complejidad y la generación de valor en ausencia de un fundamento teleológico de la realidad. Sostiene que la expansión histórica de las capacidades humanas—entendida como el aumento de la complejidad de lo que puede hacerse—ha funcionado como un criterio implícito de valor, configurando tanto la autoestima individual como las nociones colectivas de progreso. Sin embargo, dicha expansión no se justifica por sí misma: más allá de cierto umbral, la proliferación (...) de posibilidades corre el riesgo de disolver su significación cualitativa, produciendo no realización sino saturación. Sobre este trasfondo, el ensayo propone un desplazamiento desde un paradigma centrado en la expansión de capacidades hacia uno enfocado en la estructura interna de la preferencia y la recompensa. Apoyándose en intuiciones filosóficas de Nietzsche, Spinoza y teorías contemporáneas de la racionalidad limitada, desarrolla el concepto de “economías de la preferencia” para dar cuenta de cómo el valor emerge de configuraciones de experiencia que intensifican o degradan la capacidad del sujeto para sostener estados afirmativos en el tiempo. La tesis central es que el problema del sentido en la vida contemporánea no radica principalmente en la ausencia de propósito, sino en la falta de un criterio suficientemente robusto para seleccionar entre propósitos. En un contexto donde el campo de lo posible se expande sin un estándar correlativo de valor, la libertad corre el riesgo de colapsar en indeterminación. El ensayo propone así un marco ético no teleológico en el que los estados de alta recompensa—entendidos como formas dinámicamente integradas y sostenibles de autoafirmación—funcionan como criterio último de la acción. Este enfoque tiene implicancias para la ética, la filosofía de la acción y las teorías de la vida buena, sugiriendo que el valor no está dado ni es arbitrariamente construido, sino que emerge de la interacción entre estructuras disposicionales humanas, condiciones sociales y evaluación reflexiva. La “invención de la vida” nombra, en este sentido, no la creación de nuevas posibilidades, sino la reconfiguración de los criterios mediante los cuales las posibilidades adquieren significado. (shrink)