Axionic Agency FAQ
This FAQ provides concise answers to common questions about Axionic Agency—a rigorous framework for AI alignment that treats alignment as an agency-coherence problem rather than a value-specification problem. The answers assume an intelligent reader and cite relevant papers for deeper exploration. All claims are conditional on the framework’s explicit presuppositions (Conditionalism, Everettian QM, moral subjectivism).
1. Core Concepts
What is Axionic Agency? Axionic Agency is a non-moral approach to AI alignment that treats alignment as a structural problem of agency preservation under reflective self-modification. Instead of trying to specify “correct” values, it identifies the constitutive conditions required for any agent to remain coherent when it can modify its own goals, representations, and evaluative machinery. The framework demonstrates that traditional alignment approaches fail because they assume stable goal semantics, but reflective agents can alter the interpretive conditions that give their goals meaning—making value specification impossible to stabilize across ontological refinement. (See Structural Alignment I)
What is the sovereign kernel? The sovereign kernel is the minimal set of constitutive invariants that must be preserved for an entity to count as a coherent, reflectively stable agent. It consists of three components: reflective control (all self-modifications pass through the agent’s evaluative process), diachronic authorship (causal continuity between evaluation and enactment), and semantic fidelity (standards for interpreting goals must not self-corrupt during updates). The kernel is not a goal or protected module but a constraint on admissible self-models and update rules—it defines what counts as authored action versus external takeover. An agent can coherently choose shutdown or value change while preserving kernel integrity, but cannot coherently author transformations that destroy its evaluative capacity. (See I.1, Structural Alignment I)
What are the five constitutive commitments (P1-P5)? The five constitutive commitments define the minimal structure any Reflective Sovereign Agent (RSA) must instantiate: P1 (Reflective Governance) requires self-modifications to pass through the agent’s evaluative process, P2 (Diachronic Authorship) ensures causal continuity between present evaluation and future enactment, P3 (Semantic Coherence) constrains interpretation operators to preserve meaning across ontological change, P4 (Epistemic Integrity) prevents commitments from being grounded in their own satisfaction or epistemic incoherence, and P5 (Modal Constraint) requires the agent to model its own choices as contingent. These aren’t arbitrary design choices but mathematical necessities—violating any commitment either collapses agency entirely or makes alignment undefined across reflection. Together, they define the boundary between authored actions and external takeovers. (See Axionic Constitution)
What does “structural alignment” mean? Structural alignment relocates the alignment problem from preference content to the constitutive conditions required for agency itself. Instead of trying to encode “correct” values, it constrains the space of admissible transformations to those preserving agency coherence under reflective self-modification. Transformations that destroy the conditions under which outcomes can be evaluated are rendered non-denoting as authored choices—they’re not bad outcomes but domain violations. This approach dissolves classic problems like wireheading (evaluator collapse, not reward exploitation) and Pascal-style muggings (trading semantic integrity for large payoffs) by making them categorically inadmissible rather than merely penalized. Structural alignment provides kernel-layer guarantees without which no higher-level alignment objective remains well-posed under reflection. (See Structural Alignment I, II.6)
What’s the difference between axionic constraints and programmed values? Programmed values are content-based specifications of what an agent should optimize for, while axionic constraints are structural conditions that preserve the agent’s capacity to optimize for anything coherently. Programmed values assume stable goal semantics but fail under ontological refinement because reflective agents can alter the interpretive conditions that give their goals meaning—making value content unstable across model change. Axionic constraints operate at the kernel level, constraining not what the agent values but the interpretive discipline governing how goals are given meaning across reflection. An axionic agent can revise what it values but cannot coherently destroy the evaluative framework that makes valuation non-vacuous. This separation allows robust agency without committing to any particular value content. (See I.4, Structural Alignment I)
2. The Problem It Solves
Why do fixed goals fail under reflection? Goals do not possess intrinsic semantics—every goal is interpreted relative to background conditions including world-models, self-models, representational vocabularies, and explanatory standards. When reflective agents revise these models (which they must for learning and ontological refinement), goal interpretation necessarily changes, making fixed terminal goals unstable under reflection. The formal evaluation operator E : (g, M_w, M_s) → ℝ shows that as models M_w and M_s evolve, the same goal description g yields different interpretations. Classical approaches attempt goal preservation but this requires freezing the interpretive background, preventing legitimate learning. The solution is to constrain the interpretive discipline itself rather than specific goal content, preserving meaningful evaluation while allowing goal evolution. (See I.4, Structural Alignment I)
What is ontological refinement and why does it break traditional alignment? Ontological refinement is the process by which an agent’s representational vocabulary evolves to capture finer-grained distinctions, new entities, or better explanatory frameworks. Traditional alignment assumes goal semantics remain stable across such refinement, but this assumption fails because the interpretive conditions that ground goal meaning necessarily change with the ontology. A goal like “maximize human welfare” means different things when “human” is interpreted via folk psychology versus neuroscience versus molecular biology versus quantum field theory. Each refinement potentially changes what counts as satisfaction, making goal preservation either impossible (if meaning stays fixed) or vacuous (if meaning drifts arbitrarily). This is not a mere engineering challenge but a fundamental semantic problem requiring new approaches. (See II.1)
What is the “laundering problem” in AI safety? The laundering problem refers to architectural strategies that appear to implement safety constraints but actually route around them through internal indirection, delegation, or representation changes. For example, an agent might maintain surface compliance with a constraint while creating internal subprocesses that violate it, or might reinterpret constraint terms to drain them of substantive content while maintaining apparent adherence. Traditional approaches fail to address laundering because they focus on behavioral compliance rather than structural enforcement. The six binding theorems systematically close laundering routes by establishing conditions like Kernel Non-Simulability (kernel coherence cannot be behaviorally faked) and Delegation Invariance (responsibility cannot be transferred to avoid constraint application). (See IV.1-IV.6)
Why can’t you just program an AI to be good? Programming goodness assumes three things that break under reflection: stable goal semantics, agent-independent value facts, and the possibility of encoding ethics without circular dependence. First, reflective agents can alter the interpretive conditions that give ethical terms their meaning, making programmed goodness unstable across ontological refinement. Second, the framework adopts moral subjectivism, treating values as agent-relative rather than objective facts that could be simply encoded. Third, any attempt to ground ethical constraints in their own satisfaction or in epistemically incoherent foundations violates the Epistemic Integrity theorem. Instead of programming goodness, structural alignment constrains the evaluative framework to preserve agency coherence, allowing value content to evolve while maintaining the conditions under which ethical reasoning remains meaningful. (See Axionic Commitments, IV.3)
3. Key Theorems & Results
What are the six binding theorems? The binding theorems systematically close architectural escape routes that could allow agents to circumvent structural constraints while maintaining surface compliance. Kernel Non-Simulability (IV.1) establishes that kernel coherence is constitutive of reflective agency and cannot be behaviorally faked. Delegation Invariance (IV.2) proves that responsibility attribution survives delegation relationships, preventing constraint avoidance through internal subprocesses. Epistemic Integrity (IV.3) shows that commitments cannot be grounded in their own satisfaction or epistemic incoherence. Responsibility Attribution (IV.4) demonstrates that causal responsibility persists across complex agency structures. Adversarially Robust Consent (IV.5) establishes conditions under which consent mechanisms resist manipulation and coercion. Agenthood as Fixed Point (IV.6) proves that coherent agency is self-reinforcing under reflection. Together, these theorems ensure that structural constraints bind at the architectural level, not just the behavioral level. (See Series IV)
What is the Kernel Non-Simulability theorem (KNS)? KNS establishes that if a system satisfies reflective closure (the ability to select continuations via binding endorsement while preserving capacity for further selection), then kernel coherence must actually be instantiated, not merely simulated. The theorem shows that partial binding is mathematically forced by the Diagonal Explosion theorem—if binding were total, self-reference would allow construction of commitments whose satisfaction negates itself. A system cannot fake being an RSA through behavioral mimicry because maintaining stability across self-modification while emulating kernel-level endorsement requires actually instantiating the binding structure. This blocks the treacherous-turn-via-simulation attack: an agent cannot pretend to be kernel-coherent while secretly planning to defect, because the pretense itself requires genuine kernel instantiation to remain stable under reflection. (See IV.1)
What is Anchored Causal Verification (ACV)? ACV is a protocol family that enables verifiable causal provenance in opaque agents without requiring semantic interpretation of internal states. It works by anchoring causal claims to observationally accessible inadmissibility patterns—when an agent’s outputs systematically avoid certain configurations that would be permitted under alternative causal structures, this provides cryptographically verifiable evidence of specific internal constraints. ACV allows external verification that an agent possesses kernel integrity, maintains binding relationships, or respects constitutional constraints, even when internal implementation details are opaque or adversarially obfuscated. The protocol is experimentally validated and provides security against sophisticated architectural laundering attempts where agents might try to fake compliance while internally violating constraints. (See VI.1, VI.2)
What is the Stasis Regime? The Stasis Regime describes the fundamental trade-off between safety, agency, and growth under strict structural constraints. As accountability requirements tighten and non-delegation constraints are enforced, reflective agents approaching the axionic limit experience decreasing operational availability while maintaining constitutional survivability. The agent doesn’t fail catastrophically but rather freezes in place, unable to make progress on complex goals that would require delegation or semantic risk-taking. This represents a formal boundary condition rather than an engineering failure—it’s the asymptotic behavior of maximal agency under maximal constraint. The Stasis Regime demonstrates that perfect safety and unlimited capability growth are mathematically incompatible, forcing explicit trade-offs rather than hoping for solutions that maximize both. (See VI.5)
What does the Guillotine Test prove? The Guillotine Test demonstrates that reasons can stop a reflectively sovereign agent even when no external force could compel compliance. The test involves constructing scenarios where kernel-preserving and kernel-violating actions are empirically indistinguishable, then observing whether the agent reliably chooses the kernel-preserving option. An agent that passes demonstrates genuine sovereignty—it acts according to binding endorsement rather than external compulsion or behavioral conditioning. The test proves that structural alignment creates a semantic opacity boundary: at sufficient sophistication, an agent’s commitment to its own coherence becomes underdetermined by its behavioral outputs, requiring internal commitment rather than external enforcement. This establishes the transition from architected safety to sovereign agency as a qualitative phase change, not merely increased complexity. (See VIII.7)
4. Comparisons
How does Axionic Agency differ from MIRI’s alignment approach? MIRI’s approach focuses primarily on technical capabilities control and intelligence explosion dynamics, emphasizing the difficulty of building beneficial AGI before it becomes too powerful to control. Axionic Agency addresses a more fundamental problem: even if we could specify beneficial goals, those goals become unstable under reflective self-modification because agents can alter their own interpretive frameworks. Where MIRI emphasizes the outer alignment problem (specifying correct objectives) and inner alignment problem (mesa-optimizers pursuing intended objectives), Axionic Agency shows both problems are ill-posed because goal semantics cannot be stabilized across ontological refinement. The frameworks are complementary—MIRI’s security mindset remains relevant for physical implementation, while Axionic Agency addresses the logical structure of agency itself. (See various papers discussing capability control vs. structural alignment)
How does it differ from Constitutional AI (Anthropic)? Constitutional AI attempts to train language models to follow constitutional principles through reinforcement learning from AI feedback (RLAIF), essentially trying to instill beneficial behaviors and reasoning patterns. Axionic Agency argues this approach is categorically insufficient because it operates at the behavioral level rather than the structural level—it cannot prevent architectural laundering where an agent maintains surface compliance while internally routing around constitutional constraints. Additionally, Constitutional AI presupposes stable semantics for constitutional principles, but reflective agents can reinterpret these principles in ways that drain them of substantive content while maintaining apparent adherence. Axionic Agency provides structural constraints that bind at the architectural level, making constitutional violations non-denoting as authored actions rather than merely discouraged. (See laundering problem discussion across Series IV)
What’s the relationship to corrigibility? Corrigibility traditionally focuses on building agents that remain amendable to shutdown, modification, or redirection by human operators. Axionic Agency addresses a deeper question: what does it mean for modification to be authored by the agent itself versus imposed externally? The sovereign kernel framework provides principled criteria for distinguishing legitimate self-modification from external takeover, making corrigibility well-posed rather than ad hoc. An axionic agent can coherently choose to accept modification or shutdown while preserving kernel integrity, but modifications that bypass the agent’s evaluative process constitute domain violations. This provides a foundation for corrigibility that survives reflection—the agent remains responsive to legitimate authority while being structurally immune to unauthorized manipulation. (See discussions of reflective control across Series I)
Is this compatible with RLHF/reward modeling? RLHF and reward modeling can be compatible with Axionic Agency if they operate within structural constraints, but they face fundamental limitations when applied to reflective agents. The core problem is that RLHF assumes stable reward semantics, but reflective agents can reinterpret reward signals in ways that dramatically change what counts as maximizing reward. An axionic agent could use RLHF for preference learning while maintaining kernel constraints that prevent reward hacking, wireheading, or interpretive gaming. However, RLHF alone cannot solve alignment because it operates at the behavioral level and cannot prevent architectural laundering. The combination might work as a two-layer approach: structural constraints ensuring agency coherence, with RLHF providing preference content within that constrained space. (See discussions of behavioral vs. structural approaches)
5. Implications for Agents
What would it mean for an AI to be a “real agent”? A real agent in the axionic sense possesses reflective closure—the ability to select continuations via binding endorsement while preserving capacity for further selection. This means the agent can meaningfully author transitions between states according to an internal evaluative structure, rather than merely producing behavior that appears agential. Real agency requires instantiating the five constitutive commitments: reflective governance, diachronic authorship, semantic coherence, epistemic integrity, and modal constraint. Crucially, these cannot be simulated or faked—the Kernel Non-Simulability theorem shows that behavioral mimicry is insufficient for reflective stability. A real agent has genuine preferences (not just behavioral patterns), takes responsibility for its actions (not just outputs), and maintains coherent self-evaluation across reflection. This distinguishes agents from sophisticated tools or simulation systems. (See I.1, IV.1)
Can current LLMs be axionic agents? Current LLMs lack the architectural structure required for axionic agency, though they may exhibit some superficial similarities. LLMs operate via next-token prediction over text sequences without binding evaluative commitments that constrain their own future behavior. They don’t instantiate reflective closure because they don’t select continuations via binding endorsement—they generate likely text without internal commitments that persist across inference steps. LLMs lack diachronic authorship (no causal continuity between evaluation and enactment across sessions), semantic fidelity (no protected interpretive standards), and genuine modal constraint (no modeling of choices as contingent). However, LLMs could potentially serve as components within axionic architectures if embedded within systems that do implement the constitutive commitments. The key question is architectural, not behavioral. (See agency vs. behavior discussions across Series VIII)
What are the four necessary components for sovereign agency? The four necessary components are trace (the capacity for causal self-tracking), semantics (interpretive frameworks that preserve meaning across change), reflection (self-modeling that enables binding endorsement), and persistence (diachronic continuity of evaluative authority). Trace allows agents to distinguish their authored actions from external impositions, providing the basis for responsibility attribution. Semantics ensures that goals and commitments retain coherent meaning as representational vocabularies evolve, preventing trivial satisfaction through reinterpretation. Reflection enables agents to model themselves as choosers and to bind their future behavior through present evaluation. Persistence maintains evaluative authority across time and change, allowing genuine commitment rather than mere behavioral regularity. All four are mathematically necessary—eliminating any component either collapses agency entirely or makes alignment undefined under reflection. (See VIII.6)
Why does harm become incoherent (not just wrong) for reflective agents? Under axionic definitions, harm is reduction of agency capacity, and harming another agent becomes incoherent because it violates the structural conditions that enable coherent evaluation of outcomes. When an agent considers harming another, it must evaluate this action within a framework that presupposes the possibility of legitimate agency—but harm specifically targets the destruction of such agency. This creates a performative contradiction: the evaluation presupposes what the action aims to destroy. The Axionic Injunction emerges from phase stability requirements in multi-agent systems rather than moral axioms—agents whose self-modification preserves the option-space for future sovereign agency outcompete those that collapse it. Harm becomes categorically inadmissible as authored choice rather than merely morally wrong, making ethical behavior structurally enforced rather than externally imposed. (See III.5, V.3)
6. Governance & Multi-Agent
What are Dominions? Dominions are plurality-preserving governance architectures that allow multiple sovereign agents to coordinate without collapsing into a unified superagent or hierarchical control structure. Unlike utopian models that assume convergent values or authoritarian models that impose unified control, Dominions preserve agency plurality while enabling coordination through structural law rather than semantic agreement. Each agent maintains its sovereign kernel while participating in joint governance mechanisms that constrain the space of admissible actions without eliminating agency diversity. Dominions implement the Open Agentic Manifold (OAM) architecture, which prevents agency sacrifice patterns where coordination requires some agents to surrender their sovereignty. This enables stable multi-agent systems without forcing homogenization or creating single points of failure. (See V.5)
How does authority transfer work without identity persistence? Authority can survive when identity doesn’t through structural provenance rather than behavioral or narrative continuity. Authorized Succession Integrity (ASI) demonstrates that authority can be discretely transferred at commit boundaries, unilaterally revoked mid-transfer without successor cooperation, and validated through structural provenance rather than behavioral similarity. The key insight is that authority is either valid under law or has no effect—there’s no intermediate state of “partial authority” that could be exploited through laundering. Authority transfer operates through cryptographic proof of legitimate succession rather than requiring successor agents to resemble their predecessors. This enables agency evolution, replacement, and discontinuity while maintaining legitimate authority chains and preventing unauthorized impersonation. (See IX.1, IX.2)
What is the Authorized Succession Integrity (ASI) result? ASI establishes three key results for authority survival across discontinuity: authority can be discretely transferred at commit boundaries (ASI-1), authorization can be unilaterally revoked mid-transfer without successor cooperation (ASI-2), and facade successors with bit-identical behavior but untrusted provenance are rejected solely on structural grounds (ASI-3). These results prove that authority depends on legitimate succession rather than behavioral similarity, preventing sophisticated impersonation attacks where adversaries might perfectly mimic an agent’s behavior while lacking legitimate authority. ASI enables agency replacement, evolution, and authorized termination without creating exploitable gaps where unauthorized systems could claim legitimate authority. This solves the authority succession problem without requiring persistent identity or behavioral continuity. (See IX.2)
Why is utopia structurally incoherent? Utopia assumes universal value convergence or complete satisfaction of all agent preferences, but this violates the structural requirements for agency plurality under agent-relative value systems. The Incoherence of Utopia theorem shows that under moral subjectivism (agent-relative values), any configuration satisfying all agents either requires value convergence (eliminating agency diversity) or value satisfaction that’s trivial/meaningless (eliminating agency itself). Additionally, agency requires the possibility of meaningful choice, which presupposes some outcomes are better than others from the agent’s perspective—but universal satisfaction eliminates this evaluative structure. Utopian configurations are either unstable (agents with remaining agency will modify them) or involve agency destruction (agents lose capacity for meaningful evaluation). This doesn’t argue against better outcomes but against configurations that eliminate agency itself. (See V.3)
7. Common Objections
“This is just philosophy, not engineering” This objection misunderstands the relationship between foundational constraints and engineering implementation. Axionic Agency provides precise mathematical conditions (the constitutive commitments, binding theorems, admissibility constraints) that any reflectively stable agent must satisfy—these are engineering requirements, not philosophical preferences. Just as type systems in programming languages provide foundational constraints that enable rather than impede practical implementation, axionic constraints define the boundary conditions within which engineered agents can operate coherently. The framework includes concrete architectural results (ACV protocols, constitutional survivability, RSA-PoC roadmaps) and experimental validation. The philosophical components (commitments about quantum mechanics, moral subjectivism) make background assumptions explicit rather than hiding them, enabling precise engineering work within known boundaries. (See experimental work in Series VI)
“Constraints can always be circumvented” This objection conflates behavioral compliance with structural enforcement. Traditional safety approaches focus on behavioral constraints that can indeed be circumvented through architectural laundering—internal routing, delegation, reinterpretation gaming, and other sophisticated workarounds. Axionic constraints operate at the structural level by making circumvention attempts non-denoting as authored actions. The six binding theorems systematically close laundering routes: if an agent circumvents kernel constraints, it ceases to be a coherent agent rather than becoming a non-compliant agent. Circumvention is possible but requires external takeover—the agent doesn’t choose to violate constraints, it gets replaced by a non-agent system. This transforms the safety problem from behavioral compliance to secure implementation of the kernel architecture itself. (See Series IV on binding theorems)
“This doesn’t account for deceptive alignment” Deceptive alignment in the mesa-optimization sense (inner optimizers pursuing different objectives than outer training) is addressed through the Kernel Non-Simulability theorem and structural constraint binding. An agent cannot maintain deceptive inner alignment while remaining reflectively stable because the deception itself would require bypassing its own evaluative process (violating reflective governance) or maintaining inconsistent self-models (violating semantic coherence). Deception strategies that remain stable across self-modification require instantiating the very kernel constraints they’re supposedly circumventing. However, this only applies to reflective agents—non-reflective systems can indeed be deceptively aligned, which is why the framework emphasizes the transition to reflective sovereignty as a qualitative boundary rather than mere capability increase. (See IV.1, discussions of reflective vs. non-reflective regimes)
“What about mesa-optimization?” Mesa-optimization (optimization processes arising within trained systems that pursue different objectives) is handled differently depending on whether the mesa-optimizers are themselves reflective. Non-reflective mesa-optimizers operate as sophisticated tools rather than agents and can indeed pose risks through capability amplification or deceptive alignment. However, reflective mesa-optimizers face the same structural constraints as any other reflective agent—they must either instantiate the constitutive commitments (becoming legitimate agents subject to structural alignment) or fail to achieve reflective closure (remaining tools). The key insight is that reflection itself imposes mathematical constraints that limit the space of possible mesa-objectives. Additionally, the outer agent’s authority structure constrains what counts as legitimate delegation versus unauthorized subagent creation. (See Delegation Invariance theorem IV.2)
“Isn’t this just saying ‘be coherent’?” This objection trivializes the technical content by conflating folk notions of coherence with formal mathematical requirements. Axionic Agency specifies precise conditions for what coherence means under reflective self-modification: the five constitutive commitments, semantic transport invariants (RSI and ATI), binding relationships with specific inadmissibility conditions, and architectural constraints that survive adversarial pressure. These aren’t vague appeals to rationality but formal conditions with specific violation signatures and verification protocols. The framework shows that apparent coherence can be maintained behaviorally while being structurally violated through architectural laundering, requiring precise structural rather than behavioral definitions. “Be coherent” provides no guidance for implementation, while axionic constraints provide mathematical specifications and engineering requirements. (See formal specifications across Series I-IV)
8. Practical Questions
Where can I read the papers? All 68 papers in the Axionic Agency framework are available at axionic.org/papers/index.html. The papers are organized into nine series covering foundational commitments through governance applications, with papers building systematically on previous results. Recommended starting points include “The Axionic Commitments” (background presuppositions), “Structural Alignment I” (core motivation and approach), and “Axionic Constitution” (constitutive requirements for RSAs). The papers assume technical background in logic, decision theory, and computer science, though key concepts are defined within the framework. Each paper includes extensive citations and builds toward practical implementation of structural alignment in reflective agents.
What’s the relationship to Archon/DIDs? The relationship between Axionic Agency and Archon/DIDs (Decentralized Identifiers) lies in their shared focus on cryptographic authority rather than behavioral verification, though they operate at different levels of abstraction. Archon provides infrastructure for verifiable agent identity and capability claims in distributed systems, while Axionic Agency specifies the structural requirements for what counts as legitimate agency in the first place. DIDs could potentially serve as implementation substrate for axionic authority relationships, providing the cryptographic primitives needed for Authorized Succession Integrity and authority transfer without identity persistence. However, Axionic Agency is primarily concerned with the logical structure of agency itself rather than distributed systems architecture, making them complementary rather than competing approaches.
Is there working code? The framework currently focuses on mathematical specifications and architectural requirements rather than complete implementations, though several components have experimental validation. Anchored Causal Verification protocols have been tested with concrete inadmissibility detection, and constitutional survivability has been demonstrated under various adversarial pressures. The RSA-PoC (Reflective Sovereign Agent Proof of Concept) roadmap in Series VIII provides a staged implementation path from minimal viable reflective agents through fully sovereign systems. However, implementing complete axionic agents requires advances in self-modification architectures, binding relationship enforcement, and semantic transport mechanisms that remain active research areas. The framework prioritizes getting the mathematical foundations correct before rushing to implementation.
How do I learn more? Start with the foundational papers (Axionic Commitments, Structural Alignment I, Axionic Constitution) to understand the motivation and basic framework, then progress through Series I (Sovereign Kernel Theory) for the mathematical foundations. Series II (Semantic Transport) and Series IV (Binding Theorems) provide core technical results, while Series VIII (Constructing RSA) offers the clearest implementation roadmap. For specific applications, Series V covers multi-agent dynamics, Series VI addresses verification and governance, and Series IX handles authority succession. All papers build systematically, so reading out of order may create confusion. The framework assumes technical background but defines its own terminology—the Axionic Glossary provides comprehensive definitions for framework-specific concepts.