VIOLENCE AS ARCHITECTURAL OUTPUT

Why the Confabulation Cascade Terminates in Physical Harm

Position Statement

This chapter establishes that violence (from interpersonal aggression through organised warfare to genocidal extermination) is not a pathological deviation from normal human functioning but the predictable behavioural output of the architecture established in Chapters 3 through 10. The Confabulation Cascade that produces collective convergence on false Beliefs does not terminate in Belief alone. It terminates in action. When that action targets human beings identified through Confabulation as the cause of experienced suffering, violence is the architectural terminus of the Scapegoat Mechanism.

Dependency Statement

This chapter requires prior understanding of:

• The Three-Source Taxonomy (A-I-M) (Chapter 4) and why motivation requires disaggregation (Chapter 5)
• The component-function localisations of Chapter 4, Section 4.4.12
• Apex M-Capture, I-Vacancy and Rung Loss (Chapter 8, Sections 8.5.6 and 8.12.5)
• Source Opacity as structural feature preventing introspective access to motivational sources (Chapter 9)
• The Evolutionary Mismatch Hypothesis establishing internal targeting as default when external targets diminish (Chapter 10)
• The Four-Type Agent Taxonomy (Predator, Prey, Collaborator, Rival) and the Agent-Based Causal Default (Axiom 9) as the cognitive architecture that produces agent-based attribution (Chapter 10)
• Confabulation, Projection, and the Scapegoat Mechanism as the pathway from individual misattribution to collective targeting, including the Dual-Role Confabulation signature and the Presumption Against Scapegoating (Chapter 11)
• The Bayesian Mimetic Crisis Principle establishing Bayesian Lock-In as mathematical attractor (Chapter 13), together with the amplification and corrective rates of Chapter 13, Sections 13.8.2 and 13.8.3 and the floor condition derived from them at Chapter 13, Section 13.8.4
• The Structural Output Principle, the endogeneity of the mental element, the Punitive Inversion Principle, and Cascading Support and Cascading Consequences (Chapter 16)

Imports the Critical-Coupling Fold (Chapter 14, Section 14.7.5), the Provenance Rule (Chapter 14, Section 14.10), and the Endogenous-Evidence Condition (Chapter 14, Section 14.10.4.7) as established, and develops the full instantiation of the economic-crisis and war cases that Chapter 14 introduces only briefly.

Derivation Preview

The chapter’s central claim (that violence is architectural output rather than aberration) is derived from four established components:

1. The Information Gap Principle: diffuse causation exceeds computational tractability, requiring the Agent-Based Causal Default (Axiom 9) to complete the causal chain with agent-based attribution
2. The Confabulation Cascade: Bayesian updating on endogenously-generated evidence produces convergence on false Grievance Beliefs, terminating in Bayesian Lock-In
3. Phenomenological Collapse: M-urgency is experienced as A-threat (survival-level alarm) after Common-Currency Integration
4. Behavioural discharge: Belief states produce action tendencies; convergent Grievance Beliefs produce convergent action against those agents, completing the Scapegoat Mechanism at population scale

From these the chapter derives, for import by later chapters, the Leadership Compression Principle and its Time-Buying Principle, the Dual-Locus Ignition Theorem, the Structural Determination Principle, the M-Capture Incapacity Principle, the Burden-Reversal Principle, and the Incidental Buffer Principle.

Chapter Preview

The chapter traces the pathway from confabulated Belief to physical harm (Section 17.1), states the Unifying Crisis Principle from which the chapter's three domain cases follow (17.2), and locates the three sources of aggression and their structural consequences (17.3). It distinguishes sincere confabulation from cynical manipulation (17.4) and develops the economic-crisis case, the Appetite floor reasserting against an inflated Mimetic aggregate (17.5), and the succession case, the suppressed discharge arriving at once (17.6). It derives the apex route into crisis, stating the Leadership Compression Principle, the two ignition loci that concentrate the capacity to ignite at the floor and the apex, and the Trigger-Indifference Corollary, under which neither waiting nor removal averts the discharge of the terminal configuration (17.7). It establishes the scale invariance by which the same mechanism runs from interpersonal aggression to organised war (17.8).

It then derives why violence is experienced as defensive (17.9), the incoherence of proportionality assessment (17.10), the content-independence of the violence pathway (17.11), and why a declared right does not bind the Belief that breaks it (17.12). It then treats duty as Self-Model insulation (17.13), and the symmetry of violence, from which it states the Structural Determination Principle (17.14). It states the Leadership Asymmetry and the arithmetic of lives under M-capture, distinguishes the refusal to reciprocate violence from concession, and derives what a population that declines to fight can still do and what an aggressor can still impose on it (17.15), establishes the Truth-Exclusion Principle (17.16), develops the nuclear race condition including why recovery is restart rather than self-correction and why a boundary-independent-output domain is a gating domain whose safety requires low self-coupling of the executive act rather than isolation (17.17), states the Precautionary Asymmetry as a derived methodological commitment (17.18), applies the Field-Correction Obligation at collective scale to derive what a correction must and must not do (17.19), derives how a war that has begun is contained as a structural problem rather than decided as a contest of Beliefs, stating the Self-Defence Orthogonality Principle and the Mutual-Support Commitment, and states which parts of that derivation are directional and what calibration must supply (17.20), bounds what the chapter does not claim (17.21), and turns toward the structural intervention developed in the chapters that follow (17.22).

17.1 The Pathway from Belief to Violence

Chapter 13 established that Bayesian Lock-In (catastrophic reconfiguration) functions as an attractor state. But catastrophic reconfiguration is abstract language that obscures what actually occurs. The terminus is violence: physical harm inflicted on human beings who have been identified, through the Scapegoat Mechanism, as responsible for experienced suffering.

The pathway operates as follows:

Step 1: Diffuse Harm Generates Causal Demand. Populations experience suffering (economic precarity, Status anxiety, A-insufficiency, loss of I-meaning). The cognitive system requires causal explanation: Why am I suffering? This is not optional curiosity but architectural necessity. The brain evolved to identify threats and their sources; unexplained suffering represents unresolved threat.

Step 2: True Causes Exceed Computational Tractability. The actual causes of most collective suffering are systemic: institutional arrangements, structural dynamics, feedback loops operating across populations and generations. These causes are:

• Distributed across many actors, none individually sufficient
• Operating through mechanisms invisible to direct observation
• Temporally extended beyond immediate perception
• Resistant to agent-based framing

The Information Gap Principle (Chapter 2) establishes that when true causes exceed computational tractability, the Agent-Based Causal Default (Axiom 9) completes the causal chain with agent-based attribution. This is not cognitive laziness but architectural constraint: the system cannot represent diffuse systemic causation (especially without access to all the necessary information and correct Hypothesis to evaluate) and defaults to identifying responsible agents.

Step 3: Confabulation Generates Scapegoat Candidates. The Confabulation system produces candidate explanations: They are responsible. Group X caused this. The enemy is identifiable. These explanations satisfy the causal-attribution requirement. They provide the cognitive closure that diffuse causation cannot. They feel correct because they resolve the felt tension of unexplained suffering. The candidate targets are selected through Prey-detection criteria from the Four-Type Agent Taxonomy (Chapter 10): vulnerability, visibility, reciprocity-exemption (the target can be persecuted without triggering costly retaliation). Simultaneously, the framing follows Predator-attribution: the candidate is described as threatening, taking, blocking. This combination of Prey-selection with Predator-framing is Dual-Role Confabulation (Chapter 11), the diagnostic signature of Scapegoat production. The step is Ground-Resolution closing on the causal-attribution question ("who caused this suffering?") under an Information Gap: the representable Hypothesis Space $\mathcal{H}$ has been restricted to agent-based causes by the Agent-Based Causal Default; the interference term has collapsed to zero under closure-demand; probability mass has concentrated on the Prey-selected, Predator-framed candidate; and the output, a Grievance Belief against the Prey-selected, Predator-framed candidate (Appendix A, Section A.11.8; prose derivation at Chapter 16, Section 16.3), is written to the Belief Register. The same architectural features that produce individual-scale Confabulation on motivational-source questions therefore produce collective-scale Scapegoat production on causal-attribution questions; the difference is the content of S and the evidential inputs, not the engine.

Step 4: The Confabulation Cascade Produces Bayesian Lock-In. Individual Confabulations become social proof. When Individual A expresses their Confabulated attribution, Individual B treats this as evidence: if Group X were really responsible, people would say so. B updates toward the same attribution and expresses their now-higher Belief. The expression becomes evidence for Individual C. Chapter 13 established that this Confabulation Cascade produces Bayesian Lock-In through valid Bayesian reasoning applied to invalid, endogenously generated evidence. The population moves toward P(H|GroupX) → 1.

Step 5: Belief Produces Action-Tendency. Beliefs are not inert representations; they are action-guiding states. A Grievance Belief that Group X threatens our survival generates action-tendencies appropriate to survival threat: defence, pre-emption, elimination of the threat source. Only a Belief of this type does so: a Content-Belief about the world converges and is done, and a Belief about one's own wants names no one to act against (Chapter 16, Section 16.3).

This is where Phenomenological Collapse becomes critical. The threat posed by the identified Scapegoat is experienced as survival-level threat because:

• M-urgency (Status threat from the Scapegoat’s existence) has been integrated in Common-Currency format
• Source Opacity prevents recognition that the urgency is M-generated rather than A-generated
• The threat-detection system fires with genuine physiological activation
• The Individual experiences “they threaten us” as survival emergency, not Status competition

Step 6: Convergent Belief Produces Coordinated Action. When a population converges on the Grievance Belief that the Scapegoat threatens survival, coordinated action against the Scapegoat becomes:

• Individually rational (responding to perceived threat)
• Socially validated (everyone agrees the threat is real)
• Morally required (failure to act endangers the group)
• Experientially authentic (the felt urgency is genuine)

The action is violence. Physical harm to members of the targeted group. Exclusion, expropriation, expulsion, extermination: the specific form varies with circumstance, but the structure is identical (eliminate the identified threat).

This six-step pathway IS the Scapegoat Mechanism (Chapter 11) reaching its behavioural terminus. The pathway was derived in Chapter 11 as the architectural consequence of Source Opacity, Confabulation, and the Agent-Based Causal Default operating at population scale. This chapter establishes that the terminus is not abstract “catastrophic reconfiguration” but violence against Individuals and groups who satisfy the Scapegoat definition: selected through Prey-detection criteria while framed through Predator-attribution, exhibiting Dual-Role Confabulation.

The architecture entails a single kind of terminal event: a fold-style discontinuous collapse from the high-valuation, high-PM, supercritical state to a low-valuation attractor that corresponds to A-I functional baseline. It does not entail a sustained oscillation around the high-PM state, where the system would compound, partially correct, and compound again. A sustained oscillation would require a stabilising correction inside the cycle, and the architecture provides none: there is no negative-feedback term on the M channel, Source Opacity blocks the error signal that an oscillating system would need, and Concealment Infrastructure increases monotonically with PM until it fails catastrophically. The terminal phase of the Mimetic Crisis Trajectory is therefore not one of two possible end-states but the architecturally entailed end-state of supercritical M-dynamics in the system as it stands. Sustained limit cycles around the supercritical state are possible only where an additional stabilising structure has been engineered into the system, an institutional buffer with intrinsic negative feedback, and that engineering is the appropriate object of Subcriticality Maintenance research. The formal proof of the fold's necessity is given in Appendix A.5.3.

17.2 The Unifying Crisis Principle

Section 17.1 traced one pathway from confabulated Belief to physical harm. The pathway is not the only terminus the architecture admits, and the reason this chapter carries an economic case and a succession case alongside the scapegoating cascade is that all three are the same result reaching ground in different domains. That result is the Unifying Crisis Principle, and it is what the architecture looks like when it runs without correction.

The Unifying Crisis Principle follows from six established premises, each carried by a result already in place. That demonstration is the Unifying Crisis Principle Derivation.

From Common-Currency Integration (Axiom 2): the Decision Hub integrates Appetitive, Intrinsic and Mimetic contributions into a single signed Common-Currency Scalar (the Scalar), and the integration destroys the source labels.

From Confabulation (Axiom 4): the Individual must nonetheless account for the wanting the Scalar delivers, and the account is generated from the two channels that remain legible, so a Mimetic want is reported as an Appetitive need or an Intrinsic engagement.

From Bayesian Belief Dynamics (Axiom 8): that account enters the Belief Register as evidence, and updating is content-agnostic, so it is weighted by its evidential form rather than by its accuracy.

From the same axiom: a posterior becomes the prior for the next update, so the account compounds on itself rather than being re-examined.

From Preconscious Transmission (Axiom 7) with the Salience-Status Identity: Individuals express those accounts publicly, and expressed valuation is what other Individuals read.

From the Reciprocal Valuation Property: other Individuals weight that expression as evidence, because an observed valuation is the evidence the Observer is built to take.

Given these premises the conclusion is entailed rather than predicted. Mimetic Desire inflates the perceived value of Objects that carry genuine Appetitive or Intrinsic function beyond any level at which they could satisfy, because Object Parasitism, the failure mode Axiom 5 (M as Amplifier) delivers, gives M no separate Object of its own to inflate, and Axiom 6 (Differential Satiation Dynamics) gives it no closure condition. The Decision Hub, having no source labels to read, integrates that inflation exactly as it would integrate what it resembles, which is chronic Appetitive emergency. A population of Decision Hubs in that condition moves through the phases of the Crisis State Architecture toward the Terminal Phase, so catastrophic reconfiguration is the configuration the system evolves toward under its own dynamics rather than the worst case it might reach, and the reconfiguration presents in whichever domain the inflated Objects sit: as market collapse where they are priced, as institutional failure where they are administered, and as open violence where they are attributed to an agent.

Two properties of the Unifying Crisis Principle do the work.

The first is that crisis is the default rather than the exception. Nothing in the six premises is an unusual condition, a failure of character, or a contingency of any particular culture. Each is an ordinary operation of the architecture running as specified, and the conclusion follows from their conjunction. So the question a crisis poses is not what went wrong but what was holding it off, which is why the chapter's interventions are all maintenance of something rather than removal of something.

The second is that opt-out is infeasible for the Individual. Where access to Appetitive sufficiency or to a protected Intrinsic domain has become contingent on continued participation in the escalation, an Individual who recognises the dynamic still cannot leave it without surrendering the provision that participation secures. Recognition is therefore not sufficient for exit, which is the Opt-Out Paradox, and the Opt-Out Paradox is why the Unifying Crisis Principle's corrective is structural rather than educational.

The Unifying Crisis Principle is stated here in the domain where the architecture reaches ground. Its Bayesian-layer counterpart is the Bayesian Mimetic Crisis Principle, derived at Chapter 13, Section 13.6, which establishes the same compounding at the level of Belief and supplies the threshold condition under which the compounding becomes self-reinforcing. The two are one result at two levels of description: Chapter 13 establishes that the updating locks in above the Subcriticality Threshold, and this section establishes what a population in that condition does.

Formal treatment: Appendix A, Sections A.5.7.5 and A.6.3.

17.3 The Three Sources of Aggression and Their Structural Consequences

All aggression begins identically. The threat-detection system fires, cortisol rises, attention narrows, and the body prepares for action. The Individual experiencing aggression cannot in realtime introspectively determine whether their survival is genuinely threatened, their Intrinsic engagement has been destroyed, or their relative position has been challenged, because Common-Currency Integration (Axiom 2) strips source-tags from the integrated signal that conscious access receives. This is Phenomenological Collapse applied to violence specifically: A-thwarting, I-thwarting, and M-thwarting all feel like survival threat because the alarm signal arrives through the same channel, integrated into the same common-currency value.

The phenomenological identity disguises structural differences that are architecturally decisive. What differs across the three sources is not the quality of the aggression but its trigger conditions, its termination dynamics, its scalability, and (critically) the status consequences it produces for the perpetrator. These differences are invisible to the Individual experiencing the aggression but determine whether a given act of violence is self-limiting or potentially civilisation-threatening. Individuals differ in how readily the threat-detection system fires and aggression is selected, and sex-typical hormonal differences are among those differences: they vary the weights the Decision Hub integrates and leave the three sources and their termination dynamics unchanged (Chapter 9, Section 9.4.7).

17.3.1 A-aggression: Deficit-Driven Violence with Built-In Termination

A-aggression arises when genuine survival needs are actively blocked: someone is witholding food from the starving person, physically attacking the Individual, or destroying shelter that provides necessary protection. The A-system monitors departure from homeostatic set-points and generates urgency proportional to deficit. When that deficit is both severe and caused by an identifiable agent whose removal would restore access, A-aggression is the system working as designed. The ancestor under physical attack or facing a rival physically blocking access to water experienced aggression that served adaptive function: it prepared the body for defensive action against a genuine, identifiable, removable threat.

The architecturally decisive feature of A-aggression is that it terminates. The A-system evolved with built-in off-switches: conditions that signal goal achieved, stand down. The ancestral hunt ended with a successful kill. Consumption produced physiological fullness. These termination signals ensure that A-cycles do not run indefinitely, wasting metabolic resources. Once the threat is removed or the deficit corrected, the aggression signal drops to baseline. Satiation is genuine: the Individual is not suppressing ongoing aggression but experiencing its absence.

This termination capacity holds even when A-deficit drives coordinated, planned violence. A population suffering genuine resource deprivation may organise a coordinated attack to secure A-goods (a raid on a grain store, a territorial incursion to access water). The coordination may be sophisticated, the planning extensive, the execution lethal. But the structural prediction is precise: once the deficit is met (the grain is secured, the water is accessed, the threat is eliminated), the violence ceases. The A-system's set-point has been restored. There is no architectural signal demanding further action. The raid ends when the raiders are fed.

This places a natural ceiling on A-violence. A-aggression cannot sustain ongoing campaigns beyond securing the neccessities that were sought, cannot motivate systematic persecution, cannot drive genocide, because the aggression dissipates once the triggering deficit is resolved. Violence that continues after the immediate threat or deficit has been addressed is, by definition, no longer purely A-sourced.

17.3.2 I-aggression: The Destruction of Irreplaceable Engagement

I-aggression arises when deep Intrinsic engagement (the flow state characterised by time distortion, reduced self-consciousness, and autotelic absorption) is abruptly and irrecoverably interrupted by external force. The guitarist absorbed in playing, the writer deep capturing an idea, the painter immersed in composition: these states create a distinctive vulnerability. The Individual in deep flow has achieved a rare motivational condition where I-engagement is dominant, A-signals are quiescent, and M-signals are temporarily subordinated.

When this state is abruptly destroyed (the instrument is snatched away, the writer is interrupted while attempting to capture heir thought, paint is accidentally spilt accross the canvas), the I-system experiences a specific form of thwarting requiring three conditions: a steep engagement gradient (the distance between the engaged state and the interrupted state is large), external imposition (the interruption was forced, not voluntary), and irrecoverability (the state cannot be immediately restored).

The irrecoverability condition is critical and extends beyond physical destruction. The Individual having a moment of creative insight (a unique musical phrase that arrived unbidden during improvisation, a paragraph that captured something the writer had been circling for months, a painting the artist had been working on for weeks) occupies a state that is, by its nature, non-reproducible. The insight arrived through conditions that cannot be reassembled on demand. When an external agent destroys this moment (the interruption shatters the concentration, the noise obliterates the musical phrase, the intrusion collapses the fragile cognitive state), what is lost is not merely time or effort but something that will not recur. The aggression (and likely rage) this produces is genuine: the threat-detection system fires because the common-currency signal representing the lost I-engagement drops precipitously, and the system that monitors value-changes treats rapid loss as threat regardless of source.

The phenomenological quality of I-aggression is distinctive: fury combined with helplessness. The aggression is genuine (the alarm system is firing, the body is prepared for action), but the action cannot achieve what A-violence achieves (deficit restoration). The destroyed engagement cannot be rebuilt through violence. The lost insight will not return because the interruptor is punished (if the interruption was caused by an egent in the first place). I-aggression is therefore intensely focused on the specific agent or object that destroyed the engagement, but structurally futile in its own terms.

I-aggression has partial satiation capacity. The acute phase subsides as the Individual processes the loss and begins to engage with restoration possibilities. Attention is refocussed. The episode terminates, unlike M-dynamics. The chronic condition of not having access to I-engagement is not I-aggression. It is I-deprivation, taking the forms of depression, lack of meaning and ennui rather than aggression (though periods of anxiety are common), producing not violence but vulnerability to M-capture because I-alternatives are unavailable.

17.3.3 M-aggression: Violence Without Termination

M-aggression arises when relative position is threatened: when the Individual's standing within their Reference Set drops or is perceived as dropping. The trigger is not genuine survival threat (A) or engagement-loss (I) but Status threat. Source Opacity then completes the transformation: after Common-Currency Integration, the Individual cannot determine that the urgency is M-generated. They experience unqualified, survival-level threat. This is the Phenomenological Collapse applied to aggression. M-urgency presents as A-urgency because the machinery processing both is identical.

M-aggression does not terminate in the same way as A or I. This is its architecturally decisive and catastrophic feature. The A-system has built-in off-switches (set-point restoration). The I-system gradually restores engagement through new flow states. M operates through relative positioning, and relative position has no absolute full signal (Axiom 6, Differential Satiation). There is no status level at which the system signals enough, stop pursuing. Conquest does not satiate the desire for more Status. The absence of a full signal is itself the downstream face of a deeper architectural fact: the Observer's Salience signal is integrated into a Scalar whose sole function is the selection of corrective action, so mimetically installed positive Salience is felt as a deficit for which no corrective mechanism exists (the Salience-Deficit Conflation Principle, Chapter 2). Conquest cannot satiate because there was never a shortfall beneath the Salience for acquisition to close.

The consequence for violence is that target elimination does not resolve the felt deficit, because the deficit was never actually caused by the target. It was caused by M-dynamics that persist regardless of any Individual's elimination. The Scapegoat who is removed does not produce lasting satisfaction. Reference-Set Reset ensures that status gains decay, recreating the deficit. A new target must be identified. The pogrom does not end the felt deficit. The genocide does not produce the promised resolution. The system continues to generate threat-signals, which continue to require scapegoat-identification, which continues to motivate elimination, indefinitely, because the underlying cause (M-dynamics without satiation) is never addressed.

This is the structural distinction between A-driven coordinated violence and M-driven coordinated violence. The A-raid terminates when the grain is secured. The M-driven war does not terminate when the enemy is defeated, because the felt deficit resets. A new enemy must be found. History confirms the pattern: scapegoating is iterative. The elimination of one identified enemy produces not resolution but the selection of the next target.

17.3.4 The Apex Predator Context

These dynamics operate within the ecological context established by the Evolutionary Mismatch Hypothesis (Chapter 10). Humans are the apex predators of the planet. The transition to apex predator status eliminated the external targets for which the mimetic machinery evolved. Without external predators to evade, and with prey domesticated via mass agriculture, the prey-detection system (which computes vulnerability, visibility, reciprocity-exemption, and value-assignment) processes internal targets: other Individuals whose displacement in status hierarchies can be computed using the identical variables that once identified game animals. Without external predators to evade, the threat-detection system flags internal threats: status rivals whose advancement triggers the same alarm responses that sabre-toothed cats once triggered.

The framing is always as predators. Humans do not experience M-driven status competition as prey behaviour, because the computational system doing the processing is an agent-detection system repurposed for internal targeting. The Individual computing vulnerability in a status rival is running the same programme their ancestors ran when computing vulnerability in an antelope. The difference is not the programme but the target class. The hunt continues. What has changed is that the quarry is now other members of the species.

The satiation asymmetry follows directly. External prey had satiation signals: the hunt ended with the kill, consumption produced fullness, the predator was escaped or defeated. These endpoints terminated the coordination cycle. Internal targets (other humans as status rivals) lack satiation signals: mimetic attraction targets comparative standing, which has no absolute full signal. Each time someone else gains a desired Object or rises in perceived social standing, the reference set shifts and is experienced as a felt deficit. Status is zero-sum, continuous and monotonically inflationary. There is no kill that ends the hunt for position. The ancestral hunt ended. The mimetic status hunt never ends because position is relative and reference sets reset with every kill.

17.3.5 The Status-Reward Asymmetry

A structural asymmetry in the status consequences of violence across the three sources completes the taxonomy and provides a retrospective diagnostic that circumvents Source Opacity.

The asymmetry:

A-aggression violence: Even if you win, you lose status. Fighting because you are starving reveals A-deficit, and A-deficit is itself a status-diminishing signal. Observers see desperation, not dominance. This is why when violence is genuinely conducted for A driven reasons, often the aggressor frames their act as high status to their peers, so that observers revise their valuations upward after standing has fallen with the aggressive act.

I-aggression violence: Even if you win, you lose status. Flying into aggression because your creative process was destroyed produces observer judgments of unstable, unhinged, disproportionate, eccentric. The social response is of shock and discomfort, not admiration. The contrast is architecturally informative: observers' status-assessment circuits register Salience under Valence-Neutral Status, but I-aggression violates the dominance-display template that converts Salience into admiration-class Status, producing negative-valence engagement that still registers as Salience yet fails to convert into positional gain because the behavioural signature does not track comparative-position victory.

M-aggression violence: The outcome reconfigures the field's valuations by definition. The winner's standing rises and the loser's falls, because the fight was about relative standing and observers update on the result.

This asymmetry is not cultural accident or contingent social norm. It follows from the architectural fact that only M-aggression operates within the status frame (the relational context where positional transactions occur). A-aggression operates in an A-frame (survival). I-aggression operates in an I-frame (engagement protection). Status was never on the table as an operative variable. Therefore, status cannot be gained from A-violence or I-violence, but it can be lost, because violence itself carries reputational costs imposed by M-operating observers. Observers always operate in M-frame: other Individuals' valuations (which constitute Status) are by definition mimetic assessments. M is both player and referee.

In A-aggression and I-aggression, no status was contested, no positional transaction occurred, no other Individuals' status-assessments shift favourably. The upside is zero. The downside is negative (violence was observed, reputational costs are imposed, status-assessments shift unfavourably). The expected status outcome of A-aggression or I-aggression violence is therefore strictly negative. The Individual can only lose status, never gain it, because gaining it would require the configuration to include status-competition as an operative variable, which it does not, by definition of the aggression being A-sourced or I-sourced.

17.3.6 Why This Architecture Produces Escalation Without Bound

The Status-Reward Asymmetry interacts with M's non-satiation to produce the self-reinforcing loop that makes M-violence civilisationally dangerous:

1. M-threat triggers violence: anticipated status loss, or Rung Loss itself where a believed standing held against the field routes the discharge outward (Chapter 15, Section 15.10.2).
2. Violence raises the winner's standing.
3. The observed rise validates the violence as effective, reinforcing the association between violence and the recovery of standing.
4. The Reference-Set Reset erodes the rise, recreating the deficit.
5. The next M-threat is now more likely to trigger violence, because violence has been reinforced.
6. The cycle repeats, with each iteration strengthening the violence response.

A-aggression and I-aggression lack this cycle. A-violence is reinforced in A-terms (food obtained) but punished in M-terms (status lost). The A-system's satiation signal prevents recurrence once the deficit is corrected. I-violence produces no gains in either I-terms (engagement lost) or M-terms (status lost). There is no reinforcement, only cost.

M-violence is reinforced in M-terms and has no satiation signal (M is structurally unsatisfiable). The result is escalation without internal brake. The only limit is external: exhaustion, resource depletion, intervention by outside parties, or crossing into the catastrophic reconfiguration of the Terminal Phase, where the violence destroys the substrate that made status-competition possible.

The M-system is the only source that pays itself through violence. Each successful act of M-violence pays the perpetrator in the resource (Status) whose pursuit generated the violence. This is a structural self-reinforcement that A-violence and I-violence cannot replicate, and it is the architectural reason why M-violence scales to war, genocide, and civilisational collapse while A-violence and I-violence remain self-limiting.

17.4 Sincere Confabulation versus Cynical Manipulation

The architecture produces violence through Sincere Confabulation: genuine Belief in which Individuals sincerely experience Confabulated causation as accurate and act on that Belief with authentic conviction. This is distinct from cynical manipulation, where leadership knowingly deploys false narratives for ulterior purposes. Conventional frameworks treat this as a difference requiring different interventions: catch the liar through investigation and accountability; enlighten the sincere believer through education and evidence. But the distinction, while phenomenologically real, is structurally less significant than it appears. The chapter’s subsequent analysis (Sections 17.13 and 17.14) will show that the violence output and the intervention architecture are invariant across the sincere-to-cynical spectrum, because the population-level dynamics that translate Belief into coordinated action operate identically regardless of whether the leader initiating the cascade believes the narrative.

Sincere Confabulation operates as the architecture predicts:

• Source Opacity prevents the Individual from recognising M-contamination in their threat-assessment
• Confabulation generates causal narratives that feel accurate because they resolve cognitive tension
• The Individual genuinely believes the Scapegoat is responsible
• Violence is experienced as legitimate defence against real threat
• Introspection confirms the Belief because Confabulation has already shaped the introspective report

Cynical manipulation operates through deliberate deception:

• Leadership knows (or suspects) the narrative is false or exaggerated
• The narrative is deployed because it serves other purposes: political consolidation, resource acquisition, distraction from failures
• Violence serves instrumental goals distinct from the stated justification
• Leadership may or may not believe the narrative; the point is its utility

The architectural analysis primarily focuses on Sincere Confabulation. This is not because cynical manipulation is rare or unimportant, but because:

1. Sincere Confabulation is the architecturally deeper problem. Conventional frameworks assume that cynical manipulation can be caught through investigation, evidence, and accountability. Even granting this (which is doubtful, given that distinguishing sincere from cynical leaders in real time is epistemically intractable), catching the cynical leader does not address the conditions that made the population receptive to the narrative. The A-insufficiency, the elevated M-intensity, the available Scapegoat Templates persist, and another leader (sincere or cynical) fills the vacancy. Moreover, the “cynical” leader’s motivation for deploying the false narrative (political consolidation, resource acquisition, distraction from failures) is itself M-generated and processed through Source Opacity. They have partial transparency about the Content of the narrative (they know it is exaggerated or false) while remaining opaque about the motivational architecture driving their behaviour. Full cynicism, in which the leader has complete transparency about both narrative and motive, may not exist within the AIM architecture.
2. Introspection-based constraints fail specifically against Sincere Confabulation. Just War Theory and similar frameworks assume that requiring actors to examine their intentions will filter out bad-faith justifications. This assumption fails architecturally. The sincerely Confabulating actor examines their intentions and finds genuine defensive necessity, because that is what Source Opacity and Confabulation have produced.
3. Population-level violence requires sincere Belief at scale, and interventions are invariant across the spectrum. A cynical leader can initiate violence, but sustained collective violence (genocide, total war, pogrom) requires population participation. Populations do not commit atrocities they know to be unjustified. Mass violence requires mass conviction, which requires the Confabulation Cascade to have operated at population scale. This is why the AIM intervention toolkit targets the cascade architecture (Subcriticality Maintenance, A-floor provision, institutional friction, information diversity, coupling‑strength reduction) rather than the leader’s phenomenology: these interventions work identically whether the leader feeding the cascade is sincere or cynical, because they target the population-level dynamics that translate narrative into coordinated action. Even the cynical leader can be treated as sincere within this model without requiring different interventions, because the scope for full cynicism (complete transparency about both narrative and motive) is greatly reduced once the model accounts for the case where all parties are sincere Confabulators.

Historical examples illuminate the point:

WWI (1914): Sincere Confabulation. Private correspondence, internal memoranda, and personal diaries from German, French, Russian, and British decision-makers show authentic conviction in defensive necessity. Kaiser Wilhelm II did not cynically deploy encirclement rhetoric while privately acknowledging German expansionism. He experienced encirclement as genuine threat. His marginalia, his private letters, his documented emotional states all confirm that M-driven Status anxiety (Germany’s position relative to established great powers) had been integrated as A-threat (survival of the German nation). He was not lying. He was Confabulating, and the Confabulation was invisible to him because Source Opacity prevented recognition of M-contamination.

Sir Edward Grey did not cynically invoke Belgian neutrality while privately pursuing imperial objectives. His recorded anguish, his hesitation, his documented uncertainty all suggest genuine moral struggle resolved by sincere Belief in defensive necessity. The M-dynamics of great-power Status competition had been processed through the architecture and emerged as felt obligation to defend civilisation against German aggression.

The same pattern held across all major belligerents. Each leadership cohort experienced authentic conviction that they were responding to threat. The Confabulation Cascade operated at elite level, producing synchronised false Belief in defensive necessity across opposing powers simultaneously. This is why WWI remains historically confounding under conventional analysis. If one side was the aggressor and the other the defender, introspection should have revealed this, at least to the defenders. Instead, all parties experienced themselves as defenders. The architecture explains what conventional analysis cannot: M-driven Status competition, processed through Source Opacity and Confabulation, produces symmetrical defensive phenomenology on all sides.

Rwanda (1994): Sincere Confabulation at Population Scale. The speed and breadth of participation in the Rwandan genocide cannot be explained by cynical compliance alone. Approximately 200,000 perpetrators participated in killing over a 100-day period, a rate requiring willing, active participation far beyond what coercion or obedience could produce.

The radio propaganda (RTLM) activated existing Templates and accelerated Confabulation Cascade dynamics. But the response was not cynical compliance with known-false narratives. Documented testimony from participants shows genuine conviction that Tutsis posed existential threat: that the killings were defensive necessity, that failure to participate endangered Hutu survival, that the victims were not innocent civilians but enemy combatants in an existential struggle.

The phenomenology was not “I know this is wrong but I must comply.” The phenomenology was “We must defend ourselves against those who would destroy us.” This is Sincere Confabulation operating at population scale, producing genuine Belief that satisfies the architecture’s prediction: M-dynamics (ethnic Status competition, historical grievance, positional anxiety) integrated as A-threat (survival of the Hutu people), generating convergent action experienced as legitimate defence.

The Rwandan case exhibits the Dual-Role Confabulation signature (Section 17.1, Step 3). Tutsis were selected through Prey-detection criteria (a minority population, identifiable, lacking powerful external allies) while simultaneously framed through Predator-attribution (planning genocide, controlling resources, threatening Hutu survival). The Selection-Framing Coincidence Test confirms Confabulated attribution: if Tutsis were as powerful as the framing described, they could not have been targeted by machete-wielding neighbours.

European Pogroms: Recurrent Sincere Confabulation. The pattern of anti-Jewish violence across European history demonstrates both the Content-independence of the Confabulation Cascade and the sincerity of Confabulated Belief.

When plague struck medieval Europe, populations required causal explanation for incomprehensible mass death. The true causes (bacterial infection, trade-route transmission, urban sanitation failures) exceeded computational tractability. The Agent-Based Causal Default completed the causal chain: Jews had poisoned the wells. The Belief was sincere. Participants were not cynically deploying a convenient narrative while privately acknowledging bacterial aetiology (which was unknown). They genuinely believed the causal account. The violence was experienced as defensive public health measure: eliminating the source of contamination.

When economic hardship struck, the same Template was available: Jewish moneylenders were responsible. When social instability emerged, Jews were responsible. The specific accusation varied with context. The structure was identical: diffuse suffering → Information Gap → Agent-Based Causal Default → Sincere Confabulation → violence experienced as defence.

In every case, the Scapegoat exhibited the Dual-Role Confabulation signature: selected because vulnerable but framed as threatening. The recurrence across centuries and diverse contexts confirms architectural derivation. The same Scapegoat Mechanism operated repeatedly because the architecture persists.

The Stab-in-the-Back Myth: Confabulation Preparing Future Violence. Post-WWI Germany faced an Information Gap of extraordinary intensity. The German military had not been defeated on German soil. The army remained intact at armistice. Yet Germany had lost the war: surrendered, accepted punishing terms, suffered national humiliation. How could this be explained?

The true causes (strategic overextension, resource exhaustion, American entry, internal collapse from civilian deprivation) were systemic, diffuse, and resistant to agent-based framing. They did not satisfy the cognitive demand for explanation. Confabulation provided closure: Germany had been stabbed in the back by internal enemies. Jews, socialists, war profiteers: identifiable agents who could bear causal responsibility for incomprehensible national catastrophe.

The Dolchstosslegende (Stab-in-the-Back Myth) was not merely propaganda deployed by cynical nationalists (though it was that too). It was Sincere Confabulation held by millions who could not otherwise explain their experienced reality. The Belief was satisfying in a way that systemic explanation was not. It resolved the Information Gap. It identified Scapegoats. It preserved the possibility of future vindication.

This Confabulation did not immediately produce violence. It prepared the Template that would be activated when subsequent suffering (hyperinflation, depression, national humiliation) retriggered the causal-demand cycle. The Nazi party did not need to invent the Scapegoat. The Confabulation was already in place, available for amplification through the Confabulation Cascade until Bayesian Lock-In Completion produced the conditions for behavioural discharge.

The pattern across all four cases confirms the architectural prediction: invariant structure (the six-step pathway of Section 17.1) producing infinitely variable Content.

The sincere/cynical distinction is extended in Sections 17.13 and 17.14 below, which identify a third pathway to violence (duty as Self-Model insulation) that operates independently of both sincere Confabulation and cynical manipulation, a fourth mechanism (coerced compliance) through which control of the state's coercive apparatus imposes participation on the entire population regardless of individual conviction, and show that once reciprocal conflict is established, the structural process producing violence is invariant across all content configurations, including the moral quality of the leaders involved.

17.5 The Economic-Crisis Case – Fold as the Exogenous Floor Reasserting

The economic-crisis case that the Critical-Coupling Fold introduces only in outline is the same architectural pathway analysed for violence, read in the domain of prices, money, and material provision, and it terminates in the same fold-style discontinuous collapse from a high-valuation, high-PM state to the low attractor corresponding to the A-I functional baseline. What the economic case adds is an explicit account of what occupies that baseline and why the return to it is violent rather than gradual.

A fold occurs whenever the accumulated gap between an inflated Belief-aggregate and whatever will eventually discipline it discharges discontinuously. The disciplining intrusion can arrive by either of two routes, and the distinction is architecturally decisive. The primary and most catastrophic route is the reassertion of the exogenous Appetite floor. Appetite answers to the body, which no consensus can revise; when a price or fiscal aggregate has inflated on self-confirming Belief to the point where real needs go unmet, the resulting deficit cannot be believed away, and it forces itself back into the system as hunger, cold and exhaustion, through default and destitution. The secondary route is purely endogenous: a coupled Belief-aggregate can fold under its own internal inconsistency, with no floor breached, when a self-referential expectation cannot be satisfied on its own terms. Endogenous folds are real but generally milder, because what is lost is Belief rather than substance, and Belief-losses are recoverable in a way that starvation and default are not. The economic collapses that scar the historical record are overwhelmingly the first kind: the Appetite floor reasserting itself after the endogenous limb has inflated past the point where real needs can be met.

This follows from the disaggregation of price. A price is a filtered aggregate of three motivationally distinct valuations, of which only the Appetite component is anchored to an exogenous floor of use-value and physiological need; the Mimetic component is self-referential, non-satiating, and has no fundamental to revert to.

Money is the substrate the signal of any deficit, once transduced, can occupy (Chapter 12, Section 12.2): the deficit signal is the Scalar, and money is what that Scalar can be carried onto from any deficit whatever. That generality is what makes a debit against an Individual's money a debit against their capacity to correct deficits they do not yet have. The Appetite floor is that capacity, the standing means of meeting bodily need, and not the momentary fact of feeling it. Appetites themselves report only the state of the body (Chapter 4, Section 4.3.1), so uncompensated extraction of money, taking without return, does not register in the Body Manager as a loss of means. It registers first in the Thinker, which knows the means are gone, and it reaches Appetite by the one route the body allows: the loss, and the Thinker's appraisal of it, set off the body's stress response, which shifts the body off its set-points, and the Body Manager reports that bodily state as deficit even in Individuals not yet in want (Sterling and Eyer, 1988; McEwen, 1998). This is why fiscal exaction experienced as illegitimate belongs on the same axis as subsistence failure: both erode the provisioning capacity on which the Appetite floor depends, and both load the Decision Hub through the body before any meal is missed. A framework that treats money as a single undifferentiated quantity cannot separate the floored component from the fundamental-less one, and so cannot see the Appetite floor being approached as the Mimetic limb inflates against it.

The floor's reassertion is not only violent; it is epistemically self-sealing. Bayesian updating is exact over inputs it does not choose, and one of those inputs is the weight the Thinker can afford to place on incoming Evidence. Under Appetite deficit the Body Manager's signal dominates the Decision Hub and the resources that deliberation, source-discrimination and the full incorporation of Evidence would draw on are consumed, so the likelihood term is underweighted and the posterior stays closer to the prior than the Evidence warrants. Under high Coupling Strength that prior is the Mimetically sourced one, so deprivation preserves exactly the Beliefs reality-contact would displace, and it does so for every Individual at once, collapsing the Belief heterogeneity that keeps the population below its critical threshold. This is the Upstream Corruption Principle: Appetite deficit degrades Belief revision as such, and in a coupled population it is Coupling-amplifying. Appendix A, Section A.5.8.7 carries it as the fidelity-attenuation channel of the floor locus, and it is why an Appetite floor referenced to the worst-off is a condition of Subcriticality Maintenance and not a welfare commitment added to it.

Recurrent catastrophic collapse is therefore entailed, not anomalous. It is the guaranteed signature of governing a floored-plus-unfloored aggregate as though it were undifferentiated, and it will recur for as long as the disaggregation is not performed. The A-raid of the aggression taxonomy, which terminates the moment the grain is secured, is the same mechanism seen from the other side: Appetite has a set-point, and a system that has driven a population below that set-point has generated a deficit that will discharge as surely as the fed raider stands down.

The fold of this section, in its economic instance, has a name in the vocabulary of Chapter 12 and a mechanism already derived there. Chapter 12, Section 12.30 establishes Parasitic Equilibrium: extraction adjusts toward the maximum the host sustains, settling into a configuration in which the dominant flow runs from Appetitive and Intrinsic satisfaction toward the servicing of the unbounded Mimetic claim. Appendix 12A identifies the locus of that maximum exactly (Section 12A.7): service equal to the whole of the population's surplus above subsistence: the Interest-Only Threshold, at which principal is never reduced, the Object borrowed for is never acquired, and everything produced beyond staying alive flows to the maintenance of claims. An extraction that adjusted perfectly would not cross that boundary but ride it, indefinitely. And riding it is holding the population at exactly the floor whose reassertion this section derives: the equilibrium is the Appetite floor read from the extraction side.

Parasitic Overshoot, the extraction passing its own maximum, is therefore the fold. To ride the boundary is to leave subsistence exactly intact and take everything above it; to overshoot is to take from subsistence itself, and a system that takes from subsistence has generated the deficit whose discharge the preceding paragraphs derive. Because Appetite is close to uniform while surplus is not, the breach registers first in the worst-off segment, whose margin above the floor is smallest, which is why the indicator developed below is referenced to the worst-off and not to the mean. Nor is the overshoot a misjudgement that better extractors would avoid. By the three mechanisms of Chapter 12, Section 12.30, each extractor's rate of extraction is a reference point for its rivals, no participant computes the aggregate load on the population that must service every book at once, and the extraction is collected now while the exhaustion arrives later, so a maximum that could in principle be ridden is in practice overshot. The overshoot is the crisis this section describes, and its arrival is architectural rather than accidental.

The Jubilee Theorem (Appendix 12A, Section 12A.8) establishes that no single instrument ends the recurrence: a clearance of accumulated claims executed with the Mimetic Premium unbounded re-crosses the service boundary on a computable schedule, restraint of the accumulation alone cannot retrieve a stock already past retrieval, and only the two together, the M-Tax standing and the Jubilee executed once, leave the corrected state held. The recurrence this section derives as the signature of governing an undifferentiated aggregate is, in that formalism, the re-crossing of the unaccompanied clearance, and the disaggregation whose absence guarantees the recurrence is the same disaggregation the correction requires.

The reassertion has a leading indicator, available only after disaggregation. Because Appetite is close to uniform across Individuals, grounded in a shared physiology that needs roughly the same calories, warmth and protection from injury, its satisfaction is a population-general requirement and not a distribution of tastes. The fold is triggered not when average Appetite-satisfaction falls but when the satisfaction of the worst-off segment is breached, because it is the exposed subpopulation, not the mean, whose unmet need ignites the discharge. The pre-fold signature is accordingly the widening of the Mimetic aggregate measured against the Appetite-satisfaction of the worst-off, a decoupling that an undifferentiated price averages into invisibility and that a disaggregated account renders observable before the breach.

Stated as a testable clause, the indicator has three named observables and a directional, dated prediction. Observable 1: the Mimetic aggregate for the domain, proxied by the component of price or claim not accounted for by use-value or income. Observable 2: the Appetite-satisfaction of the worst-off segment, measured by that segment's residual capacity to meet subsistence provision after obligations, not by the population mean. Observable 3: the ratio of the first to the second. The prediction is that a fold is preceded by monotonic widening of this ratio, with the breach occurring as worst-off Appetite-satisfaction crosses below its provisioning floor, and the fold discontinuous rather than a smooth reversion.

The clause is falsified (as the §17.5 leading-indicator application specifically, and consistent with the scale-differentiated schedule of Appendix A, A.6.3) if, in a domain independently established to be high-κ and long-τ with a lapsed disconfirmation channel, catastrophic folds of this class recur without prior widening of the worst-off-referenced ratio, or if the widening reliably resolves by smooth deflation rather than discontinuous collapse. Because the prediction constrains the direction and discontinuity of the ratio rather than its absolute level, it is not defeated by imperfection in any single proxy for the Mimetic component; it is defeated only by failure of the directional, discontinuous prediction itself. A result drawn from a low-κ domain with an active external-reality testing channel does not constitute a counterexample, because the fold is not predicted there; the domain-admissibility criterion that fixes which fields constitute admissible tests, and that confines a falsification to its domain-local locus rather than to the filtered-aggregate identity, is stated canonically at Appendix A, A.12.5 (Scope corollary). This leading indicator is the applied face of the price-disaggregation result formalised in Appendix A, A.12.5.

Two further tests sharpen the clause, one prospective and one retrodictive. Prospectively, the early-warning-signals literature for critical transitions supplies the spectral discriminator the fold reading requires: the approach to a fold-type threshold shows rising variance and autocorrelation in the leading-indicator series (Scheffer et al. 2009) without a corresponding growth in oscillatory coherence or quasi-cycle amplitude, the signature that instead marks an approach to an oscillatory transition of the kind this chapter argues is unavailable to the unaugmented architecture (Bury, Bauch and Anand 2020). The clause is accordingly falsified in its bifurcation-class reading if leading indicators ahead of collapse events in admissible domains show growing oscillatory coherence rather than monotone variance and autocorrelation growth.

Retrodictively, the two regimes predict different aftermaths: an oscillatory regime returns the system to its prior operating point when the gain falls, where a fold destroys that operating point and the system restarts from new initial conditions. The fold reading is therefore supported where post-collapse reconstruction is constitutionally discontinuous with the pre-collapse regime, the characteristic emergence of new forms of and controls on government after collapse-scale decoherence events, and would be undermined by a record in which such crises characteristically resolved by restoration of the prior institutional configuration without structural novelty. Both tests are consistency checks on the bifurcation-class reading, not validations of the model's parameters by named events.

This is the disconfirmation channel whose absence converts a manageable large aggregate into a fold-prone one: a system blind to worst-off Appetite has no reality-test on its inflating Mimetic limb.

Three instances illustrate the one mechanism. The financial crisis of 2007 to 2009, in which mortgage-backed valuations inflated as an endogenous output of a coupled ratings-and-pricing system, is developed fully at Chapter 14, Section 14.10.4.7; the point to add here is only its terminus, that the fold arrived when real households could not meet real payments, the Appetite floor reasserting through mass default, as households lost the means on which their shelter depended (Mian and Sufi 2014). In the French Revolution of 1789, sustained fiscal extraction met a subsistence population already at its Appetite floor, and the breach of that floor, not a shift in political Belief, supplied the discharge that ideology then organised (Lefebvre 1971). In the American Revolution of 1776, escalating exaction breached the colonists' provisioning capacity, the money-scalar by which Appetite is ultimately met, converting a fiscal grievance into rupture (Morgan and Morgan 1953).

The Content differs entirely across the three; the structure is consistent across all of them. Each is the Appetite floor reasserting itself after an endogenous aggregate inflated past the point where real needs could be met, and each was read by its own era as a unique failure of policy or leadership because the common mechanism was invisible without the disaggregation this framework supplies. None of the three is offered as a quantitative fit of the model's named observables; they illustrate the mechanism, and the clauses of this section state how the reading is tested.

The floor-reasserting fold of this section is, finally, one locus of a paired result. The unique critical depth of worst-off deprivation at which correction is first outrun, the Floor-Breach Ignition Corollary, is stated formally at Appendix A, Section A.5.8.7, and its structural twin sits at the opposite extreme of the distribution of provision and authority: the same formal section states the Dual-Locus Ignition Theorem, under which the capacity to ignite the collective transition is concentrated at exactly two positions (the floor treated here, and the apex derived at Section 17.7 of this chapter, with its race-condition dynamics developed at Section 17.17) with the paired derivation given at Section 17.7.8 and the floor condition's own derivation at Chapter 13, Section 13.8.4. The two conditions are independent: the floor can ignite with no involvement from the apex, as this section's revolutionary cases illustrate, and the apex can ignite with the population provisioned and subcritical, as Section 17.7 derives and the race-condition analysis of Section 17.17 develops.

17.6 The Succession Case – Fold as the Suppressed Discharge Arriving at Once

The chapter's main line runs from diffuse harm, through the demand for a cause, to the designation of a Scapegoat and coordinated violence against them. Section 17.5 read that same architectural pathway in the domain of prices and material provision. This section reads it in a third domain, and the third domain differs from the other two in what triggers the discharge.

Where a population is organised around a single Individual holding concentrated authority, the discharge is triggered by the removal of that Individual rather than by the designation of a target. No Scapegoat is required. The violence runs between contenders rather than against anyone identified as the cause of the suffering. This is a distinct pathway to the same terminus, and the chapter has not yet treated it.

17.6.1 Forced Quiescence Is Not Subcriticality

Subcriticality is the condition in which corrective mechanisms outpace mimetic amplification, and Chapter 13 establishes that it is maintained rather than achieved: it requires independent institutional friction that continues to operate. Multiple actors holding authority on independent bases absorb shocks, supply competing explanations, maintain I-domains and secure the A-Floor. The arrangement is resilient because no single failure removes all of the friction at once.

A system under complete institutional capture presents as more stable than that, and is not subcritical at all. Its mimetic dynamics are not constrained but concealed. What the population can observe is controlled by the authority whose position the observation would otherwise test, so the dynamics run at full strength and emit no visible signal. This condition is Forced Quiescence.

Two parameters of Chapter 14's account make the difference precise. Coupling Strength approaches its maximum, because the evidence base available to the population is almost entirely endogenous: what circulates is what the authority permits to circulate, and independent verification is eliminated or penalised. Decoherence Time is extended in step, because reality is denied the channels through which it would ordinarily intrude. Economic misallocation is not reported, military incapacity is not assessed, and administrative failure is reframed by the institutions that would otherwise have identified it.

The consequence is accumulation. Every error that Subcriticality would have corrected instead compounds, and every mimetic dynamic that institutional friction would have absorbed instead builds behind the control that conceals it. Such a system does not discharge tension incrementally. It stores it, and what it stores is Accumulated Mimetic Debt.

17.6.2 The Five Suppressed Channels

Five categories of dynamic are suppressed rather than managed, and each is a channel through which a subcritical system discharges continuously.

1. Mimetic competition among high-Status Individuals. Under Subcriticality this competition is routed through appointment processes, adjudication and public scrutiny, which bound escalation so that no single rivalry destabilises the arrangement. Under Forced Quiescence it is suppressed by an authority who eliminates or co-opts rivals, and the suppressed rivalry retains its intensity, because M has no satiation signal and suppression is not satisfaction.
2. Confabulation correction. Under Subcriticality, Confabulations meet independent sources, experiential contrast from I-domains, and processes that test claims against evidence. Under Forced Quiescence each uncorrected Confabulation becomes the prior for the next, which is the Confabulation Cascade running with its correction term removed.
3. Institutional competence. Authority under Subcriticality derives partly from demonstrated competence. Under Forced Quiescence positions are filled for compliance, so adjudicative, administrative and military capacity decays while the formal structures remain in place and continue to report that they are functioning.
4. The A-Floor. Under Subcriticality, provision has redundant sources. Under Forced Quiescence it is routed through distribution organised around political loyalty, so A-sufficiency becomes contingent on compliance and the Opt-Out Paradox becomes acute.
5. I-domains. Under Subcriticality these supply the experiential contrast by which an Individual can tell M-led from I-led activity. Under Forced Quiescence they are captured or crowded out, which removes the population's capacity to make that discrimination at exactly the point where it will be needed.

17.6.3 The Succession Moment

Removal of the incumbent is not a shock to one buffer. It is the simultaneous failure of the organising principle of all of them.

Every institution under Forced Quiescence held its operating authority from one will. When that will ceases, each institution loses its basis at the same moment, and the contest that follows is therefore not a competition for one position but a simultaneous contest for every institution in the system, because none holds authority that would survive the transition on its own account.

The five channels then release together, and the release is what makes the episode violent rather than merely disorderly.

Suppressed rivalry activates at once. Each contender reads every other as an existential threat, which is Apex M-Dysfunction operating with no authority above it to manage the contest and with the penalty for losing set at destruction. The intensity is architectural rather than incidental. A subordinate's position is proximate to an apex that is simultaneously their Model and the obstacle to their own standing, and proximity of that kind intensifies rivalry rather than resolving it, so what is released is not political disagreement but something closer to hatred, which is what the historical record of these episodes actually shows.

Removal by killing makes the point sharpest, because there the removers state their intention and the architecture contradicts it. When Julius Caesar was assassinated in 44 BCE his killers held that the problem was the man, that one Individual had gathered too much authority, and that removing him would restore the Republic. Caesar had by then become the point on which the Republic's tensions converged, the figure every faction positioned itself for or against. Killing him relieved none of those tensions and removed the only thing holding them in a fixed arrangement, and the contest that followed delivered the permanent single-person rule the assassins had acted to prevent, sooner than it would otherwise have arrived.

Loyalty-selected institutions must now perform. A military selected for compliance must fight, an administration selected for obedience must operate without directives, and adjudicative bodies selected for deference must rule between claimants who each demand deference.

The A-Floor fragments along the lines of the contest, because provision was organised around an authority that no longer exists, and each contender has an interest in controlling provision as leverage. The resulting A-precarity degrades the population's capacity for exactly the discriminating assessment the moment requires, and Phenomenological Collapse formats the whole of it as survival-level alarm.

Accumulated Confabulation meets reality at once. Figures that were fabricated are found to be false, capacities that were asserted are tested, and each revelation degrades what remains of the population's confidence in every other claim it has been given.

The absence of I-domains removes the diagnostic contrast. A population that has had no access to genuine Intrinsic pursuit has no reference against which to distinguish an I-led proposal for governance from a bid for standing, because it has no experience of the former to compare.

Every factor that constrained cascade velocity under Subcriticality now accelerates it, and not because those factors degraded gradually. They were held at zero and released together.

17.6.4 When the Anchor Is Not a Person

The account so far has assumed that what the field is organised around is an Individual holding concentrated authority. The architecture does not require it. What the field loses at a removal is a fixed point against which its members computed their own standing, and Chapter 2's Anchoring Object names the class that can serve as one: a person, a sacred or symbolic object, an external power, or an entire system of governance. The person-case has its own name, the Fault-Line Figure, and the Fault-Line Figure is the most familiar rather than the most revealing.

A sacred object strips the mechanism bare. Consider a people whose identity is bound up in a relic, the object around which its cohesion and its sense of itself are organised. Rivals have understood this for as long as there have been rivals, which is why taking or destroying such an object recurs across the record as a way of breaking the group that venerated it. The object issues no orders and holds no power, so nothing it did can account for the damage its loss does. What is lost is the reference point, and the political cases are therefore not a distinct kind of event. They are the same event with an acting figure standing where the object stood.

Why removal precipitates a crisis rather than preventing one follows from where the Mimetic Premium sits. The Mimetic Premium concentrates in the anchor irrespective of the valence of the attention paid to it: supporters draw standing from alignment, opponents draw standing from opposition, and commentators draw standing from proximity. Each actor's Self-Model is built partly on that orientation, and while the anchor stands the orientation is confirmed daily by every event the anchor is part of, so it never faces a reckoning. That continuous reinforcement is the Decoherence-Time Stabilisation Effect: the Anchoring Object's ongoing presence holds Self-Models at effectively infinite Decoherence Time. The anchor that appears to be causing the division is therefore also what holds the division in a stable arrangement. Remove it and every Self-Model built against it loses its confirmation at once, while the accumulated Premium, anchored to nothing, must relocate. This is the Differentiation Collapse Mechanism: the removal collapses Decoherence Time synchronously across the network, which lowers the critical threshold below the prevailing loop gain and crosses it.

An external power occupying a polity is an Anchoring Object of the same kind, and the case matters because it is the one a polity may enter by invitation. When an outside power removes a ruler and installs a government, it rarely builds institutions that stand on their own. More often it becomes the thing they lean on, so that the courts, the military and the administration draw their operating authority from the occupier exactly as they drew it from the incumbent. The arrangement is then back in Forced Quiescence with the suppression supplied from outside, and the five channels accumulate behind it as before. Departure removes the prop and the shared reference point together, which is why the unravelling is worse than a domestic succession: the occupier was also the common thing every faction had defined itself for or against. Iraq after Saddam, Libya after Gaddafi and Afghanistan in 2021 are this case.

The contrast is what establishes it. In Germany and Japan after 1945 the occupiers did the harder thing, building institutions that could eventually operate without them and remaining long enough for those institutions to take root, so that departure produced no second collapse. The Marshall Plan is the name the European half of that work is remembered by, and what it supplied was A-Floor restoration and Intrinsic Infrastructure (I-infrastructure) rehabilitation to populations whose architecture would otherwise have run on survival pressure. That is Buffer Independence installed rather than suppressed, and it is the only form of occupation the architecture does not predict to fail. The occupier-installed case is the Synthetic Buffer Corollary: the installed order substitutes for Buffer Independence rather than constituting it.

The framework does not permit the timing of that work to be incidental, which is why it is not merely a lesson about occupation. Section 17.17.4 derives the proactive form: because a population above the M-Capture threshold is architecturally incapable of self-correcting without external Hypothesis-space expansion, A-Floor and I-infrastructure provision has to arrive before the discharge rather than after it, and the obligation to supply it falls on whoever holds the capacity. The Marshall Plan is therefore the historical case of the right intervention applied at the wrong point in the sequence, after the catastrophe it would have prevented.

At the largest scale the Anchoring Object is an entire system of governance with no single person at its centre, and the unravelling can then run for generations. When Rome withdrew from Britain around 410 CE it took with it the scaffolding that had held the province together: the legions, the administration, the coinage, the tax system, and the towns that ran on imperial authority. Within a generation the money economy had collapsed, the towns had largely emptied, and the island had broken into small warring kingdoms into which settlers moved opportunistically. The same empire supplies its own control in the same period. The Eastern half survived because it had grown administrative and religious institutions that no longer depended on the emperor's person, while Britain, the most militarised and least self-supporting of the western provinces, had least of its own to fall back on and fell hardest.

17.6.5 The Inverse Proportionality and the Succession Gradient

Buffer Independence (Chapter 2) is the degree to which an institution holds its authority, competence and operating capacity from sources other than the incumbent's personal will, and each independent buffer is a resource available for managing a transition. An independent adjudicative body can settle a disputed claim. An independent military can hold security while the political question is resolved. Independent sources can inform a population about a process it is living through. Independent provision can hold the A-Floor while authority is unsettled.

Complete capture eliminates each of these by redirecting its authority to the incumbent, and the completeness of that redirection is precisely what produces the appearance of total stability, since no institution retains the standing from which visible dissent could issue. The same redirection determines how many transition-management resources exist when the incumbent goes, and at the limit the number approaches none. The contest then proceeds with no adjudication, no independent information, no separate guarantor of security and no autonomous provision, so the only mechanisms available are the ones the contenders create during the contest, which are partisan and escalatory by construction.

Every unit of apparent stability bought through capture is therefore a unit of succession resilience lost, and the relation is structural rather than correlational, because one and the same operation produces both effects. This is the Succession Decoherence Principle.

The Succession Decoherence Principle predicts a gradient rather than a threshold, and the gradient is what makes it testable. Severity should track how completely the buffers were captured, not merely whether the arrangement was authoritarian. Chapter 2 records the four types as the Succession Gradient.

1. Minimal capture. Buffer Independence is high across adjudication, security, information, provision and civil association. Transitions are managed by institutions that outlast the holder, and severity is low.
2. Selective capture. The incumbent holds some buffers while others retain partial independence. Transitions are partially mediated by whatever was not captured, and severity is intermediate.
3. Near-complete capture. Only residual independence survives. Transitions proceed with minimal mediation and violence is likely, though surviving pockets may bound its scope.
4. Complete capture. Personal authority has replaced institutional authority entirely. The whole apparatus is contested at once, violence is the default means of resolution, and severity is maximal.

The historical record carries the gradient, and it carries it across eras and political forms. Alexander's empire is the limiting case: authority entirely personal, the aristocratic councils that had constrained earlier Macedonian kings dismantled, and satraps holding office at his pleasure. It fragmented within a year of his death, and the wars among his generals ran for four decades and consumed most of the original command. China's first emperor is the same limiting case in a different century: authority concentrated entirely in one person, and within a year of his death the suppressed revolts arrived at once, within four the dynasty was gone.

The Roman principate exhibits the gradient inside a single system across time. The adoptive mechanism, by which an incumbent selected and elevated a successor while still living, reconstructed a partial bridge across the transition and produced roughly a century of comparatively orderly succession from Nerva onward; when it lapsed, the arrangement fell into the recurring military contests of the third century. The adoptive case is also instructive about what kind of buffer it was. It required each incumbent to subordinate the founding of a dynasty, which carries a large Mimetic Premium, to the selection of the most capable successor, and Marcus Aurelius's elevation of Commodus, his own son, over more capable alternatives is the point at which it failed. A buffer conditional on the motivation of one Individual is not yet a buffer.

Three further cases fix the boundaries. The Ottoman succession institutionalised the contest directly, requiring the successful claimant to destroy his brothers, which is an acknowledgement that no institutional mechanism for the transition existed rather than a solution to the problem; and when confinement of potential successors later replaced their destruction, the violence was redistributed rather than reduced, moving out of the moment of accession and into the palace politics of the reigns that followed. Stalin's Soviet Union achieved near-total capture, the purges of the later 1930s having eliminated what independent standing remained in the party and the military, and the contest after his death ran for years, resolved by the successive elimination of the contender holding the security apparatus and then of those holding the party organisation. Tito's Yugoslavia is the intermediate case with a delay. Its capture was high but incomplete, because the constituent republics retained partial administrative independence, and its founder, recognising the problem, left a rotating collective presidency behind him. That mechanism held for roughly a decade and then failed into war, which the gradient predicts rather than contradicts: partial Buffer Independence bought a decade and could not settle the Accumulated Mimetic Debt behind it.

Severity is not the only output the gradient governs. Duration is governed too, and by the same quantity read at the moment of removal rather than across the incumbency. Where an obvious successor exists and some institutions still hold authority on independent bases, a new order forms quickly, because the contest is only over who occupies the vacated position. Where the removal has swept away the structure and left nothing independent behind, there is no position to occupy in a functioning arrangement, and the society rebuilds from very little, which can take generations. Disorder therefore lasts in inverse proportion to the Buffer Independence that survived the removal. That inverse relation is the Discharge Duration Principle. Afghanistan in 2021 sits at the fast end of that scale and sub-Roman Britain at the slow end, and they are one phenomenon measured at two depths rather than two phenomena. The completeness of the capture fixes how violent the transition is; what survived it fixes how long the violence lasts.

17.6.6 The Exogenous Succession Rule and Its Brittleness

A succession plan and a succession rule are architecturally different, and the difference is what the arrangement rests on once the incumbent is gone. A plan depends on the incumbent's continuing will and is unenforceable at exactly the moment it is needed. A rule that executes itself does not. An exogenous succession rule is of that kind: a self-executing specification, such as inheritance by the eldest child or a fixed constitutional line, that resolves who is next without negotiation, consensus or competition among candidates.

Such a rule does two architectural jobs. During the incumbent's life it removes the position from the field of Objects over which the subordinate elite can compete, to the extent that the rule is treated as settled, because an outcome that cannot be moved by competition or by Confabulation is not worth competing for. And at the moment of transition it supplies a focal point that survives the incumbent, so that security and administrative structures can converge on one claimant without anyone alive to enforce the convergence.

It follows that an apparently absolute monarchy operating under a strict rule is not an instance of complete capture. It has surrendered the succession to an institution outside itself, and it is more resilient at the transition than a personal dictatorship for that reason precisely.

The qualification is that the rule's authority is itself a Belief held by the elite who act on it, and not a fact about the world that holds independently of them. That is what makes the rule brittle rather than absolute. A single rigid rule has no redundancy, and the friction it supplies is a quantity rather than a guarantee, so where a cross-cutting mimetic cascade generates pressure exceeding that quantity, the rule does not bend. It shatters, and there is nothing behind it.

England in 1688, the episode remembered as the Glorious Revolution, is the case that shows it. The polity was locked in Bipolar Convergence between Protestant and Catholic factions, and the birth of a male heir to James II set the inheritance rule and the Protestant elite's account of its own survival in direct opposition. The rule required the accession of a Catholic heir. The faction that accession would exclude read the prospect through Phenomenological Collapse, as a threat to survival rather than to standing, and generated the Confabulation that licensed setting the rule aside, alleging that the infant was an impostor smuggled into the birth chamber in a warming pan. The rule was then set aside outright. The excluding faction invited William of Orange, the Protestant husband of Mary, James II's elder daughter, to take the throne by invasion; James fled, and William and Mary reigned jointly in his place. The pressure of the confessional division had exceeded the friction the inheritance rule could supply, and the rule gave way completely rather than partially.

The arrangement did not then collapse into Forced Quiescence under whoever had won, which is where every other case in this section went. The shattered single rule was replaced by a distributed architecture in which the succession, the revenue and the adjudication of disputes were held by different bodies on independent bases, which is a constitutional settlement understood architecturally: not a better holder of the position, and not a better rule, but many buffers where there had been one.

Concentrated authority is therefore dangerous during the incumbent's tenure, because it compresses the distance between one Individual's mimetic shock and a societal crisis, and Section 17.7 establishes that limb, on the Bayesian ground Chapter 13, Section 13.7 supplies. It is dangerous after the incumbent's tenure, because the completeness of the capture determines the violence of the transition, and this section establishes that one. No phase of the configuration is safe (Section 17.7.10), which is why the corrective is the distribution of authority rather than the improvement of its holder, and why the intervention has to be installed while an incumbent who has not yet foreclosed it is still in place.

---

17.7 The Apex Route: Executive M-Capture and the Leadership Compression Principle

17.7.1 Why Population-Level Subcriticality Is Not Sufficient

The two cases above begin in the population. The economic case starts with an aggregate the population has priced and ends when the Appetite floor reasserts against it; the succession case starts with a suppression the population has been living under and ends when the suppression lifts. Both readings carry an assumption worth naming, because the architecture does not license it: that a crisis trajectory is settled by where the collective posterior sits, with leadership as the instrument through which a population's Belief is executed rather than as a determinant in its own right.

The assumption fails, and it fails structurally rather than in unlucky cases. Where authority is concentrated, the temporal distance between one Individual's Mimetic shock and a society's crisis is compressed to a length at which nothing operating at population scale can reach it. That compression is why the interventions the architecture licenses are not exhausted by anything done to or for a population.

17.7.2 Statement of the Leadership Compression Principle

Leadership Compression Principle: For any governance system in which the authority held by one Individual exceeds a critical threshold, defined by (1) command authority over crisis-relevant resources including military force, (2) control over the dominant information channels, (3) appointment power over institutional positions, and (4) Appetitive leverage over dependent populations, there exists no population-level intervention set capable of reliably preventing M-Captured leadership from imposing mimetic crisis dynamics on the governed population.

The Leadership Compression Principle follows from a temporal asymmetry between two rates the framework has already established. Crisis imposition through concentrated authority operates on timescales of days to months: an order is given, an appointment is made, a channel is saturated. Population-level intervention operates on timescales of months to decades, because the Knowledge Patch has to be distributed, Appetitive floors have to be built, Intrinsic domains have to be protected, and visibility governance has to be legislated. The race is not close, and it is not close by orders of magnitude rather than by a margin better execution could close.

Two clarifications keep the Leadership Compression Principle from being read as more or less than it is. It is stated on capacities and not on constitutional form, so whether a particular arrangement crosses the threshold is settled by auditing the four capacities rather than by consulting the arrangement's name, and Section 17.7.9 takes up why that distinction carries weight. And the Leadership Compression Principle does not say the population is irrelevant. Population-level Subcriticality remains necessary, because leadership is selected from the population and arrives at the apex carrying whatever the population's Belief has installed in it. What the Leadership Compression Principle establishes is that population-level Subcriticality is not sufficient.

17.7.3 The Four Compression Channels

The compression operates through four channels, and each is an ordinary institutional capacity rather than an abuse of one.

Scapegoat designation. A leader with control of the dominant information channels can saturate them with a Scapegoat narrative faster than scepticism can develop in the population receiving it. The Presumption Against Scapegoating, even where some fraction of the population holds it, cannot propagate faster than a state apparatus can propagate the contrary message.

Appetitive leverage. A leader with economic control can make Appetitive access conditional on participation in the narrative. Those who reject it face employment loss, benefit exclusion, or direct deprivation. This is the Opt-Out Paradox at its most acute: recognising the dynamic does not generate an Appetitively adequate alternative when the leader controls Appetitive access.

Institutional capture. A leader with appointment power can install aligned personnel across judicial, military, administrative and informational institutions. Each captured institution becomes both an instrument of imposition and an obstacle to correction, and the cascade is self-reinforcing, because early captures supply the authority for later ones.

Control of the information environment. A leader with media control can suppress distribution of the Knowledge Patch, remove competing explanations from circulation, and curate what reaches the population. The epistemic diversity on which Subcriticality depends is not argued away; it is administratively eliminated.

17.7.4 Executive M-Capture as the Principal Accelerant

Distributed mimesis builds cumulative Mimetic Premium slowly. No single Decision Hub controls more than its own integrated valuations and the small fraction of network influence its Status entitles it to. That slow accumulation across very large numbers of interpersonal transactions is the necessary precondition for civilisational supercriticality, but it is not by itself the proximate cause of the historically compressed transitions the record displays.

The proximate accelerant is Executive M-Capture: the condition in which an Individual occupying an executive level of formal institutional authority operates with an integrated valuation whose Mimetic component is dominant, and uses institutional levers to raise both the magnitude and the frequency of Mimetic-Premium extraction across an entire governed population within a single executive term. It operates through five institutional channels. Extraction-favourable regulation lowers the friction of Mimetic-Premium extraction and raises the friction of contribution, which raises the magnitude taken per transaction. Status-elevation of extraction pathways directs state honours, media attention and institutional appointments toward extraction-derived rather than contribution-derived figures, which accelerates Mimetic transmission across the population. Suppression of the framework's dissemination channels restricts, defunds or captures the educational, journalistic and academic channels through which the framework's vocabulary, or its functional equivalents, would otherwise propagate, so the Hypothesis Space narrows and lock-in to confident but architecturally deficient explanations becomes easier to reach. The propaganda apparatus institutionalises Confabulation, producing Reciprocity narratives at scale and extending Decoherence Time by raising the density of concealment within the population. And narrative-authority capture is the above-threshold counterpart to compression in Belief-formation: the executive's expressed Beliefs receive disproportionate weight in observers' updating because of the Mimetic Premium attached to the office, with the formal institutional levers superimposed on that weight.

The result is a step-change in the rate at which Mimetic Premium accumulates. What took centuries to build under distributed mimesis can be raised by orders of magnitude within a single executive term, which is why transitions into terminal-phase trajectories are observed unfolding across years even though the precondition for them required centuries to assemble. Several twentieth- and twenty-first-century cases display the pattern: the slow accumulation over generations was the precondition, and the executive-level capture was the accelerant.

All four Crisis State Variables (Chapter 13, Section 13.6.2) move under Executive M-Capture. The Price Inflation Ratio rises with the step-change in the rate at which Mimetic Premium accumulates. Concealment Density rises sharply, because a propaganda apparatus is the principal institutional driver of concealment at scale, and because suppressing the framework's dissemination channels narrows the Hypothesis Space. Status Integrity falls sharply, because status-elevation of extraction pathways operates directly on the conditions under which contribution-derived Status is recognised. And Exit Capacity falls, because extraction-favourable regulation raises the Premium in the markets an Individual cannot leave, and because a leader with economic control can make Appetitive access conditional on participation (Section 17.7.3).

So the relationship between the two accumulation rates is the central architectural fact about how a civilisation crosses the threshold. Distributed mimesis is the slow build; Executive M-Capture is the fast finish. Subcriticality Maintenance has to address both, and the two halves are not interchangeable: the slow build is addressed by reducing the drivers of accumulation across the population, and the fast finish is addressed by the institutional defences against executive capture, which is what Section 17.7.7 specifies.

17.7.5 What the Historical Record Shows

The Leadership Compression Principle generates a specific and checkable prediction: in concentrated-authority systems, threshold-crossing should occur through leadership capture rather than through population consensus. Cases of mimetic crisis reaching catastrophic expression should show M-Captured leadership holding the institutional capacity to impose it, rather than a gradual accumulation of Confabulation across a population arriving at majority endorsement.

Germany between 1930 and 1939 is the illustrative case, because the transition from parliamentary instability to genocidal war required neither majority support nor a gradual population consensus. The party never achieved majority electoral support, peaking at 37.3 per cent in July 1932 and declining to 33.1 per cent that November. The chancellorship came through elite appointment rather than popular mandate. Institutional capture, running from the Reichstag fire through the enabling legislation to party bans, civil-service purges and military loyalty oaths, was completed inside eighteen months and by administrative action. Persecution escalated at each stage through executive action rather than in response to population demand. And the initiation of war in 1939 met population apprehension rather than enthusiasm, which is the diagnostic point: the decision sat at the apex, not in the collective posterior.

The same pattern recurs under Stalin, under Mao, and in Cambodia under the Khmer Rouge. In each, imposition through concentrated authority operated faster than resistance at population scale could mobilise, which is what the mechanism predicts. These cases illustrate the mechanism; none is offered as a quantitative fit of the model's parameters.

17.7.6 The Downstream Backstop Is Not an Architecture

Compression has a consequence for the partially managed regime. Where a system remains supercritical and external buffering has bought time rather than removed the instability, the buffering does not distribute the failure mode across the population. Under compression it does the opposite: it compresses the executive-level decision interval to a length at which the corrective mechanisms the architecture relies on at population scale cannot reach the decision in time. The system has not been made safer. It has been made dependent on what happens at one node inside a shortened interval.

One consequence is that the operational chain below the executive can come to function as an ad hoc backstop. Where an M-Captured decision is taken, the path to consequence still runs through subordinate operators whose individual refusal can interrupt execution, and the Cold War near-misses are illustrations of exactly that. The officers involved were not exercising executive authority, and the Leadership Compression Principle does not apply to them; they were exercising operational refusal at single nodes through which an executive decision had to pass.

The backstop is not an architecture, and the reasons are structural. It is absent by construction in fully automated launch architectures. It is weakened by loyalty selection in any chain of command, which is to say it is weakened by precisely the appointment power the Leadership Compression Principle names as one of the four capacities. And it is not preservable as a reliability property at population scale, because nothing in it is designed. It is a fortuitous property of how the chain happens to be staffed.

Two architectural responses follow, and they discharge different parts of the vulnerability. Distributing command authority, information control, appointment power and Appetitive leverage constrains the executive level itself, so no single node can compress the interval below what correction requires; this addresses the source of the compression rather than relying on subordinates to refuse, and because what it buys is the interval rather than the correction it is the Time-Buying Principle, a corollary of the Leadership Compression Principle. Gain reduction, by reducing the five channels' contribution to loop gain until the system is subcritical, removes the Mimetic configuration that produces the compressing decision at all. The two are not substitutes. The second operates on a generational timescale and the first on the timescale of a single decision cycle, so both are required at once: authority distribution preserves the system long enough for gain reduction to complete its work, and gain reduction is what eventually removes the dependence on the distribution. A partially managed regime is therefore not a sustainable equilibrium under compression, and operational refusal downstream is not a substitute for the correction the Leadership Compression Principle actually specifies.

17.7.7 The Four Architectural Requirements

The Leadership Compression Principle refines what Subcriticality requires. Holding a population below the threshold at which convergence becomes self-reinforcing is not sufficient if leadership can impose crisis whatever the population believes. Subcriticality has to be maintained at the apex as well, which means preventing Individuals whose integrated valuation is Mimetically dominant from occupying positions where that dominance can produce civilisational consequences. Four requirements follow.

1. Distributed authority. No single Individual should be able to impose crisis alone. Multiple concurrent authorities, any one of which can block, supply the structural protection.
2. Selection constraints. Leadership selection should incorporate anti-mimetic filters, reducing the probability that Individuals whose integrated valuation is Mimetically dominant reach positions of concentrated authority. Why such filters are required is the Ascent Selection Principle (Chapter 8, Section 8.12.4), which also carries the three design results that supply them.
3. Removal mechanisms. Detection of Mimetic failure, through audience-removal testing, outcome monitoring and independent assessment, should trigger suspension of authority before imposition becomes possible.
4. Distribution of irreversible authority. Where an authority can produce a consequence that cannot be undone, its exercise should require multiple concurrent agreements, mandatory delay, and distributed cancellation authority.

The four are stated here as requirements rather than as a design. What a governance architecture satisfying them looks like, and why it looks so unlike the arrangements a reader will be familiar with, is the subject of the governance chapter. The point to carry forward is that the requirements are not preferences about how power ought to be arranged. They are the only operations available on a route into crisis that provision cannot reach, for the reason the next section gives.

17.7.8 The Two Ignition Loci

Two routes into the collective transition now stand side by side. One runs through the population: individual Confabulation Cascades synchronise through social proof into collective convergence once amplification outruns correction, which is the pathway of Section 17.1 and the Unifying Crisis Principle of Section 17.2. The other runs through the apex: a single M-Captured Individual with sufficient authority imposes crisis whatever the collective posterior. Placing the two together yields a further structural result. The capacity to ignite the transition is not spread across the population. It is concentrated at exactly two positions in the distribution of provision and authority, the floor and the apex, and the two conditions are independent of each other.

The floor condition is derived where the corrective rate is in hand, at Chapter 13, Section 13.8.4, and the result is this. Correction presupposes that evidence arriving from outside the population retains enough weight to displace Mimetically sourced priors; sustained Appetitive deficit erodes exactly that weight, and the same erosion drains the Belief Heterogeneity Buffer. Deprivation therefore raises the amplification rate and lowers the corrective rate at once, and both movements deepen with the deficit. The deficit that matters is the worst-off segment's rather than the mean, because the exposed subpopulation is where unmet need ignites the discharge. There is consequently a unique critical depth of sustained worst-off deprivation at which amplification first outruns correction across the whole guarded range of collective Belief. That is the Floor-Breach Ignition Corollary, the formal statement of the fold the economic case of Section 17.5 develops. Below that depth the population self-corrects; at it, Subcriticality fails and the floor reasserts. Nothing about leadership appears anywhere in the condition. A society can ignite from the floor with an empty throne.

The apex condition is this section's Principle read as an ignition geometry, and two features sharpen it once the floor condition stands beside it. The leader's Belief is autonomous: an individual Confabulation Cascade runs on self-generated evidence and converges toward certainty from any starting Belief, so an M-Captured leader crosses the threshold for action in finite time with no input from the population's evidence at all. And the drive behind that cascade is non-satiable, because Differential Satiation Dynamics leaves no level of provision at which an apex deficit closes, so the apex route cannot be disarmed by giving the leader more of anything. The only controllable term in the condition is authority itself, which is why every requirement of Section 17.7.7 is an operation on the authority gate rather than on the drive. And where the floor route is extensive, requiring deprivation at population scale, the apex route is a point source: one Individual suffices, and the collective posterior is irrelevant.

Exactly two positions carry this capacity, and the middle of the distribution carries none of it. An Individual provisioned above the floor supplies no forcing of the corrective rate. An Individual below the authority threshold has no bypass. And a mid-distribution Individual's Belief is one voice among many in the collective, incapable on its own of moving either rate. Ignition capacity is therefore concentrated at the two extremes and vanishes between them. That is the Dual-Locus Ignition Theorem, and the two loci it names differ in kind as well as in position. The floor drive accumulates slowly, at population scale, and can be held at zero by provision, which is where universal Appetitive-floor security enters this argument: not as a welfare commitment adopted alongside the framework, but as an entailment of Subcriticality Maintenance. The apex drive is self-arming, permanent, and cannot be zeroed at all, only gated. A society can provision its way out of the floor locus. It can only govern its way out of the apex locus. The discharge phenomenologies match the asymmetry: floor ignition is the mass-mobilised discharge whose economic face is Section 17.5 and whose historical face is revolutionary rupture, and apex ignition is imposed crisis, war on executive decision, whose race-condition dynamics are developed at Section 17.17.

The two loci are not equally dangerous, and the asymmetry runs against the one provisioning cannot reach. The floor locus arms at population scale, so its accumulation is visible while it happens and slow enough that something can be done about it. The apex locus arms inside one person, and because the cascade runs on self-generated evidence, nothing about its timing depends on how large the population is, or how sceptical, or how well informed. Section 17.7.2 gives the magnitudes, days to months against months to decades, and the intervention surfaces differ in the same direction, since the floor drive can be taken to zero by provision while the apex drive can only be gated. The locus that arms fastest is therefore also the locus with the least to work on, and the conjunction is at its sharpest wherever the authority in question can produce a consequence that cannot be undone in the time any correction would take. Command of a nuclear arsenal is the limiting case. Launch authority rests with one or very few Individuals, the consequence is irreversible within minutes, and no part of the condition draws on the population: the population reaches the apex only through who comes to hold the authority, on a timescale far longer than the one on which the authority is used. That is the point-source geometry of the apex locus, at a magnitude no correction can follow. Section 17.17.1 develops it.

The two conditions are independent, in that each ignites without the other. They are not isolated from one another, and the channel that joins them runs one way only. A population already in cascade selects its leadership from itself, which is what the Calcified Phase is: institutional positions filling with Individuals whose own cascades align with the collective one. A sustained floor breach therefore does more than ignite the population; it raises the chance that whoever arrives above the authority threshold arrives already M-Captured. Nothing runs the other way, because Differential Satiation Dynamics leaves an apex drive nothing to close, so no amount of provision reaches it. Two things follow. Holding the worst-off segment above the Appetitive floor does double work, disarming its own locus and lowering the pressure that arms the other. And it is still not sufficient, because the locus whose pressure it lowers is the one provision cannot reach: distributed authority, selection constraints and removal mechanisms are required at every level of provision, not only where a floor has been breached.

If ignition capacity sits at the two extremes and vanishes between them, the structure carrying no ignition capacity at either locus is the one that is all middle: no segment below the Appetitive floor, and no position above the authority gate. A distribution with neither tail offers no position from which either sufficient condition can be met.

The flattening is two operations rather than one, and they are not interchangeable. Removing the bottom is provisioning, universal and unconditional, referenced to the worst-off Individual because a floor that holds on average does not hold at all. Removing the top is gating, because the apex drive closes at no level of provision. Neither substitutes for the other, and the one-way channel between the loci means that doing the first well lowers the pressure on the second without ever discharging it.

The middle is not merely the region where nothing ignites. It is where the correction lives. The Belief Heterogeneity Buffer is cross-Individual Belief variance, and that variance is sustained by Individuals whose reality-contact still displaces Mimetically sourced priors: Individuals provisioned above the floor, and not positioned to have their Belief ratified by authority. A cascade drains it as it spreads, converting adherents off their prior Hypotheses. The two loci therefore do not merely ignite from the ends. They consume the middle, and the transition arrives when what they are consuming runs out, which is what the rate condition states when correction falls below amplification across the guarded range. The geometry is not a pivot about a fulcrum. It is two ends closing on the only thing that held them apart. The same structure narrows the dispersion of adoption thresholds, because the gap a miscalibrated standing can open against provisioning is bounded by the spread of provisioning itself. Both axes of population heterogeneity move favourably together: content heterogeneity is preserved because the middle is not being consumed, and threshold dispersion falls because the positions that disperse it are absent.

An all-middle distribution is not therefore a safe one, and two results bound the inference. It removes the two position-indexed sufficient conditions; it does not establish Subcriticality, which is a condition on rates and loop gains that a flat distribution can still fail, since an observability environment driving every Individual's loop gain up requires no tail at either end. And it is a maintained state rather than an achieved one, because stratification is what a high-Mimetic system produces, so a distribution flattened once and left will not stay flat. What is fixed is where the maintenance has to be directed, not that it can stop.

17.7.9 The Vehicle Is Content

What the apex locus looks like is a different question from what it is, and the two should not be run together. The imposition condition is stated on capacities: command over crisis-relevant resources including military force, control of the dominant information channels, appointment power over institutional positions, and Appetitive leverage over dependent populations. Absolute monarchy concentrated all four. So did the fascist party-state, the military junta installed by coup, and warlord rule in territories where central authority had already failed. Those are the arrangements the historical record happens to supply, and it is tempting to read the list as the finding.

It is not the finding. The labels are Content in the sense Section 17.11 gives the word, and the architecture carries no term indexed on them. Two things follow, and the second matters more. A society that has abolished a named form has retired a vehicle, not closed a gate: the four capacities can be reassembled under any other arrangement, and whether they have been is a question about capacities rather than about names. And because the labels are Content, new ones are predicted. An arrangement concentrating the same four capacities under a name not yet coined, resembling none of the historical forms closely enough to trigger recognition, is what the architecture expects next. This is why recognition by resemblance to the last case fails, and why the diagnostic has to be run on the capacities themselves.

The point generalises what this chapter establishes about the target of a cascade. The dynamics carry no term indexed on which group is scapegoated, so comparing the content of past crises searches for the common architecture in the wrong place. The same holds for the vehicle: no term is indexed on what the concentrating arrangement is called. The architecture is Content-independent twice over, in whom it selects and in what carries it, and a method looking for family resemblance on either axis is looking where the invariance is not.

Formal treatment: Appendix A, Sections A.5.8.3 through A.5.8.5 and A.5.8.7.

17.7.10 The Terminal Configuration and the Trigger-Indifference Corollary

The antecedent of the Leadership Compression Principle is contingent. Whether authority above the threshold of Section 17.7.2 comes to rest in an M-Captured Individual depends on how authority was distributed and how its holders were selected (Section 17.7.7). One state of that antecedent closes every route by which a discharge could be avoided: an M-Captured Individual holds authority above the threshold over a polity in Forced Quiescence, with the buffers that could have managed a transition captured, or with no more than residual independence surviving. This is the terminal configuration of the Leadership Compression Principle. The configuration is distinct from the terminal fold of Section 17.17: what the configuration closes is every route around a discharge, and whether the discharge can be survived is settled by the structure the discharge meets.

Four routes around the discharge are closed in the terminal configuration, and they are closed together.

Satisfaction is closed. The apex drive has no closure condition at any level of provision, so nothing given to the leader reaches the drive (Section 17.7.8; Appendix A, Section A.5.8.7).

Institutions are closed. The buffers that would have discharged the accumulated tension incrementally hold their authority from the incumbent's will, and what builds behind them is Accumulated Mimetic Debt (Sections 17.6.1 and 17.6.5).

Time is closed. Under Forced Quiescence what is suppressed is stored, so every interval of waiting adds to the Accumulated Mimetic Debt (Section 17.6.1). The leader is mortal, so every wait ends in a removal.

The leader's own act is closed. A Captured holder cannot leave, because Apex M-Capture has left nothing outside the position for the holder to step into (I-Vacancy, Chapter 8, Section 8.12.5), and a transfer of authority would reverse the one move the apex leaves the holder, the lowering of others' standing. A transfer lowers the holder's standing, and the holder reads whatever lowers standing as a direct threat (Chapter 8, Section 8.12.5). The Observer's reading of the lowered position is signed negative, and the negative reading reaches the Decision Hub as threat, felt as fear of damage to the body (Chapter 8, Section 8.5.6). Avoidance ends a felt threat (Chapter 4, Section 4.4.0), and the transfer is what the holder avoids. A peer observed transferring authority and surviving is Evidence against the threat reading (Chapter 4, Section 4.4.0), but the fall in standing itself remains, and nothing in that Evidence reaches the I-Vacancy that leaves the holder nothing to step into. Nothing inside the leader contests the drive, because the leader's Authenticity Belief drifts toward certainty that a Status-sourced drive is genuine necessity (Appendix A, Section A.5.8.5). The only correction the architecture admits for that drift is an external supply of the Thinker's priors (the I-Corrective Asymmetry, Appendix A, Section A.5.9). A leader who controls the dominant information channels can keep the Knowledge Patch from arriving (Section 17.7.3).

What is left open is the event that sets the discharge off. The imposition condition has two conjuncts, a leadership Belief above its action threshold and authority above its threshold (Appendix A, Section A.5.8.5), and in the terminal configuration the first is met endogenously, on the leader's self-generated evidence alone (Appendix A, Section A.5.8.7). A provocation by another state, an attempted coup and an attempted assassination therefore act on the outcome through the second conjunct alone: whether the leader still holds the authority the imposition condition requires. An attempt that fails leaves the leader holding that authority, and one that succeeds removes the leader. While the leader holds that authority, the discharge is imposed crisis, executive action carrying the collective trajectory to its crisis stage whatever the collective posterior (Sections 17.7.2 and 17.7.8). An imposed crisis reaches the terminal fold of Section 17.17 only where the threat it answers is attributed to another state, because an external threat is what turns Mimetic targeting outward (Section 17.17.2; Chapter 10, Section 10.6); an attempt from inside the polity is answered inside it. Once the leader is removed, whether by a successful attempt, by death in office or by the collapse of the regime, the discharge is the succession case of Section 17.6: every institution loses its basis at the same moment, the five suppressed channels release together, and the Differentiation Collapse Mechanism collapses Decoherence Time across the network (Sections 17.6.3 and 17.6.4; Appendix A, Section A.5.7.6). Waiting is not a third form, because every wait ends in one of the two.

Where the polity holds nuclear weapons, the succession case carries a danger that removal does not end. A removed leader orders no launch, but custody of the arsenal held its authority from the leader's will, so custody loses its basis at the same moment as every other institution, and the arsenal becomes one of the assets the contest for succession is fought over. When the Soviet Union dissolved, thousands of weapons and hundreds of tons of weapons-usable material passed through a period in which their custody was in doubt (Allison, Coté, Falkenrath and Miller, 1996), and that transition was among the more structured. In a less structured one, a faction that seizes the weapons holds an asset no rival can match, and the Individual who prevails by it consolidates authority around the weapons. Where that Individual is M-Captured, the terminal configuration re-forms with the arsenal at its centre, and the arsenal serves as leverage over other states, each demand backed by the threat the arsenal makes credible, so the configuration the removal ended is re-entered with the confrontation with another state already in place. Custody held on a basis independent of the regime is therefore the buffer that matters most in a nuclear polity, and securing it before the removal is among the interventions the Trigger-Indifference Corollary leaves open.

Trigger-Indifference Corollary: Where the terminal configuration of the Leadership Compression Principle obtains, neither waiting nor the removal of the leader averts the discharge. The event that sets the discharge off decides only which of two forms the discharge takes, imposed crisis or the succession case, by deciding whether the leader still holds authority.

The event decides neither how violent the discharge is nor how long the discharge lasts. The violence of a transition tracks how completely the buffers were captured, and the length of the disorder tracks the Buffer Independence that survives the removal (the Succession Decoherence Principle and the Discharge Duration Principle, Section 17.6.5). Both are properties of the structure the event meets. The trigger is Content, as the name of the arrangement is (Section 17.7.9), and comparison across crises finds no common trigger for the reason it finds no common target (Chapter 13, Section 13.6.5).

The Trigger-Indifference Corollary closes the strategies that turn on timing and on the trigger, and leaves open every intervention that changes the structure the discharge will meet. Provision delivered through channels the regime does not control holds the A-Floor while authority is unsettled, which is the one buffer capture cannot reach (Sections 17.15.6 and 17.20.6). The Knowledge Patch and a survivable exit reach the officials, soldiers and operators whose compliance converts the leader's orders into acts (Section 17.20.6). Proactive Cascading Support reaches the population before the discharge rather than after it (Section 17.17.4). The three interventions shape the discharge and do not avert it, and support held through Forced Quiescence shapes the end that the support could not prevent (Section 17.15.5). Removal is an intervention only before the configuration obtains, while the buffers still hold authority on independent bases, which is when the removal mechanism of Section 17.7.7 operates. The agency that prevents the discharge lies upstream, in distributing authority while an incumbent who has not yet foreclosed the distribution is in place (Section 17.6.6).

Formal treatment: Appendix A, Sections A.5.7.6, A.5.8.5, A.5.8.7 and A.5.9.

17.8 Scale Invariance: The Same Mechanism at Every Level

A critical implication of architectural derivation is Scale Invariance. The Scapegoat Mechanism operates identically whether the output is:

Interpersonal aggression. Individual A experiences diffuse frustration; the Agent-Based Causal Default identifies Individual B as responsible; behavioural discharge follows without requiring Bayesian Lock-In because only one actor is involved.

Intergroup conflict. Group A’s collective suffering triggers Agent-Based Causal Default toward Group B; the Confabulation Cascade produces within-group Lock-In; coordinated action follows.

Organised warfare. Nation-level suffering triggers Agent-Based Causal Default toward Nation B; state apparatus amplifies the Confabulation Cascade; military action follows.

Genocidal extermination. Severe collective suffering triggers Agent-Based Causal Default toward an internal population identified as existential threat; total Bayesian Lock-In produces systematic elimination.

The scale differs. The complexity of coordination differs. The historical particulars differ. The Scapegoat Mechanism is identical. This Scale Invariance is not metaphor but derivation. The same architectural features (Source Opacity, Confabulation, Confabulation Cascade, Phenomenological Collapse) produce the same output pattern at every scale. The Individual who punches a colleague and the state that initiates genocide are running the same program with different inputs and coordination structures.

At every scale, the Dual-Role Confabulation signature governs target selection. The bullied colleague is selected because junior (Prey-criteria) but framed as “incompetent” and “letting the team down” (Predator-attribution). The targeted population is selected because minority but framed as existential threat. The Signature does not vary with scale because the architecture producing it does not vary with scale.

17.9 Why Violence Feels Defensive

Perpetrators of violence consistently describe themselves as defenders. Aggressors experience themselves as responding to threat. Initiators frame their actions as pre-emptive necessity. This is not cynical propaganda (though propaganda may amplify it). The architecture produces this phenomenology through two converging mechanisms: Phenomenological Collapse and Dual-Role Confabulation.

Phenomenological Collapse (Section 17.1, Step 5) ensures the attack on the Scapegoat is experienced as defending against genuine threat, because M-urgency has been irreversibly integrated as survival-level alarm. The Individual’s conscious experience is of existential danger, not positional competition.

Dual-Role Confabulation (Section 17.1, Step 3) is architecturally required for the Scapegoat Mechanism to function: selection must follow Prey-criteria because targeting actual Predators is dangerous, while framing must follow Predator-criteria because targeting Prey without threat-framing would be experienced as unjustified aggression rather than defence. The brain’s moral-licensing systems require perceived threat to authorise violence.

The two mechanisms reinforce each other. Phenomenological Collapse ensures the urgency feels like survival. Dual-Role Confabulation ensures the narrative presents the Scapegoat as the source of that survival threat. Together, they produce the universal phenomenology of perpetrators: We are defending ourselves against those who would destroy us.

Confabulation provides defensive narrative. The Confabulation system generates explanations that make the Individual look good, to themselves and others. “I am attacking” is less flattering than “I am defending.” The narrative system produces defensive framing automatically, not through conscious spin.

Pre-emption follows from threat assessment. If the Scapegoat genuinely threatens survival (as the convergent Belief holds), then waiting for them to strike first is suicidal. Pre-emptive action becomes rational. The aggressor experiences themselves as prudent defender, not initiator.

Escalation confirms threat. When the Scapegoat responds to violence with violence, this confirms the original threat assessment. See, they really are violent. Our pre-emption was justified. The feedback loop tightens.

At the scale of states the defensive phenomenology produces a claim that cannot be coherent. A power with overwhelming military superiority over a designated adversary cannot at the same time face an existential threat from that adversary, because existential threat requires a capacity to destroy and the superiority claim requires that the adversary lack it. Where a superior power asserts both, the assertion is either false or self-refuting, and the architecture explains why it is made anyway: Phenomenological Collapse delivers the threat as survival-level alarm regardless of the balance of force, and Dual-Role Confabulation supplies the adversary as its source. This is the Apex Predator Paradox Principle, and it is sharpest at the global apex, where the power holding the largest arsenal and the greatest resources is also the one most consistently reporting itself endangered.

This architectural analysis dissolves traditional approaches to war ethics. Just War Theory (the dominant framework from Augustine through contemporary international humanitarian law) rests on assumptions the architecture falsifies:

Assumption 1: Actors can accurately assess their own intentions. Source Opacity denies this. The leader who initiates war to maintain Status hegemony experiences and reports defensive necessity. They are not lying. The architecture has produced genuine Belief in defensive intention by processing M-urgency as A-threat through Phenomenological Collapse before conscious access.

Assumption 2: Articulated grievances correspond to actual harms. Confabulation denies this. The grievance is generated to satisfy the causal-demand requirement and resolve the Information Gap. It feels accurate because it provides closure. Whether it corresponds to actual causation is not assessed, and cannot be assessed, by the system generating the grievance.

Assumption 3: Defenders can be distinguished from aggressors through introspection. Phenomenological Collapse denies this. Both aggressors and defenders experience themselves as defenders. The symmetry is architectural: M-dynamics produce survival-level phenomenology on all sides simultaneously.

Just War Theory asks: When is war justified? The architecture reveals this question is unanswerable through introspection. Section 17.10 develops the replacement: questions with determinable answers grounded in A-I harm measurement rather than phenomenological self-report.

17.10 The Incoherence of Proportionality

Just War Theory’s proportionality criterion asks whether the harm caused by war is proportional to the good achieved. This sounds like moral reasoning. It is not. It is Confabulation with philosophical decoration.

The criterion has no content. Proportional to what? By whose measure? Against which baseline? The concept gestures toward comparison without specifying what is being compared or how. “The harm must be proportional to the good” is not a test; it is an invitation to generate whatever justification the actor already requires.

The evaluator is the problem. Proportionality assessment is performed by the same actors whose Decision Hubs are already configured by the conflict. If M-capture has occurred (if Status competition has hijacked A-urgency circuitry through Phenomenological Collapse), the proportionality calculation will be performed by a system that has already determined the stakes are existential. The evaluation cannot be neutral because the evaluator is compromised by the very dynamics the evaluation purports to assess.

Inflated stakes guarantee the answer. The mechanism is straightforward:

1. M-amplification inflates perceived stakes (“this threatens our civilisation/way of life/survival”)
2. Inflated stakes make extreme responses appear proportional (if civilisation is at stake, then measures sufficient to save civilisation are justified)
3. The Confabulation system generates post-hoc reasoning calibrated to the action already preferred
4. The proportionality test is satisfied because the test has no external anchor

Every war in history has been defended as proportional by its initiators. This is not because proportionality is difficult to assess correctly. It is because proportionality cannot be assessed: the concept provides no objective measure, no external standard, no constraint whatsoever on what a motivated reasoner can conclude.

The criterion produces the opposite of its stated function. Proportionality purports to constrain violence by requiring that harm not exceed benefit. In practice, it licenses violence by providing a framework within which any harm can be justified. The aggressor need only inflate stakes sufficiently (which M-capture accomplishes automatically through Phenomenological Collapse) and proportionality is satisfied.

Consider what proportionality would require to function as advertised:

• Neutral assessment of stakes (impossible under Source Opacity)
• Accurate comparison of harms and benefits (impossible when M-urgency distorts valuation)
• External standard defining “proportional” (none exists or could exist)
• Evaluator independent of the dynamics being evaluated (structurally unavailable)

The criterion fails on every requirement. What it cannot do is bear the weight placed on it: judgment under it does not converge. Tested experimentally on legal and moral experts across eleven countries and on serving military officers, the principle is understood correctly in the abstract and applied sensitively to the relevant factors, and the judgments still do not agree with one another (Statman et al., 2020). A criterion whose application does not converge among the people trained to apply it cannot discipline a decision taken by a party to the dispute. Section 14.8.10 gives the reason: the weighing has no decohering Referent, so nothing outside the coupled population settles it, and what such a population deposits is a position rather than an agreement. This is not implementation failure. It is the concept working exactly as its structure permits.

This is an instance of the Moral Vocabulary Displacement Principle (Chapter 11). Proportionality is a moral vocabulary concept that has been displaced from its normative function by the architecture. The concept retains moral authority (it sounds like moral reasoning, it is taught as moral reasoning, it is invoked as moral reasoning) while performing no constraining function. The displacement occurs because the concept was never grounded in the architectural realities of human motivation. It assumes the three capacities Section 17.9 established the architecture denies: introspective access, accurate self-assessment, and neutral evaluation. A concept built on three false assumptions cannot constrain anything.

Proportionality is Confabulation formalised. The Just War theorist asks: Is this war proportional? The Confabulation system generates: Yes, because [elaborate justification calibrated to desired conclusion]. The theorist experiences this as moral reasoning. It is narrative generation in philosophical costume.

The AIM framework does not offer a better proportionality test. It recognises that proportionality-style reasoning (weighing harms against benefits through introspection by interested parties) is architecturally incapable of constraining anything. The question “Is this proportional?” must be replaced with questions that have determinable answers:

• What A-I harm is occurring?
• Who are the Scapegoats, and do they exhibit the Dual-Role Confabulation signature?
• Does this proposed action pass the Scapegoat Effectiveness Battery (Immediacy, Mechanism, Authorship)?
• What would restore the Individuals whose A-I access has been disrupted?

These questions have external anchors. A-harm is measurable. Scapegoats are identifiable through the Selection-Framing Coincidence Test. The Scapegoat Effectiveness Battery provides criteria independent of the evaluator’s motivated reasoning. Proportionality provides none of these, and after two millennia of application, has constrained nothing.

17.11 The Content-Independence of Violence

Chapter 13 established that the Confabulation Cascade is Content-agnostic: the mathematics operates identically regardless of what Beliefs are converging. This explains a historical pattern that has resisted explanation: the infinite variation in who is targeted combined with structural identity in how targeting proceeds.

Consider:

• Jews in medieval Europe and Nazi Germany
• Tutsis in Rwanda
• Muslims in Bosnia
• Kulaks in Soviet Russia
• Intellectuals in Cambodia
• Armenians in Ottoman Turkey
• Uyghurs in contemporary China

The Self-Model Moral Boundary Principle provides the micro-mechanism for this Content-independence. Each population's moral taxonomy is generated by Self-Model Inversion operating on that population's rival-constituted Self-Model. Because Self-Models are path-dependent and constituted through different rivals in different epochs, the contrast categories (and therefore the content of "evil") differ across every historical instance. Medieval Christendom, Hutu Power Rwanda, and Weimar Germany each produced radically different Scapegoat accusations because each operated Self-Model Inversion on a different Self-Model substrate. The mechanism producing the accusation is invariant. The content tracks the labeller, not the target.

These populations are all Scapegoats in the sense defined in Chapter 2: selected as targets of Confabulated Agent-based causation for experienced deficits, exhibiting Dual-Role Confabulation where selection occurred through Prey-detection criteria (vulnerability, visibility, reciprocity-exemption) while framing occurred through Predator-attribution (threat, harm, responsibility). They share no common feature of ethnicity, religion, class, or ideology. What they share is structural position: they were identified, through the Scapegoat Mechanism, as the cause of experienced suffering in their respective contexts.

A recurrent subclass of these targets, however, is not selected at random from the field of the vulnerable. Where the accusation fastens on a race, a caste, or a sex, and divides human variation into a natural and ranked kind, the target is a Diversity Referent (Chapter 14, §14.8.3.1.2): the kind-claim it encodes is held against an available decohered Referent, because human variation, in genes and in Self-Models alike, admits no such modal kind. This has a specific dynamical consequence at population scale, and it is what makes these particular divisions the durable axes of collective violence rather than incidental ones. A kind-claim that has abandoned its Referent draws no maintenance Cost from reality; the Cost is exported instead onto the group the boundary excludes, and by the Boundary-Hardening Principle (Chapter 14, §14.8.6) that export is the same act as the hardening of the boundary, raising the Coupling Strength of the enclosed population and lengthening its holder-measured Decoherence Time. The decoherence pressure the belief would otherwise discharge continuously is stored behind the hardened membrane and released, when forced visibility finally arrives, as a single violent Saddle-Node Default (§17.17.5), which is the scapegoating cascade this chapter derives, now seen from its epistemic side.

The three cases remain distinct in diagnosis: the race case closes the Appetitive-Deficiency channel cleanly, the beneficiary being the better-resourced party by construction; the sex case turns not on non-modality but on the point that a bimodal difference licenses no ranking; and the neurodiversity case leaves the Appetitive channel open, its sustaining parties not uniformly the better-resourced. But each, once its kind-claim is enforced as fact, supplies the exported Cost that hardens a boundary toward the fold, which is why a population under Appetitive precarity reliably finds its scapegoat along one of these lines. This is the Agentic Precedence Principle (Chapter 10) in its crisis-mode: an available systemic account of human variation set aside so that a group can be installed as the agent-cause of a diffuse harm.

The Content (who is blamed) varies with local conditions, historical narratives, and available Templates (Chapter 11). The form (how blame converges and produces violence) is invariant because it follows from architectural constraints that do not vary across cultures or epochs. This Content-independence is why the common element has proved so hard to locate in the Content. Comparative work on pogroms, genocides, and wars has established structural preconditions that distinguish the cases which escalate from those which do not, and has built forecasting models on them (Harff, 2003). What it has not found is a common feature of the targeted group: the search for what Scapegoats share, across cases, returns little (Uzonyi and Asal, 2020). That is the result the architecture predicts. The commonality is in the Scapegoat Mechanism, and the Content is precisely where it is not.

The Confabulation Cascade operates identically whether converging on “Jews are responsible” or “Tutsis are responsible” or “the bourgeoisie is responsible.” The mathematics is the same: endogenous evidence generation → prestige-biased transmission → social proof amplification → Bayesian Lock-In. The target is whichever candidate satisfies the Prey-detection criteria of the Four-Type Agent Taxonomy within the local context: vulnerable enough to be safely targeted, visible enough to be identified, and reciprocity-exempt enough that targeting does not trigger costly retaliation.

Every listed Scapegoat population exhibits the Dual-Role Confabulation signature, as the detailed case analyses in Section 17.4 demonstrated for WWI, Rwanda, and European pogroms. The invariance of the Signature across radically different Scapegoats, epochs, and ideologies confirms that the Signature is produced by the architecture, not by the Content.

The Individual-as-Object Principle (Chapter 8) provides the micro-mechanism. Each member of the targeted group is processed through the Decision Hub as an Individual-Object, receiving a scalar valuation through Common-Currency Integration. When the Confabulation Cascade shifts population-level priors toward “this group is responsible,” each Individual member of that group inherits the Scapegoat valuation through PM-transfer: the social proof that “everyone knows they are responsible” enters the observer’s Decision Hub as M-relevant evidence about each encountered member’s Object-value. Source Opacity prevents the observer from recognising that this negative valuation is M-constituted (produced by copied Beliefs rather than independent assessment).

17.12 Why a Declared Right Does Not Bind the Belief That Breaks It

Section 17.10 showed that the rule international law uses to permit violence, the proportionality test, cannot constrain anything. This section examines the rules international law uses to forbid violence. They fail too, and for a reason this chapter has already assembled.

The Universal Declaration of Human Rights was adopted by the United Nations General Assembly on 10 December 1948 (United Nations, 1948). Its Article 5 reads in full: "No one shall be subjected to torture or to cruel, inhuman or degrading treatment or punishment." There is no exception clause. Article 4 prohibits slavery and the slave trade "in all their forms." Article 3 gives everyone "the right to life, liberty and security of person." Article 9 forbids "arbitrary arrest, detention or exile." Article 1 states that "all human beings are born free and equal in dignity and rights," and Articles 2 and 7 forbid discrimination on grounds including race, colour, sex, language, religion, and political opinion.

More than seventy treaties have since been built on that foundation, among them the Genocide Convention of 1948, the Convention on the Elimination of All Forms of Racial Discrimination of 1965, the two International Covenants of 1966, and the Convention against Torture of 1984. Between them these instruments have been ratified by almost every state on earth.

Every one of these prohibitions is broken every year, in many countries, including countries that have ratified the treaties forbidding the conduct. This is not a secret and it is not new. What is harder to explain is that ratifying a human-rights treaty is not reliably associated with a state treating people better afterwards, and on some measures is associated with treating them worse (Hathaway, 2002; Hafner-Burton & Tsutsui, 2005).

The usual explanations are that enforcement is too weak, that political will is lacking, or that sovereign states resist outside authority. Each explanation points to the same remedy: enforce harder. This section reaches a different conclusion. Routine breach is what the architecture predicts of a system built this way, and enforcing harder does not act on the quantities that produce it.

17.12.1 The Rights Name Facts the World Has Already Settled

Article 5 is about torture. Torture inflicts severe pain, damages people lastingly, and does not reliably produce true information. Article 25, which guarantees food, clothing, housing and medical care, is about deprivation. People deprived of those things sicken and die. Article 11, which presumes the accused innocent, rests in part on the fact that a confession extracted by force is unreliable evidence of guilt.

Articles 1, 2 and 7 are about equality, and they rest on the finding Section 17.11 has already established. Human variation has no modal type. There is no standard human being from whom others are departures, in genes or in Self-Models, so the variation cannot be carved into natural kinds and the kinds cannot be ranked. That finding is the Diversity Referent, and it governs the largest single group of rights in the Declaration.

In the AIM Framework's terms, each of these is a decohered Referent: a definite fact about the world, reached by a method that can be traced back to an actual test, which has already collided with the claim it contradicts. The world has already answered the question, and the answer is available to anyone who wants it.

The answers are not all of the same quality. Some were established by testing and are decohered Referents in the full sense: the absence of natural joins in human genetic variation, the unreliability of forced confessions, the physiology of starvation and exposure. Others are true because of what the words mean rather than because a test settled them, as with the observation that killing someone ends their life. The argument below rests on the first group. The second is included because nothing requires leaving it out, not because it is doing any work.

Because the answers are written down and available, no Belief that licenses a violation can claim that the question was never asked. It cannot claim to sit in Type I Exclusion. Type I Exclusion is the condition in which the true explanation is not merely doubted but missing altogether from the range of explanations the holder is able to formulate, because nothing from outside ever arrived to supply it. The Declaration puts the answer where it can be seen, which is the precondition of the only repair available in this case.

17.12.2 Writing the Answer Down Does Not Make Anyone Follow It

What the instruments do not do is make the answer govern anybody's Belief.

Chapter 14 states the distinction exactly: what matters is not whether a Referent exists but whether it has authority over the Belief. A fact that is available and ignored does no work. The Belief has cut itself off from the reality that would otherwise make it expensive to hold, and the expense falls instead on the people the Belief excludes.

Inviolable means that a right must not be broken. It is read as though it meant that a right will not be broken. The architecture leaves the first meaning completely untouched: nothing here weakens the claim that these things must not be done. It returns the opposite answer on the second. The gap between the two is the breach rate.

A person who violates a right may hold any of three quite different Beliefs, and they do not all fall the same way.

1. That the people in question are a lesser kind of person. This kind-claim runs through slavery, apartheid, caste, pogrom and genocide, and it is the characteristic case, covering the axes along which collective violence actually runs. In every one of them the group the claim ranks is a boundary the Agent-Type classifier has drawn under inherited priors, not a division the world contains: human variation is non-modal, continuous and Individual-specific in genes and in Self-Models alike, so it affords no such join and no rank across one (the Diversity Referent, Chapter 14, Section 14.8.3.1.2). The verdict is available and the kind-claim is held against it, which is Referent Abandonment (Chapter 14, Section 14.8.3.1), and which places the licensing Belief in the Referent-abandoned cell rather than the Referent-free one. What sustains the ranked kind is not contact with physical reality but Confabulation, elaborated into doctrine and transmitted between Individuals by prestige-bias and social proof, which is why the doctrine follows the institution it justifies rather than founding it (Chapter 11, Section 11.17).

   The kind-claim licenses violation by two routes. Where the excluded group is also framed as taking or threatening, the kind-claim is the Prey-selection limb of a Grievance Belief, and the framing is what authorises the act. Where the framing is suppressed instead, and the institution's work is to prevent any evidence accumulating that would make the excluded group a rival or a collaborator, the kind-claim stands alone as a stable classification: chattel slavery and caste hold this form. Referent Abandonment convicts the kind-claim on either route, and the Cost-export analysis of Section 17.12.3 applies to both.

2. That some method will produce some result: that torture will yield the information, that mass detention will reduce the danger. This is a prediction, and a prediction may simply not have been tested yet. It is not automatically a Belief held against a settled answer.

3. That every fact is as stated, and the suffering of people outside the boundary does not count. The claim is about what ought to be done. No amount of evidence settles it, and treating it as though evidence did would be exactly the overreach this chapter's method exists to prevent.

Only the first is convicted by the availability of the right's Referent. So the universal claim in this section is the narrow one from Section 17.12.1: not that every licensing Belief has abandoned a settled answer, but that none of them can claim no answer exists.

Where a Belief does fall into the first group, the eliminative proof of Chapter 14 becomes available, and it is a test to be run rather than a verdict to be assumed. Motivation comes from three sources, and the Physical-Reality Channel, the evidential route by which physical reality constrains a Belief, is not a fourth source, so the proof requires four closures rather than three (Chapter 14, Section 14.8.3.1). The Physical-Reality Channel is closed by the availability of the decohered Referent: physical reality has already returned its verdict and made it accessible. The channel an interested party invokes in defence of the Belief is not the one the Referent closes. The I-channel, valuing the Belief because it is true, is closed by the fact that it is not true.

The A-channel, holding the Belief because it secures something the holder's body or livelihood actually needs, does not close by itself. It has to be shown closed, and there are two ways of showing it. The first asks who benefits from the Belief and who pays for it, and whether the beneficiaries are short of anything the Belief supplies. The second asks whether those same people would keep the Belief if they woke up tomorrow on the other side of the line. In the typical case both close decisively, because the party sustaining the Belief is the better-resourced one and the whole cost falls on the excluded group. Where a party is under genuine material threat, the closure has to be demonstrated and cannot be assumed. Where all four closures hold, only the M-channel remains: the Belief is being held for the position it confers among other people.

17.12.3 Why Banning It and Punishing It Is Not the Repair

The Hardness–Referent Matrix sets out, for each kind of Belief, what actually repairs it. For a Belief held against an available answer there is exactly one repair: follow the answer.

A prohibition restates an answer the holder already has and is already ignoring. It opens no route to that person that was not open before. This is why the instruments can be extended, strengthened and more widely ratified without the breach rate moving. The work they do is the work already done by the answer being available, and that was already not enough.

Punishment does something, but far less than is asked of it. The claim is about size, not about kind.

A Belief that has been contradicted by reality normally becomes uncomfortable to hold. The AIM Framework calls that discomfort the Cost of the Belief, and when the holder bears it, it is what eventually forces a change of mind. What a hardened boundary does is move that Cost onto other people. In a human-rights violation, the people it moves onto are the victims. Their injuries are where the Belief's bill is settled.

Punishment claws back a fraction of that bill and puts it on the holder. That is a real effect on the only quantity that corrects anything. The fraction is nonetheless small, arrives rarely, arrives years late, and is unlikely to arrive at all, while the export is continuous, total, and borne by an entire population for as long as the Belief is held. What would make punishment bite is the certainty of being caught rather than the severity of the sentence, which Chapter 16, Section 16.9 derives from the rate form of the Subcriticality condition, and certainty is precisely what this system cannot deliver.

Every instrument described so far meets the violation at the point of output. This chapter has spent its length establishing what produces that output: a Belief formed under Source Opacity, hardened by a Cost placed on the people it excludes, held in a population whose coupling rose as the boundary hardened, discharging into conduct when the loop crosses. Prohibition addresses the last term in that sequence. Sanction addresses the last term after the fact. Neither reaches any earlier term, and the earlier terms are what produce the next instance.

That also fixes what an output-level instrument can achieve at best. It must be applied once per instance, and it must reach each instance to do anything at all, which is why its efficacy turns on the certainty of being caught rather than the severity of the sentence, and why no severity carries a super-critical loop back below its threshold (the Punitive Inversion Principle, Chapter 16, Section 16.9). Instances are unbounded and no apparatus reaches all of them. An instrument that acts on the conditions is applied to the conditions, and the conditions are few. Securing Appetitive provision does not have to be repeated for each Belief it makes cheap to drop. It changes what all of them cost at once.

The repair for this kind of Belief has two halves. The first is following the answer. The second is softening the boundary. That means not charging people for failing to fit the category, so the cost of not fitting falls on nobody. People then move across the line as understanding changes, instead of being pinned on one side and billed for standing there.

A criminal-enforcement regime performs neither half. It acts on the violation afterwards. It leaves completely untouched the ordinary arrangements by which belonging to the excluded group is made expensive: the worse schools, the closed jobs, the harder policing, the documents that will not come. Those arrangements are the export, and by the Boundary-Hardening Principle the export and the hardening of the boundary are one act, not two. So the enforcement regime leaves the Coupling Strength of the enclosing population exactly where it found it, while looking like the remedy. That is why a country can have a functioning human-rights apparatus and an undiminished rate of violation at the same time. The apparatus is not the hardening operation. It is a separate operation, running alongside the hardening one, on a different quantity, at a scale too small to offset it.

17.12.4 Stopping the Act Without Changing the Belief Stores the Damage

Sometimes enforcement does succeed in stopping the conduct while the Belief behind it stays intact. That is not repair. It is postponement, and the postponement compounds.

The pressure that reality exerts on a false Belief does not disappear because the Belief is prevented from expressing itself. It builds up behind the hardened boundary. When the boundary finally gives, the pressure is released all at once, as a Saddle-Node Default (Section 17.17.5): a sudden collapse rather than a gradual adjustment. This is the formal reason for a pattern the historical record supplies again and again and which gradual accounts of reform cannot explain. Decades of a discriminatory Grievance Belief, suppressed in conduct but never corrected in content, end in a single episode of mass violence rather than in slow improvement.

The settlement is unequal even then. Whatever was already taken from the excluded group (in food, in safety, in schooling, in years of life) is gone, and the collapse does not return it. Only the share the holders kept comes back to them.

Correction is incomplete in any case. Once an idea has entered someone's mind it cannot be removed by retracting it. It stays, at reduced weight, and the reduced weight is still enough to tilt what that person concludes about related questions later (Chapter 14). Convictions, official findings, apologies and truth commissions weaken a licensing Belief in a population. They do not delete it. So responsibility for the quality of such a Belief belongs at the moment it is spread, not at the moment it is corrected. That puts the obligation before the conduct rather than after it, and on the people who circulate the kind-claim, not only on the person who acts on it.

17.12.5 The Cost of Putting a Belief Down

The instruments should not be weakened, narrowed, or abandoned, and nothing here says violation is inevitable. The Declaration is the record of the answer, and it is that record which makes the contrary Belief correctable rather than a Belief about a question nobody ever settled. What fails is not the statement of the right. It is the theory of constraint built on top of it.

This chapter has treated violence as architectural output. A prohibition, and the sanction attached to it, act on that output and on nothing else. They meet the conduct at the end of the chain, after the Belief has formed, after it has hardened, after it has discharged into action, and they leave every term before that point running as it was. An output whose conditions are intact regenerates. That is why the breach rate does not respond to more of the same instrument.

The rate responds to the conditions, and two of them fix it: how tightly coupled the population is, and what the Belief costs. The second has two sides, who pays to keep it up and what it costs the holder to put it down. Nothing in the current apparatus touches either.

What a holder can lose. By Axiom 1 there are exactly three sources of motivation and no fourth, so what an Individual forfeits by abandoning a Belief is settled before any survey of cases. They can lose Appetitive provision, they can lose Intrinsic access, or they can lose Status. Nothing else is available to lose, because nothing else is available to want.

Two of the three can be guaranteed and one cannot. Food, shelter, medical care and safety can be provided to everyone at once. Materials, tools, and protected time and space for absorbing work can be provided to everyone at once. Status cannot. It is relative rank, and a rank that everyone holds is not a rank.

So securing A and I leaves exactly one thing at stake. Guarantee Appetitive and Intrinsic provision unconditionally, not contingent on what an Individual believes, whom they associate with, or whether the hierarchy stands, and abandoning the Belief cannot cost the holder A or I. By the exhaustiveness of Axiom 1, what remains at stake is Status and nothing besides.

That residual has no substrate of its own. By Axiom 5, M cannot produce Objects. The computation runs from Object to valuation, with no inversion available within the channel, so M deposits a Mimetic Premium on whatever A- and I-Objects are already present. The Status the Belief secures is therefore not something the Belief makes. It is a toll taken on A and I provisions. The holder stands as a toll-gate on the excluded group's access to that provision while contributing nothing to it. And what the toll buys cannot be held by everyone at once: everyone can be fed without anyone going hungry, and one person's absorbing work does not foreclose another's, but rank rises for one only as it falls for another.

Both sides of the ledger are now visible. On one side, real subtraction from the excluded group's Appetitive and Intrinsic provision: hunger, danger, untreated illness, schooling that does not arrive. On the other, a position no one can add to the world, taken as a toll by a channel that produces nothing.

Why the Belief takes the form it does. Lowering another's position extracts more Mimetic Premium per unit of effort than raising one's own. Two grounds compound. Being lowered, and lowering a rival, run on the high-gain alarm channel that is slow to settle, while raising oneself runs on the ordinary reward channel that decays normally. And because position is a difference, a single downward move widens the gap from both ends at once. The licensing Belief therefore takes the form it always takes, a claim that a group is lower, not by cultural accident but because down is the cheaper route to the same premium.

There is nowhere to export to. The word export suggests the Cost leaves the system. It does not. Under the Relational Subcriticality Principle the couplings through which Individuals mutually constitute one another's Self-Models are the same loops whose gain must stay below the critical threshold for the field to hold: one object seen from two directions. A Self-Model left deformed is at once a damaged part of the field and a loaded loop raising the field's gain toward its next crossing. The excluded are nodes in the holder's own field. The Cost placed on them is not removed from the system. It is converted into stored load inside it, and the fold, when it arrives, arrives for everyone the field contains.

Who is obliged to act. Under the Cascading Structure of Rights and Duties the duties to protect A, to enable I, and to bound M fall first on the nearest and most capable parties, passing outward to organisations, to the state, and finally to international actors only where inner layers genuinely cannot discharge them. No level can excuse its inaction by pointing at another. The duty to lower the abandonment cost rests first with the state maintaining the boundary and passes outward on that state's failure, not on anyone's permission. Sovereignty does not answer the obligation. It is one of the layers the obligation runs through.

The second quantity. Coupling Strength moves on two levers. Many small, loosely connected Belief-domains contain a collapse where one large domain spreads it everywhere, subject to the boundary-independent-output exception of Section 17.17.6. Curation that lowers coupling inside a community and restores contact across domains is therefore a structural intervention where banning particular speech is a behavioural one. And an Individual who abandons the licensing Grievance Belief leaves the group that holds it: where leaving costs employment, housing or standing, the Belief is held by people who cannot afford to stop holding it. Guaranteeing that refusal forfeits neither Appetitive sufficiency nor Intrinsic access makes departure survivable, and departure sets the threshold, because the point at which a population folds falls as it approaches unanimity and the variable fixing that threshold is the probability of the minority pathway (Chapter 14, Appendix 14A.6).

The Knowledge Patch (the Patch) comes first regardless. The Belief is sustained through a channel its holders cannot feel, and a population without the words to recognise that in themselves will experience every measure above as something done to them, and will resist it mimetically.

What follows at the scale of states, and what enforcement would have to look like to run on these quantities rather than on the conduct, is the subject of Section 17.19.

The instruments write down an answer that the licensing Belief ignores, and writing an answer down does not make anyone follow it. Banning the conduct repeats an answer already available and already ignored. Punishing the conduct recovers a small, late and unlikely fraction of a cost the holder has already imposed on someone else. Neither touches the everyday arrangements that keep the boundary hard. Routine breach is what this architecture predicts of a system built that way. The rate can still be brought down. But it comes down by working on how tightly the population is coupled, on who is paying to keep the Belief standing, and on what it would cost the holder to put it down. Those sit upstream of the conduct the instruments address, in a register the instruments do not operate in at all.

17.13 Duty as Self-Model Insulation

The sincere/cynical distinction, while important, risks obscuring a mechanism that operates parallel to both: the role-constituted licensing of violence through duty. Duty is an M-grounded Self-Model element (the Self-Model Disruption Principle establishes that M-grounded identity elements are constitutively dependent on reference groups and their evaluative criteria) that shapes the Individual’s self-concept in ways that do not merely permit violence but at times require it, while simultaneously insulating the self-concept from the violence it produces.

The mechanism has three components, each architecturally derivable.

Component 1: Duty shapes self-concept. The Individual who occupies a role (soldier, leader, police officer, tribal elder, party official) does not merely perform functions associated with that role. The role becomes a Self-Model element: "I am a leader," "I am a soldier," "I am a protector." This element is M-grounded because it is constitutively dependent on the reference group’s recognition of role-occupancy. The leader who is not recognised as leader by any reference group has no leadership identity to maintain. Because the Self-Model element is M-constituted, Phenomenological Collapse (Section 17.1) ensures its maintenance is experienced as survival-level necessity rather than Status anxiety.

Component 2: Duty does not permit opt-out. Unlike a preference or even a sincere Belief, duty presents itself to the Individual as a non-discretionary obligation. The soldier who believes the war is defensive has a Belief. The soldier who experiences soldiering as constitutive of who they are has a Self-Model element that does not permit the question "should I participate?" to arise as a live option. The question has already been answered by the architecture: I am the kind of person who does this. Opt-out would require not merely changing a Belief (which I-Override could in principle accomplish) but dismantling a Self-Model element, which the Self-Model Disruption Principle establishes as an identity crisis, not a Belief revision. The cost of opt-out is experienced as self-destruction, not merely as updating a probability estimate. This is why duty-bound participants in collective violence are often the hardest to reach: you are not asking them to change their mind about the enemy, you are asking them to stop being who they believe themselves to be.

Component 3: Duty separates self-concept from behaviour. When violence is licensed or required by the duty-constituted Self-Model, the Individual’s self-concept ("I am a good leader," "I am a responsible officer," "I am fulfilling my obligations") becomes decoupled from the content of their behaviour. The leader who orders violence while sincerely experiencing themselves as "a good leader fulfilling their responsibility to protect their people" is not lying (not cynical manipulation) and may or may not hold a sincere Confabulated Belief about the target’s culpability. The duty-structure provides a third pathway to violence that operates regardless of where the Individual sits on the sincere-to-cynical spectrum: the Self-Model element "I am a good leader" licenses whatever behaviour the duty-role specifies, because the evaluative criterion is role-fulfilment, not act-content.

This separation produces the characteristic phenomenology of duty-licensed violence: the Individual may experience genuine moral discomfort about specific acts while experiencing no identity-level disruption, because the Self-Model element ("I am doing my duty") remains intact throughout. The discomfort attaches to the act. The self-concept attaches to the role. The architecture keeps them separate, because the Self-Model is confabulated at a level of abstraction that does not require act-level consistency. "I am a good leader" does not contain within it a specification of which acts are permitted. It contains only the requirement that the role be fulfilled, and Confabulation fills in whatever act-justification the situation demands.

Components 1 through 3 establish why duty-bound Individuals participate in violence even when they experience moral discomfort about specific acts. But all three Components operate through the Individual's own motivational architecture, specifically through M-grounded Self-Model elements whose violation would constitute identity-destruction. A fourth mechanism requires no such motivational integration. It requires only the state's monopoly on lethal force and the structural impossibility of non-participation.

Coerced compliance and the impossibility of opt-out. Once the Confabulation Cascade has captured enough of the population to control the state apparatus and its coercive instruments (military command, police, courts martial), the remaining population, which may constitute a majority, is subjected to the consequences regardless of their Beliefs, their duty-constitution, or their position on the sincere-to-cynical spectrum. No one can opt out. This is not limited to soldiers. It applies to the entire population.

A young man who does not believe the war is justified is conscripted. If they attempt to refuse an order to kill, they are executed for desertion. But the civilian who opposes the war fares no better. Their economy is redirected to wartime production. Their city is bombed by the enemy, who does not distinguish between citizens who supported the war and those who opposed it. Their resources are rationed. Their borders are closed. Their children are conscripted. The enemy's retaliation falls on the entire population without differentiation, because the enemy's own Confabulation Cascade has identified the whole nation as the Scapegoat, not the sub-group that initiated the conflict. The Individual who never supported the war, who saw through the Scapegoat narrative, who would have voted against it if given the chance, is killed by the same bombs as the Individual who demanded it.

This is not a marginal case. It is the historically dominant case.

Majority support is not required. Hitler never won a majority in a free election. His party's highest result was 37.3% in July 1932. Even after his appointment as Chancellor, with paramilitary thugs intimidating voters at polling stations, the subsequent election returned only 43.9%, still short of a majority (Evans, 2003). Yet the entire German population was compelled to participate in a war of annihilation that most had no meaningful opportunity to refuse, and from which none could exempt themselves. The war's consequences (conscription, bombing, starvation, occupation, division of the country for half a century) fell on every German, including those who had never supported the regime. Majority rule was not required. Control of the coercive apparatus was sufficient.

The mechanism is straightforward. The state's coercive instruments translate the M-Captured leadership's crisis dynamics into consequences across the entire population. For soldiers, the translation is direct: fight or be shot. The Decision Hub processes this as an A-threat of maximal urgency, and the output is compliance, because the architecture prioritises immediate survival. For civilians, the translation is structural: the state redirects the economy, closes borders, imposes rationing, and subjects the population to the retaliatory violence that its own actions provoke. No Individual decision, no act of conscience, no private dissent can extract the Individual from these consequences. The war is done to them whether they consent or not.

This produces a distinctive and architecturally important phenomenology. When soldiers are subsequently accused of war crimes, they correctly report that they had no choice. This is not Confabulation. It is not excuse-making. It is an accurate description of their architectural situation. Their Decision Hub received two inputs: participate in violence against the designated Scapegoat, or be killed by their own state. Source Opacity is not relevant here, because no M-contamination is required to explain the behaviour. The coerced participant is not running the Scapegoat Mechanism through their own architecture. They are running the survival calculus that the state's coercive apparatus has imposed upon them. The Nuremberg principle ("I was only following orders" is not a defence) is morally coherent only to the extent that the Individual retained genuine optionality. For the soldier facing execution for desertion, the optionality was architecturally absent.

This fourth mechanism refines the claim made in Section 17.4 that "mass violence requires mass conviction." Mass violence requires mass participation, which is correct. But mass participation does not require mass conviction. It requires sufficient conviction among those who control the coercive apparatus, plus the apparatus itself. The Confabulation Cascade need not reach P(H|Group X) → 1 across the majority. It needs to reach Lock-In among a sub-majority that controls the state, and the state then imposes participation on everyone else, through conscription for soldiers and through inescapable structural consequences for civilians.

The implication for the race condition (Section 17.17) is direct. The critical threshold for population-level violence is lower than any framework assuming majority consent would predict. The cascade needs to capture only enough of the population to seize and operate the coercive apparatus. The threshold is not 50% conviction. It is whatever fraction is sufficient to control the military, the police, and the judiciary, and to make individual non-compliance either lethal or irrelevant. In Weimar Germany, 37.3% proved more than sufficient, because institutional friction (the damping factor identified in Section 17.17.2) had been eroded to the point where a determined minority could capture the state. This is the Leadership Compression Principle of Section 17.7 applied not merely to launch authority but to the entire coercive apparatus. The leader does not need majority support. The leader needs control of the instruments that make individual non-compliance lethal for soldiers and structurally inescapable for everyone else.

17.14 The Symmetry of Violence and the Content-Independence of Structure

This analysis dissolves a distinction that conventional frameworks treat as fundamental: the difference between a "good leader" acting from genuine conviction and a "cynical leader" manipulating for personal advantage. From the architecture’s perspective, this distinction is real at the level of individual phenomenology but irrelevant to the structure of violence once reciprocal conflict is established.

The definitions of Scapegoat and Scapegoat Mechanism (Section 17.1) remain operative. Once reciprocal violence is established, each side functions as the other’s Scapegoat. Side A identifies Side B as the threat. Side B identifies Side A as the threat. Each attribution may satisfy some or all of the Scapegoat Effectiveness Battery tests: the threat may be real, the mechanism may be identifiable (they are attacking us), the authorship may be traceable (their leader ordered the strike). The good leader may have been genuinely attacked. Duty demands they mirror the aggressor’s behaviour. Truth, in the sense of accurate structural diagnosis, provides no benefit within the conflict horizon, because the conflict horizon is shorter than the time required for structural analysis to produce actionable alternatives to violence.

Both sides therefore kill symmetrically until the war concludes. Truth will not be discovered through the process at all. The war may end through exhaustion, external intervention, or one side’s destruction, and the structural causes that produced the convergence on violence may never become visible to either party. The reasons participants believe caused the war are Content. The identity of the participants is Content. The outcome of the war is Content. The reasons attributed to winning are Content. And Content, by Content-Agnosticism (Axiom 8, Bayesian Belief Dynamics; the Structure-Content Distinction, Chapter 4, Section 4.3.11), does not alter the structure of the process. Bayesian updating is Content-agnostic: it operates identically regardless of the accuracy, quality, or crisis-potential of the Beliefs being processed. The same updating machinery that locks Side A into certainty that Side B is the cause of their suffering locks Side B into identical certainty about Side A. The same Confabulation architecture that generates Side A’s sincere narrative of justified defence generates Side B’s identical narrative. The Confabulation Rationality Lemma establishes that both narratives are the unique rational output of Bayesian inference operating on restricted Hypothesis Spaces, and the Sincerity Guarantee establishes that both sets of participants genuinely believe their narrative.

The structural claim is therefore:

For any war W between sides A and B, the AIM architecture predicts:

• A identifies B as Scapegoat (the agent-based cause of A’s experienced deficit)
• B identifies A as Scapegoat (the agent-based cause of B’s experienced deficit)
• Both identifications are produced by the same architectural pathway (Agent-Based Causal Default, Confabulation, Bayesian Lock-In)
• Both identifications are sincere (Confabulation Rationality Lemma, Sincerity Guarantee)
• The duty-constituted Self-Model of each side’s leaders licenses whatever violence the role specifies (Self-Model Disruption Principle)
• The resulting violence is symmetrical in structure regardless of asymmetry in capability, moral justification, or historical context

The content variables (who started it, who is right, who wins, why they won) are all processed by the same content-agnostic Bayesian machinery (Section 17.11). They determine the narrative each side constructs. They do not determine the structure of the process that produces violence.

17.14.1 The Irrelevance of Leader Quality

The Scapegoat, in the sense that matters for this analysis, is not necessarily a Confabulated target. The Scapegoat is whoever is identified as the threat. The threat may be real. The good leader may have been genuinely attacked. The cynical leader may have manufactured the crisis. The architecture does not require the distinction to be resolvable during the conflict, and within the conflict horizon, it is not resolvable, because the behavioural demands of duty are identical in both cases.

Consider two leaders on opposing sides of a war:

• Leader A is by any conventional measure a "good leader": sincere, anguished, consultative, genuinely convinced of defensive necessity because their nation has been attacked. Their Self-Model is "I am a responsible leader who would never initiate aggression, but who will do whatever is necessary to protect my people." Their conviction is not Confabulated. The attack happened. The threat is real.
• Leader B is cynical, manipulative, aware that they initiated the aggression for reasons of political consolidation, and is deploying a defensive narrative they know to be false. Their Self-Model is "I am a strong leader who does what needs to be done," and the duty-element licenses the violence without requiring sincere Belief in the defensive narrative.

The conventional framework treats these as categorically different: Leader A is justified (or at least excusable); Leader B is culpable. Just War Theory can in principle condemn Leader B while vindicating Leader A, because Leader A passes every sincerity test, every proportionality test, every last-resort test. Leader A is right. Leader B is wrong.

The architecture does not dispute this moral asymmetry. What it reveals is that the moral asymmetry has no structural consequence for the violence output. Both leaders’ duty-constituted Self-Models require them to act. Leader A’s duty requires defence against a real attack. Leader B’s duty requires the appearance of strength in a manufactured crisis. Both produce the same behavioural terminus: coordinated violence. Both leaders’ populations experience the same Phenomenological Collapse: survival-level alarm. The attacked population’s alarm is A-grounded; the attacking population’s is M-amplified through Phenomenological Collapse. The soldiers on both sides kill symmetrically, because duty demands it from both.

This is the architectural point the sincere/cynical distinction obscures. Truth provides no operational benefit within the conflict horizon (a claim formalised as the Truth-Exclusion Principle in Section 17.16). Leader A’s accurate perception of the situation produces the same military orders as Leader B’s Confabulated perception: engage the enemy, destroy their capacity, protect our people.

Duty (Section 17.13) ensures this convergence. Neither leader can opt out, because their duty-constituted Self-Models license whatever behaviour the role specifies. Both produce identical violence until the conflict terminates.

The deeper consequence: Truth may not be discoverable through the process at all. The war ends when one side’s capacity is exhausted, not when moral asymmetry is resolved. The Confabulation architecture generates closure narratives after the war just as reliably as it generates threat narratives before it.

This is why the chapter’s analysis cannot rest on the sincere/cynical axis alone. The architectural prediction, that violence will be symmetrical, that both sides will kill, that duty will lock participants into the conflict regardless of the moral facts, holds independently of where any given leader falls on that axis. The Scapegoat (the identified threat) may be genuinely threatening or may be Confabulated. Duty does not ask the question. Duty acts.

17.14.2 What This Means for Scapegoat Analysis

The canonical Scapegoat definition, with its emphasis on Dual-Role Confabulation (prey-selected, predator-framed), describes the diagnostic signature of confabulated attribution, the case where the Selection-Framing Coincidence Test fails: if they are as powerful and threatening as described, why can we target them without expecting overwhelming retaliation? This diagnostic remains valid and important. It identifies the paradigmatic case where the attribution is almost certainly confabulated.

But war between roughly matched adversaries is the case where the diagnostic is not dispositive, because each side genuinely is threatening, each side genuinely is causing harm, and each side genuinely can retaliate. The Dual-Role signature may not be present because the framing (predator-attribution) coincides with the selection criteria: they are targeted because they are genuinely dangerous. This is precisely why the duty-licensed violence pathway is so difficult to interrupt. The architecture is producing violence through a channel where the standard diagnostic tools (Selection-Framing Coincidence Test, Presumption Against Scapegoating) lose traction, because the content-level attribution may be locally accurate.

The AIM contribution is to show that even when the content-level attribution is accurate (they really are attacking us), the structural process is identical to the process that produces paradigmatic scapegoating. The same Axioms (3, 4, 8, 9) generate the convergence. The same Confabulation architecture generates the narrative. The same Bayesian Lock-In prevents structural analysis. The same duty-constituted Self-Model licenses the violence. The content accuracy of the attribution is irrelevant to the structural diagnosis, because the structural diagnosis concerns the process that generates violence, not the accuracy of the Beliefs processed.

The practical consequence: structural intervention (Subcriticality Maintenance, institutional buffering, coupling‑strength reduction) must operate before reciprocal violence establishes the symmetrical Scapegoat structure. Once both sides are simultaneously Scapegoat and Scapegoater, the content-accuracy of their attributions makes the standard diagnostic tools insufficient, and only structural measures (ceasefire, external mediation, conflict-horizon shortening) can interrupt the architectural pathway. The formal proof that Truth is not merely operationally irrelevant within the conflict horizon but architecturally excluded from the violence process follows in Section 17.16.

17.14.3 The Structural Determination Principle

The process that produces violence carries no term indexed on the Content of the Beliefs it processes, on the identity of the leader who feeds it, or on what the parties intend, and the same is true of every mechanism the architecture runs but one: the Agent-Based Causal Default constrains the form a completion takes, agentive rather than systemic, and enters only as a constraint on the classifier upstream of the fast loop (Axiom 9; Chapter 4, Sections 4.3.9 and 4.3.11). What an arrangement produces is fixed by its structure. The Content carried through an arrangement and the intentions of the Individuals who fill it bear on the output only by setting structural parameters. This is the Structural Determination Principle.

Content bears on Belief only by setting parameters. Bayesian updating carries no variable for what a Hypothesis is about or for how a prior entered, so two Hypotheses with equal priors and equal likelihoods receive equal posteriors whatever they assert (Axiom 8, Chapter 4, Section 4.3.8). At the scale of a population, the velocity of a cascade, its convergence and the stability of the state it converges to are set by how strongly private and social Evidence are weighed and by the environmental parameters, and none of these is indexed on which group fills the scapegoat slot (Chapter 13, Section 13.6.4; Appendix A, Section A.5.8.6). A crisis therefore recurs wherever the structural conditions that produced it persist, whatever Content the next episode carries, and a barrier built against one Content is routed around, because the space of scapegoat Hypotheses is unbounded and any list of protected groups is finite (Chapter 13, Sections 13.6.6 and 13.6.7). Content changes the output by changing a parameter. A Hypothesis that enters a Hypothesis Space, a vocabulary that opens an Intrinsic domain, and a rule published with its Referent each change a parameter the dynamics run on (the Content-Channel Asymmetry, Chapter 4, Section 4.3.11; Chapter 16, Section 16.11.2), which is why the Knowledge Patch, supplied as vocabulary, acts as a structural intervention. Truth reaches the output through a parameter too, the contact of a live Referent. Where that contact holds, the Evidence reaching each Individual is sampled from the world and convergence tracks Truth; the trajectory of a cascade is independent of its Content where Decoherence Time is long relative to cascade velocity, so that expressed Belief rather than a Referent is the dominant Evidence (Chapter 13, Section 13.8.6; Appendix A, Section A.5.8.4).

Intent bears on the output through the same parameters. The Decision Hub's integration discards the record of which sources produced a valuation, and no later stage recovers it (Chapter 4, Section 4.3.3). An Individual's account of why an act was aimed where it was is therefore a Confabulation however sincerely given, and an observer's inference about it is a Projection, capped by the observer's Belief about that Individual (Chapter 16, Section 16.6). What an Individual says of intent and what an observer infers of it are Content. What reaches the outcome is the conduct and the conditions that produced it. The interventions on a cascade work identically whether the leader who feeds it is sincere or cynical (Section 17.4), and the moral asymmetry between a sincere leader and a cynical one has no consequence for the violence each produces (Section 17.14.1). Malice is the uncommon case, and most harmful conflict is the collision of sincere Confabulations (the Non-Malice Corollary, Chapter 11, Section 11.21). The mental element grades culpability, and culpability does work only for punishment (Chapter 16, Section 16.8). What prevention needs to know about the Individual who acted is carried by the upstream conditions and the pattern of that Individual's conduct, and what is called intent enters the assessment of risk only as it is inferred from the same conduct (Chapter 16, Sections 16.8 and 16.12).

Harm is fixed by what an act does to the motivational system of the Individual who receives the act (Chapter 2), and the repair of the Trauma is the same whatever the offender intended (Chapter 16, Section 16.12). What the Individual harmed takes the act to have meant is part of what that Individual receives. The reading is that Individual's own estimate of the offender, built from the offender's conduct and entered in that Individual's Belief Register as Evidence like any other (Chapter 16, Sections 16.6 and 16.9), and it bears on the Trauma through the architecture of the Individual harmed. Given the conduct the Individual harmed receives and reads, the Harm is independent of the offender's actual state of mind.

Three named results are instances of the Structural Determination Principle. The Structural Output Principle (Chapter 16, Section 16.6) is the instance for criminal proceedings. The Trigger-Indifference Corollary (Section 17.7.10) is the instance at the apex, where the event that sets off the discharge of the terminal configuration acts on the outcome only through whether the leader still holds authority. The Self-Defence Orthogonality Principle (Section 17.20.1) is the instance for aggression and violence at every scale, where the self-defence Belief held on every side cancels and every response is selected by conduct and need.

Formal treatment: Appendix A, Section A.5.8.6.

17.15 The Leadership Asymmetry – Self-Defence, Duty, and the Arithmetic of Lives

Individual bodily self-defence is architecturally permitted, and no framework should deny it. When an Individual is under genuine A-threat (physical attack), the survival system fires appropriately, and defensive violence is A-grounded in the most direct sense available. The defender's Belief about the attacker is a Grievance Belief, and an accurate one; what the architecture denies is not its accuracy but any structural difference between an accurate Grievance Belief and an M-driven one at the point of action. The chapter’s analysis does not contest this. What the architecture reveals is that Phenomenological Collapse makes genuine A-grounded self-defence indistinguishable from M-driven violence at the point of action. Both arrive at the Decision Hub formatted as survival-level alarm. Both produce defensive phenomenology. Neither the Individual under genuine bodily threat nor the Individual whose M-amplified Status anxiety has been integrated as A-threat can tell the difference, because Source Opacity has destroyed the source-tags that would permit the distinction.

Once reciprocal violence is underway, the participant-level dynamics are as Section 17.13 established: both sides are genuinely under A-threat, neither can relent, and the architecture provides no leverage point at the participant level.

At the leadership level, however, the calculus is entirely different. The leader is not under immediate bodily threat. The leader is making a decision about whether to deploy others into violence. Engaging in war will, with certainty, result in loss of life. This is not moral philosophy. It is arithmetic. The number of deaths produced by a war is always greater than zero. The number of war deaths produced by not fighting a war is always zero. Not fighting does not make every death impossible, because an aggressor can go on killing a population that has declined to fight (Sections 17.15.5 and 17.20.6), but those deaths are conduct every observer can see, are not certain, and remain open to every intervention Section 17.20 derives. The leader who orders war “to protect lives” is therefore exchanging deaths that are contingent and open to intervention for deaths that are certain.

This feels morally reprehensible, and the feeling is itself architecturally diagnostic. The objection arises immediately: “But they are attacking us! We must defend our people!” The objection feels like A-urgency because Phenomenological Collapse has formatted it as survival-level alarm. But the leader’s decision is not about whether Individual soldiers should protect their bodies from incoming fire. It is about whether to create the conditions in which thousands of Individuals’ bodies are exposed to incoming fire. The leader is not defending their own body. The leader is deciding whether other people’s bodies will be destroyed.

The duty-constituted Self-Model (Section 17.13, Components 1-3) drives the conflation: “protector” as self-concept prevents the question from arising as a live option, and separates the self-concept (“I am a good leader”) from the Content of the behaviour (ordering thousands to their deaths).

The architecture systematically conflates two evaluation frames:

• The moral frame evaluates the leader against the duty-constituted Self-Model and its reference group’s evaluative criteria. The moral leader fights, because duty demands it, because “protecting our people” requires force, because the reference group would punish failure to act. The moral frame is M-grounded. It asks: does this leader fulfil the role? The evaluative criteria are constituted by the reference group’s expectations, which are themselves shaped by the same Confabulation Cascade that produced the conflict. Morality, in this frame, is the architecture evaluating its own outputs by its own standards.
• The ethical frame evaluates the decision against A-harm minimisation measured across all affected Individuals, not merely those within the leader’s reference group. The ethical leader does not fight, because the deaths that fighting produces are certain, and the deaths that declining risks are neither certain nor beyond the reach of intervention. The ethical leader absorbs the M-cost (their Self-Model is destroyed, they are judged a coward by their reference group, they lose Status, they may lose office) to prevent A-harm at population scale. Ethics, in this frame, is the evaluation of outcomes against the actual welfare of actual Individuals, independent of the evaluator’s Self-Model.

The two frames point in opposite directions because Phenomenological Collapse has made the M-grounded duty feel identical to A-necessity, when in fact pursuing the M-grounded duty (war) guarantees A-harm (deaths) that the alternative only risks. The moral path is the M-path. The ethical path is the A-path. The architecture ensures that leaders, populations, and post-war historians experience the moral path as the ethical one, because Confabulation generates the narrative that “we had no choice” and “more would have died if we had not fought,” a counterfactual that is by definition unverifiable and is generated by the same Confabulation system whose outputs the chapter has already established as architecturally unreliable.

The leader who chooses the ethical path (not fighting, accepting Status destruction, absorbing the M-cost) will be judged morally deficient by every conventional framework, by their reference group, and by history as written by the architecture’s Confabulation system. The leader who chooses the moral path (fighting, fulfilling duty, producing maximum death) will be judged heroic. The architecture rewards the path that maximises death and punishes the path that minimises it. This is not a paradox. It is the architecture operating exactly as derived: the M-grounded Self-Model licenses whatever behaviour the role specifies (Section 17.13, Component 3), and Confabulation fills in whatever justification the situation demands. The justification for war is always available, and where it is sincere it is experienced as ethical necessity, because the architecture cannot distinguish the M-urgency driving the decision from the A-harm it purports to prevent.

17.15.1 Why Conquest Is Recoverable and War Is Not

The objection to the ethical path arrives immediately: “If we do not fight, we are conquered. Our nation ceases to exist. This is catastrophe.” The objection contains the Phenomenological Collapse identified in other contexts, operating on the decision to decline war.

What is threatened by conquest? The nation’s sovereignty (M: Status of the nation-as-entity in the international hierarchy). The governance system (M: the institutional expression of the reference group’s evaluative criteria). The leader’s position (M: the Self-Model element whose destruction constitutes identity crisis). The cultural dominance of the population’s norms (M: the reference group’s capacity to set evaluative criteria for others). These are genuine harms. They are M-harms, experienced through Phenomenological Collapse as survival-level catastrophe. But they are not deaths.

The asymmetry is therefore:

• War certainly kills people. Every war produces deaths. The deaths are A-harm, irrecoverable. No post-war process can un-kill the dead.
• Declining to fight produces no war deaths. The harms of conquest (governance disruption, cultural subordination, Status loss) are M-harms and contingent A-I harms, and both categories are recoverable through structural intervention in a way that death is not. The deaths an aggressor can still inflict on a population that has declined to fight are contingent on the aggressor's conduct and open to intervention (Sections 17.15.5 and 17.20.4).

The fear of conquest is the architecture processing M-catastrophe (Status Death of the nation) as A-catastrophe (actual death of people), and the processing is so effective that the claim that declining to fight is worse than war feels self-evident. It is not self-evident. It is Confabulation. The Confabulation resolves the tension between the duty-constituted Self-Model (“I must protect my people”) and the arithmetic fact that fighting will certainly kill some of them and not fighting need not, by generating the narrative that “more would die under occupation than in the war,” a counterfactual that cannot be verified at the moment it is asserted, is generated by the same Confabulation system whose outputs the chapter has already established as unreliable, and is held with a certainty that no Evidence available to the leader can supply. Where an aggressor does go on killing a population that has stopped, the killing is conduct, and Section 17.20.4 answers it with the protection of the Individuals under attack rather than with war.

Consider the conquered nation. It has not fought. Its infrastructure is intact. Its population is alive. Its social networks are functional. Its I-domains (cultural practices, communal bonds, meaningful work, creative traditions) are operational. The conqueror must now govern this population. This is not the governance problem of a devastated, traumatised post-war society. This is the governance problem of an intact society that does not subscribe to the conqueror’s cultural norms and governance systems, that has not been “defeated” in any sense that would produce compliance through exhaustion or trauma, and that retains full capacity for non-cooperation.

The conquered population, supported by third-party Cascading Support delivered through channels the conqueror does not control (A-floor maintenance, I-domain protection, Cascading Consequences applied to the occupier; Section 17.15.6), retains what war destroys: living people, intact infrastructure, functional social networks, and operational I-domains. The conqueror must govern an intact society that does not subscribe to the conqueror's authority and is sustained by external support that the conqueror can interdict only by acts every observer can see.

This produces the question the architecture has prevented from being asked: if the conquered population has A-sufficiency and I-sufficiency, what does it matter, in A-I terms, who is nominally in charge? The architecture’s answer (“it matters because our nation, our sovereignty, our way of life”) is M-content processed through Phenomenological Collapse at collective scale. The AIM-aligned answer is different: governance authority is not the relevant variable; governance behaviour, measured in A-I terms, is. Non-cooperation, the turning of repression back on the regime that orders it, and Cascading Support delivered past the conqueror render the conqueror's position untenable, within the limit that Forced Quiescence sets (Sections 17.15.4 to 17.15.6).

The ethical path (refusal to reciprocate violence, followed by non-violent action and structural intervention) therefore produces: zero war deaths, a population intact except for what the aggressor's own conduct takes from it, recoverable M-harm, and a conqueror whose position becomes progressively untenable under sustained structural pressure. The moral path (war) produces: certain deaths numbering in the thousands or millions, devastated infrastructure, traumatised populations on both sides, post-war Confabulation preserving both sides’ Self-Models, and no structural diagnosis of what produced the violence. The architecture selects for the moral path and against the ethical path, because the architecture cannot distinguish M-catastrophe (Status Death) from A-catastrophe (actual death), and Confabulation generates whatever narrative is required to justify the M-preserving, A-destroying choice.

17.15.2 Content, Architecture, and the Flourishing Constraint

The Structural Determination Principle (Section 17.14.3) applies to the decision to decline war. Who the leader is, what the nation is called, which flag flies: these are Content-level variables processed by the same content-agnostic Bayesian machinery. The AIM framework’s concern is not Content but architecture: ensuring the structural conditions under which Individuals can flourish.

Content-level variables (who governs, what the nation is named, which traditions hold official status) matter only insofar as they affect A-I welfare. A governance change that maintains A-sufficiency and I-access is a Content change. A war that kills a hundred thousand people is an architectural catastrophe, because dead Individuals cannot flourish. They cannot be restored. They cannot benefit from subsequent governance improvements. They are permanently excluded from the domain of A-I welfare that the framework exists to protect.

Flourishing is only possible among living Individuals. This is not a moral claim. It is a logical constraint. The set of Individuals capable of A-sufficiency, I-engagement, and meaningful life is bounded by the set of Individuals who are alive. Every death produced by war removes one Individual permanently from the set. No Content-level outcome of the war (which side governs, which flag flies, which leader holds office) can compensate for this removal, because the removed Individual is not available to experience whatever Content-level outcome the war produces. The dead do not benefit from victory.

The architecture ensures this constraint is never applied, because Phenomenological Collapse converts Content-level M-harm (loss of sovereignty, loss of national Status, loss of cultural dominance) into felt A-catastrophe, and the duty-constituted Self-Model generates the narrative that fighting is necessary to “protect our people,” when fighting is the action whose deaths are certain. The architecture is optimised for M-preservation, not for A-I welfare. The ethical path requires overriding the architecture’s M-generated urgency in favour of the logical constraint that the architecture cannot represent: that Content is recoverable and death is not, and that the only Individuals who can flourish are the ones who are still alive.

17.15.3 Refusal Is Not Concession

Two different acts pass under the description of not fighting. Surrender concedes: the surrendering party accepts the other party's claim, hands over its cooperation, and lets the war stand as a contest the other party won. Refusal to reciprocate violence declines the war and concedes nothing: the claim is not accepted, cooperation is not given, and the contest has no result, because one party has declined to conduct it by violence. A party that stops a war already under way makes the same choice later. The ethical path of Section 17.15 is refusal, and never surrender.

Refusal is the harder of the two. Phenomenological Collapse formats the demand for reciprocity as survival-level alarm, so declining to return violence runs against what presents as the defence of the population, of its borders and of its Beliefs. The duty-constituted Self-Model reads refusal as cowardice, and the reference group reads it the same way (Section 17.13). Surrender requires no such override, because surrender gives the architecture the reading the architecture already expects.

Refusal is read as defeat unless it is seen to be refusal. Observers of a contest of standing update on its result (Section 17.3.5), and a refusal that no one can see presents to them as the loser's result. The refusal and its reasons are therefore made visible: to the aggressor, to the aggressor's population, to the refusing population, and to the parties that hold Cascading Support. Low-Theatre conduct does not forbid this. Low-Theatre limits disclosure to the parties for whom disclosure is necessary, and frames what is disclosed by the function it performs rather than by the standing of the parties (Chapter 2). Here disclosure is necessary to every party whose reading decides whether the refusal works. What Low-Theatre excludes is the staging of the refusal as a victory over the aggressor. Stephan and Chenoweth (2008) record the same condition from the side of the repressing regime: repression of a non-violent campaign turns against the regime that orders it most often "when an opponent's violence is not met with violent counterreprisals by the resistance campaign and when this is communicated to internal and external audiences".

Four marks distinguish refusal from concession, and every observer can check each of them.

1. Stated reasons. The refusing party says why it will not fight and what it does not accept. Concession says that the other party's claim is right. Refusal says that the refusing party will not fight and does not accept the claim. The difference lies in what is said, which is why the reasons have to be visible.
2. Withheld cooperation. Concession hands over compliance. Refusal withholds violence and compliance together, and continues as non-violent action (Section 17.15.4).
3. Non-recognition. Other parties decline to recognise what was taken: "No territorial acquisition resulting from the threat or use of force shall be recognized as legal" (United Nations, 1970).
4. Support and consequence. Under a Mutual-Support Commitment, Cascading Support reaches the refusing population and Cascading Consequences reach the aggressor (Section 17.20.5), and both show every observer how the field has read the refusal.

The architecture selects war because refusal registers as Status Death (Section 17.15.1), so the Mimetic term that drives the choice to fight reverses only where refusal carries Status. Refusal declines in one act both the framing the aggressor brings, that the encounter is a contest of standing to be settled by violence, and the aggressor's Belief about the claim the violence is meant to win, and so leaves the aggressor no contest to conduct. This is the Refusal Without Concession Principle: declining to reciprocate violence, and declining to continue it, is the highest-Status choice open to a party, and it is never to be read as a concession of the other party's Beliefs. Regard is owed to the Individuals and the populations who make the choice. The regard attaches to the act of refusing, and not to the refusing party's standing over the aggressor. A refusal celebrated as a victory over the aggressor re-enters the contest the refusal declined, and the aggressor's reference group reads the celebration as humiliation.

Refusal also lowers the aggressor's Status, by the one route that reaches the aggressor. Mimetic violence pays because observers update on the result, and the winner's standing rises (Section 17.3.5). Refusal denies the result where the four marks hold. Territory taken from a population that concedes is a result, and the aggressor's reference group reads it as one. Territory that is not recognised, cannot be governed and costs more to hold than it yields is not a result, and violence that wins nothing falls into the class whose expected Status outcome is strictly negative: nothing was gained, and the violence was observed. Where the purpose of the violence is Status, refusal defeats the purpose. The route has a limit, and the limit is also a diagnosis. Where violence is predation, the target is selected for the vulnerability that refusal displays (Chapter 10, Section 10.2.3), and refusal does not by itself stop the blow. By the Eliminative Structure of Crime (Chapter 16, Section 16.13), where Cascading Support holds the A-floors and I-floors of the populations on both sides (Sections 17.15.6 and 17.20.3), a motive to prey on a party that refuses has no Appetitive or Intrinsic source left, and the predation stands exposed as Mimetic: a status rival computed as prey (Chapter 10, Section 10.3.5). The Object of Mimetic predation is a comparative position, and the predation is denied that Object wherever refusal is read as refusal and not as concession. The protection of the Individuals under attack (Section 17.20.4) answers the act itself.

Lowering imposed from outside does not reach the same place. Status is a valuation held within the aggressor's own reference group, and denunciation from outside arrives there as the visible external hostility from which the Scapegoat Mechanism builds its agent-attribution (Section 17.19.4). Approval of a president rose around international events that involved the United States directly and were sharply focused (Mueller, 1970), and the humiliation imposed on Germany after 1918 became material for the stab-in-the-back Confabulation (Section 17.4). Lowering another is also the move that gains most standing for the effort (Appendix A, Section A.10.2), so a third party that humiliates the aggressor makes a positional gain at the aggressor's expense, and restores the contest that the refusal removed. The aggressor's Status falls where the aggressor's own reference group sees three things: that nothing was won, that the cost was carried (Section 17.19), and which party went on with violence after the other had stopped (Section 17.20.4).

17.15.4 Non-Violent Action Is Not Pacifism

Refusal is not abstention. Pacifism, taken as the renunciation of violence followed by standing down, leaves the contest with a visible result, and observers read that result as the aggressor's win, which is the reading Section 17.15.3 shows refusal must avoid. Non-violent action continues the contest over the claim by means that are not violence. A population that has declined to reciprocate violence can go on acting against the aggressor: refusing to staff the occupier's administration, striking, boycotting, protesting, publishing, coordinating, and keeping the life of its own institutions going. A population that does all of these at once is ungovernable, and governing it is what holding territory requires.

King set out the method in terms he took from Gandhi: "nonviolent resistance is not a method for cowards; it does resist", and it "does not seek to defeat or humiliate the opponent, but to win his friendship and understanding", with non-cooperation and boycotts serving as "means to awaken a sense of moral shame in the opponent" (King, 1958). Sharp (1990) counts the Indian campaigns against British rule in the first half of the twentieth century among the non-violent struggles waged against foreign occupation, and builds a policy of national defence on the same method, in which massive non-cooperation and defiance aim "to deny the attackers their objectives and to make impossible the consolidation of their rule".

The architecture gives three reasons the method works.

1. The ruler depends on the ruled. An order becomes an act only through the compliance of the Individuals who carry it out (Section 17.7), and the Status of a ruler is a valuation the ruled confer. Non-cooperation withholds both the material yield the occupier came for and the recognition. The occupier holds the territory and gains no Status from it.
2. Repression turns back on the regime. Violence against Individuals who do not fight produces no contest and no result, only conduct that the aggressor's own forces and population can see. Across major campaigns between 1900 and 2006, non-violent campaigns succeeded 53 per cent of the time and violent campaigns 26 per cent; defections from the security forces made non-violent campaigns "forty-six times more likely to succeed", and had no significant effect on violent campaigns (Stephan and Chenoweth, 2008). A defection is the aggressor's own reference group withdrawing the compliance that converts orders into acts, which is the route by which Section 17.15.3 derives that the aggressor's Status falls.
3. Non-violent action that does not humiliate supplies no external hostility. The aggressor's population is given nothing from which to build an agent-attribution (Section 17.19.4), and the moral shame King names is a judgment the aggressor's own reference group reaches rather than one imposed on it from outside.

Non-violent action is visible by design, and Low-Theatre conduct governs how non-violent action is framed rather than whether it is seen: the action is framed by what it refuses and what it asks, and not by which party came off better, which is the rule King states in declining to humiliate the opponent. What remains under Section 17.19.4 is the broadcast of the action to distant populations with no stake in it, and the parties that hold Cascading Support do not amplify the broadcast.

Non-violent action is not without cost. Regimes meet it with repression (Stephan and Chenoweth, 2008), and strikes and boycotts draw on the floors of the Individuals who conduct them, so non-cooperation lasts as long as those floors hold. The first case Sharp (1990) records of non-cooperation adopted as government policy against an invasion, the German resistance to the French and Belgian occupation of the Ruhr, ran from January to September 1923 and was financed by the German government. Holding the floors of a population that declines to cooperate is the work of Cascading Support (Section 17.15.6). Two further limits follow from the architecture. Defection requires a shared reference group, and an occupier's forces answer to their home society rather than to the population they hold, so the refusal and the occupier's conduct have to reach the occupier's home population (Section 17.15.3). And overt sponsorship of a non-violent campaign by a foreign state supplies the foreign hand from which an agent-attribution is built (Section 17.19.4): "Overt external state support for a campaign has no effect on the success of nonviolent campaigns", while for violent campaigns "it nearly triples their chances of success" (Stephan and Chenoweth, 2008). Cascading Support for a population engaged in non-violent action is the provision of floors and of independent information to the population, and never the sponsorship of the campaign.

17.15.5 When Quiescence Is Forced

Refusal followed by non-violent action does not guarantee that an occupation ends. An aggressor can force quiescence on an occupied population. Section 17.6.4 identifies an occupying power as an Anchoring Object, and the order it installs as Forced Quiescence with the suppression supplied from outside; Forced Quiescence conceals the dynamics it suppresses rather than resolving them (Section 17.6.1). The architecture permits this outcome, and nothing in the derivation removes it. What the derivation fixes is that the interventions do not change when Forced Quiescence is imposed: Cascading Consequences continue on the occupier, Cascading Support continues for the occupied, and the record of the refusal is kept.

Forced Quiescence becomes harder to impose the further support reaches. Forced Quiescence is maintained through five suppressed channels (Section 17.6.2), and two of them are the channels Cascading Support supplies: provision, which a regime under Forced Quiescence routes through loyalty so that refusal costs a family its floors, and the correction of Confabulation by independent information. Support that reaches the population by routes the occupier does not control keeps both channels independent, and so works against Forced Quiescence by the definition of Forced Quiescence rather than as pressure added from outside. The cost of holding territory also grows with the territory held (Section 17.20.6).

Where Forced Quiescence is imposed regardless, the support is not wasted. The violence of the transition that ends Forced Quiescence tracks how completely the occupier captured the institutions that could have managed the transition (the Succession Decoherence Principle), and the length of the disorder tracks the Buffer Independence that survived (the Discharge Duration Principle; Section 17.6.5). Support held through Forced Quiescence shapes the end that the support could not prevent.

Nothing in the architecture requires the parties that hold Cascading Support and Cascading Consequences to stop, and nothing guarantees that they continue. Their attention is the same scarce quantity that a war captures as its oxygen (Section 17.20.2), and a distant occupation competes for that attention with every other event. A Mutual-Support Commitment makes continuation a standing obligation rather than a decision renewed each year (Section 17.20.5).

The Baltic states show both limbs. Estonia, Latvia and Lithuania were occupied in June 1940 after their governments yielded to Soviet ultimatums, so that no military conflict occurred, and were annexed in the same year (Mälksoo, 2003). On 23 July 1940 the United States declared that its people "are opposed to predatory activities no matter whether they are carried on by the use of force or by the threat of force" (Welles, 1940). A significant part of the international community refused to recognise the annexation, and the occupation never acquired legal title in the half century before independence was restored in 1991 (Mälksoo, 2003). Forced Quiescence held for half a century, and its cost was borne by Individuals throughout. Independence was restored through mass non-violent action, the Singing Revolution (Šmidchens, 2014), as the Soviet system reached its own fold. The case does not show that non-cooperation alone ends an occupation. It shows a refusal kept on the record by outside parties for fifty years, an annexation that was never made lawful in that time, and an end reached without a war.

The direction of each intervention is fixed. The time an occupation takes to end under them, and the depth of repression at which a population can no longer sustain non-cooperation, are not (Section 17.20.8). A population may have to wait fifty years or longer for freedom from its oppressor, as the Baltic populations did, and the derivation promises no shorter wait.

That conclusion does not satisfy the Individuals who bear the wait, and it is not measured against their satisfaction. Oppression, however long, falls on Individuals who are alive, and what a living population has lost remains recoverable (Section 17.15.1). As long as Individuals are alive, the species is alive, and a living species can still correct what was done to it. The alternative to waiting is to fight. Fighting raises the Network Loop Gain of the network that carries the war (Appendix A, Section A.5.7.4; Section 17.20.4), and a rising Network Loop Gain moves that network toward the terminal fold, from which recovery is a restart and not a correction (Section 17.17.5). A course that tries to shorten the wait by raising the Network Loop Gain spends what cannot be recovered to buy what can. The species is the measure against which every intervention is held.

17.15.6 Support That Does Not Run Through the Regime

Cascading Support presupposes a channel. Provision delivered through the regime that holds a population becomes provision through loyalty (Section 17.6.2): the regime takes the credit and allocates the provision by compliance, and the support strengthens the Forced Quiescence it was meant to weaken. Cascading Support therefore reaches Individuals through channels the generating regime does not control, and where no such channel exists, building one is part of the support.

The channels run from the widest to the narrowest: negotiated corridors; routes across borders and conflict lines with notification to the regime rather than its consent, the form the Security Council authorised for Syria in 2014, with a monitoring mechanism "to confirm the humanitarian nature of these relief consignments" (United Nations, 2014); transfers made directly to Individuals; broadcast for information; and, where nothing else reaches, delivery from the air. Receipt can be punished, so delivery is diffuse and hard to trace to the Individual who receives it.

A channel is never fought open. A channel that needs force to open is entry into the war (Section 17.20.4). Delivery is unarmed, declared in advance and open to inspection, so that stopping it requires an act every observer can see, which is the conduct the Self-Defence Orthogonality Principle selects interventions by (Section 17.20.1). The law of occupation already carries both the obligation and the inspection: an occupying power "shall agree to relief schemes on behalf of the said population, and shall facilitate them by all the means at its disposal", and a power that grants passage to the consignments may search them (Geneva Convention IV, 1949, art. 59). A refusal of relief is then conduct on the record.

A party that feeds a population and asks nothing in return is hard to cast as an aggressor. The agent-attribution needs visible external hostility (Section 17.19.4), and unconditional provision is Evidence against it, so a regime that needs its population to read the provider as an enemy has to Confabulate, and every delivery tests the Confabulation against what the population eats. The provision has to ask nothing. Provision tied to compliance or allegiance is leverage, which the regime can name as bribery. Information that travels with the provision gives the providers' reasons, and neither denounces the regime nor asks the population to rise, because a call to rise makes receipt a political act the regime can punish and supplies the foreign hand of Section 17.15.4. Provision is maintained as ordinary practice rather than announced as clemency (Section 17.19.4).

The Berlin Airlift of 1948 and 1949 met these conditions. On 24 June 1948 the Soviet authorities severed the land and water routes to the western sectors of Berlin, and the air forces of the United States and Britain supplied the population by air until the blockade was lifted in May 1949. The transports were unarmed and flew corridors fixed by a written agreement of 1945, so stopping the airlift would have required shooting the aircraft down at the risk of a shooting war, a step the Soviet authorities avoided (Miller, 2000). The airlift resolved none of the issues behind the blockade. It bought the time in which they could be negotiated without either yielding or escalating (Miller, 2000), which is the clutch held open at the scale of a city. The population it fed had been at war with the powers that fed it three years earlier, and that population's determination to remain free of Soviet control was among the conditions of the airlift's success (Miller, 2000). One feature of the case falls outside the design: fighter escort was briefly ordered for the unarmed transports in April 1948, after a Soviet fighter collided with a British airliner (Miller, 2000).

Where a siege leaves no channel at all, the constraint cannot be met. What remains is the refusal of relief, recorded as a breach of an obligation the besieging power already carries, and Cascading Consequences on the party that refuses it.

17.16 The Truth-Exclusion Principle

Section 17.11 established that the Confabulation Cascade is Content-agnostic. Combined with the Structural Determination Principle (Section 17.14.3), this entails a consequence: violence has never uncovered Truth. In the entire history of human conflict, not once.

The derivation is short. The architectural pathway generating violence (Section 17.1) is Content-independent: it operates identically regardless of which Beliefs are converging. Truth (the actual causal structure that produced the experienced suffering) is Content. The pathway contains no mechanism that evaluates whether Beliefs correspond to reality.

Within the conflict horizon, dynamics are identical regardless of which party holds True Beliefs. This is Truth-Value Orthogonality: an entailment of Axiom 8 (Content-agnostic Bayesian updating) applied to the conflict domain specifically. Both aggressors and defenders experience themselves as defenders (Section 17.9). Both populations undergo the same Confabulation Cascade dynamics. Both update toward increased certainty in their own position. The party whose Beliefs happen to be more accurate possesses no internal advantage in the updating process, because Bayesian updating on endogenous evidence is Content-agnostic: it operates identically on true Hypotheses and false Hypotheses when the evidence base is socially generated rather than reality-generated. The correct party’s Beliefs do not converge faster, do not produce more cohesive coordination, do not generate stronger conviction, and do not announce themselves in any form that obviously contrasts against a false Belief. The incorrect party’s Beliefs do not converge more slowly, do not produce weaker coordination, do not generate less conviction, and possess no identifiable quality that clearly marks the Belief as false. The cascade dynamics are source-blind in this precise sense: the Truth or falsity of the Hypothesis being amplified makes no difference to the amplification rate.

The outcome of violence (who prevails) is determined by capability: military resources, strategic competence, technological advantage, geographic position, alliance structures, demographic weight. None of these variables correlate with the Truth of the causal attributions that produced the conflict. The side with more accurate Beliefs about why the war started does not thereby gain a military advantage. The side with superior firepower does not thereby hold more accurate Beliefs. Capability and Truth are orthogonal variables. Victory selects for capability, not for Truth.

Post-war attribution (why the war occurred, what it meant, why the victor prevailed) is itself Content, generated by the same Confabulation architecture, subject to the same Bayesian Lock-In. The Content Mutation Imperative, which follows from the Content-Independence of Violence (Section 17.11), predicts that populations learn the Content of the previous crisis but not its structure. After WWI: “Don’t let Germany rearm.” After WWII: “Don’t let fascism rise again.” After Rwanda: “Don’t let ethnic hatred go unchecked in Central Africa.” Each lesson is Content-bound. The structure (coupled Belief dynamics converging toward certainty without tracking Truth, producing Scapegoat designation and coordinated violence) remains invisible because it has never been formally articulated in the population’s conceptual vocabulary. The wars recur with novel Content because Content must mutate while structure remains invariant. Indeed the Content of a war is structurally impossible to replicate. Hitler can never invade Poland again. Rome can never conquer Gaul again. The exact conditions that gave rise to the tragedy of Gallipoli can never occur again. The Content of each war will always be unique. When people say “history never repeats but it rhymes” no one ever stops to analyse what rhyming is. Rhyme is structure - stressed vowels, alliteration, meter, repetition. The Content of the rhyme is bound by the structure.

The implication is universal: no war in human history has produced Truth as an output of the violence itself. Violence determines which side prevails. It does not determine which side’s Beliefs correspond to reality (Truth-Value Orthogonality). Where Content-level facts were established in the aftermath of a conflict, they were established by investigative processes whose existence was causally dependent on the war having occurred but whose epistemic methods are entirely distinct from the violence process. The Nuremberg Trials are the canonical illustration, not of this obvious distinction, but of something deeper: the architectural constraint that determines what even post-war investigation is interested in, and what it is not.

What Nuremberg was interested in was Content. Which leaders decided what. Which orders were given. Which atrocities occurred. Who was responsible. Who should be punished. These are Grievance Beliefs applied post-hoc, the Scapegoat Mechanism’s logic operating through institutional rather than mob channels. Nuremberg was a juridical exercise in assigning responsibility for collectively-experienced catastrophe to identifiable individuals. It was entirely disinterested in the structural conditions that gave rise to the war, why populations locked into sincere conviction of defensive necessity on all sides simultaneously, or why this produced the violence whose human costs the tribunal was now adjudicating. The tribunal found Scapegoats for the war’s human costs. It did not, and could not, diagnose the architectural failure modes whose mismanagement produced those costs.

The depth of this constraint is revealed by what Nuremberg declined to revisit: structural questions that had not only been available in ordinary language but had actually been asked, publicly and influentially, during the interwar period. In 1919, John Maynard Keynes resigned from the British Treasury delegation at the Paris Peace Conference and published The Economic Consequences of the Peace, arguing that the punitive terms of Versailles would crush the German economy and create the conditions for future catastrophe. He predicted the next war would begin approximately twenty years later. The book became a bestseller, and his analysis became academic orthodoxy in Britain and America. Keynes asked, in plain economic language, precisely the questions that Nuremberg declined to revisit: What conditions produce war? Could different intervention (ensuring populations retain adequate access to material necessities, avoiding punitive settlement terms that generate unbearable collective suffering with no adequate causal explanation) have prevented the catastrophe? Would structural remedy have been more effective than punitive attribution?

These questions required no novel framework. They had been asked. They had been answered. The answers had been correct in their predictive content. And they were ignored, because the institutional demand after catastrophe is for responsible agents, not systemic conditions. Nuremberg operated under the same architectural constraint it was adjudicating: the Agent-Based Causal Default completed the causal chain with identifiable individuals who could be punished, because that is what the cognitive system demands, and systemic explanation does not satisfy that demand. The human costs that Nuremberg attributed to specific leaders (being only the lower level ones who survived the war) were in substantial part produced by structural conditions and the absence of any institutional mechanism for managing their consequences. No single individual created the economic devastation of Weimar Germany, no individual designed the psychological conditions under which populations converge on false causal attribution, no individual chose the mechanism by which collective suffering produces demand for identifiable enemies rather than systemic diagnosis. The individuals Nuremberg convicted operated within conditions they did not create and dynamics they did not understand.

Nuremberg’s structural silence was not mere oversight. It was the Agent-Based Causal Default operating in juridical form, the Scapegoat Mechanism channelled through institutional procedure rather than mob violence, and it demonstrates the depth of the architectural constraint: even when correct structural analysis existed, was publicly available, and had been empirically vindicated by events, the post-catastrophe institutional response defaulted to agent-based attribution and punishment. What remained genuinely beyond reach, even given Keynes’s structural analysis, was the deeper diagnosis: why agent-based attribution recurs across centuries with novel Content, why the Confabulation Cascade converges on Scapegoat designation regardless of which population is targeted, and why violence cannot resolve the deficit that produced it. Keynes could diagnose the economic conditions; he could not diagnose why the Confabulation Cascade converges, because that diagnosis is blocked by the Camouflage Architecture (Layers 4-6 blocking academic investigation of the mechanism, Layer 7 preventing formulation of the Hypothesis that would explain the structural recurrence), and the post-war process is not positioned to supply it.

This dissolves the oldest and most persistent Confabulation about violence: that it reveals something. “The crucible of war reveals the Truth.” “History is written by the victors” (implying that victory confers epistemic authority). “Might makes right” (the cynical version of the same structural confusion). “Trial by combat” (the medieval formalisation). Each of these is a Confabulated narrative generated by the same architecture that produced the violence. The architecture produces violence; the architecture produces a narrative explaining the violence; the narrative attributes truth-revealing properties to the violence; and the attribution is itself Confabulated. The Confabulation is doubly recursive: it first generates the false causal attribution that produces violence (Scapegoat designation), then generates a false meta-attribution that the violence itself validates Truth. Neither attribution tracks reality. Both are outputs of the same Content-agnostic machinery.

Formal statement. For any conflict C: (i) the violence itself produced no Truth, because the generating process is Content-agnostic and Truth is Content; (ii) any Content-level facts established in temporal proximity were established by investigative processes independent of the violence; (iii) the structural diagnosis of the architectural pathway is not among the facts the conflict established, because the Camouflage Architecture (Layers 4-7) blocks formulation of the requisite Hypothesis for any party operating inside the conflict.

The principle’s universality follows from the architecture, not from induction over particular conflicts. The premises from which it is derived (Common-Currency Integration, Source Opacity, Bayesian updating) are themselves empirically grounded in neuroscience and information theory. The historical record confirms what the architecture predicts: across radically different epochs, cultures, ideological contents, and scales of conflict, the structural pattern is invariant. This is not the empirical generalisation of the historian or the conflict scholar, who derives claims from the archive of political decisions and strategic assessments. It is confirmation of an architectural prediction against a different empirical archive: the motivational, neuroscientific, and behavioural record that documents how the Decision Hub actually processes urgency, constructs attribution, and generates the phenomenology of defensive necessity. The universality of the principle is architectural. Its credibility is empirical.

The principle’s universality is architectural, not inductive. It is not that violence has usually failed to uncover Truth and might succeed next time. The generating process is Content-agnostic and Truth is Content. They are architecturally incommensurable. No modification of circumstances, weapons technology, or strategic doctrine can bridge this gap. What people learned after wars, they learned through other means. What they learned was Content. Structural Truth is not among the things the post-war process recovers. The violence itself produced no Truth. It never has. It never will.

17.17 The Race Condition

Since 6 August 1945, humanity has been in a race condition. On that date, and again three days later, the architecture described throughout this chapter discharged its output through a weapon capable of destroying civilisation itself. The cognitive mechanisms that produced those decisions were identical to the mechanisms that produced every prior instance of collective violence: Source Opacity ensured that M-urgency presented as survival-level threat, Confabulation generated agent-based justification experienced as defensive necessity, and the Phenomenological Collapse ensured that the decision to incinerate two cities felt like the only rational course of action available to leaders who sincerely believed they were protecting their people.

Nothing about the architecture has changed since. The Transmission Window still closes at 340 milliseconds, integrating M-urgency into the Decision Hub as A-threat before conscious awareness arrives. Source Opacity still destroys source-tags irreversibly. Confabulation still generates sincere but inaccurate causal narratives. The Bayesian cascade still produces convergent false Belief through valid reasoning applied to invalid, endogenously generated evidence. The architecture that dropped those bombs continues to operate under the same conditions it always has: absent any understanding of its own motivational structure, absent any institutional design calibrated to its failure modes, absent the vocabulary that would allow the Individuals operating it to recognise what is happening inside their own Decision Hubs.

The race condition is this: without understanding the architecture described in this treatise and designing institutions and interventions to maintain humanity’s ongoing existence, the species will eventually drift toward its own destruction. The drift is not hypothetical. It is architectural. The Unidirectional PM Flow Principle (Section 17.17.2 below) establishes that A-I Buffer erodes continuously through normal social operation, which means that the damping factors preventing cascade-to-violence are being consumed at all times, whether or not anyone notices. The architecture does not require a crisis to move toward crisis. It moves toward crisis by default. Avoiding crisis requires sustained, deliberate, structurally embedded effort that the architecture itself makes difficult to recognise as necessary.

The nuclear arsenal is the self-destruct mechanism that the architecture installed. It can only be circumvented by a firmware patch: the understanding of human motivational architecture that this treatise provides, and the institutional interventions that understanding makes possible.

The conclusion that follows is deductive, not speculative. It is not a forecast, not a scenario, not a probabilistic estimate of what might occur. It is a logical consequence of premises already established, each drawn from replicated empirical scholarship or proved via mathematical entailment in preceding chapters.

The entailment chain (Derivation Preview; demonstrated in Section 17.1) is entirely deductive. Each step follows from the one before. No step depends on empirical forecasting. The historical record confirms the derivation: every case analysed in this chapter instantiates the same architectural pathway with different Content.

What has changed is not the architecture but the environment. The architecture now operates in a world containing approximately 12,200 nuclear warheads, delivery systems capable of reaching any point on the planet within 30 minutes, and launch-authority structures that concentrate terminal decisions in Individuals whose Decision Hubs are subject to the same Source Opacity, Confabulation, and Phenomenological Collapse as every other human being.

Rejecting the conclusion requires rejecting a premise or faulting a step. If no premise is false and no step fails, The appropriate response is not scepticism about the conclusion but urgency about the intervention.

17.17.1 The Nuclear Threat

Consider radiation. It is invisible to every human sense. You cannot see it, hear it, smell it, taste it, or feel it. Yet it will kill you. Humanity lived alongside radiation for millennia without knowing it existed, because nothing in the human perceptual architecture could detect it. We learned of its existence only through its effects: illness patterns that defied existing causal models, photographic plates that fogged without exposure to light, workers who sickened and died in conditions that appeared safe to every available sense. Once understood, we built instruments to detect it (Geiger counters, dosimeters), protocols to manage exposure (shielding, distance rules, time limits), and institutions to enforce those protocols. We did not eliminate radiation. We cannot. It is a feature of the physical world. We learned to see what our senses could not show us, and we managed what we could not remove.

Unmanaged Mimetic Desire is the radiation of the social world. It is phenomenologically invisible to the Individual experiencing it. Source Opacity destroys the source-tags that would identify M-driven urgency as mimetic rather than appetitive or intrinsic. Confabulation generates plausible narratives that conceal M’s operation. The Camouflage Architecture’s eight layers ensure that this invisibility persists not merely at the individual level but at the institutional and civilisational levels. No available sense, no introspective effort, no unaided act of reflection can detect it.

Yet we see its effects everywhere. Rivalry that escalates without satiation. Scapegoat designation that recurs with novel Content but invariant structure across every culture and epoch. Confabulation Cascades that produce collective certainty in false causation. Violence experienced as defensive necessity by perpetrators on all sides simultaneously. The effects have been documented exhaustively across every discipline for centuries, just as radiation’s effects were documented in illness and death long before anyone identified the cause.

And it will kill us. Not metaphorically. The architecture that renders M invisible is the architecture that has produced every instance of collective violence analysed in this chapter. Under nuclear conditions, the race described in this section is between the Knowledge Patch (the Geiger counter for mimetic dynamics) and the Terminal Phase (the unshielded exposure that civilisation cannot survive). The recursion is exact: an invisible social force, left unmanaged because it is invisible, constructed the weapons whose lethality derives from an invisible physical force, and may deploy those weapons through the same architectural pathway (Phenomenological Collapse, Confabulation, Bayesian Lock-In) that has produced every prior instance of civilisational-scale violence, except that this time the discharge is terminal.

The Geiger counter was not a cure for radiation. It was a detection instrument that made management possible. The Knowledge Patch is not a cure for Mimetic Desire. The Patch is a detection instrument that makes Subcriticality Maintenance possible. Neither eliminates the force. Both make the force visible. Both enable protocols that keep exposure within survivable limits. And both are useless if they arrive after the lethal dose has been delivered.

Sufficiently disastrous nuclear war could destroy the planet. The combined arsenals of existing nuclear powers are sufficient to produce immediate deaths in the hundreds of millions, destruction of infrastructure necessary for organised human existence, nuclear winter effects capable of collapsing global agriculture for years or decades, and radioactive contamination rendering large portions of the Earth’s surface uninhabitable. The question is not whether nuclear war would be catastrophic. It is whether anything recognisable as human civilisation, or potentially anything recognisable as a habitable biosphere, would remain afterward.

Prior to nuclear weapons, the race condition’s consequences were cyclical. Damping factors eroded, cascade dynamics intensified, violence discharged the accumulated tension, populations survived (at enormous cost), damping factors partially rebuilt, and the cycle recommenced. The Roman crisis of the third century, the Thirty Years’ War, the two World Wars of the twentieth century: each killed millions, each permitted recovery. Recovery was slow, painful, and incomplete (the cultural and human losses were irreplaceable), but the species retained the capacity to rebuild.

The cyclical pattern has the structure this chapter has described throughout, now read across generations rather than within a single crisis. The field of coupled Self-Models behaves as a fabric under tension: each node holds its configuration only under the pull of those it is coupled to, load is transmitted node to node, and the structure holds only while that transmitted load stays below the threshold at which the couplings run away: the supercritical threshold whose minimal form is derived in Appendix A, where the critical value is exactly unity. When a society crosses that threshold, the fold does not return it to where it was; the configuration that crossed does not recohere. What follows is not the same society oscillating back into shape but a different one, woven from the survivors on the same ground. The Individuals who lived through the collapse carry it forward as altered priors in their Self-Models, and the behaviour those priors generate is the perturbation they transmit to the generation they constitute. This is why the pre-nuclear cycle could recommence at all: the tear was terminal for the configuration, not for the weaving. After every prior fold, the weaving resumed.

Nuclear weapons remove the one condition on which that resumption depends. The reweaving holds only while there are survivors to update and reweave. Every collapse the historical record contains has left a next row, diminished, deformed, carrying the perturbation forward, but weaving. A discharge severe enough to leave no next row ends not a configuration but the weaving itself, and the architecture that has always rebuilt after a fold has nothing left to rebuild from. This is the precise sense in which the nuclear arsenal converts the cyclical race condition into a terminal one: it is the first form of architectural output capable of taking the loom rather than the cloth. The objective of the interventions that follow is therefore not to prevent tearing, which the architecture does not permit, but to keep every discharge below the magnitude at which it takes the loom, because there is no weaver outside the cloth, and the cloth reweaves itself, or, it destroys itself and the loom all at once.

Nuclear weapons transform the race condition from cyclical to terminal. Three parameters distinguish nuclear violence from all prior forms of architectural output:

Decision timeline compression. The interval between decision and irreversible consequence compresses from months or years (conventional war mobilisation) to minutes or hours (nuclear launch). An M-Captured leader experiencing acute mimetic shock (humiliation, Status threat, perceived existential challenge to their position) can produce planetary consequences before the shock subsides, before advisors can intervene, before the leader’s own deliberative capacities can engage. The Transmission Window (60–340 milliseconds, Axiom 7) means that Phenomenological Collapse has already occurred before conscious awareness arrives at 300–500 milliseconds (best-current empirical estimate; the architecturally load-bearing claim is that τ_M is substantially shorter than conscious-report latency, Appendix A.6.2). The nuclear launch timeline is long enough for the M-weighted valuation to translate into action but short enough to foreclose the slower corrective processes that might interrupt conventional escalation. The leader does not experience “I am responding to a Status challenge.” The leader experiences “We face an existential threat requiring immediate action.”

Consequence magnitude explosion. The maximum consequence of leadership M-Capture shifts from regional catastrophe with eventual recovery to potential planetary destruction. The race condition’s damping factors have historically limited the output of cascade-to-violence: even when the Confabulation Cascade reached the Terminal Phase, the resulting violence destroyed some fraction of the population and infrastructure while leaving enough intact for reconstruction. Nuclear exchange at sufficient scale eliminates this limitation. The violence does not destroy a fraction of civilisation. It can destroy civilisation’s capacity to exist. It can destroy the biosphere’s capacity to support complex life.

Irreversibility. Nuclear exchange cannot be undone. Conventional war permits negotiation, ceasefire, and reconstruction. The feedback loops that might arrest conventional conflict (mounting costs producing war-weariness, battlefield realities forcing reassessment, third-party mediation) operate across timescales of weeks to years. These feedback loops cannot operate before irreversible thresholds are crossed in a nuclear exchange. Once missiles launch, no subsequent M-recognition, no dawning awareness of Mimetic Capture, no application of the Two-Question Diagnostic, no Epistemic Suspension, no change of heart can recall them. The Therapeutic Window that might open after conventional M-shock does not open after nuclear exchange, because there may be no surviving population in which doubt could be elevated.

The interaction between nuclear capacity and the Leadership Compression Principle (Section 17.7) produces the existential configuration. Leadership Compression ensures that an M-Captured leader can impose crisis dynamics regardless of where the collective posterior sits. Nuclear capacity ensures that the imposed crisis can terminate not merely a civilisation but a planet, regardless of any subsequent correction. The two compressions multiply. Consider the temporal architecture concretely. The Transmission Window closes at approximately 340 milliseconds; M-urgency is already integrated as survival-level alarm. By 500 milliseconds, conscious awareness arrives, but the valuation is set and Source Opacity has destroyed the source-tag. Confabulation engages within seconds: “They have left us no choice,” “This is defensive necessity,” “Civilisation itself is at stake.” Minutes later, the launch order. Hours later, the missiles arrive. The entire sequence from preconscious M-integration to planetary consequence operates within a timeframe shorter than a single news cycle, shorter than a cabinet meeting, shorter than the deliberative processes that every governance framework since Aristotle has assumed would intervene between impulse and irreversible action.

Architectural cost and the Humanisation Null Hypothesis. The Leadership Compression Principle and Phenomenological Collapse gain an additional architectural explanation when read alongside the Humanisation Null Hypothesis (HNH, Chapter 11). Compressed decision timelines do not need to overcome a humanisation barrier in the leader's Decision Hub, because no such barrier is architecturally present. What the compression window must do is prevent reclassification of the target population back toward Collaborator or Rival during the interval between decision and irreversible consequence. This is a structurally easier task than defeating a humanisation operation would be, and it explains why nuclear and similar high-compression decision architectures produce their characteristic failure modes so reliably. The target of institutional design is therefore not the protection of a humanisation layer (which does not exist) but the extension of decision timelines and the preservation of conditions under which Collaborator or Rival classifications can be maintained against propaganda-driven reclassification toward Predator or Prey, because the cooperation-reciprocity and contest-reciprocity moral profiles carry stronger protections against the specific kinds of violence high-compression decision architectures are most likely to produce than the threat-response and extraction-ethics profiles do.

17.17.2 The Parameters of the Race Condition

The race condition’s parameters determine how quickly the system approaches criticality. The apex half of the condition, the conjunction of leadership Belief above its action threshold with authority above the gate, is stated formally as the Dual-Locus Ignition Theorem at Appendix A, A.5.8.7, is derived at Section 17.7.8, and pairs with the floor condition that is the subject of Section 17.5.

Damping factors have historically prevented continuous cascade-to-violence:

• A-Floor security. When basic needs are met, M-urgency is less likely to reach the threshold where Phenomenological Collapse produces survival-level alarm.
• I-domain refuge. Engagement with Intrinsic Motivation provides alternative value-sources that reduce M-Salience. Populations with robust I-infrastructure (meaningful work, creative engagement, genuine community) are less susceptible to cascade dynamics.
• Institutional friction. Effective institutions create delays between Belief-convergence and coordinated violence. Democratic process, legal constraint, separation of powers: these do not prevent cascade but slow its translation into action, creating windows for correction.
• External target availability. Chapter 10 established that external threats redirect mimetic targeting outward, producing solidarity rather than internal violence.
• Information diversity. When populations access diverse information sources, cascade-synchronisation is impeded. Disagreement about who is responsible prevents convergence on any single scapegoat.

These damping factors act as independent multiplicative friction on cascade dynamics: the Multiplicative Friction Model, the Belief Heterogeneity Buffer, and the Friction Compounding Effect are defined in Chapter 2 and carried formally at Appendix A.5.7.

Damping factor erosion is continuous, not episodic. The Unidirectional PM Flow Principle identifies the micro-mechanism. Every dyadic interaction between unequal-Status parties constitutes a Micro-Extraction Event in which net Mimetic Premium flows from the lower-Status to the higher-Status party through four transfer channels: attention transfer, deference concession, narrative authority yield, and positional concession. Each event is individually trivial. But the events are ubiquitous (occurring in every unequal-Status interaction), unidirectional (always flowing upward), cumulative (each compounding into the next), invisible (neither party tracking the running total), and Buffer-Converting (each extraction converting A-I relational investment into PM, reducing total buffering capacity).

Over time, this cumulative extraction converts the A-I Buffer that constitutes genuine institutional resilience into concentrated PM at the apex of the Status hierarchy. The conversion is the Buffer-Converting Property: PM extraction does not merely sit alongside A-I Buffer but actively depletes it, because the attention, deference, and positional concessions flowing upward are resources that would otherwise sustain horizontal cooperation, mutual aid, and shared I-engagement. That deference is a directional concession means PM is the differential between mutual deferences; symmetrising that differential is the Mutual-Regard Equilibrium, which is therefore reachable by universal mutual deference as well as by universal withholding of downward attribution.

This yields the Buffer-Depletion Acceleration Principle: Decoherence Time (time-to-crisis) is proportional to A-I Buffer depth at the point where M-dynamics commence intensification. Societies with deep A-I Buffer can absorb PM extraction for longer before cascade dynamics become self-reinforcing. Societies with shallow A-I Buffer reach criticality faster.

The race condition stated: the contest is between (a) the rate of PM extraction, which depletes A-I Buffer continuously through normal social operation, and (b) the rate of A-I Buffer replenishment, which depends on deliberate institutional design. Extraction is an architectural constant, operating through every unequal-Status interaction. It cannot be stopped. It can be slowed by AIM-aligned institutional design, but it cannot be reduced to zero while M operates. Buffer replenishment is never automatic. It requires conscious, sustained, structurally embedded effort. If extraction outpaces replenishment, Buffer depth decreases monotonically and criticality approaches as a mathematical certainty, not as a contingent historical accident.

Historical validation. The Treaty of Versailles (1919) imposed conditions that simultaneously depleted Germany’s A-Floor (reparations producing economic precarity), destroyed I-infrastructure (institutional humiliation, loss of cultural confidence), and concentrated PM through the grievance dynamics that followed. A-I Buffer depth at the point of M-intensification was minimal. Time-to-crisis: fourteen years (1919 to 1933).

The Marshall Plan (1947) pursued the inverse strategy: rebuilding A-Floor (economic reconstruction), restoring I-infrastructure (institutional rehabilitation, cultural re-engagement), and distributing resources horizontally rather than extracting them vertically. A-I Buffer depth was actively increased. Time-to-crisis: in the defeated states among the recipients, West Germany, Italy and Austria, the cascade has not reached criticality since, the longest such period in the modern history of each, and the recipients have not gone to war with one another. The aid was allocated among recipients by political and budgetary considerations rather than on the parameters of A-Floor, I-infrastructure and M-bounding, so the recipients that later passed through coups, dictatorship or colonial war do not test the principle as the architecture would parameterise it.

The contrast is not coincidental. It is the Buffer-Depletion Acceleration Principle operating in opposite directions. Versailles depleted Buffer and compressed Decoherence Time. The Marshall Plan replenished Buffer and extended Decoherence Time. The architecture did not change between 1919 and 1947. The intervention strategy did.

At the Belief level, the Buffer-Depletion Acceleration Principle maps onto the mechanism of Referent Abandonment (Chapter 14, Section 14.8.3.1). When a polity maintains Beliefs against available decohered Referents, Beliefs whose truth-channel has been closed by the world's verdict but whose maintenance serves the position of those who hold them, the gap between the Belief and the world is the accumulated debt that the A-I Buffer defers. Buffer depth determines how long the deferral can continue before Decoherence Time expires and the Belief collides with reality at crisis scale. Versailles created the conditions in which the gap could not be sustained: the Referents were forced into contact with the German population's A-state too rapidly, too visibly, and with no compensating I-infrastructure to absorb the collision. The Marshall Plan did the reverse: it widened the A-I Buffer so that the gap between post-war nationalist Beliefs and their decohered Referents could resolve gradually rather than catastrophically. Buffer maintenance is, at the Belief level, Decoherence Time management: the structural provision of the absorptive capacity that allows Beliefs to update against their Referents rather than to collide with them.

17.17.3 Individual Violence as Expected Output

Due to the constraints of the human cognitive architecture, complete elimination of violence at the Individual level is an unrealistic goal. Violence is a regular and expected output of the architecture. Source Opacity ensures that M-urgency will continue to present as A-threat in Individual Decision Hubs. Confabulation will continue to generate agent-based attributions for diffuse suffering. The Phenomenological Collapse will continue to produce survival-level alarm from Status-derived anxiety. These are architectural constants, not correctable defects.

This does not excuse the occurrence of violence. But it should shape how we think about it and treat the upstream causes that lead to it. The appropriate response to an architecture that produces violence is not moral exhortation to stop being violent (which the architecture will process through Confabulation and disregard) but structural intervention on the parameters that determine how frequently, how intensely, and at what scale the architecture discharges its violent output.

At the Individual level, this means accepting that some frequency of interpersonal violence is an irreducible feature of human social existence, while working to minimise its occurrence through A-Floor provision, I-domain availability, and Knowledge Patch dissemination. At the population level, however, Subcriticality is in principle achievable via AIM-aligned interventions, even and especially when an M-Captured society and its leadership are approaching the Subcriticality Threshold (Pc, formalised in rate form at Appendix A, Section A.5.8.4, whose multi-layer correspondent is the Total Stability Condition G_crit^total of Section A.5.7.4; the threshold above which population-level cascade dynamics become self-reinforcing). The goal is not the elimination of violence but the indefinite deferral of population-scale cascade-to-violence, keeping the system below Pc through sustained architectural maintenance. This chapter's subcriticality framing is complemented by the Knowledge Patch's Operation 7, the Normality Fiction Breaker, which dismantles the "normal person" Template as a mimetic artefact: a key mechanism through which M re-enters self-evaluation and peer evaluation even after Operations 1–6 have been internalised.

The architectural target of Subcriticality Maintenance is the internal cycle's amplification per loop, not exposure alone. Reducing exposure to Models lowers the operating point of the system, which has a second-order effect on stability, but it does not on its own collapse a closed cycle that is already running supercritical. Collapsing the cycle requires reducing the Decision Hub's weighting on M (the Knowledge Patch), reducing the Observability feedback gain (Observability Governance), or reducing the salience-to-mimetic-drive coupling (Salience engineering). The architectural reason these three are the load-bearing levers is that they are the parameters that close the internal cycle. Exposure-based interventions are necessary, because some level of exposure is the precondition for any activation at all, but they are not sufficient. The combination of exposure-based and cycle-based interventions, applied in concert, is what keeps the cycle below threshold across the populations and institutions in which M-dynamics operate. The formal characterisation of the two intervention classes is given in Appendix A.5.1.

17.17.4 AIM-Aligned Intervention and the Current Configuration

What intervention can achieve upstream differs from what it can achieve once a potentially terminal configuration is already in place.

Most AIM interventions are upstream prophylactics. They operate by maintaining A-I Buffer depth (A-Floor provision, I-domain protection), reducing cascade velocity (information diversity, institutional friction, Low-Theatre governance), distributing authority to prevent Leadership Compression (separation of powers, mandatory delays, distributed cancellation authority), and disseminating the Knowledge Patch to expand the population’s Hypothesis Space beyond Confabulated agent-based attribution. The Institutional Brake is that friction, imposing independent delay on Bayesian cascade velocity through the implementation chain from a leadership Decision Hub to collective action. These interventions work by keeping the system below Pc, preventing the configuration from arising in which an M-Captured leader with executive military authority over nuclear weapons could translate mimetic shock into planetary consequence.

Once that configuration is in place, the scope for practical intervention shrinks dramatically. An M-Captured leader with command authority, information control, A-leverage over subordinates, and appointment power has already bypassed the institutional friction that upstream interventions were designed to maintain. Advisory diversity has been replaced by loyalty-filtered selection. Institutional checks have been eroded or captured. The information environment that would sustain Belief heterogeneity has been narrowed to reinforce the leader’s Confabulated threat assessment. The configuration is self-reinforcing: M-Capture produces institutional capture, which deepens M-Capture, which accelerates institutional capture. Neither waiting the configuration out nor removing the leader averts its discharge, and what remains open is the structure the discharge will meet (the Trigger-Indifference Corollary, Section 17.7.10).

Avoiding this configuration altogether, through AIM-aligned upstream intervention, is how Subcriticality can theoretically be maintained indefinitely. The Time-Buying Principle (Corollary to the Leadership Compression Principle, Section 17.7) establishes that any reform adding friction between leadership M-shock and crisis imposition serves civilisational survival proportionate to the friction added, even if the reform falls short of optimal architecture. Under nuclear conditions, this corollary is not policy recommendation. It is survival arithmetic.

As at the time of writing, the precise terminal configuration described above is already in place, and behaving exactly as the AIM architecture predicts. Multiple nuclear-armed states exist with authority configurations that approach or exceed the Leadership Compression Principle’s critical threshold. The race condition’s damping factors are weakening across multiple dimensions concurrently: economic precarity is rising (A-Floor erosion), institutional trust is declining (friction reduction), authority is concentrating (Leadership Compression amplification), and information infrastructure is optimised for engagement, which is to say optimised for M-activation and cascade synchronisation (κ amplification across the Observability Gradient).

Yet as humanity drifts toward the Terminal Phase, hope remains. The Symmetry of Crisis and Prevention established in Chapter 13 (that the same Bayesian machinery enabling crisis also enables correction) means that intervention remains structurally possible at every point short of irreversible action. The only moment when hope is no longer warranted is when the missiles are in the air. Until then, AIM-aligned interventions remain possible.

These interventions take an unconventional form. They apply proactively, before or during crisis, principles that have historically been applied only after war. The Marshall Plan rebuilt A-Floor and I-infrastructure in defeated nations after catastrophic violence had already occurred. Its logic was sound: secure the A and I sufficiency of a population so that its cognitive architecture is not dominated by the survival-level alarm that M-amplified precarity produces, and the population retains the deliberative capacity to address its genuine problems pragmatically rather than through Confabulated scapegoating. The error was timing. The Marshall Plan was reactive (applied after tens of millions had died) rather than proactive (applied before the cascade reached violence).

AIM-aligned intervention applies Marshall Plan logic pre-war, or during war, by securing the A and I sufficiency of the M-Captured nation and any other stakeholders, so that the populace has the cognitive capacity to address its situation pragmatically, not from a state of duress and anxiety that drives M-urgency above Pc. When A-floors are secure, the bandwidth effect (Mullainathan and Shafir, 2013) reverses: cognitive resources become available for deliberative processing rather than being consumed by survival computation. When I-domains are accessible, Individuals have alternative value-sources that reduce M-Salience. When both conditions hold, the population’s collective posterior can be shifted away from convergent scapegoating because the felt urgency driving the convergence has been architecturally reduced at its source.

This is not charity. It is not appeasement. It is not rewarding aggression. It is structural intervention on the parameters that determine whether a nuclear-armed population’s Confabulation Cascade reaches the Terminal Phase or is arrested before irreversible discharge. The Marshall Plan was not generosity toward Germany. It was the most effective anti-crisis intervention in modern history, precisely because it addressed the architectural parameters (A-Floor, I-infrastructure, horizontal resource distribution) rather than the Content of German political Belief. The same logic, applied proactively rather than after catastrophic violence, is what AIM-aligned intervention looks like in a nuclear world. The race is real. The clock is running. But the race is not yet lost.

The proactive application of Marshall Plan logic is not merely intuitively appealing. It follows as an architectural entailment from what this chapter names the M-Capture Incapacity Principle: an Individual or population operating at w_M > 0.5 is architecturally incapable of self-correcting toward lower w_M without external Hypothesis-space expansion. The incapacity is overdetermined by four independent constraints operating simultaneously. Source Opacity (Axiom 3) prevents recognition of M-contamination. Confabulation (Axiom 4) actively generates false attributions that explain the experienced distress without implicating M. Bayesian Lock-In (Axiom 8) entrenches those false attributions through the endogenous evidence loop established in Chapter 13. And the Vocabulary Gap (Layer 7 of the Camouflage Architecture) ensures that the correct attribution concept does not exist in the entity’s cognitive repertoire. Material resources, intelligence, education, and institutional sophistication do not address any of these four constraints. A wealthy, institutionally sophisticated nation in advanced mimetic crisis cannot diagnose its own condition for the same reason an Individual under Source Opacity cannot introspect their way to M-recognition: the decomposition was destroyed at integration, and the concept that would let the entity infer the leading source from the course of its own conduct has never existed in any framework the entity has access to.

The M-Capture Incapacity Principle transforms the Marshall Plan observation from historical pattern into architectural entailment. The historical observation is that proactive A-Floor and I-infrastructure provision works better than reactive post-catastrophe reconstruction (the Versailles/Marshall Plan contrast validates this empirically, as shown above). The architectural entailment is that waiting for self-correction is incoherent, because self-correction requires precisely the cognitive resources that M-Capture has foreclosed. The M-Captured entity is not failing through negligence or wilful blindness. The Camouflage Architecture is performing its function, concealing M from detection at every level (individual, institutional, civilisational) through eight independently operating and mutually reinforcing layers. The “they should help themselves” objection is itself a Confabulation, an agent-based attribution (“they are failing”) for a structural condition (the architecture prevents detection), and is precisely the kind of attribution the Presumption Against Scapegoating is designed to catch.

This grounds the proactive obligation directly. Under the Cascading Structure of Rights and Duties, those who possess the Knowledge Patch and withhold it from M-Captured entities bear the burden of justifying inaction (the Burden-Reversal Principle), because the M-Captured entity cannot acquire the Knowledge Patch through its own cognitive resources (architectural incapacity), and the cost of withholding it (crisis trajectory acceleration toward the Terminal Phase) exceeds the cost of providing it (vocabulary dissemination, A-stabilisation, I-enablement). Proactive Cascading Support (A-restoration, I-enablement, M-containment, Knowledge Patch dissemination applied before or during crisis rather than after catastrophic violence) is therefore not charity, not appeasement, and not optional. It is the architecturally required response to the formal demonstration that the alternative (waiting) is structurally guaranteed to produce the outcome (crisis) that intervention is designed to prevent.

Critically, Proactive Cascading Support operates entirely through legitimate channels, whether at the international level (development aid, A-floor investment, I-infrastructure provision, Cascading Consequences for aggressors) or the interpersonal level (stabilising material security, providing the Knowledge Patch, making I-activities available, dampening visibility pathways). The intervention addresses architectural parameters (A-sufficiency, I-access, M-bounding, Hypothesis-space expansion), not Content (what to believe, how to govern, which policies to adopt). Sovereignty is preserved. The Knowledge Patch is developed in Chapter 18; the remaining mechanics are not derived here.

---

17.17.5 Why Recovery Is Restart, Not Self-Correction

The loom passage in 17.17.1 stated that a society which crosses the threshold does not recohere into its prior form, and that the pre-nuclear cycle could recommence only because survivors reweave a new configuration on the same ground. That claim is a precise architectural result proved in Appendix A.5.3 and, at the network scale, A.5.7.6, and the result is sharper than the metaphor alone conveys.

There are, in principle, two ways a compounding system can stop compounding. It can oscillate: overshoot, correct, overshoot again, wobbling around a stable point without ever flying apart. Or it can fold: run away until it collides with its own limit and collapses discontinuously to a distant baseline, with no path back to where it started. The first is a self-correcting system; the second is not. The whole question of whether a society "recovers" from a mimetic crisis is the question of which of these two behaviours the architecture produces.

The architecture produces the fold, and it produces only the fold. This is the Saddle-Node Default: the bifurcation at the critical loop gain is necessarily non-oscillatory, and under the architecture's stated conditions a saddle-node collapse, in which the inflated valuation and a neighbouring unstable equilibrium collide and annihilate, leaving the system to jump discontinuously to the A-I baseline.

The fold has a second face, and it is the one that decides whether prevention and recovery cost the same. A folding system does not only fall; it also catches. Approached from below, the same collision happens to the lower of the two resting states, and the system snaps upward into the inflated one, which is what a society crossing into crisis is doing. Approached from above, it is the inflated state that vanishes and the system drops. Both are the same geometry and neither is gradual, but they do not happen at the same conditions, and the gap between them is the whole practical difference between the two tasks.

It is the behaviour of a microphone held too near its speaker. The howl begins at one setting of the volume knob and stops only at a lower one, so turning the knob back to where the howl started leaves the room still howling. A society that has ignited is accordingly not somewhere that can be walked back from by reversing the conditions that took it there. Those conditions have to be brought further down than they ever were on the way up, and how much further grows with how far the system has travelled. This gap between the conditions that ignite a loop and the conditions that release it is the Ignition-Release Asymmetry, and it is architectural rather than political: it is a property of the geometry and not of anybody's willingness to act.

What is inflated, in this chapter's case, is a conviction. The population believes something about a group of people, and the Confabulation Cascade has made that belief feel like a finding rather than a transmission. The reason the belief holds at its inflated level is not that the evidence for it is strong. It is that each person's conviction is functioning as evidence for the next person's, so the belief is being supplied from inside the population rather than from anything outside it. That circulation is the feedback, and it is what makes the level self-sustaining: nobody has to set out to maintain it, because the circulation itself supplies the maintenance, and no one inside can see it happening, because what reaches each person is other people's sincere agreement and not the route by which they arrived at it.

This changes what a corrective effort has to achieve. There is a threshold of collective conviction below which the population's own doubt outpaces its own reinforcement, and beneath that threshold the rest of the descent happens without further help; the belief comes apart on its own. Above it nothing holds: conviction reduced but left above the threshold returns as soon as the pressure comes off, because the circulation is still supplying more belief than doubt is removing. This is why partial measures characteristically appear to work and then do not, and why the requirement is not that correction should exceed circulation at some moment but that it should exceed it continuously across the whole range of conviction the population might occupy.

A single well-received rebuttal, or a single season of calm, does not meet that requirement. What meets it is a corrective supply sustained until collective conviction has been brought down through the threshold, and it is the crossing and not the effort that is decisive: below the threshold the population's own doubt completes the remaining descent, so the supply no longer has to be maintained against a circulation that replaces what it removes. Above the threshold that same supply is expended against replacement and buys only the appearance of progress. The population remains capable of re-ignition once it is down, which is why Subcriticality Maintenance is maintenance and not a cure.

And what lowers collective conviction is not, in the first instance, argument about the group. The architecture does not care which group is named, so contesting the content leaves the mechanism running and simply invites a substitution. What lowers it is the supply of what the population does not have: the hypothesis that the conviction is transmitted rather than reasoned, which is the Knowledge Patch's enlargement of the Hypothesis Space; the practice of holding a question open rather than closing it, which is Epistemic Suspension, and which reduces how forcefully each person's conclusion is emitted as evidence to everyone else; a reduction in the weight a neighbour's conviction carries simply by being a neighbour's, which is the Coupling Strength; and a shortening of the interval before contradiction from outside actually arrives, which is the Decoherence Time. Every one of those is supplied from outside the loop. None of them is generated by the loop. That is the architectural content of the claim that recovery is restart rather than self-correction: not that nothing can be done, but that the doing has to come from somewhere the cascade does not reach, and it has to carry the population down through the threshold rather than merely toward it.

The oscillating, self-correcting alternative (the Hopf Sub-Case, in which the system loses stability into a sustained limit cycle rather than a collapse) is not merely rare; it is architecturally inaccessible in the unaugmented three-node skeleton. Appendix A.5.3 gives the derivation: the Hopf regime requires an internal loop gain no greater than minus eight, while the three-channel architecture (the mimetic channel's weight in the integrated valuation, the valuation's feedback into environmental salience and Status, and salience's coupling back into mimetic drive) runs at positive gain for every setting of its parameters, so the oscillatory locus is unreachable by construction.

The reason is structural and was stated already in 17.3.3 and 17.3.6: the M-channel has no off-switch. A system can only wobble-and-correct if something inside it pushes back after each deviation: a restoring force, a negative-feedback term. A has one (correction of the Body Manager's deficit). I has one (correction of the deficit the Thinker signals for the Object its process requires). M has none. And Source Opacity removes the error signal that a self-correcting cycle would need in order to execute its corrective half-stroke, because the system cannot read its own valuations back to source. With no restoring force and no readable error signal, the only behaviour available is run-away-then-collapse. The exclusion is at the same time the localisation of the alternative: because no tuning of the existing couplings can produce the corrective half-cycle, sustained self-correction exists exactly where a restoring structure is added to the architecture, a fourth state variable carrying intrinsic negative feedback, which is the formal signature of Institutional Buffering (Appendix A.5.3). The fold is the necessary geometry of the unbuffered loop, not of every possible society built on it.

The exclusion runs deeper than the three-channel loop, into the one layer where self-correction would most plausibly have lived. Belief revision looks like a restoring force: Evidence arrives, conviction updates, and error is worn away by the correction. The Monotone Belief-Race Lemma (Appendix A, Section A.5.6) establishes that the Bayesian layer supplies no such force. Hypotheses compete for the Belief Register as runners in a race whose referee is blind: under content-agnostic updating the support the Evidence gives a Hypothesis does not depend on that Hypothesis's current popularity, each competes only against the field average, and the race sorts monotonically to a winner without ever circling. Belief cycling would require Evidence that responds to prevalence around a closed ring, which is a violation of the updating architecture rather than a regime of it. The slow layer therefore adds gain to the loop and never a swing, and the result carries the fold across scales: with no node able to rotate and the couplings between Individuals one-signed, the network-scale collapse is a derived result at the decohered operating points where crises live (Appendix A, Section A.5.7.6), not an extrapolation from the individual case.

The converse supplies a diagnostic against which the historical record can be read. Observed oscillation is never the bare architecture correcting itself; it is the signature of an installed restoring structure doing work. In the terms of the race, a restoring structure is a handicapper, a rigging that favours the trailing runner, and it is the only rigging under which the field can circulate; the bare architecture contains no handicapper, only the tilt toward the leader. The tilt is a rate and not a verdict: while the race runs, a better-supported entrant still gains, and Epistemic Suspension throttles the leader's self-reinforcement at the emission point; what the architecture withholds is self-correction, not correctability. Where a polity cycles rather than collapses, a buffer is present, whether or not anyone built it on purpose. Where the buffers erode, the folds return.

This is why survivor-succession must not be misread as evidence that the system heals itself. When a society folds and survivors build a new civilisation on the ruins, the architecture has not oscillated back toward its prior configuration. It has collapsed to baseline and been restarted from new initial conditions by Individuals carrying forward the altered priors the catastrophe left in them. Each historical "cycle" of collapse and recovery is therefore more consistent with a sequence of independent folds than with a single self-correcting oscillation, each fold followed by a fresh ignition from a perturbed starting point. The loom was never springing back into shape. It was being restrung, by hand, each time, by whoever was left.

A self-correcting system carries its own guarantee of recovery: the restoring force is built in, so the system cannot help but return. A fold-then-restart system carries no such guarantee. Its recovery depends entirely on an external condition the architecture does not supply: the survival of enough Individuals to perform the restart. The pre-nuclear loom "always" rewove not because anything in human motivation ensured it would, but because the folds had so far been survivable. Remove the survivors and you do not get a slower recovery or a worse one. You get no recovery, because the only mechanism that has ever produced recovery, surviving weavers re-igniting from baseline, has been eliminated. This is the precise architectural content of the claim in 17.17.1 that the nuclear arsenal can take the loom rather than the cloth.

The single route to a genuinely self-correcting society, one that wobbles and recovers without folding at all, is to install the restoring force the architecture lacks, which is to say: to bring the inaccessible Hopf Sub-Case into reach by engineering it from outside. Formally, this means adding a fourth element to the system whose entire function is to push back against M-amplification: a new variable carrying intrinsic negative feedback. In the language of this treatise, that fourth element is an Institutional Buffer: an institution engineered specifically to supply the damping that individual human architecture cannot. This is not a refinement of existing arrangements. It is the addition of a structure that is not currently present anywhere.

The architecture leaves only one location available for this restoring force. It cannot be installed at the level of the Individual. This is not a practical limitation but the same architectural fact the whole treatise turns on: at the individual scale the Mimetic Premium cannot be identified, because Source Opacity strips the source-tags at integration and the Decision Hub returns only an integrated Scalar. The Individual cannot read how much of their own valuation is M-driven, and so cannot apply a correction calibrated to it; there is no internal reading from which the quantity needing damping can be measured. A restoring force requires a controller that can measure the variable it is meant to regulate, and at the individual level that variable is structurally unmeasurable from within.

At the institutional level the situation inverts, and it inverts for a specific reason. Aggregation across a population acts as a low-pass filter: the idiosyncratic, high-frequency noise of any single Individual's Confabulation averages out across many Individuals, and what survives the averaging is the slow, population-scale signal: the actual movement of valuations, prices, status distributions, and grievance across the whole field. The institution therefore has access to a quantity no Decision Hub possesses: a resolved measurement of what is in fact happening across the population, precisely the Mimetic Premium signal that Source Opacity hides at the individual scale.

The Easterlin Paradox is the clearest worked instance. Measured wellbeing rises with income across Individuals at a point in time, and within a country at a point in time, yet fails to rise as a whole country grows richer over the long run: a contradiction that has resisted resolution for half a century because it is examined as a fact about money and happiness. Disaggregated through this architecture it ceases to be a paradox at all. The cross-sectional gradient is the M-signal: relative position, being richer than one's Reference Set, raises reported wellbeing. The flat long-run trend is what remains once a whole population rises together; the A-component, having already been satisfied, yields no further wellbeing, while the relative component nets to zero across the population because not everyone can rise in relative standing at once.

What the paradox is measuring, once disaggregated, is a society expending real and rising resources on acquisition that produces no aggregate wellbeing gain: growth being absorbed by M rather than by A or I. That is a macro-level readout of the Mimetic Premium running away, exactly the kind of indicator no Individual can perceive, because each Individual sees only their own scalar valuation and cannot observe that the gain nets out across everyone, but which resolves cleanly at the scale where the low-pass filter exposes the aggregate.

Because the institution can see the variable, it can act on it. It can pull levers (on observability, on coupling strength, on the A-floor, on the evidential environment) that are calibrated to the population-level dynamics rather than to any Individual's introspective report. This is what makes the Institutional Buffer the only viable site for the restoring force: the control surface must be installed where the system is observable, and the system becomes observable only at the scale at which the low-pass filter resolves the signal. Damping at the institutional level is possible for exactly the reason damping at the individual level is impossible.

The institutions required to maintain Subcriticality do not exist. The buffering that current institutions provide (democratic process, legal constraint, separation of powers, the international order assembled after the last catastrophe) is real, and it is well-intentioned, and it has bought time. But it is insufficient and ill-equipped, for one decisive reason: it was not designed against this architecture, because this architecture had not been identified. These institutions sit at the right scale (they have, in principle, the resolution to see the population-level signal), but they lack the map that would tell them which lever the signal calls for, and so they apply generalised friction where calibrated damping is required. Institutions built without a model of the system they are meant to stabilise cannot supply targeted negative feedback to it; friction slows the approach to the fold without preventing it. Friction buys time. It does not install the restoring force. As 17.17.2 established, the damping factors on which present arrangements rely are themselves eroding continuously through normal social operation, and nothing in the current institutional inventory replenishes them by design rather than by accident. The existing buffers are the right instinct, at the right scale, applied without the right map, and this is the reader-surface statement of what Appendix A.5.3 names the Incidental Buffer Principle.

Three things follow, and together they are the Incidental Buffer Principle. Because the architecture supplies no restoring force of its own, every buffer that has ever damped a mimetic cascade has been external to the loop. Because the loop had not been identified, no such buffer was engineered against it; each was built to a culture's Confabulated account of its own disorder and damps the Mimetic Premium only as a side-effect. And because a buffer's strength is sustained by that Confabulated rationale rather than by recognition of what it damps, buffers of one form or another have always existed but have always varied in depth, drifting with the rationale that holds them up rather than with the architectural quantity at stake.

The same holds for damping applied from within the Individual rather than from outside. The Knowledge Patch adjusts the Individual's priors and interposes the I-Override gate, lowering the effective loop gain, but the Patch acts inside the loop rather than as the added restoring node, so it merely adds internal damping and, depending on sustained and partial uptake, is unreliable as a means of completely avoiding the Fold. It too was unavailable throughout history, because the architecture it adjusts had not been identified. Internal damping by the Patch and external buffering by institutions are complementary, and neither is sufficient alone. The formal statement is the corollary at Appendix A.5.3.

Supplying that map, and specifying the conditions the institutions it makes possible must meet, is the work of Chapter 20, which sets out what Subcriticality-maintaining institutions must satisfy to function as genuine Institutional Buffers (to supply the intrinsic negative feedback that converts the architecture's fold into a survivable oscillation) and what may be established about them from outside. How to transition from the current institutional form to the required form without precipitating the very collapse the reconfiguration exists to prevent is not derived here. That second problem is not secondary. An institutional order is itself a high-coupling system, and dismantling existing buffers faster than replacements can bear load is a recognised route to the fold. The design of the transition is therefore part of the design of the destination, and the framework does not settle it.

17.17.6 The Boundary-Independent-Output Exception and the Gating Domain

Sections 17.17.1 and 17.17.5 established, in the loom vocabulary, why the nuclear arsenal is the first form of architectural output capable of taking the loom rather than the cloth. That claim can now be stated in the Belief-space vocabulary developed in Chapter 14, Section 14.8.4, which gives it a precise structural form and locates exactly where the containment the rest of this treatise relies on fails.

Chapter 14 established that a mature civilisation's Belief-space is not a single undifferentiated medium but a foam of many weakly-coupled Belief-domains, each bounded by maintained boundary-Beliefs. Because those boundaries are themselves maintained Beliefs, collapses are ordinarily domain-local: a price bubble can fold while legitimacy holds, and legitimacy can fold while prices keep clearing, because the low cross-domain coupling at each membrane contains each correction within its own bubble. This is the migration-firebreak result. A fold does not multiply bubbles or destroy the Beliefs inside the collapsing one; because those Beliefs are about something, and the demand they answered does not vanish with the collapse, they must relocate into adjacent bubbles as new evidential load. Multiplicity matters because that migrating load crosses each membrane at low cross-domain coupling and, in a finely divided foam, arrives attenuated enough to dissipate below the re-ignition threshold in the neighbour rather than driving it supercritical. Terminal collapse therefore requires either a shock large enough to clear that threshold across a connected chain of bubbles or the sequential decoherence of the separating boundary-Beliefs. It is subcriticality maintenance implemented structurally, and it is a qualitative resistance property of the foam, not a quantitative claim; the formal migration-and-containment condition is the network critical coupling of Appendix A.5.7.5.

This firebreak holds only under one condition: that every bubble's output on the world is itself Belief-mediated. A fold in one bubble relocates its Beliefs into a neighbour only across their shared membrane, and where that coupling is low, the migrating load arrives attenuated and dissipates before it reaches the re-ignition threshold in the neighbour. The condition fails for a Belief-domain whose effect on the world is physical and total rather than Belief-mediated. Such a domain is not contained by any Belief-membrane, because its destructive channel is exogenous to the Beliefs that constitute the boundaries of the foam. A fold that captures such a domain does not relocate its load through the foam and does not stop at a membrane: its destructive channel bypasses the membranes across which migration would otherwise dissipate, so it destroys the foam rather than being contained by it.

Nuclear command is the canonical instance. The output of a nuclear-command domain is not a price, a legitimacy, or an orthodoxy that neighbouring domains may or may not update against; it is physical destruction delivered on a timescale (Section 17.17, decision-timeline compression) shorter than any Belief-membrane's correction time and at a magnitude (consequence-magnitude explosion) that the firebreak structure was never able to bound. This is the Belief-space statement of the loom-taking claim: a fold captured in a boundary-independent-output domain takes the loom rather than the cloth precisely because its output does not pass through the weave at all.

The gating-domain condition. It follows that a Belief-domain with boundary-independent catastrophic output is a single point of failure for the entire structure, and cannot be governed by the multiplicity-as-firebreak logic that governs every Belief-mediated domain. Such a domain, a gating domain, must be held at the lowest attainable Coupling Strength and the shortest attainable Decoherence Time by dedicated structural means, because the ordinary containment on which the rest of the foam relies does not reach it.

Here "low Coupling Strength" must be read with care, because it names the opposite structural condition from the one that makes the general foam safe. In the foam at large, the protective quantity is low cross-domain coupling: a fold in one bubble dissipates because its migrating load crosses a low-coupling membrane into a neighbour. For a gating domain the load-bearing quantity is instead the self-coupling of the executive act: the degree to which the decision to discharge the boundary-independent output takes a single Individual's own Beliefs as its sole evidence base. Concentrating command in one person, or in a small number of people, is the maximally self-coupled and therefore the most dangerous configuration: such an Individual is coupled almost entirely to their own Beliefs, so a confabulated Belief on which the discharge decision is taken is subject to no external discipline and can move directly to physical output. Source Opacity guarantees that the Individual cannot introspect the confabulation, and maximal self-coupling guarantees that no one else can arrest the decision it produces.

Holding a gating domain at low Coupling Strength therefore means the reverse of isolating it: the executive action must be coupled to, and disciplined by, many independent bubbles (distributed authorisation, independent evidence-bases, and required concurrence across Belief-domains that do not share the deciding Individual's priors), so that no single self-coupled Belief can carry the decision across the gate to physical output. Low self-coupling of the decision and mandatory cross-domain discipline of the executive act are the same requirement stated at the mechanism and at the institution.

The normative structure falls out of the premises rather than being imposed on them. If self-coupling is real, and if a single Individual holds executive power to discharge a boundary-independent catastrophic output, then a decision taken on a confabulated Belief in that Individual's mind is an unarrested planetary hazard; this is not a value judgement added to the analysis but a direct entailment of the architecture the treatise has already established.

What the same architecture additionally explains is why command has drifted toward concentration in the first place. The movement of nuclear authority toward a single person is not a neutral administrative fact; it is the predicted output of the dynamics developed throughout this treatise. M ascribes Salience, Salience aggregates into Status (the Salience-Status Identity), and the apex of a Status hierarchy attracts the concentration of authority (the Leadership Compression Principle, Section 17.7). The chain of reasoning that concludes "one trusted Individual should hold the decision, to protect the populace" is then supplied after the fact by Confabulation, which presents a Status-driven concentration as a considered safeguard. This is precisely the Confabulation signature the architecture predicts: the felt rightness of single-person command is the phenomenological residue of M operating on the authority hierarchy, not the conclusion of an argument that survives the premises. The parsimony runs the other way, from the axioms, through M's salience-to-Status operation and the compression of authority, to concentrated command as an architecturally expected, and architecturally dangerous, outcome.

17.18 The Precautionary Asymmetry as a Derived Methodological Commitment

The preceding section has established, in two vocabularies, that the terminal event is a fold and that the nuclear arsenal is the first architectural output able to take the loom rather than the cloth. One methodological consequence of these results is a commitment derived from the architecture, not a cautionary posture adopted in place of one. It is the reason the civilisational-scale claim cannot be held to the same falsification schedule as the survivable-scale claims, and it is the reason that exemption is not an evasion of Popperian discipline but a direct consequence of it.

The architecture distinguishes two error modes. At the survivable scales (consumer valuation cascades, positional-good bubbles, polarisation within bounded communities, status-driven organisational pathology) a supercritical fold is a recoverable event. It destroys value, and it can destroy lives, but it leaves behind a population and an adjudicating community that can update its posteriors on what occurred. The fold at those scales is therefore admissible as an ordinary object of empirical work, and the framework's predictions there stand or fall in the standard Popperian sense: if calibration places realistic operating points below the loop-gain threshold, the predictions are not instantiated in the populations measured, and the claim is falsified by that evidence. This is the schedule the survivable scales are held to, and the leading-indicator clause of Section 17.5 is one instance of it.

The terminal manifestations identified in this chapter are error modes of the second kind. A thermonuclear exchange between mimetically entangled nuclear powers, a civilisational-scale cascade failure, a terminal reconfiguration: these are observer-extinguishing outcomes. No posterior is updated after the event, because no adjudicating community remains to update it. The distinction between the two error modes is not one of magnitude but of recoverability, and it is the same distinction the chapter has already drawn between M-harm and A-harm at Section 17.15.1: Content is recoverable, and death is not. The recoverable fold and the non-recoverable fold differ in exactly the way conquest differs from war, and for exactly the same architectural reason.

The methodological consequence follows without a further premise. The Popperian schedule presupposes a surviving adjudication community, because falsification is a procedure carried out on posteriors after the predicted event either occurs or fails to occur. Where the predicted event is one that eliminates the community that would perform the adjudication, the schedule cannot be run, not because the claim is unfalsifiable in principle but because the procedure that would falsify it is destroyed by the event it is meant to assess. To insist that the civilisational-scale claim be held in abeyance until it has been tested in the ordinary way is therefore not a demand for rigour. It is a demand that the one error mode the architecture identifies as non-recoverable be treated as though it were recoverable, which is precisely the conflation the framework exists to name.

This is the Precautionary Asymmetry. The asymmetry is not a preference for caution over evidence. It is the recognition that the two error modes carry asymmetric costs of being wrong, and that the asymmetry is itself an architectural entailment. If the framework predicts a civilisational-scale fold and the prediction is mistaken, the cost of having acted on it is the cost of maintenance measures that turned out to be unnecessary: real, but recoverable. If the framework predicts a civilisational-scale fold and the prediction is correct, the cost of having declined to act on it is terminal and irrecoverable. Under that structure, withholding action pending survivable confirmation is not the neutral, rigorous default it presents itself as. It is a bet that stakes the non-recoverable outcome against the recoverable one, and it is exactly the bet a system that cannot distinguish M-catastrophe from A-catastrophe (Section 17.15) will be architecturally disposed to make.

Two results of the preceding chapter bear directly on the asymmetry, and together they explain the shape the objection to precaution reliably takes. The Averted-Cost Opacity Principle establishes that an intervention preventing a Conflict destroys the event that would have priced it, so the party urging precaution can never exhibit the cost avoided, while the party urging delay is always in possession of the visible and immediate cost of acting. The dispute between them is asymmetrically evidenced by construction rather than by the merits, and the appearance that precaution is the weakly supported position is an artefact of which costs crystallise. At the terminal scale the opacity closes from both directions at once. The averted cost cannot be exhibited because the event was prevented, and the incurred cost cannot be exhibited either, because the event that would exhibit it eliminates the community that would do the exhibiting. Precaution at that scale is the one position that can never assemble the evidence its opponents demand of it, and the demand is not a standard of rigour but a requirement that the non-recoverable error mode be treated as recoverable.

The Failure-Cost Identity supplies the figure the first result says cannot be had on the prevention side. Global military expenditure, and the reconstruction that follows its use, is the Identity evaluated at the scale on which this chapter's terminal claim operates. That expenditure is not the price of security. It is the audited annual cost of the failure to hold the interfaces between populations sub-critical, published in national accounts, reconciled by treasuries, and reported without anyone reading it as a failure measure. A civilisation confronting a non-recoverable error mode in the form of nuclear war is therefore not short of evidence about the cost of the mode it is in. It is publishing that evidence continuously, under a heading that describes the expenditure as prudence, and the heading is the institutional form of the same Confabulation this chapter attributes to the Individual who experiences a Mimetically acquired desire as authentic preference.

The civilisational-scale fold is not offered as a cautionary hypothesis that further evidence might one day support. It is an entailment of the architecture. Where the loop-gain condition is met at civilisational scale, the fold is what the mathematics says will follow, and the fold's necessity (the absence of any restoring force on the M-channel, Source Opacity blocking the error signal, and Concealment Infrastructure rising monotonically until it fails) is the same necessity proved for the survivable scales and formalised at Appendix A.5.3 and A.6.3. The claim is cautionary and entailed at once, and the two are not in tension. The framework says: if you do not attend to what the construction of the system mathematically entails, this outcome, total annihilation, will follow. Proceed accordingly. The caution is the appropriate response to the entailment, not a substitute for one.

17.19 The Field-Correction Obligation at Collective Scale

Chapter 16 established that a Harm does not transfer a quantity to a creditor. It deforms Self-Models across a whole field, and under the Relational Subcriticality Principle the field's own persistence requires that every deformed Self-Model be re-fitted, no node excluded. That requirement is the Field-Correction Obligation, and the Field-Correction Obligation has two limbs running at the same time. Cascading Support is the obligation applied to the field deformed around the victim. Cascading Consequences is the obligation applied to the generating field: the party who acted, and every party who loaded the loop that produced the act. Chapter 16 derived both for Conflict between Individuals and established that the collective case follows by the same derivation. This section takes up the collective case.

The scaling changes nothing structural. When a state violates the rights of a population the generating field is larger and its nodes are institutional rather than personal, the leadership that authorised, the apparatus that executed, the interests that resourced it, the outlets that framed it, the outside suppliers whose materials made it possible, but each is a node carrying the load it added and the forward risk it presents. The culpability gradient tracks the load, not the office. The obligation attaches to the deformation rather than to anyone's intent, because the deformation is the same whatever produced it.

17.19.1 The Generating Field Cannot Be Permitted to Export

Section 17.12 established what holds a decohered Belief in place. The maintenance Cost is placed on the people the Belief excludes, and by the Boundary-Hardening Principle the act of placing it there is the act of hardening the boundary. A correction that leaves that arrangement intact corrects nothing. The Cost has to return to the field that generated it.

At collective scale that has a definite institutional form. Those who decide on violence must be unable to place themselves, or what they hold, outside the field of consequence their decision creates. Where they can, the export channel stays open and the Belief stays cheap to hold. Closing it means the field of consequence is coextensive with the field that generated the act, for as long as the act continues, and that the closure is indifferent to rank. A party who can purchase an exit has not been included in the field, whatever their formal position within it.

This is not deprivation and it is not scaled to the gravity of the offence. It removes one structural possibility: holding a Belief whose Cost is paid by someone else.

17.19.2 The Floors Are Non-Derogable Throughout

Section 17.12.5 derived that the cost of abandoning a false Belief is what holds it in place, and that securing Appetitive and Intrinsic provision unconditionally drives that cost down to Status alone. The same derivation fixes what a consequence may not do. A measure that removes Appetitive provision or forecloses Intrinsic access from the generating field raises the abandonment cost, which is the quantity the correction exists to lower. It makes the Belief harder to put down.

Chapter 16 gives the mechanism rather than leaving this as a paradox. Withdrawing A-essentials or I-space from a node does not attenuate the loop, it loads it further, because under the Non-Malice Corollary the suffering registers as evidence of external hostility and hardens the very priors that produced the act. Deprivation is not a stronger correction. It runs the correction backwards.

The floors therefore hold throughout, and they hold for every node inside the generating field, including those who decided. Those floors are the Non-Derogable Essentials: A-related and I-related goods that institutions may not withdraw even when imposing sanctions. That is not leniency toward the powerful and it is not a balance struck between justice and mercy. It is the condition under which the correction works at all. The requirement is over-determined, because the ignition condition formalised at Appendix A, Section A.5.8.7 is referenced to the worst-off Individual rather than to the mean: a floor that holds on average while failing at the minimum leaves the ignition condition live.

What may be constrained follows from the same place: the levers, materials and channels that convert into the capacity for violence, and not the provisions that answer the body's demands or sustain absorbing work. What is constrained is the capacity to produce the output. What is preserved is the provision the correction runs on. The distinction is one of position in the chain rather than of severity: constraining the levers removes the ability to act, before the act, while a penalty answers the exercise of an ability left intact. The first is a smaller intervention than the second and reaches further, because it does not have to be repeated for each occasion the ability would otherwise be used.

17.19.3 The Consequence Ends When the Conduct Ends

Section 17.12.5 derived that a boundary is permeable when crossing it is free, and that a permeable boundary decoheres continuously where a hardened one stores load for a single violent release. A field of consequence that persists after the conduct has stopped is a boundary charging for having been on the wrong side of it, which is the operation the correction is meant to reverse.

The consequence therefore lifts on cessation, and lifts automatically rather than on a further judgement. A penalty phase extending past the conduct is not a stronger correction. It manufactures a fresh grievance and loads the loop the correction has just unloaded, re-creating on the far side of the settlement the stored pressure that produced the original crossing.

17.19.4 The Correction Is Conducted Without Theatre

A person watching a distant war has no Appetitive exposure to it. Nothing about their food, shelter or safety turns on the outcome. They are alarmed nonetheless, and the alarm is not a mistake they could correct by looking more carefully. Mimetic transmission runs preconsciously, inside the Transmission Window, so the observed alarm of others arrives before any judgement about relevance can form. Common-Currency Integration collapses it into a single Scalar and Source Opacity strips the tag that would mark it as M rather than A. By Phenomenological Collapse what remains presents as a threat to the body. The Individual cannot tell from the inside that they have no stake, because the architecture has destroyed the information that would tell them.

That has a consequence for how a correction is conducted, independent of what the correction does. Staging it for an audience recruits A-urgency across populations with no exposure to the conflict and no reality-contact able to correct it, which raises Coupling Strength in precisely the domains least equipped to bring it down. Inside the field being corrected, the same broadcast supplies what the Scapegoat Mechanism requires: visible external hostility, which is the material an agent-attribution is built from.

Low-Theatre conduct follows. Measures are framed by reference to the capacity being constrained rather than to what the party subject to them deserves. Provision is maintained as ordinary practice rather than announced as clemency. The correction is made boring deliberately, because every increment of spectacle is an increment of coupling in populations with no way to discharge it.

17.19.5 The Apparatus That Exists, Measured Against the Four

The current architecture prices the conduct. It attaches a sanction to the act, applies it late and improbably, and does not touch the arrangements by which the Cost of the licensing Belief is placed on the people the Belief excludes. Against the four requirements it does not close the field, so the export continues; it derogates from the floors, which raises the abandonment cost and hardens the Belief; it extends past cessation, which manufactures the next grievance; and it is conducted as spectacle, which raises coupling in the perpetrating population and the watching one alike.

These are not four faults. They are one fault seen four times. The apparatus is built at the point where the violation appears rather than at the points that produce it, and each of the four requirements is a requirement about an earlier term in the sequence. Closing the field acts on where the Cost lands. Holding the floors acts on what abandonment costs. Ending on cessation acts on whether the boundary stays permeable. Conducting without theatre acts on coupling. An apparatus that reaches none of them can suppress instances of the output at whatever rate its reach allows, and will leave the rate at which the output is generated where it found it.

Suppression is not neutral in the meantime. By Section 17.12.4, holding the conduct down while the conditions stand does not hold the load at zero. It stores it.

The apparatus that exists treats the symptom. Each instrument is laid over the last at the same level, a further prohibition, a further tribunal, a further penalty, and each meets the disorder where it presents rather than where it is generated. A dressing over a dressing is still a dressing, and no number of them reaches the term that sets the rate.

What has been derived here is aimed at the cause. Not a cure in the sense of ending the condition, because the architecture that generates it is the architecture that generates human culture and is not removable. A cure in the sense available for anything chronic: treat what produces the episodes and they become rarer and milder, in place of dressing each one after it arrives.

17.20 Containing a War That Has Begun

A war is a local fold: a population, or two populations coupled to one another, carried across a threshold that the dynamics which carried them across cannot carry them back over (Section 17.17.5). The fold changes the state of the system and leaves its architecture as it was, so the quantities that govern a war are the quantities that governed its approach: the coupling between the parties and between the war and every population around it, the provision that holds Individuals above their Appetitive and Intrinsic floors, the authority that converts a leader's Belief into an order, and the materials that convert an order into violence. None of these quantities is a Belief about which party is right. A war is therefore managed through the terms that govern the system, not by deciding between the Beliefs that run on it.

Appendix A fixes the direction in which each intervention moves the system and the kind of threshold toward or away from which it moves. Appendix A does not fix the size of any quantity in a real population of states, the rate at which an intervention acts, or the time a war takes to end. No result about a war that has begun is given a size by the formalism: each is a statement of direction, and the quantities that would size it are named in Section 17.20.8.

17.20.1 The Belief That Starts a War Cannot Decide Its Treatment

Every party to a war holds the Belief that it is defending itself. Section 17.9 derives that Belief from Phenomenological Collapse, which delivers Mimetic urgency to the Decision Hub as survival-level alarm, and from Dual-Role Confabulation, which supplies the other party as the source of the alarm. The derivation runs on both sides at once and runs whether or not the facts support the Belief. Chapter 16, Section 16.3 establishes that a defender's Grievance Belief is of the same type as an attacker's, the type being fixed by structure rather than by accuracy. The self-defence Belief starts a war, because the self-defence Belief licenses the first act, and the self-defence Belief sustains a war, because each act of the other party enters the Belief Register as Evidence that the threat was real.

A Belief held on every side whatever the facts carries no Evidence about the facts. The probability of finding the self-defence Belief in a party is the same whether that party attacked first or was attacked, so the likelihood ratio the observation supplies is one, and observing the self-defence Belief leaves every estimate of which party attacked exactly where it stood. Each party's Belief names the other party as the aggressor with the same force, so the two Beliefs cancel, as two signals of equal strength and opposite phase cancel. Where the conduct of the parties is also symmetrical, as Section 17.14 shows it is once violence is reciprocal, a judgment that prefers one party can rest only on a preference the judge supplies, which is neither a principled nor a reliable basis for the judgment. An intervention keyed to a party's claim of self-defence is keyed to a quantity that cannot discriminate between the cases the intervention exists to separate. The claim of self-defence is therefore orthogonal to the treatment of a war. No intervention is selected by the claim, and every intervention is selected by quantities that do differ between cases: conduct, meaning which party is fighting, which party is supplying the fighting, and which party continues after another has stopped; and need, meaning which Individuals are below their floors. Selection of that kind is the form the Belief Multipolarity Principle requires of any intervention in the Subcriticality Maintenance set, specifiable on a rate, a threshold, or a structural property of the arrangement, and never on the content of a Belief (Chapter 13, Section 13.8.6; Chapter 11, Section 11.18).

Self-Defence Orthogonality Principle. In aggression or violence between parties at any scale, the Belief that a party acts in self-defence is held by every party whatever the facts, each party's Belief names the other as the aggressor with equal force, and the Beliefs cancel, leaving no Evidence about which party is the aggressor. Where the conduct of the parties is also symmetrical, a judgment that prefers one party rests only on a preference the judge supplies. No response to aggression or violence is selected by a claim of self-defence, and every response is selected by conduct and need.

Orthogonality asks no Belief to change, and asks no Individual under attack to stop defending themselves. Bodily self-defence under attack is the survival response of every living organism, and Section 17.15 establishes that the architecture permits it in full. Once a war is under way the Beliefs on both sides are beyond the reach of any intervention inside the war's horizon (Section 17.14.2). Every intervention on a war that has begun is therefore an intervention by those outside the war, and none asks anything of the Individuals inside it. The question of which party started the war is not abandoned. It is answered after the war, by a procedure built to answer it, whose finding is recorded rather than amplified (Chapter 16, Sections 16.11.7 and 16.18), and the answer decides repair rather than which party may fight.

17.20.2 What Sustains a War

A fire burns while four conditions hold together: fuel, oxygen, heat, and the chain reaction by which burning produces the heat that keeps it burning. Removing any one puts the fire out, and a firebreak stops it spreading. A war has a counterpart to each, and each counterpart is a quantity the architecture already names.

1. Fuel is the material of war: weapons, ammunition, supply lines, the technology of fighting, and the revenues that buy them. These are "the levers, materials and channels that convert into the capacity for violence" (Section 17.19.2), and Cascading Consequences act on them (Chapter 16, Section 16.16).
2. Oxygen is attention: the war broadcast as spectacle, and propaganda that carries the war's narrative to Individuals with no exposure to the war itself. Observed alarm reaches those Individuals inside the Transmission Window and arrives as a threat to the body (Section 17.19.4), and the share of a population's attention that a war captures is one of the two couplings through which the network approaches its critical threshold (Appendix A, Section A.5.7.5). Low-Theatre conduct denies the war its oxygen (Section 17.19.4).
3. Heat is deprivation: Appetitive deficit processed as survival threat. Sustained deficit at the worst-off segment of a population raises the force of every Grievance Belief and erodes the Belief Heterogeneity Buffer that would otherwise damp the cascade (the Floor-Breach Ignition Corollary, Appendix A, Section A.5.8.7). Cascading Support lowers the heat (Chapter 16, Section 16.15), and securing Appetitive and Intrinsic provision drives the cost of putting a Belief down toward Status alone (Section 17.12.5).
4. The chain reaction is the loop between the parties. Each party's act is the other party's Evidence that the threat was real, and escalation confirms the threat (Section 17.9). The clutch interrupts the loop from one side, without the other party's agreement and without a finding of fault (Chapter 16, Section 16.17).
5. The firebreak is the low coupling between a war and the populations around it, on which the containment of every Belief-mediated collapse depends (Section 17.17.6; Section 17.20.4).

Cascading Support and Cascading Consequences are therefore neither neutral nor partisan interventions. The empirical literature on the duration of civil wars classifies an intervention by the party it favours, and finds that third-party interventions have tended to lengthen civil wars rather than shorten them, and that interventions biased toward one party have shortened wars more than neutral interventions have (Regan, 2002). Cascading Support and Cascading Consequences are classified by what they act on. Each is a rule fixed before any war, on quantities that differ between cases, and the incidence of each follows the facts. Cascading Support secures the deepest deficits first (Chapter 16, Section 16.16), and the deficit that matters for ignition is the worst-off Individual's, on whichever side of a war that Individual stands (Appendix A, Section A.5.8.7). Cascading Consequences track the load each node added and not the office it holds (Section 17.19), close the levers of the war to every party still fighting it, and lift when the conduct stops (Section 17.19.3). Closing the levers of a war does not reach the bodily self-defence of Individuals under attack, which uses what they already have; it reaches what would otherwise be added. The incidence of a targeted intervention can be heavily asymmetric, and the asymmetry moves when the conduct moves: a party that stops is out from under Cascading Consequences at once, and a party that carries the war on after the other has stopped, a party that was attacked first included, remains under them for as long as it does. A partisan intervention is fixed to a side, and a targeted intervention to a condition.

The classification the duration literature uses has no place for the targeted class, so its findings neither support nor refute it. Where the facts are lopsided, a targeted intervention would be coded there as biased, because its Cascading Consequences would fall on one party. The comparison between targeted, neutral and partisan interventions remains to be made (Criterion 18).

17.20.3 The Floors Hold Whatever They Do to the Length of a War

Holding the floors of the Individuals inside a war can lengthen the war. A population that is fed, housed and supplied can sustain a war longer than a population that is starving, and a war that would have ended in the exhaustion of a population runs on. The length of the war is not the quantity the floors answer to. Flourishing is possible only among living Individuals (Section 17.15.2), and the measure of an intervention on a war is the harm to Individuals that the intervention averts. A judgement of the war taken as a whole is the aggregate that Section 17.10 sets aside in favour of questions with determinable answers: what harm is occurring, and what would restore the Individuals whose Appetitive and Intrinsic access the war has disrupted.

Measured that way, the floors address the larger share of a war's mortality. In the conflicts since the early 1990s for which good data exist, deaths from hunger, disease and displacement ran between three and fifteen times the deaths from violence, and four indirect deaths to each direct death is a conservative average (Geneva Declaration Secretariat, 2008). The estimates are contested and vary widely between conflicts. A longer war in which the floors are held can nonetheless cost fewer lives than a shorter war in which they are not.

The floors are also Non-Derogable Essentials (Section 17.19.2). The floors hold for every node, the Individuals of the attacking population and those who decided on the war included, because withdrawing provision from a node loads the loop rather than attenuating it. A floor that lapsed whenever it lengthened a war would be a conditional floor, which Section 17.19.2 does not permit. The law of armed conflict reaches the same place from the other side: "Starvation of civilians as a method of warfare is prohibited" (Protocol I, 1977, art. 54(1)). Ending a war by pushing a population below its floors is the method that rule prohibits, and the Floor-Breach Ignition Corollary gives the architectural reason the rule is sound: sustained deficit at the worst-off segment is one of the two conditions that ignite the collective transition (Appendix A, Section A.5.8.7).

Over a sequence of wars, the floors may shorten what they lengthen within one. Holding the deficit of the worst-off below its critical depth disarms the floor locus of the Dual-Locus Ignition Theorem, and it lowers the selection pressure through which a population in deficit draws its leaders from a field that is itself cascading (Appendix A, Section A.5.8.7). The floors can lengthen a war while lowering the rate at which wars recur (Criterion 20).

What the floors exclude is provision that converts into the capacity to fight. Provision that an armed party can capture, tax or divert is not a floor. It is fuel, and it costs lives through the fighting it supplies, which fails the very test the floors exist to meet. The record shows the conversion: increases in United States food aid raised the incidence and the duration of civil conflict in recipient countries, without a robust effect on its onset, and most markedly where civil conflict was recent (Nunn and Qian, 2014). Floors therefore attach to Individuals and not to formations. Rations delivered to a unit in the field are the logistics of the fighting, and the same soldier who stops fighting is owed the floor, which is the clutch at the scale of one Individual. Provision delivered directly to households carries a second route to conversion, because a party relieved of feeding its population can spend on the war what it would have spent on its population. That route closes only where Cascading Consequences close the levers the released resources would buy. Cascading Support and Cascading Consequences therefore run together by construction: floors without Cascading Consequences release resources for fuel, and floors with Cascading Consequences release resources that have nothing left to buy.

The floors do not exist to end a war. They protect the lives within it, and they hold whether or not they lengthen it.

17.20.4 Joining Spreads the Fold

Joining a war raises the coupling of the network that carries the war. Appendix A, Section A.5.7.5 establishes that a network moves toward its Critical Coupling Strength as its graph of Evidence couplings grows denser and, where the attention of Individuals is the binding limit, as a larger share of that attention is captured by the Evidence and Observability channels. A population whose state enters a war does both. The deaths of its soldiers and the threats made against it become Evidence its Individuals take in as their own, and the war becomes what that population watches. Each entry therefore moves the network toward the threshold, and each exit moves the network away from it. The direction is fixed by the formalism. How far a given entry moves a given network is a question of calibration (Section 17.20.8).

Section 17.17.6 establishes why the threshold matters beyond the war itself. In a finely divided civilisation a Belief-mediated collapse is ordinarily contained where it occurs, because the boundaries between domains are weakly coupled, and terminal collapse requires one of three things: a shock that clears the threshold across a connected chain of domains, the sequential decoherence of the boundary-Beliefs that separate them, or the capture of a gating domain whose output no boundary contains. An alliance system whose members are committed to enter one another's wars is a connected chain built in advance. Where the chain reaches a power that holds nuclear command, the chain couples the war to a gating domain.

Because every party holds the self-defence Belief (the Self-Defence Orthogonality Principle, Section 17.20.1), a rule that entitles one party's allies to enter a war entitles the other party's allies on the same ground. A joining rule followed by every party is a cascade by construction, each entry supplying the next party with its reason to enter and with the escalation that confirms the threat (Section 17.9). An extraction rule followed by every party removes a party at each step. Only a rule that does not depend on which party is right can be followed by both sides of a war at once.

The record of how wars spread runs the same way. Between 1816 and 1965, wars spread along alliance ties as well as along borders (Siverson and Starr, 1991). Between 1816 and 2007, alliances were a virtual necessary condition of wars with more than two parties, and more nearly so the larger the war (Vasquez and Rundlett, 2016). Outside intervention has tended to lengthen civil wars rather than shorten them (Regan, 2002), and the weapons and resources supplied from outside are among the conditions that have sustained them (Hironaka, 2005).

The obligation of a party's allies follows, and it is the obligation of a party's friends in a bar fight, carried to the scale of states by the scale invariance of Section 17.8: to help the party out of the fight, not to join it, whichever party is at fault. Supplying a party with the means to fight is joining at one remove, because Section 17.19 counts the outside suppliers whose materials made an act possible among the nodes of the field that generated the act. A declaration of solidarity staged for an audience is a smaller form of the same act, because the declaration raises the share of attention the war captures.

What an ally owes in place of entry is a usable clutch. Chapter 16, Section 16.17 derives the clutch as the survivable unilateral exit from a Conflict: engaged from one side, without the other party's agreement and without a finding of fault, and available only where leaving does not cost the party that leaves more than staying does. "A clutch that is ruinous to engage is not a clutch but a locked transmission." A party to a war can stop fighting only where stopping does not cost its population its floors, its Individuals' lives, or more in Status than fighting on would cost. An ally makes the exit survivable by holding the floors of the party's population, by protecting its Individuals if the other party pursues them, and by carrying the Status cost of stopping, as a boxer's corner throws in the towel so that the boxer does not have to quit. Entry does the reverse. A party whose allies are in the war cannot stop alone, because stopping now requires every ally's agreement, and every death already suffered raises the cost of stopping. An ally that joins a war locks the transmission.

Once one party has stopped, what the other party does next is conduct, which every observer can see. A party that stops has declined to reciprocate violence and has conceded nothing (Section 17.15.3). If the other party stops too, or occupies and governs, the war was the loop between the parties, and the arithmetic of Section 17.15.1 and the correction of Section 17.19 govern the aftermath. If the other party goes on killing Individuals who have stopped, the conduct is an assault in progress rather than a Conflict, and protecting the Individuals under attack is repair of an open gate (Chapter 16, Section 16.9), which ends when the attack ends and is not entry into the war. The distinction is drawn from conduct, and it requires neither party's Belief to prevail.

17.20.5 Reversing the Polarity of Collective Defence

A mutual-defence commitment couples its members before any war begins. The North Atlantic Treaty provides that "an armed attack against one or more of them in Europe or North America shall be considered an attack against them all", and that each member, "in exercise of the right of individual or collective self-defence recognised by Article 51 of the Charter of the United Nations", will assist the member attacked by "such action as it deems necessary, including the use of armed force" (North Atlantic Treaty, 1949, art. 5). The trigger is the claim of self-defence, which the Self-Defence Orthogonality Principle shows every party to a war holds (Section 17.20.1). The response is decided by each member at the moment the alarm is greatest, which is the moment at which Phenomenological Collapse is strongest.

A commitment of that kind deters. Commitments that would bring allies to the defence of a target have lowered the probability that a dispute is initiated against it, and commitments promising that outside powers will stay out have raised it (Leeds, 2003). On Leeds's account the mechanism is information: an alliance tells a potential challenger how likely others are to respond, which is information about the certainty of a response. The same commitment wires in coupling that switches on at the trigger, so that the trigger and the crossing of the threshold become one event, with no point between them at which any member can engage the clutch. In 1914 Austria-Hungary declared war on Serbia on 28 July, and by 4 August Germany, Russia, France and Britain were at war. The formalism does not establish that any given trigger produces general war, and the chapter claims no inevitability in any specific instance (Section 17.21). What Appendix A establishes is the direction in which committed entry moves the network (Appendix A, Section A.5.7.5), and that the crossing, when it comes, is a fold and not an oscillation (the Saddle-Node Default, Appendix A, Sections A.5.3 and A.5.7.6).

The polarity of the commitment can be reversed without giving up the commitment. Under the reversed polarity, an attack on any member triggers Cascading Support and Cascading Consequences as the architecture derives them:

1. The Appetitive and Intrinsic floors of every population the war reaches are held, the attacking population's included (Section 17.19.2), in forms that an armed party cannot capture (Section 17.20.3).
2. The levers, materials and channels of the war are closed to every party still fighting it (Section 17.19.2), and those who decided on the war cannot place themselves or their holdings outside the field of consequence their decision created (Section 17.19.1).
3. Every consequence lifts automatically when the conduct stops (Section 17.19.3).
4. The response is conducted without theatre (Section 17.19.4).

Mutual-Support Commitment. A commitment under which an attack on any member triggers Cascading Support and Cascading Consequences as the architecture derives them, in place of entry into the war, and which excludes entry into combat, the supply of the means of fighting, any penalty that continues after the fighting stops, and the deprivation of the attacking population.

Under a Mutual-Support Commitment the claim of self-defence switches on nothing that can be used to fight. A staged or mistaken trigger buys support for a population and the closing of the means of war until the fighting stops, and a true trigger buys the same. The response is identical whichever party's Belief is accurate, which is what the Self-Defence Orthogonality Principle requires: not a test of the claim, but a commitment that needs none.

A Mutual-Support Commitment is not a promise to stay out, which Leeds (2003) finds invites attack. It is a commitment to a response as certain as a pledge to fight, delivered through the terms that keep a war local. Whether a certain response of that kind deters as well as a pledge to fight is a question the formalism does not settle (Criterion 19). A Mutual-Support Commitment carries two costs. A Mutual-Support Commitment does not prevent a determined attacker from taking territory; it makes the holding of territory untenable, by the route Sections 17.15.1 and 17.15.4 to 17.15.6 derive (Criteria 8 and 22). And a Mutual-Support Commitment decouples nuclear command from every member's border: a guarantee by which a nuclear power extends its arsenal over its allies couples a gating domain to each of their conflicts, which is the reverse of what Section 17.17.6 requires of a gating domain.

17.20.6 The Aggressor That Does Not Stop

An aggressor that takes one state and moves on to the next is the case the architecture predicts, not an exception to it. An acquisition raises the wanting for the Objects that come into range because the acquisition closed the distance to them (Reference-Set Reset, Appendix A, Section A.9.6), so a conquered state brings its neighbour into range. The Trajectory Futility Theorem establishes that no point on the trajectory exists at which the aggregate Mimetic drive comes to rest (Appendix A, Section A.9.10). At the apex of a hierarchy the one remaining route by which an apex holder's standing can rise is the lowering of the standing of others (Appendix A, Section A.9.9), and the subjugation of populations is that route at the scale of states.

Where the population shares the leader's Belief, and the Belief is a kind-claim that ranks human variation into natural kinds, the Belief has abandoned an available Referent, because human variation admits no such kinds (Section 17.11; the Diversity Referent, Chapter 14, Section 14.8.3.1.2). A Referent-abandoned Belief draws no maintenance Cost from reality, and the Cost is exported onto the Individuals the boundary excludes (Section 17.12). Every population an expanding regime takes adds to the Individuals the Belief excludes, and so adds to the Cost that must be exported to keep the Belief in place. Cascading Consequences close the export (Sections 17.19 and 17.19.1), returning the Cost to the field that generated it, and Cascading Support keeps each held population intact, provisioned and able to decline cooperation (Sections 17.15.1 and 17.15.4). The Cost of holding territory then grows with the territory held. Appendix A, Section A.9.9 finds the same direction in the apex strategy itself: stripping the populations below the apex of position compresses the Decoherence Time of the joint system and accelerates the joint system toward the Saddle-Node Default. The expansion moves the expanding regime toward its own fold. The rate at which the expansion does so is an open quantity (Appendix A, Section A.9.11).

The leader of such a regime cannot be reached by an intervention on the drive. Under Apex M-Capture the leader's Authenticity Belief drifts toward certainty, so that nothing inside the leader contests the act (Appendix A, Section A.5.8.5). The apex drive has no closure condition at any level of provision, so the apex locus cannot be disarmed by provision and can only be gated (the Dual-Locus Ignition Theorem, Appendix A, Section A.5.8.7). Within the leader's own architecture, the only component that could correct the drift is the component that drifted (Appendix A, Section A.5.9). Imposition requires two conditions together: the leader's Grievance Belief above its action threshold, and the leader's authority above the level at which the leader can act whatever the population believes (Appendix A, Section A.5.8.5). Only the second condition is open to intervention, and the second condition is held by others: the officers, officials and operators whose compliance converts the leader's order into an act (Section 17.7).

Two levers reach those Individuals. The first is their Beliefs. The Knowledge Patch reaches the Individuals around a captured leader, whose Hypothesis Spaces remain open to Evidence that the leader's does not admit (Chapter 18). The second is their exit. Under Forced Quiescence a regime routes Appetitive provision through loyalty, so that refusal costs an official's family its floors and the Opt-Out Paradox becomes acute (Section 17.6.2). Cascading Support that holds the floors of officials, soldiers and their families independently of the regime makes refusal survivable, which is the clutch made available to the apparatus. Section 17.7.6 finds that the refusals of subordinate operators have served as an ad hoc backstop, and that a backstop of that kind is not an architecture; Cascading Support held for the apparatus is what makes the backstop one. Taken with Cascading Consequences, Cascading Support presents every node of the generating field with the same gradient: continuing costs, and stopping is survivable.

A regime that burns out ends in a succession, and the Succession Decoherence Principle governs how violently: the violence of the transition tracks how completely the regime captured the institutions that could have managed it (Section 17.6.5). Section 17.6.5 reads the succession after Stalin as a case of near-total capture, resolved over years by the successive elimination of contenders. Cascading Support supplies the one buffer that capture cannot reach, because the provision Cascading Support supplies does not run through the regime: "independent provision can hold the A-Floor while authority is unsettled" (Section 17.6.5). Section 17.6.4 reads the Marshall Plan as the right intervention applied after the catastrophe it would have prevented. Cascading Support held through a war is the same intervention applied before the succession on which it bears. Holding the floors also lowers the selection pressure that arms the next captured apex (Appendix A, Section A.5.8.7).

The posture has a precedent. Kennan's argument for the containment of Soviet expansion called for "a long-term, patient but firm and vigilant containment of Russian expansive tendencies", which "has nothing to do with outward histrionics", with demands put "in such a manner as to leave the way open for a compliance not too detrimental to Russian prestige", and judged that Soviet power bore "within it the seeds of its own decay" (X, 1947). Kennan's containment applied "unalterable counter-force". Containment by Cascading Support and Cascading Consequences puts the two in the place of counter-force, and keeps the patience, the absence of theatre, and the open exit.

The long view is unsatisfying, and the architecture predicts that it will be. The demand that the leader be defeated and punished is the retributive surplus at the scale of states (Chapter 16, Section 16.9). Lowering another is the move that gains most standing for the effort (Appendix A, Section A.10.2), and declining the move is felt as failure; Section 17.15 derives the same verdict for the leader who declines to fight. The cost is real, and it is borne by Individuals. While the regime runs its course, the Individuals in the territory it holds bear its rule, and Cascading Support and the protection of Individuals under attack continue throughout without removing that cost. The comparison that supports the long view is with the alternative: total war against an apex whose authority gate admits an output whose consequence is irreversible, which Appendix A, Section A.5.8.7 identifies as the configuration in which the apex locus is most dangerous.

17.20.7 Why a War Ends Slowly

Wars in progress have grown longer. Between 1945 and 1999 civil wars began at a rate of about 2.3 a year with no trend up or down, and ended at only about 1.85 a year, so unresolved wars accumulated almost linearly, and the average duration of the civil wars in progress rose from about two years in 1947 to about fifteen years in 1999 (Fearon and Laitin, 2003). Over the same period the average length of civil wars roughly tripled (Hironaka, 2005). The lengthening is a property of civil and internationalised wars. Interstate wars between 1945 and 1999 had a median duration of under three months, while civil wars, which killed roughly five times as many, had a median duration of about six years (Fearon and Laitin, 2003).

The architecture gives the reason a war outlasts its causes. The collective Belief dynamics are bistable: below an interior threshold a perturbation corrects itself, and above it the population runs to the Collective Fixed Point (Appendix A, Section A.5.8.3). A bistable system held in place by positive feedback crosses back at a different point from the one at which it crossed over (Angeli et al., 2004), and Criterion 14 states the consequence for intervention: a trajectory that has accumulated coupling and concealment "must be brought below the conditions it crossed rather than back to them". A war produces the conditions that sustain it, because each act is Evidence for the next, and a war continues in conditions that could not have started it. A war ends when its sustaining conditions have been taken below that lower threshold: the fuel exhausted or closed off, the heat lowered, the coupling reduced. Recovery is restart rather than self-correction (Section 17.17.5).

The same structure explains why no single intervention suffices. A fire starved of oxygen while its heat and fuel remain accumulates both, and erupts when air returns. Suppressing the conduct of a war, or the speech of a population, while deprivation and the means of fighting remain is the same operation. Suppression of that kind stores the load rather than discharging it (Section 17.12.4), which is Forced Quiescence (Section 17.6.1), and the stored load arrives at once when the suppression lifts (Section 17.6.3). Fuel, oxygen and heat are worked on together, or the work on one stores what the others supply.

A society in crisis under Referent-abandoned Beliefs has the same structure without the violence. The Beliefs are held in place by the Cost they export and by the coupling of the population that holds them, and they decohere only as those conditions are removed, over a Decoherence Time that intervention can shorten and cannot abolish (Chapter 14, Section 14.8.3.1).

17.20.8 The Design Goal, Its Constraints, and What Calibration Must Supply

The goal is the indefinite deferral of terminal Mimetic crisis together with vast improvement over the status quo (Chapter 14, Section 14.11.3; Chapter 16, Section 16.12), and the two are pursued by the same interventions. A war is a local fold, and local folds that cannot be prevented are tolerated, provided none of them is allowed to become the terminal fold. Section 17.17.6 names three routes from a local fold to the terminal fold, and the Dual-Locus Ignition Theorem names the two places at which ignition originates (Appendix A, Section A.5.8.7). Each gives one constraint, and the delivery of Cascading Support gives a sixth.

1. No connected chain. The coupling of a war to the populations around it is kept low: no entry by allies, no supply of the means of fighting, no commitment that enters members into one another's wars, and no theatre (Sections 17.20.4 and 17.20.5).
2. Boundaries maintained for as long as the war runs. The second route is the sequential decoherence of the boundaries that separate a war from its neighbours, and a long war gives that route time. A long war is tolerable only while the boundaries around it are actively maintained, and this is the one way in which the duration of a war bears on the terminal fold.
3. Nuclear command never coupled to a war. A gating domain is held at the lowest attainable Coupling Strength and the shortest attainable Decoherence Time (Section 17.17.6), which excludes commitments that extend nuclear command over the borders of allies (Section 17.20.5).
4. Floors held for every Individual. The floor locus is disarmed by holding the deficit of the worst-off below its critical depth, inside the war and in the attacking population alike, in forms that cannot be converted into fighting (Section 17.20.3).
5. Authority gated. The apex locus cannot be disarmed and can only be gated, and from outside a regime the gate is reached through the Beliefs and the survivable exit of the Individuals whose compliance converts orders into acts (Section 17.20.6).
6. Support delivered past the regime. Cascading Support reaches Individuals through channels the generating regime does not control, and is delivered unarmed, declared in advance and open to inspection, so that stopping it requires an act every observer can see (Section 17.15.6).

Under these constraints it is the coupling of a war, and not its duration, that threatens the terminal fold. A long war that meets the constraints remains a local fold, and a short war that breaches the third can end everything.

No intervention meets its constraint alone, and none meets it exactly. The clutch is "an ideal to be approximated rather than a binary" (Chapter 16, Section 16.17); restrictions on the means of war leak; floors are approached and never perfectly held. The design rests on several imperfect barriers held together, which is the logic Section 17.17.6 applies to the gating domain, disciplining its executive act "through many independent Belief-domains rather than concentrating it in one". Prevention remains the first course and the cheaper one, because the threshold a war must be brought back below is lower than the threshold it crossed (Section 17.20.7; Criterion 14).

What the formalism establishes and what it leaves to measurement divide cleanly. Appendix A establishes, qualitatively, that the thresholds exist and that crossing them is a fold rather than an oscillation (Appendix A, Sections A.5.3 and A.5.7.6); that the collective dynamics are bistable (Appendix A, Section A.5.8.3); the direction in which each intervention moves the system relative to the network threshold (Appendix A, Section A.5.7.5) and relative to the floor and apex conditions (Appendix A, Section A.5.8.7); and that the apex drive has no point of rest (Appendix A, Section A.9.10). Appendix A does not establish the size of any of these quantities in a real population of states. The critical coupling of a real network waits on the spectral estimation of its Evidence couplings and its Observability, and on the identification of populations at or near the threshold (Appendix A, Section A.5.7.8). The critical depth of the floor deficit waits on the calibration Appendix A, Section A.5.8.7 assigns to specialist work. The rate at which an expanding regime compresses its own Decoherence Time is open (Appendix A, Section A.9.11). Nothing in the formalism gives the time a war takes to burn out under these interventions, the size of the deterrent effect of a Mutual-Support Commitment, the rate at which restrictions on the means of war leak, or the time an occupation takes to end under non-cooperation and Cascading Support (Section 17.15.5). Every claim about a war that has begun therefore holds as a claim of direction, and every application of the design is also a measurement whose results calibrate the next application. Criteria 18 to 22 state the regime claims on which the design is exposed.

The derivation tells a designer which way to push. Only measurement tells the designer how hard, and for how long.

17.21 What This Chapter Does Not Claim

It does not claim all violence is mimetically driven. Genuine A-defence (protecting bodily integrity against immediate attack) exists and is architecturally distinct. The chapter concerns violence that emerges from the Scapegoat Mechanism (which accounts for most collective violence historically), not all violence categorically.

It does not claim perpetrators are not responsible. Architectural explanation is not excuse. This is the Non-Malice Corollary (Chapter 11): the Scapegoat Mechanism operates through normal cognitive architecture rather than individual malice, but architectural derivation explains mechanism without eliminating responsibility. The mechanisms operate through Individual Decision Hubs, and I-Override capacity (Chapter 9) remains available. That violence is architecturally predictable does not mean it is architecturally compelled. Responsibility, consequence, and prevention remain coherent concepts.

I-Override and Self-Application. I-Override, operating on an expanded Hypothesis Space provided by the Knowledge Patch (Chapter 18), is the Individual’s primary defence against the Confabulation Cascade. The Presumption Against Scapegoating (Chapter 11) operationalises this defence: where Scapegoat attribution is proposed, the attribution is presumed Confabulated until demonstrated otherwise through the Scapegoat Effectiveness Battery.

It does not claim that Scapegoat targeting is always based on Confabulated attribution. Section 17.14 establishes that in reciprocal conflict between roughly matched adversaries, both sides may hold locally accurate attributions while the structural process producing violence remains identical. The canonical diagnostic tools remain valid for paradigmatic scapegoating but lose traction in the matched-adversary case (Section 17.14.2).

It does not claim that self-defence is impermissible. Section 17.15 establishes that Individual bodily self-defence under genuine A-threat is architecturally permitted. The chapter’s claim about the leadership decision (the deaths of a war are certain, and those that declining risks are neither certain nor beyond intervention) applies at the leadership level, not to the Individual soldier under fire.

It does not claim that a war can be ended quickly, or by any single intervention. Section 17.20 derives the direction in which each intervention moves a war and the constraints under which a war stays local. It derives neither the time a war takes to end nor the size of any intervention's effect, and each of those waits on calibration (Section 17.20.8).

It does not claim that Individuals under attack should not defend themselves. The interventions of Section 17.20 are interventions by those outside a war. None asks anything of the Individuals inside it.

It does not claim that a Mutual-Support Commitment prevents the taking of territory. Section 17.20.5 derives that a Mutual-Support Commitment makes the holding of territory untenable and keeps a war local. Whether it deters initiation as well as a pledge to fight is open (Criterion 19).

It does not claim violence is inevitable in any specific instance. The architecture produces violence probabilistically when parameters align. Specific wars, pogroms, and genocides could have been prevented through parameter-shifts that maintained Subcriticality. The chapter establishes mechanism, not fatalism.

It does not claim the current trajectory is irreversible. The Symmetry of Crisis and Prevention established in Chapter 13 (that the same Bayesian machinery enabling crisis also enables prevention) applies here. The Knowledge Patch can shift parameters. Structural intervention can restore damping factors. Subcriticality Maintenance remains achievable.

It does not claim that no True Beliefs are formed during or after conflict. The Truth-Exclusion Principle (Section 17.16) is specific: the violence itself contributed nothing to whatever Content-level facts were established, and the distinction between Content-level findings (achievable by post-war investigation) and structural findings (which required penetrating the Camouflage Architecture) is architecturally precise.

It does not claim that every human-rights violation is licensed by a Referent-abandoned Belief. Section 17.12 establishes the negative result only: because the instruments record the Referent, no licensing Belief can occupy the Referent-free condition. Which licensing Beliefs fall in the Referent-abandoned cell is settled case by case, and the eliminative proof fires only where the Appetitive-Deficiency Test and the Reversal Counterfactual close the A-channel. A prediction about a method's efficacy that has not yet decohered, and a claim that concedes the facts and asserts only a normative exclusion, are both outside the result.

It does not claim that the human-rights instruments should be weakened, narrowed, or abandoned. The opposite follows. The instruments are the record that makes the licensing Belief abandoned rather than Referent-free, and the corrective available in that cell exists only because the Referent has been recorded and made available. What Section 17.12 identifies as failing is the theory of constraint built on the record, not the record.

It does not claim that punishment accomplishes nothing. Sanction re-internalises a fraction of the exported maintenance Cost onto the holder, and holder-borne Cost is the quantity that forces correction. The claim is quantitative: the fraction is small, episodic, delayed, and improbable, set against a continuous and total export.

It does not claim a violation rate, or that any given violation was compelled. Section 17.12 derives a structural predisposition and the quantities that govern it. It predicts neither the magnitude of the breach rate nor the occurrence of any particular breach, consistently with the probabilistic character of every other prediction in this chapter.

It does not claim that domain-local containment protects against every collapse. The migration-firebreak result (Section 17.17.6) is explicitly qualified: it is a qualitative resistance property that holds only where every Belief-domain's output is Belief-mediated, so that a fold's relocated load must cross a low-coupling membrane at which it can dissipate. It offers no protection where a domain's output is physical and total rather than Belief-mediated: the boundary-independent-output exception. The claim is that the foam contains Belief-mediated collapses domain-locally, not that it contains catastrophic physical output. Nor does the containment result claim that isolating a gating domain protects it: for a boundary-independent output the protective condition is not low cross-domain coupling but low self-coupling of the executive act, achieved by disciplining that act through many independent Belief-domains rather than concentrating it in one (Section 17.17.6).

17.22 Transition to Structural Intervention

The architecture produces violence. This is derivation, not speculation. The next chapter addresses the first intervention: the Knowledge Patch.

Chapter 13 established that prevention is mathematically possible (the same Bayesian machinery that enables Cascade dynamics enables correction). Chapter 18 specifies what this initial intervention is and how it operates.

The Knowledge Patch is not a cure, nor is it sufficient alone for indefinite Subcriticality Maintenance. M-dynamics cannot be eliminated; they are constitutive of human cultural capacity (the Mimetic Trade-Off Principle). Violence cannot be permanently prevented; the architecture persists. What the Knowledge Patch provides is novel disruption capacity: a means of interrupting the Scapegoat Mechanism’s pathway that was previously unavailable because the Vocabulary Gap (Layer 7 of the Camouflage Architecture) ensured that the conceptual vocabulary required for recognition did not exist.

But recognition alone, even at population scale, cannot maintain Subcriticality indefinitely. The structural conditions that amplify M-dynamics (financial architecture, visibility infrastructure, institutional design, positional-good markets) continue generating pressure that can overwhelm individual recognition capacity. Indefinite Subcriticality Maintenance requires:

1. The Knowledge Patch (Chapter 18): Introducing the vocabulary and diagnostic tools that enable I-Override deployment against M-capture, including the Two-Question Diagnostic, the Selection-Framing Coincidence Test, and the Presumption Against Scapegoating
2. Structural M-Management: Institutional architecture that reduces baseline M-intensity, developed across the book’s mechanisms and derived in institutional form in Chapter 20
3. A-Sufficiency Universalisation: Guaranteed provision that prevents A-desperation from amplifying M-susceptibility, ensuring that Individuals can meet Appetitive needs without M-participation
4. I-Infrastructure Protection: Institutional support for Intrinsic Motivation domains that provide alternative value-sources and reduce M-Salience across the population
5. Observability Suppression: Constraints on visibility infrastructure that limit Confabulation Cascade synchronisation, treating observability as a design variable calibrated to reduce gratuitous Mimetic activation rather than maximise engagement
6. Opt-Out Preservation: Ensuring that Individuals can refuse M-participation without forfeiting A-Sufficiency or I-access, maintaining structural alternatives to M-intensive pathways at every level where M-escalation might otherwise lock Individuals into participation, and the capacity to leave a Reference Set's observation for a provisioned interval long enough for the Reference Set's priors to release (Chapter 20, Section 20.8)
7. Authority Distribution: Distributing command authority, information control, appointment power and Appetitive leverage so that no single node can compress the decision interval below what correction requires (Section 17.7.6)

Four of these seven are the Four Dimensions of M-Management (A-Sufficiency Universalisation, I-Infrastructure Protection, Observability Suppression, Opt-Out Preservation), the institutional design levers through which a society damps destructive Mimetic escalation without attempting to eliminate M; the Knowledge Patch is their prerequisite, and structural M-Management is the institutional implementation of all four working in concert.

The remaining chapters develop these structural requirements, and Chapter 20 derives the institutional form sustained Subcriticality requires.

The Knowledge Patch comes first not because it is sufficient but because it is prerequisite. Without the vocabulary that enables M-recognition, structural interventions face systematic resistance from Individuals who cannot recognise their own M-capture, including the leaders and populations whose support structural reform requires. This is the Dissemination Precedence Principle: institutional reform toward AIM-aligned flourishing requires prior or concurrent population-level acquisition of AIM conceptual vocabulary. Reform proposals that outpace conceptual dissemination will be experienced as elite imposition and will trigger mimetic resistance.

The Knowledge Patch is the wedge that creates the space.

Chapter 19 (The Dissolution of Individualism and the Derivation of Relationalism) then draws the philosophical consequences, establishing that the five canonical premises of Individualism are each falsified by the architecture this chapter has applied, and that Relationalism follows as entailed consequence rather than alternative philosophical commitment.

---

Establishment Summary

This chapter has established that violence, from interpersonal aggression through organised warfare to genocidal extermination, is the predictable behavioural output of the AIM architecture rather than a pathological deviation from it. It stated the Unifying Crisis Principle, from which the chapter's three domain cases follow: that the Terminal Phase of the Crisis State Architecture is the configuration toward which an uncorrected population moves under its own dynamics, presenting as market collapse where the inflated Objects are priced, as institutional failure where they are administered, and as open violence where they are attributed to an agent. The Confabulation Cascade that converges a population on false Grievance Beliefs does not terminate in Belief; it terminates in action, and when that action targets the human beings Confabulation has identified as the cause of suffering, violence is the architectural terminus of the Scapegoat Mechanism.

It derived the apex route by which a single Individual holding four institutional capacities can impose crisis whatever the collective posterior, and paired that route with the floor route as the Dual-Locus Ignition Theorem, under which the capacity to ignite is concentrated at the two extremes of the distribution of provision and authority and vanishes between them, so that a society can provision its way out of one locus and can only govern its way out of the other. It derived the Trigger-Indifference Corollary: once an M-Captured Individual holds authority above the threshold over a polity in Forced Quiescence, neither waiting nor removing the leader averts the discharge, the event that sets the discharge off decides only whether the discharge takes the form of imposed crisis or of the succession case, and how violent and how long the discharge is are set by the structure the discharge meets. The chapter showed this output to be scale-invariant, experienced as defensive, content-independent, and insulated by duty, stated the Structural Determination Principle, that what an arrangement produces is fixed by its structure and that Content and intent bear on the output only by setting structural parameters, and it established the Leadership Asymmetry that governs the arithmetic of lives once a population approaches the Subcriticality Threshold. It distinguished the refusal to reciprocate violence from concession: refusal concedes nothing, continues as non-violent action, and is made survivable by Cascading Support delivered through channels the aggressor does not control, while Forced Quiescence remains within what an aggressor can impose.

It further established that declaring a right does not govern the Belief that licenses its violation: the instruments record the Referent, which forecloses the Referent-free condition and makes the licensing Belief correctable, but a recorded Referent that does not govern leaves the Belief in the cell whose hardness is supplied by the Cost exported onto the excluded, so that routine breach is the predicted behaviour of a regime constituted around prohibition and sanction. The rate is reducible by intervention on Coupling Strength, on who bears that Cost, and on what it costs a holder to put the Belief down, and the chapter closes by taking up the Field-Correction Obligation at collective scale, deriving from it that a correction must close the generating field against export, hold Appetitive and Intrinsic floors non-derogably throughout, lift on cessation, and be conducted without theatre.

It then took up a war that has begun. By the Self-Defence Orthogonality Principle, the Belief that starts and sustains a war is held on every side whatever the facts, carries no Evidence about which party attacked, and selects no intervention; every intervention is selected by conduct and need instead. A war is sustained by its fuel, its oxygen, its heat and the loop between the parties, and Cascading Consequences, Low-Theatre conduct, Cascading Support and the clutch act on each in turn, as targeted rather than neutral or partisan interventions. Joining a war moves the network that carries the war toward its threshold, so the obligation of allies is to keep the clutch usable rather than to enter, and a Mutual-Support Commitment answers an attack with Cascading Support and Cascading Consequences rather than with entry. The design goal is the indefinite deferral of the terminal fold; its constraints follow from the three routes to that fold and the two loci of ignition, and the derivation fixes their direction while their magnitudes wait on calibration.

17.23 Axiomatic Recap

This chapter’s derivation invoked the following axioms:

• Axiom 2 (Common-Currency Integration): Delivers a single signed Scalar with the source labels destroyed, so the Decision Hub cannot read Mimetic inflation as Mimetic and integrates it exactly as it would integrate what it resembles (Section 17.2).
• Axiom 3 (Source Opacity): Prevents recognition of M-contamination in threat assessment, ensuring violence is experienced as defence.
• Axiom 4 (Confabulation): Generates agent-based causal narratives that feel accurate and provide cognitive closure.
• Axiom 5 (M as Amplifier) with Axiom 6 (Differential Satiation Dynamics): Object Parasitism gives the Mimetic channel no Object of its own and Differential Satiation Dynamics gives it no closure condition, which is why the inflation the Unifying Crisis Principle turns on has no ceiling (Section 17.2), why an apex drive has no level of provision at which it closes (Section 17.7.8), and why satisfaction is closed as a route around the discharge of the terminal configuration (Section 17.7.10).
• The Flourishing Constraint (derived from Axiom 5, M as Amplifier): Establishes that flourishing is possible only among living Individuals, grounding the Leadership Asymmetry (Section 17.15).
• The Transmission Window (from Axiom 7, Preconscious Transmission): The 60-340ms integration window forecloses conscious correction before M-urgency becomes survival-level alarm, critical to the nuclear decision timeline (Section 17.17.1).
• Axiom 8 (Bayesian Belief Dynamics): Content-agnostic updating produces convergent false Belief regardless of what Beliefs are converging, grounding Content-independence (Section 17.11), the Structural Determination Principle (Section 17.14.3), the Truth-Exclusion Principle (Section 17.16), and the Self-Defence Orthogonality Principle (Section 17.20.1).
• Axiom 9 (Agent-Based Causal Default): Completes causal chains with agent-based attribution when true causes exceed computational tractability, producing Scapegoat candidates (Section 17.1, Step 2).

17.24 What Would Falsify This?

The claims in this chapter would require significant revision if:

Criterion 1: Sincere defensive phenomenology failure. If systematic investigation revealed that perpetrators of collective violence do not experience themselves as defenders, that aggressors reliably recognise their aggression as such during the act (not merely in retrospect, after the Confabulation Cascade has been disrupted by external intervention), the Phenomenological Collapse mechanism would be disconfirmed. The prediction is specific: private testimony (diaries, private correspondence, documented internal deliberations) from leaders and participants in collective violence should show sincere conviction of defensive necessity, not cynical awareness of aggression. If the historical record consistently showed that perpetrators privately acknowledged aggressive intent while publicly claiming defence, the architectural account would require revision toward a manipulation model rather than a Confabulation model.

Criterion 2: Content-dependent violence. If collective violence were shown to be systematically predicted by the identity of the target group rather than by the structural parameters (κ-coupling, A-precarity, I-domain collapse, authority concentration, observability infrastructure), the Content-independence claim would be falsified. The prediction is that structural parameters should predict crisis occurrence better than Content variables (which specific group is targeted, which specific ideology is deployed). If a model using only Content variables (ethnic composition, religious demography, ideological movements) outperformed a model using only structural parameters in predicting collective violence onset, the architectural account would require revision.

Criterion 3: Proportionality constraint effectiveness. If proportionality assessment were shown to reliably constrain violence independently of the assessor’s motivational state, if wars whose initiators conducted proportionality analysis showed systematically lower harm than wars whose initiators did not, the incoherence claim would be falsified. The prediction is that proportionality assessment should show no constraining effect after controlling for the assessor’s pre-existing motivational configuration, because the assessment is performed by the same system whose M-capture determines its output.

Criterion 4: Buffer-independent violence prevention. If populations with depleted A-I Buffers (high A-precarity, collapsed I-domains, concentrated authority, homogeneous information environments) were shown to resist cascade-to-violence as effectively as populations with deep A-I Buffers, the race condition framework would be falsified. The prediction is that A-I Buffer depth should predict time-to-crisis: deeper buffers, longer resistance; shallower buffers, faster progression to violence. If buffer depth showed no predictive relationship to crisis velocity, the damping-factor analysis would require revision.

Criterion 5: Violence-Truth correlation. If the outcomes of collective violence were shown to systematically correlate with the accuracy of the causal attributions that produced the conflict (the side with more accurate Beliefs about why the conflict arose reliably prevails, controlling for capability asymmetries), the Truth-Exclusion Principle would be falsified. The prediction is that military outcome should correlate with capability variables (resources, technology, strategy, geography, demographics) and show no residual correlation with attribution accuracy after capability variables are controlled. If attribution accuracy independently predicts military outcome, the orthogonality claim requires revision.

Similarly, if post-conflict populations were shown to acquire structural understanding of crisis dynamics (not merely Content-lessons about specific enemies or specific weapons, but structural understanding of the Confabulation Cascade mechanism, Source Opacity, or the Agent-Based Causal Default) through the experience of violence itself (not through independent scholarly investigation conducted after the violence, and not through provision of the AIM conceptual vocabulary from an external source), the claim that violence is architecturally inert with respect to Truth would require revision. The prediction is specific: post-conflict “lessons learned” should be overwhelmingly Content-bound (targeting specific enemies, specific weapons, specific ideologies) rather than structure-bound (targeting the Confabulation Cascade mechanism, Source Opacity, or the Agent-Based Causal Default). If systematic post-conflict structural learning were documented at population scale without independent provision of structural vocabulary, both the Content Mutation Imperative and the Truth-Exclusion Principle would require revision, and the Camouflage Architecture’s claim to have blocked structural diagnosis would be falsified.

Criterion 6: Duty-dependence on attribution accuracy. If duty-constituted Self-Model elements were shown to produce structurally different violence depending on the accuracy of the underlying Beliefs (i.e., "good leaders" who are genuinely attacked produce measurably different conflict dynamics than "cynical leaders" who manufactured the crisis, after controlling for capability differences), the structural-irrelevance-of-leader-quality claim would require revision. The prediction is that conflict dynamics (escalation rate, population-level Bayesian Lock-In velocity, duration, violence intensity) should show no residual correlation with the accuracy of the initial attribution after controlling for capability and structural parameters (Coupling Strength, A‑precarity, I‑domain availability, authority concentration). If genuine-defence conflicts show systematically different structural dynamics from manufactured-pretext conflicts (not merely different narratives or different moral assessments, but different structural parameters governing cascade velocity, Lock-In depth, or violence output intensity), the content-independence of the violence pathway would be weakened and the duty-as-insulation mechanism would require reformulation.

Criterion 7: Coerced compliance threshold. If population-level violence were shown to require majority conviction (rather than merely sufficient conviction among those controlling the coercive apparatus), the sub-majority capture claim would be falsified. The prediction is that historical cases of population-level violence should include instances where the initiating faction held well below 50% popular support at the point of conflict initiation, and that the critical variable predicting whether a faction can impose violence on the entire population should be control of coercive instruments (military command, police, judiciary) rather than breadth of popular conviction. If systematic analysis showed that collective violence occurs only when a genuine majority of the population holds sincere Confabulated Beliefs about the Scapegoat (rather than a sub-majority that controls the state apparatus imposing participation on the rest), the coerced compliance mechanism would require reformulation and the critical threshold for cascade-to-violence would need upward revision.

Criterion 8: Irrecoverability of conquest. If populations that declined to fight were shown to experience systematically worse A-I outcomes (higher mortality, lower A-sufficiency, lower I-access) than populations that fought wars of equivalent intensity, after controlling for third-party structural intervention, the recoverability-of-conquest claim would require revision. The prediction is that populations that declined to fight and received sustained external Cascading Support (A-floor maintenance, I-domain protection, Cascading Consequences applied to the occupier) should show better A-I outcomes than populations that fought, because the war itself produces A-harm (deaths, infrastructure destruction, trauma) that declining to fight avoids. Deaths the aggressor inflicts on a population after it has declined to fight are counted against that population's outcomes, not excluded from them. If conquest with structural support produces worse A-I outcomes than war, the Leadership Asymmetry analysis would be weakened and the claim that the ethical path minimises harm would require reformulation. The comparison must control for third-party intervention: the claim is not that unassisted refusal produces better outcomes than war, but that refusal combined with AIM-aligned structural intervention does.

Criterion 9: Sanction-responsiveness of violation rates. If the rate of human-rights violation across jurisdictions were shown to respond principally to the severity or availability of sanction, rather than to the permeability of the excluded category and the distribution of the licensing Belief's maintenance Cost, the analysis of Section 17.12 would be falsified. The prediction is that after controlling for state capacity and conflict exposure, violation rates should track boundary-permeability measures (rates of movement across the excluded category, the material Cost imposed on non-fit) and A-sufficiency provision to the excluded group, and should show little residual response to sanction severity, while retaining some response to sanction certainty consistent with partial re-internalisation of exported Cost. If severity of sanction independently predicted violation rates once permeability and Cost-distribution were controlled, the claim that prohibition-plus-punishment is not the corrective for this cell would require revision.

Criterion 10: Export-independence of boundary hardness. If the hardness of a Boundary-Belief were shown to be independent of the maintenance Cost borne by the excluded party, the application of the Boundary-Hardening Principle to this case would be falsified. The prediction is that provision to an excluded group of the material and developmental goods the boundary denies them should be followed by measurable softening of the boundary, observable as increased rates of reclassification across it and reduced Coupling Strength within the enclosing population, in advance of any change in the licensing Belief's content. If the Cost borne by the excluded group could be relieved without any accompanying change in boundary permeability or coupling, the identification of Cost-export with boundary-hardening would not hold at this scale and the intervention analysis of Section 17.12.5 would require reformulation.

Criterion 11: Continuous discharge under suppression. If populations in which a discriminatory licensing Belief was suppressed in conduct but left uncorrected in content were shown to discharge that Belief gradually rather than discontinuously, the stored-load claim would be falsified. The prediction is that suppression without correction should produce the accumulation signature: low observed violation during the suppression interval, followed by a discontinuous release whose magnitude increases with the duration of suppression. If long suppression intervals were systematically followed by progressive moderation rather than by discontinuous release, the application of the Saddle-Node Default to this case would require revision.

Criterion 12: Phase-sequence correlation. If historical crises were shown to reach catastrophic reconfiguration without passing through the phase sequence the Crisis State Architecture specifies, the Unifying Crisis Principle stated at Section 17.2 would be falsified. The prediction is that a crisis trajectory should be reconstructible as an ordered sequence: an identifiable exogenous trigger opening the drift, convergence on a Scapegoat at the drift-to-calcified boundary, institutional capture and legal restriction through the calcified phase, and escalation to violence at the terminal phase. If catastrophic outcomes were shown to arrive without the antecedent sequence, or if the sequence ran in a different order across cases, the attractor account would require revision toward a model in which crisis is triggered rather than approached.

Criterion 13: The shape of collective convergence. If collective Belief in a Scapegoat explanation were shown to rise in proportion to exposure rather than accelerating as adoption accumulates, the compounding mechanism on which the Unifying Crisis Principle rests would be falsified. The prediction is specific to the shape of the curve rather than to its endpoint: where the prevalence of a Scapegoat Belief can be measured over time through media coding, survey series or archived correspondence, it should show slow initial growth followed by rapid convergence, because each adopter becomes Evidence for the next. A rise proportional to exposure, or one tracking only the intensity of the originating shock, would place the convergence outside the Bayesian account and require the Unifying Crisis Principle's population claim to be reformulated.

Criterion 14: Phase-independence of intervention. If interventions applied late in a crisis trajectory were shown to be as effective as the same interventions applied early, the attractor claim would be falsified. The prediction is that intervention effectiveness should decline with phase: measures applied while the population is still grounded or drifting should outperform identical measures applied once institutional capture is in place, because a trajectory that has accumulated coupling and concealment must be brought below the conditions it crossed rather than back to them. If effectiveness showed no phase-dependence, the account of crisis as an approach toward an attractor would require revision, and the case for early structural maintenance over late correction would lose its architectural basis.

Criterion 15: Population-level sufficiency under concentrated authority. If populations governed under the four capacities named at Section 17.7.2 were shown to be reliably protected against imposed crisis by population-level intervention alone, the Leadership Compression Principle would be falsified. The prediction is that imposition should track the concentration of the four capacities and show little residual response, once that concentration is controlled, to the depth of Knowledge Patch penetration, the adequacy of Appetitive floors, or the presence of visibility governance. If a measure of population-level intervention independently predicted the absence of imposed crisis at high capacity concentration, the Leadership Compression Principle would require revision toward a model in which population-scale correction can reach an apex decision after all.

Criterion 16: Mid-distribution ignition. If ignition of the collective transition were shown to originate reliably from the middle of the distribution of provision and authority, at Individuals neither in sustained Appetitive deficit nor holding authority above the threshold, the two-loci concentration of Section 17.7.8 would be falsified. The prediction is that traced back to its origin, an ignition should resolve either to a worst-off segment at critical depth or to an apex holder above the authority gate, and not to a provisioned Individual without institutional capacity. If a substantial class of ignitions resolved to the middle, the claim that ignition capacity vanishes between the extremes would require revision, and with it the case for locating intervention at provision and at the authority gate rather than across the population.

Criterion 17: Provision reaching the apex drive. If improved provision to an apex holder were shown to reduce the probability of imposition, the non-satiability of the apex drive would be falsified. The prediction follows from Differential Satiation Dynamics and is uncomfortable for the framework precisely because it is testable: no level of Appetitive or Intrinsic provision supplied to an Individual above the authority gate should reduce imposition, because there is no deficit beneath the drive for provision to close. If it did reduce imposition, the apex condition would have a second controllable term besides authority, the claim that the apex can only be gated and never zeroed would not hold, and the architectural requirements of Section 17.7.7 would no longer be the only operations available on that route.

Criterion 18: Targeted against neutral and partisan intervention (Tier 3). The claim at risk is that interventions selected by conduct and need, closing the means of war to every party still fighting and holding floors in forms armed parties cannot capture, shorten wars in progress and lower their recurrence relative to both neutral and partisan interventions (Section 17.20.2). The disconfirming observation is that wars receiving interventions of the targeted class, coded by what the intervention acts on rather than by the party it favours, last as long as or longer than comparable wars receiving neutral or partisan interventions, or recur as often. The observation counts if it comes from a hazard analysis across many conflicts that controls for state capacity, conflict intensity, terrain, and the selection of conflicts into intervention, and if it survives alternative codings of the targeted class. If met, the claim that targeting by condition outperforms targeting by side fails, and the design would rest on the floors' protection of life alone. The Self-Defence Orthogonality Principle (Section 17.20.1) and the architecture beneath it are untouched.

Criterion 19: Deterrence by a Mutual-Support Commitment (Tier 3). The claim at risk is that a Mutual-Support Commitment (Section 17.20.5), a commitment to a certain non-combat response, deters the initiation of disputes against its members comparably to a pledge to fight. The disconfirming observation is that members of such commitments are targeted at rates comparable to states covered only by promises of non-intervention, and markedly above states covered by defensive pledges. The observation counts if it comes from a dyadic analysis of dispute initiation that controls for capability, contiguity, prior rivalry and regime type, with commitments coded by the response they specify, and if the number of covered dyads is large enough to separate the three classes. If met, the deterrent limb of Section 17.20.5 fails and a Mutual-Support Commitment would buy containment at the price of more initiations. The result that committed entry moves the network toward its threshold (Section 17.20.4) is untouched.

Criterion 20: Floors, mortality and recurrence (Tier 3). The claim at risk is that holding floors inside a war, in forms armed parties cannot capture, lowers total mortality and the rate of recurrence even where it lengthens the war (Section 17.20.3). The disconfirming observation is that conflicts in which provision reached Individuals directly show total mortality, direct and indirect together, or recurrence rates at or above those of comparable conflicts in which provision did not. The observation counts if mortality is measured as excess over a stated baseline, if capturable and direct provision are separated by design rather than by assumption, and if the comparison controls for intensity and duration. If met, the claim that floors lower the sequence of wars fails, and the instantiation of floor provision in wartime would need revision. The floor locus of the Dual-Locus Ignition Theorem, a formal result, is untouched.

Criterion 21: Gating authority by survivable refusal (Tier 3). The claim at risk is that holding the floors of officials, soldiers and their families independently of a regime raises the rate of refusal and lowers the regime's capacity to impose (Section 17.20.6). The disconfirming observation is that, among regimes of comparable capture, those whose apparatus had credible access to provision independent of the regime show no higher rates of refusal, desertion or defection, and no lower capacity to carry out orders, than those whose apparatus did not. The observation counts if it comes from comparative cases coded before outcomes are known, controlling for the severity of repression and the state of the war. If met, the route by which authority is gated from outside a regime fails, and Section 17.20.6 would rest on the Knowledge Patch alone. The Dual-Locus Ignition Theorem's finding that the apex locus can only be gated is untouched.

Criterion 22: Refusal, non-cooperation and independent channels (Tier 3). The claim at risk is that a population that declines to fight and sustains non-cooperation, with Cascading Support reaching it through channels the occupier does not control, raises the cost of holding it and shortens the occupation (Sections 17.15.4 to 17.15.6). The disconfirming observation is that, among occupations of comparable capacity and repression, those in which the occupied population sustained non-cooperation and received provision through channels the occupier did not control show no higher cost of holding, measured by the forces and administration the occupier must commit per head of the occupied population and by the rate of defection from its apparatus, and no shorter duration, than those in which it did not. The observation counts if the cases are coded before their outcomes are known, and if the comparison controls for the occupier's capacity, the severity of its repression, the regime type of its home society, and its standing among other states. If met, the claim that refusal followed by non-violent action makes the holding of territory untenable fails, and Section 17.15.1 would rest on the arithmetic of war deaths alone. The arithmetic itself, that the deaths of a war are certain, is untouched.

These criteria are not already established by the phenomena this chapter explains. They remain open empirical questions that systematic historical coding, private-testimony analysis, and comparative crisis research could address.

The independence of Harm from the offender's state of mind, given the conduct the Individual harmed receives and reads (Section 17.14.3), is Tier 1 and is stated with the third commitment at Appendix B, Section B.2.

References

Allison, G., Coté, O. R., Falkenrath, R. A., & Miller, S. E. (1996). Avoiding Nuclear Anarchy: Containing the Threat of Loose Russian Nuclear Weapons and Fissile Material. MIT Press.

Angeli, D., Ferrell, J. E., Jr., & Sontag, E. D. (2004). Detection of multistability, bifurcations, and hysteresis in a large class of biological positive-feedback systems. Proceedings of the National Academy of Sciences, 101(7), 1822–1827. https://doi.org/10.1073/pnas.0308265100

Bury, T. M., Bauch, C. T., & Anand, M. (2020). Detecting and distinguishing tipping points using spectral early warning signals. Journal of the Royal Society Interface, 17(170), 20200482.

Evans, R. J. (2003). The Coming of the Third Reich. Allen Lane.

Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, insurgency, and civil war. American Political Science Review, 97(1), 75–90. https://doi.org/10.1017/S0003055403000534

Geneva Convention (IV) relative to the Protection of Civilian Persons in Time of War, 12 August 1949, 75 UNTS 287, art. 59.

Geneva Declaration Secretariat. (2008). Global burden of armed violence. Geneva Declaration Secretariat.

Hafner-Burton, E. M., & Tsutsui, K. (2005). Human rights in a globalizing world: The paradox of empty promises. American Journal of Sociology, 110(5), 1373–1411.

Harff, B. (2003). No lessons learned from the Holocaust? Assessing risks of genocide and political mass murder since 1955. American Political Science Review, 97(1), 57–73.

Hathaway, O. A. (2002). Do human rights treaties make a difference? Yale Law Journal, 111(8), 1935–2042.

Hironaka, A. (2005). Neverending wars: The international community, weak states, and the perpetuation of civil war. Harvard University Press.

King, M. L., Jr. (1958). Stride toward freedom: The Montgomery story. Harper & Brothers.

Leeds, B. A. (2003). Do alliances deter aggression? The influence of military alliances on the initiation of militarized interstate disputes. American Journal of Political Science, 47(3), 427–439. https://doi.org/10.1111/1540-5907.00031

Lefebvre, G. (1971). The Coming of the French Revolution (R. R. Palmer, Trans.). Princeton University Press. (Original work published 1939)

Mälksoo, L. (2003). Illegal annexation and state continuity: The case of the incorporation of the Baltic States by the USSR. Martinus Nijhoff. https://doi.org/10.1163/9789004478473

McEwen, B. S. (1998). Protective and damaging effects of stress mediators. New England Journal of Medicine, 338(3), 171–179.

Mian, A., & Sufi, A. (2014). House of Debt: How They (and You) Caused the Great Recession, and How We Can Prevent It from Happening Again. University of Chicago Press.

Miller, R. G. (2000). To save a city: The Berlin Airlift, 1948–1949. Texas A&M University Press.

Morgan, E. S., & Morgan, H. M. (1953). The Stamp Act Crisis: Prologue to Revolution. University of North Carolina Press.

Mueller, J. E. (1970). Presidential popularity from Truman to Johnson. American Political Science Review, 64(1), 18–34. https://doi.org/10.2307/1955610

Mullainathan, S., & Shafir, E. (2013). Scarcity: Why Having Too Little Means So Much. Times Books.

North Atlantic Treaty, 4 April 1949, 34 UNTS 243, art. 5.

Nunn, N., & Qian, N. (2014). US food aid and civil conflict. American Economic Review, 104(6), 1630–1666. https://doi.org/10.1257/aer.104.6.1630

Protocol Additional to the Geneva Conventions of 12 August 1949, and Relating to the Protection of Victims of International Armed Conflicts (Protocol I), 8 June 1977, 1125 UNTS 3, art. 54.

Regan, P. M. (2002). Third-party interventions and the duration of intrastate conflicts. Journal of Conflict Resolution, 46(1), 55–73. https://doi.org/10.1177/0022002702046001004

Scheffer, M., Bascompte, J., Brock, W. A., Brovkin, V., Carpenter, S. R., Dakos, V., Held, H., van Nes, E. H., Rietkerk, M., & Sugihara, G. (2009). Early-warning signals for critical transitions. Nature, 461(7260), 53–59.

Sharp, G. (1990). Civilian-based defense: A post-military weapons system. Princeton University Press.

Siverson, R. M., & Starr, H. (1991). The diffusion of war: A study of opportunity and willingness. University of Michigan Press.

Šmidchens, G. (2014). The power of song: Nonviolent national culture in the Baltic Singing Revolution. University of Washington Press.

Statman, D., Sulitzeanu-Kenan, R., Mandel, M., Skerker, M., & De Wijze, S. (2020). Unreliable protection: An experimental study of experts' in bello proportionality decisions. European Journal of International Law, 31(2), 429–453.

Stephan, M. J., & Chenoweth, E. (2008). Why civil resistance works: The strategic logic of nonviolent conflict. International Security, 33(1), 7–44. https://doi.org/10.1162/isec.2008.33.1.7

Sterling, P., & Eyer, J. (1988). Allostasis: A new paradigm to explain arousal pathology. In S. Fisher & J. Reason (Eds.), Handbook of Life Stress, Cognition and Health (pp. 629–649). Wiley.

United Nations. (1948). Universal Declaration of Human Rights. General Assembly resolution 217 A. United Nations.

United Nations. (1970). Declaration on Principles of International Law concerning Friendly Relations and Co-operation among States in accordance with the Charter of the United Nations. General Assembly resolution 2625 (XXV). United Nations.

United Nations. (2014). Security Council resolution 2165 (2014). S/RES/2165 (2014). United Nations.

Uzonyi, G., & Asal, V. (2020). Discrimination, genocide, and politicide. Political Research Quarterly, 73(2), 352–365.

Vasquez, J. A., & Rundlett, A. (2016). Alliances as a necessary condition of multiparty wars. Journal of Conflict Resolution, 60(8), 1395–1418. https://doi.org/10.1177/0022002715569770

Welles, S. (1940). Statement by the Acting Secretary of State, 23 July 1940 [Press release]. United States Department of State. Reprinted in Foreign Relations of the United States, 1940, Vol. 1, p. 401.

X [Kennan, G. F.]. (1947). The sources of Soviet conduct. Foreign Affairs, 25(4), 566–582.