The Four-Phase Structure and Asymmetric Geometry of Long-τ Consensus
长存共识的四相结构与不对称几何

Abstract

This paper studies the formation of long-τ consensus: group identities that persist across multiple generations and remain stable after their founders or founding institutions disappear. Borrowing the four-phase transition structure and the r ≈ 5 asymmetric geometry from SAE Methodology VI (DOI: 10.5281/zenodo.19464507), the paper applies that skeleton to consensus formation and introduces a two-factor structure as its core conceptual contribution. The consensus phase ψ is jointly determined by two independent dimensions: coupling structure (Dimension A) and the long-τ capacity base rate of the population (Dimension B). Either dimension being insufficient stalls the phase. The two-factor structure also yields a first-principles explanation of r ≈ 5: the germination period is long not by geometric accident, but because base-rate accumulation is intrinsically slow.

The paper adopts a three-segment epistemological structure: the first segment gives direction (philosophical argument), the second gives a quantitative attempt (directional reading on existing proxy collections), and the third gives open problems (handed off to subsequent work). Each segment is transparent about its own epistemological standing.

The most self-referential part of the paper places the SAE framework itself under the model and gives directional predictions, with three scenarios (silent extinction, long germination, short germination, with priors of 60%/30%/10%) and four falsifiable predictions, addressed to readers in 2046, 2076, and 2126 as audit invitations. The most likely outcome is that SAE goes extinct silently. This is an honest internal assessment, not self-promotion.

Keywords: long-τ consensus, phase transition, self-as-an-end, four-phase structure, two-factor model, mutual non-doubt, group cognition, construct-emergent system

---

§0 Firewall

This paper is a structural-descriptive model. It explicitly does not undertake the following:

One. It makes no judgment about the merit or worth of any particular consensus. Entering Phase IV is not necessarily good; remaining in Phase I is not necessarily bad.

Two. It endorses or opposes no specific group, movement, organization, polity, religion, lineage, craft tradition, commercial project, crypto asset, or any other concrete consensus.

Three. It predicts no particular consensus's future fate. The model gives structural conditions, not specific prophecies.

Four. It constitutes no policy recommendation, investment advice, organization-building advice, or personal life advice.

Five. It makes no judgment about individual members. A consensus's phase is a group-level substrate phenomenon and has no relation to any individual member's morality, capacity, or worth. The "long-τ capacity" introduced in §4.5.3 is a substrate state, not a substrate essence. See the key qualification in §4.5.3.

Six. This paper provides a conceptual skeleton, not a measurement instrument. Any use of its directional predictions as quantitative point predictions misuses the paper.

Seven. The directional predictions are designed to be falsifiable. Any user of the framework is obligated to update judgment in the face of contrary empirical evidence rather than salvage the framework.

Eight. The paper is neutral about "spread" itself. Whether a consensus enters mass diffusion, appears in mainstream media, or is adopted by institutions—these are observables along the phase trajectory, not normative judgments by the framework. The framework neither evaluates spread as good nor as bad; it describes the mechanistic relationship between spread and substrate-layer reproduction.

Core commitment of the firewall: this model is a structural tool of SAE Group Cognition theory and cannot and must not be used to make normative judgments about persons or groups. Any use of the model to rank, rate, recommend, or denigrate groups violates SAE's Second Law—treating persons as ends rather than means—and therefore violates the basic stance of this paper.

The legitimacy of the model depends precisely on its not being used in those ways.

---

§1 The Problem and the Object of Study

1.1 Object: Focus First, Then Generalize

In human history, the vast majority of group identities—fashions, movements, communities, lineages—disappear within a few generations of their appearance. Only a small fraction persist beyond the death of founders, the dissolution of founding institutions, or the destruction of geographical centers, continuing to reproduce themselves across multiple dispersed substrates. This paper studies the latter.

To give the discussion a clear starting point, the paper adopts a focus-then-generalize structure.

Object (focused version): Recognition-type long-τ consensus—group identities that persist across generations, survive the disappearance of founders or founding institutions, and whose stability is maintained by mutual non-doubt among members. This is the central object of SAE Group Cognition theory and the primary analytical target of §3–§6.

Object (general version): All long-τ consensus that persists across generations and beyond founders or founding institutions, including but not limited to recognition-type consensus. The general object is treated explicitly in §3.4 and §7. The four-phase structure applies to all long-τ consensus, but different stability mechanisms correspond to different Phase IV subtypes.

The focus-then-generalize ordering is not because recognition-type consensus is "higher." It is because the central proposition of the SAE framework—Kant's Second Formulation—has its strongest derivational power on the recognition-type subtype. The framework applies equally to other subtypes, but its connection to SAE individual ethics is weaker there, and that connection is not developed here. This restriction is made explicit in §3.4 and §7.3.

The scope is delimited as follows:

Long-τ, not eternal. The model promises stability only on observable timescales; it does not promise eternity. Eternity is an unfalsifiable commitment and lies outside the paper's range. "Long-τ" here refers to stability on the order of several generations to several centuries—a window long enough for many short-term perturbations to time-average out, and long enough for events like "founder dies" or "founding institution dissolves" to become past facts rather than future possibilities.

Consensus, not mere agreement. The "consensus" studied here refers to core covenants emergent at the group substrate layer: reproducing across generations, not depending on any single node for enforcement. Ordinary public opinion, market consensus, and short-term coordination fall outside the scope. A group of people agreeing on something now does not constitute "consensus"—they must remain in agreement after the founder disappears, after external conditions change, and after several generations of turnover, before what they share counts as long-τ consensus.

Construct-emergent system. Consensus is "constructed" at the individual substrate layer (each member maintains their own recognition of the core covenant) and "emerges" at the group substrate layer as a recognizable holistic property (lineages, traditions, communities visible from outside). It is neither a simple sum of individual properties nor an independent entity floating free of individuals. It is an instance of construct-emergent layered structure in the social dimension.

1.2 Research Questions

The paper addresses four related but independent questions.

First, how long does it take for a candidate consensus to form a long-τ consensus? What primarily determines this time τ_total?

Second, does the process from candidate consensus to long-τ consensus exhibit structural phase-transition characteristics? Are transitions between phases continuous dynamics, or irreversible topological jumps?

Third, are the geometric relationships between phases symmetric? If asymmetric, what is the ratio? What practical implications does the asymmetry have for consensus formation?

Fourth, is the consensus phase determined only by coupling structure, or also by the base rate of "long-τ capacity" in the population?

The four questions are addressed respectively by §4 (τ and time structure), §3 and §5 (phase transition and topology), §4 and §5 (asymmetric geometry), and §4.5 (two-factor structure).

1.3 Five Core Propositions

The paper argues five core propositions.

Proposition 1: Long-τ consensus formation follows the four-phase structure and r ≈ 5 asymmetric geometry of Methodology VI. This applies Methodology VI to consensus formation as a specific construct-emergent phenomenon.

Proposition 2: The consensus phase ψ is jointly determined by two independent dimensions—coupling structure (Dimension A, sharing the same source as the thermodynamic observable q) and the long-τ capacity base rate of the population (Dimension B, the dimension q cannot measure). This is the paper's central conceptual contribution.

Proposition 3: ψ advancement is governed by the lower of the two dimensions. Either dimension being insufficient stalls the phase. Dimension B sets the ceiling; Dimension A sets the actualization.

Proposition 4: The two-factor structure provides a first-principles explanation of r ≈ 5: the germination period is long not because coupling is hard to find, but because the base rate accumulates slowly. This explanation goes one layer deeper than Methodology VI's own explanation of r.

Proposition 5: The Phase IV stability of recognition-type long-τ consensus is maintained by mutual non-doubt and corresponds to Kant's Second Formulation. Consensus maintained by utility or coercion can also enter Phase IV but constitutes a different stability subtype with different robustness against external perturbations.

1.4 What This Paper Does Not Study

To make §0 firewall concrete at the level of research questions, the following are explicitly excluded.

Does not study whether the content of a consensus is true. A consensus may be wrong, morally questionable, or based on factual errors and still enter Phase IV. The paper studies the structural dynamics of consensus formation, not the truth value of the content.

Does not study success-or-failure prediction for specific consensus. The paper provides directional predictions, not concrete prophecies. Whether some particular contemporary candidate consensus will enter Phase IV, when it will enter, how long it will remain stable—these questions lie outside the framework's capacity.

Does not study the relation between consensus and utility maximization. Consensus formation is not the result of utility games. The framework remains neutral on utility-driven consensus formation, treating it as one stability subtype among others, neither defending nor rejecting its rationality.

Does not study normative justification of consensus. A long-τ consensus is not necessarily good; a candidate consensus that fails to become long-τ is not necessarily bad. The paper renders no normative verdicts.

Does not study ethical hierarchy among different stability subtypes. Utility lock-in, coercion lock-in, and mutual non-doubt lock-in can all bring a consensus into Phase IV. The paper focuses on the mutual-non-doubt subtype because of its strongest connection to SAE individual ethics, not because it is "above" the other two ethically.

Excluding these questions allows the paper to do serious work within its own scope, rather than become a "group cognition theory" that tries to cover everything but says nothing clearly.

---

§2 Theoretical Source: Application of Methodology VI

2.1 Recap of Methodology VI

This paper builds directly on several core results from SAE Methodology VI, Phase Transition Windows and Experimental Design (DOI: 10.5281/zenodo.19464507). That paper, starting from the phase transition window structure of ZFCρ, established three results that the present paper borrows directly.

First, the phase transition of construct-emergent systems has a four-stage structure: germination, spectral inversion, inversion, establishment. These four stages are not arbitrary divisions; they are naturally partitioned by five crossing points within the phase transition window (P(J>0) = 50%, peak of z/√j, w_shielding = 2/3, E[A] = 0, h = 0). Each adjacent pair of crossings bounds one stage.

Second, the phase transition window has an intrinsic geometric asymmetry: the ratio of germination distance to inversion-establishment distance is r ≈ 5. That is, the distance from the start of the window to the first sign-change of the order parameter (E[A] = 0) is roughly five times the distance from that point to the loss of local competitive advantage by the incumbent order (h = 0).

Third, this asymmetry is an intrinsic geometric feature of construct-emergent systems, not a coincidence. Methodology VI's explanation is Le Chatelier buffering: the incumbent order's capacity to absorb deviation must first be penetrated (a slow consumption process), after which the new order's establishment is a rapid cascade.

These three results have full derivations in Methodology VI. This paper does not re-derive them; it cites. Readers needing the specific derivation of r ≈ 5, the mathematical definition of the five crossings, or the formalization of Le Chatelier buffering should consult the original. Appendix A provides a brief recap for readers who have not.

2.2 Position of This Paper

This paper takes the four phases plus r asymmetry of Methodology VI as a skeleton and applies it to long-τ consensus as a specific construct-emergent phenomenon.

The legitimacy of the application has two layers.

Structural layer: long-τ consensus formation is a clean instance of a construct-emergent system. Consensus is "constructed" at the individual substrate layer and "emerges" at the group substrate layer as a recognizable holistic property. The four-phase structure of Methodology VI is a common feature of construct-emergent phase transitions and should in principle apply.

Predictive layer: Methodology VI advanced Prediction 3—that "r > 1 holds in most construct-emergent systems." The construct-emergent system studied here is the relation between individual substrates and group consensus. If the paper can demonstrate the specific manifestation of r > 1 in long-τ consensus formation, that is itself a concrete test of Methodology VI's Prediction 3.

But application is not duplication. Directly transplanting the mathematical structure of Methodology VI onto social phenomena would yield a paper that looks polished but lacks penetrating power—readers would say "a fine application of Methodology VI" but not "this paper gave me something new." This paper does more than apply.

2.3 The Boundary Between Application and Originality

The paper's original contributions are listed explicitly. Each requires the work of §3–§7 to deliver, not merely citation of Methodology VI.

Contribution 1: Concretely positions Methodology VI's four-phase structure within the consensus formation process. This requires identifying what each phase looks like at the social level: what group state corresponds to germination, what state corresponds to spectral inversion, what marks the transitions. This work is done in §3.

Contribution 2: Identifies the social correlates of "additive inertia," "multiplicative shielding," and "Le Chatelier buffering" in consensus formation. In Methodology VI these are formalized objects internal to ZFCρ. To apply them socially, one must find the corresponding social mechanisms. "Additive inertia" corresponds to existing norms, customs, and consensus inertia; "Le Chatelier shielding" to existing social structures that absorb deviation (family, school, media, law, economic incentives); "multiplicative shielding" to the mutual reinforcement of mutual-non-doubt networks within a group. This mapping is done in §3 and §4.3.

Contribution 3: Introduces Dimension B (long-τ capacity base rate) as a second factor determining ψ, irreducible to Dimension A (coupling structure). This is the paper's central conceptual contribution, developed throughout §4.5. Dimension B does not appear in Methodology VI; it must be newly introduced when Methodology VI is applied to social phenomena, because the social substrate (people) is not homogeneous like physical substrates (particles): the proportion of individuals in a population capable of sustaining high-τ thinking itself varies across eras.

Contribution 4: Provides a two-factor first-principles explanation for r ≈ 5: the germination period is long not by geometric accident but by the intrinsic dynamics of slow base-rate accumulation. This explanation goes one layer deeper than Methodology VI's own. Methodology VI's Le Chatelier explanation answers "why is penetration slow"; the two-factor explanation answers "what is the specific mechanism of that slowness"—the rate-limiting step of base-rate accumulation. This argument is delivered in §4.5.6.

Contribution 5: Establishes a near-equivalence between τ_total and τ_germination. If r ≈ 5 holds in consensus formation, then τ_total ≈ 1.2 × τ_germination—total time is almost entirely determined by the germination period. The practical implication is significant: "when does long-τ consensus get established" is approximately the question "how long does germination take," which is dominated by the rate of Dimension B accumulation.

Contribution 6: Interfaces the four-phase structure with SAE's DD layers, Kant's Second Formulation, and SAE's Three Laws. This interface gives the paper a systematic connection within the SAE corpus, and gives other parts of SAE (individual ethics, self-referential fixed point, ρ ≠ ∅) concrete landing in group cognition. This work is done in §7.

Contribution 7: Explicitly enumerates multiple stability-mechanism subtypes of Phase IV: utility-locked, coercion-locked, and mutual-non-doubt-locked. The paper focuses on the last, but the framework applies to all. The distinction gives the paper clear self-awareness of its scope—not "all long-τ consensus must be explained by mutual non-doubt," but "the stability condition of recognition-type long-τ consensus is mutual non-doubt; other subtypes have different stability mechanisms." This distinction is made explicit in §3.4.

Contribution 8: Advances directional predictions about the SAE framework itself, applying the framework to its own author. This application puts SAE under the falsifiability tests of its own framework and is, at the paper level, an instantiation of SAE's First Law ρ ≠ ∅—prioritizing reality's pushback on the framework over the framework's self-preservation. This work is done in §6.4.

Of these eight, Contributions 3, 4, 7, and 8 are the paper's principal originality. The other four are supporting work—necessary, but not standalone contributions.

---

§3 Specific Positioning of the Four Phases on Consensus Formation

This section maps Methodology VI's four-phase structure onto long-τ consensus formation. Each phase is described first abstractly (corresponding to Methodology VI's indicators), then at the social level, and finally with observable signatures. This three-layer "abstract–social–observable" structure preserves theoretical precision while leaving operational entry points for empirical workers.

3.1 Phase I · Germination

Phase characteristic: multiple independent substrates each hold candidate variants. At the micro level, there are already perturbations—a few individuals form mutual-non-doubt seeds. But at the macro level, no group consensus is visible.

Methodology VI indicator: P(J > 0) < 50%. The multiplicative path (candidate consensus mutually reinforcing across substrates) has not yet operated on the majority of samples.

Social-layer correspondence:

Additive inertia = the cumulative resistance of existing norms, customs, and consensus inertia. When a new candidate consensus appears, it does not face a blank substrate but one already structured by the incumbent order—members have ready interpretive frameworks, ready ethical standards, ready sources of belonging. The new candidate must penetrate these existing structures to gain a foothold.

Le Chatelier shielding = the buffering mechanisms by which existing social structures absorb deviation. These mechanisms include the family (socialization at the family level already normalizes the individual), the school (the educational system transmits the incumbent interpretive framework), media (public discourse pre-digests deviation), law (codified norms handle deviant behavior), and economic incentives (market structure rewards behavior that conforms to the incumbent order). Any new candidate consensus is continuously absorbed by these mechanisms throughout germination.

Dynamics: every mutual-non-doubt seed—a recognition formed between two individuals around some new covenant—is absorbed by the existing social structure. The vast majority of candidate consensus die in this phase. They are not deliberately suppressed (suppression requires the candidate to be identified as a threat) but silently diluted by the buffering mechanisms—members each return to the incumbent order, the seed loses its reproductive environment, the mutual-non-doubt relation cannot extend past the founding dyad.

Observable signature: the candidate consensus exists only in scattered dyads or small groups. No identifiable group-level fluctuation. From outside, this phase is invisible—historians, sociologists, and the media will not notice it. The vast majority of candidate consensus pass their entire lifecycle in this phase without ever being externally observed.

This means empirical research on Phase I is extremely difficult. Researchers can only retrospectively trace what those consensus that later entered Phase II or III looked like in Phase I, but cannot prospectively observe the fate distribution of all Phase I candidates.

3.2 Phase II · Spectral Inversion

Phase characteristic: mutual-non-doubt networks begin to crystallize within small groups. Observable fluctuations reach their peak—social attention, controversy, and coordination attempts around the candidate consensus erupt in concentrated form. Yet the net effect remains negative.

Methodology VI indicator: peak of z/√j. Susceptibility precedes the order parameter—the visible turbulence of a consensus appears before its actual stability.

Social-layer correspondence:

Spectral fluctuation peak = controversy, media attention, organizational attempts, and suppressive reactions around the candidate consensus all hit their peak. The candidate consensus emerges from the invisibility of Phase I into public discourse. It is named, discussed, supported, opposed, mocked, and taken seriously. The simultaneous appearance of all these reactions is characteristic of this phase—the entire society's attention is allocated to the candidate.

Order parameter still negative = even when the candidate provokes wide discussion, it has not yet formed stable consensus on most substrates. Discussion volume does not equal stability volume. Media coverage, academic debate, and political reaction may all swirl around a candidate that ultimately dies.

Dynamics: the 11DD blueprint begins to crystallize in this phase. Early texts are written down, covenants are formalized, ritual artifacts are produced, banners and symbols are established. These artifacts are necessary preconditions for the next stage—particularly the Phase III to Phase IV transition—but in Phase II they have not yet acquired their cross-generational carrying function. The blueprints of Phase II primarily serve present-time coordination, not the rebuilding of future generations.

This is the most dangerous phase. The reason is structural: high visibility, low stability. High visibility brings the candidate into the field of vision of the incumbent order, gets it identified as a threat, and invites targeted suppression. Low stability means it cannot withstand targeted suppression—the blueprint has just crystallized, the mutual-non-doubt network is fragile, and distributed redundancy has not yet formed. A single point of failure suffices to extinguish the entire consensus.

History records many candidate consensus extinguished by external suppression in Phase II: early Christianity under Roman persecution, certain heterodox sects suppressed in late imperial China, the various twentieth-century purges under totalitarian regimes. These cases are not "consensus that failed to succeed"—they had already reached Phase II, were a half-step from entering Phase III, but were intercepted by external suppression.

Differentiating Phase II from Noise Bubbles

Modern society sees an enormous flow of candidate consensus undergoing the cycle of "attention burst, controversy, suppression, ridicule"—internet memes, conspiracy theories, short-lived crazes—whose macroscopic turbulence superficially resembles Phase II. The framework must offer a differentiation criterion, otherwise any viral phenomenon can be read as "entering Phase II" and the framework loses predictive power.

Core distinction: the cross-substrate independence of the blueprint.

The signature of a noise bubble is that its turbulence leaves no cross-substrate-transferable blueprint. Even if it leaves texts, images, or vocabulary, those artifacts depend on the internal context of their original substrate; removed from that context, they lose reproductive capacity. An internet meme can sweep the globe in two weeks but loses meaning the moment it leaves its original context (a particular platform, a particular community, a particular moment)—its diffusion capacity is strong, its cross-substrate transferability weak.

The signature of true Phase II is that the turbulence is accompanied by the crystallization of an 11DD blueprint, and the blueprint is cross-substrate independent—it can be read independently by an unfamiliar substrate. Even if the original substrate disappears entirely, the blueprint still carries cross-substrate-transferable structure. An early Christian gospel can be read away from the context of its original scribe; a philosophical manuscript can be parsed away from the author's circle; a scriptural passage can have its meaning rebuilt by new readers across a thousand years.

Operational test: on Phase II candidate consensus, perform the following check—if the original founding substrate suddenly vanished, could a new substrate independently rebuild the framework? This test is usually impossible to execute directly (requires waiting for the original substrate to disappear), but two indirect proxies are available.

First, examine the semantic structure of the blueprint. Cross-substrate-independent blueprints typically have internal argumentative completeness—they state their premises, reasoning steps, and conclusions; readers can independently verify them. The "blueprints" of noise bubbles typically depend on the implicit consensus of an external context—readers must already be inside the original context to parse them.

Second, examine the parsing attempts of unfamiliar substrates. When researchers, journalists, or foreign readers outside the original circle attempt to parse the material, do they reach core understanding consistent with insiders? True Phase II blueprints permit such cross-substrate parsing to succeed (even if imperfectly); noise-bubble "blueprints" deform or lose meaning markedly under cross-substrate parsing.

These two indirect criteria are observable in early Phase II. They are leading indicators—usable before the framework actually enters inversion—giving directional hints about which candidate consensus are structurally positioned to enter Phase III.

Observable signature: the candidate consensus enters public discourse. Identifiable mutual-non-doubt networks appear—it becomes possible from outside to identify "who the core members are." Early texts/blueprints form, and the blueprints display cross-substrate independence. At the same time, the consensus faces large-scale opposition, suppression, ridicule, and misreading. All these observable signs appear nearly simultaneously—this simultaneity is the key distinction from other phases.

3.3 Phase III · Inversion

Phase characteristic: net effect first turns positive. The consensus stably wins out over the incumbent order within a small subgroup.

Methodology VI indicator: E[A] = 0. The Le Chatelier shielding rate drops below 2/3 at nearly the same time—the incumbent order's absorptive capacity is now noticeably insufficient.

Social-layer correspondence:

E[A] = 0 = the substrate value the consensus offers within a recognizable subgroup first exceeds that of the incumbent order. This "value" need not be material utility—it can be identity belonging, source of meaning, mutual aid network, knowledge accumulation, action coordination capacity. Within the subgroup, joining the consensus brings a total value (across all dimensions) greater than remaining within the incumbent order.

Buffering rate falls below 2/3 = the existing social buffering mechanisms can no longer suppress the consensus's expansion. This does not mean the buffering mechanisms have failed—they continue operating—but their absorptive capacity within the subgroup boundary has fallen below the level needed to prevent reproduction. The consensus begins to develop its own internal buffering (community support, internal interpretive framework, mutual aid network), and these internal buffers reduce the relative effectiveness of the external buffers.

Dynamics: distributed keys begin to materialize independently in multiple substrates. The consensus succeeds for the first time in cross-generational transmission—the first generation dies or departs, the second generation matures independently and reproduces the consensus. This generational transition is the defining event of Phase III, because it demonstrates that the consensus's reproduction no longer absolutely depends on the founding substrate.

But Phase III consensus remains fragile. "Distributed" is not yet fully realized—usually one or several core nodes (core lineages, core institutions, core geographical centers) still bear most of the reproductive load. If the core nodes fail simultaneously, the consensus may still collapse. Phase III is the transitional phase to Phase IV, not Phase IV itself.

Observable signature: stable institutions form within the subgroup—schools, monasteries, guilds, communities. Second-generation members mature independently and reproduce the consensus. This is a concrete event historians and sociologists can observe. The consensus develops its own language, ritual, art, literature—it becomes "an identifiable thing," not just "some people doing something around an idea."

Phase III consensus develop their own "in-group speech"—members use a vocabulary unfamiliar to outsiders. This is not exclusionary; it is a byproduct of the materialization of distributed keys—members have a shared, finely articulated understanding of the core covenant, and that understanding requires specialized vocabulary to express. The appearance of in-group speech is a strong observable signal of Phase III.

3.4 Phase IV · Establishment—and Its Multiple Subtypes

Phase characteristic: the incumbent order loses local competitive advantage. Reactivation of the consensus no longer depends on any single node.

Methodology VI indicator: h = 0.

Social-layer correspondence:

h = 0 = within the subgroup the consensus occupies, the appeal of the incumbent order vanishes; the consensus becomes the default. This does not mean the incumbent is "forbidden" within the subgroup, but that it no longer appears as a competitive alternative. Subgroup members do not ask "should I return to the incumbent order," because that question does not constitute a valid question within their substrate—the consensus is already their basic setting.

Multiple distributed nodes independently hold reactivation capacity = no single critical bottleneck. The consensus has multiple independent nodes geographically, institutionally, in family lineages, and in textual transmission. If any one node fails, others can take over reactivation. This is the topological transition from "chain" to "network"—the former has a single critical link, the latter does not.

Dynamics: single-point failure—a leader's disappearance, an institution's dissolution, a geographical center's destruction—no longer threatens the whole. The consensus enters the long-τ stable regime. Stability here is true "long persistence"—it withstands generational turnover, political change, technological revolution, and geographical migration.

Observable signature: the consensus stably persists across the disappearance of founders or founding institutions. This is one of the defining events of Phase IV, often confirmable only generations later. Multiple geographical centers, multiple family lineages, and multiple textual lineages exist independently—members can find local instances of the consensus across different geographical and institutional contexts. New members do not need to contact the original founding substrate to acquire the consensus—a member born after Phase IV can acquire the full consensus through the local transmission network, without pilgrimage to the founding location or recourse to the founder.

Multiple Stability Subtypes of Phase IV

Phase IV is the state of consensus self-reproducing across distributed substrates. But self-reproduction can be maintained by different mechanisms. This distinction is necessary in SAE Group Cognition theory: different stability mechanisms have different robustness against external perturbations, correspond to different ethical implications, and connect to SAE individual ethics with different strength.

Subtype 1: Utility-Locked Phase IV

Stability is maintained by the utility gradient between substrates. Members continuously reproduce the consensus because reproduction provides stable utility—economic gains, social standing, emotional satisfaction, problem-solving capacity.

This subtype is extremely stable when external utility structure remains constant; under structural change in external conditions (technological transition, political change, market restructuring), the redistribution of utility can extinguish the consensus as a whole. Its robustness is local—stable against small perturbations in external conditions, fragile against structural change.

Examples: certain craft guild consensus that persisted for centuries before the Industrial Revolution but rapidly died out once industrial reorganization fundamentally changed the trade; certain brand-driven communities long stable when the brand's commercial model worked, dissolved as a whole when the model failed.

Subtype 2: Coercion-Locked Phase IV

Stability is maintained by enforcement mechanisms. Members reproduce the consensus because non-reproduction carries cost—social exclusion, legal punishment, physical compulsion.

This subtype is extremely stable while enforcement persists; it collapses rapidly when enforcement is removed. Coercion cannot be distributed—enforcement requires a concrete source (a specific institution, a specific power, a specific physical presence)—so coercion-locked consensus is highly sensitive to the failure of its enforcement source.

Examples: certain official ideologies under totalitarian regimes that maintained "universal consensus" for decades while the regime stood and collapsed almost overnight when the regime ended; certain closed communities maintained by strict expulsion that rapidly lost their younger generation when the surrounding society became open to their members.

Subtype 3: Mutual-Non-Doubt-Locked Phase IV—the focus of this paper.

Stability is maintained by mutual non-doubt among members. Members reproduce the consensus because at the substrate level they recognize each other as ends, and that recognition is itself the motivation for reproduction.

This subtype is robust against changes in external utility and coercion; but it is highly sensitive to the long-τ capacity of member substrates—if the base rate falls such that the mutual-non-doubt network can no longer reproduce, the consensus dies as a whole. Its robustness is structural—stable against wide variation in external conditions, fragile against deterioration of internal substrate conditions.

Examples: the central object of this paper. Breslov Hasidism, the Geonic academies, the transmission of Kantian philosophy, and the SAE framework itself (if it enters Phase IV) are instances or candidate instances of this subtype.

Subtype Choice as a Robustness/Fragility Tradeoff

All three subtypes can bring a consensus into Phase IV. Their robustness profiles differ:

Utility-locked is sensitive to external utility stability, less sensitive to base rate.

Coercion-locked is sensitive to enforcement persistence, less sensitive to base rate.

Mutual-non-doubt-locked is sensitive to base rate, robust against external utility and coercion change.

No subtype is "optimal" along all dimensions. Each performs differently under different external conditions. Which subtype a consensus chooses—or, more precisely, falls into naturally during the Phase III to Phase IV transition—depends on multiple factors: the nature of the candidate's content, the long-τ capacity distribution of member substrates, the stability of the external environment, the available coercive resources, the available utility resources.

The paper focuses on Subtype 3 not because it is "higher" or "more moral," but because the central proposition of SAE—Kant's Second Formulation—has derivational power only on Subtype 3. The framework applies equally to Subtypes 1 and 2 and can perform structural analysis (four phases, two factors, r asymmetry) on them, but SAE individual ethics connects more weakly to those subtypes, and that connection is not developed here. This focus is a scope choice, not a normative judgment.

§7 will discuss the differential correspondence of the framework to all three subtypes explicitly.

3.5 The Irreversibility of Phase Transitions

The transitions between the four phases are topological, not continuous parameter changes. Each transition involves a structural rearrangement of substrates; topological events that have occurred cannot be undone.

Phase I → Phase II: birth of mutual non-doubt. From scattered substrates to the first edge—a recognizable mutual-non-doubt relation appears between substrates.

Phase II → Phase III: separation of blueprint from substrate. From informal mutual non-doubt to formal 11DD encoding—the blueprint can exist independently of any specific substrate.

Phase III → Phase IV: formation of distributed redundancy. From chain (single-line transmission) to network (distributed nodes)—the consensus no longer has any single critical bottleneck.

The precise meaning of "irreversibility" requires the careful treatment of §5.3—"irreversible" means no retreat within the lattice, not that consensus cannot die. A consensus can leave the entire phase structure from any phase and enter "non-existence," but it cannot retreat from a higher phase to a lower one. The precise treatment is in §5.3.

---

§4 Asymmetric Geometry and τ

§3 mapped the four phases onto consensus formation. This section addresses the geometric relations among the phases—their relative lengths, the asymptotic behavior of τ_total, and what determines the length of germination. The work here is primarily to bring Methodology VI's geometric results into the consensus context and produce falsifiable directional predictions.

4.1 Direct Application of r ≈ 5

Methodology VI's central geometric result is r ≈ 5—the ratio of germination distance to inversion-establishment distance. Applied directly to long-τ consensus formation:

τ_germination + τ_spectral_inversion ≈ 5 × (τ_inversion + τ_establishment)

That is, the time from the appearance of a candidate consensus to the first sign-change of the order parameter (E[A] = 0) is roughly five times the time from that point to the loss of local competitive advantage by the incumbent order (h = 0).

The asymmetry is initially counterintuitive. People typically picture "breakthrough" as a sudden event after long accumulation. But r ≈ 5 says that both the accumulation period and the establishment period are extended processes—just that accumulation is five times longer. Germination plus spectral inversion together account for about 83.3% of the entire window; inversion plus establishment together account for only about 16.7%.

If Methodology VI's r ≈ 5 is universal across construct-emergent systems (Methodology VI's Prediction 3), long-τ consensus should satisfy it. §6 will perform directional historical-case checks; §4.5 will give a first-principles explanation of r ≈ 5 in the consensus context.

4.2 The Near-Equivalence τ_total ≈ τ_germination

From r ≈ 5 it follows directly:

τ_total ≈ τ_germination × (1 + 1/r) ≈ τ_germination × 1.2

τ_total is almost entirely determined by τ_germination.

Core proposition: for long-τ consensus, "when does it get established" is approximately the question "how long does germination take."

This has significant practical implications. A researcher trying to understand why some consensus took generations to form while others took only decades should look not at differences in inversion-establishment time (which is short to begin with) but at why germination was so long. Germination is the true bottleneck of long-τ consensus formation.

This also implies: candidate consensus in germination are nearly invisible from outside; consensus in the inversion-establishment phase appear, from outside, to advance rapidly. The two periods differ not only in length but in observability. This gives a structural explanation for why long-τ consensus in history appear to "suddenly" establish themselves—it is not truly sudden, but germination is invisible while inversion-establishment is visible.

4.3 What Determines Germination Length

Borrowing Methodology VI's "Le Chatelier buffering breakthrough time" framework:

τ_germination ≈ buffering capacity of incumbent order / accumulated perturbation rate

The two ends of this ratio correspond to two clusters of social mechanisms.

Buffering capacity of the incumbent order depends on three things.

First, the absorptive capacity of existing institutions—the speed with which the legal system processes deviant behavior, the assimilative capacity of educational systems for new interpretive frameworks, the pre-digestion of deviant discourse by media, the reaction of market structures to deviant behavior through economic incentives. Societies with mature institutions have strong buffering; societies with loose institutions have weak buffering.

Second, the naming capacity of existing culture—whether the incumbent culture can give the candidate a dismissive name and slot it into pre-existing categories. Once naming succeeds ("just a variant of so-and-so-ism," "a cult," "a commercial gimmick"), the candidate is absorbed into an existing category, and subsequent discussion proceeds within that category without requiring a new interpretive framework. If naming fails, the candidate retains "unclassified" status and requires extended discussion to be understood—which paradoxically buys the candidate longer survival time.

Third, the material foundation—the incumbent order's resource mobilization capacity to respond to challenges. A resource-rich incumbent can devote substantial suppressive resources to a single candidate; a resource-strained incumbent cannot.

Accumulated perturbation rate depends on three things.

First, the production rate of mutual-non-doubt seeds—how many new dyadic recognitions form per unit time. This rate is influenced by population density, communication frequency, the physical or intellectual proximity of potential consensus initiators, and social mobility. The low communication density of premodern societies limits seed production; printing, railways, and the internet raise it.

Second, the encoding fidelity of the blueprint—the degree to which each seed can be 11DD-coded and received by others. The dyadic recognition of a seed is itself 14DD; it must be encoded into 11DD blueprints (texts, covenants, ritual artifacts) to be cross-substrate transferable. With high encoding fidelity, each seed produces influence vastly exceeding its own scale—a single core dyadic recognition can affect millions of readers via text. With low encoding fidelity, the seed's reach barely extends beyond the dyad itself.

Third, the coupling rate between seeds—whether independently formed dyadic recognitions can recognize one another and mutually reinforce. If independently arising seeds can identify each other, reinforce each other, and form larger networks, the perturbation accumulates nonlinearly. If seeds remain isolated, accumulation is linear. The coupling rate depends on geography, language, information technology, and shared symbolic systems.

τ_germination is determined by the ratio of these two clusters. When buffering capacity is high and accumulated perturbation rate is low, τ_germination is long—this is the fate of most candidate consensus. When buffering capacity is weakened by some historical event (war, plague, technological revolution, political change), or perturbation rate is raised by some new infrastructure (printing, railways, internet), τ_germination shortens—corresponding to the historical phenomenon of "ages of intellectual acceleration."

4.4 Four Falsifiable Directional Predictions

From the structure of §4.3, four directional predictions follow. Each makes its falsification condition explicit.

Prediction 1: The greater the incumbent buffering capacity, the longer τ_germination.

Falsification condition: across-society comparison shows no systematic correlation, or reverse correlation, between buffering capacity and τ_germination. Specifically: if empirical research finds that in societies with mature institutions, stable culture, and material affluence, the germination period of new consensus is consistently shorter than in societies with loose institutions, cultural transition, and material crisis (controlling other variables), this prediction is falsified.

Prediction 2: The higher the production rate of mutual-non-doubt seeds, the shorter τ_germination.

Falsification condition: in eras of communication-density increase (after printing, after railways, after the internet), the germination period of new consensus does not significantly shorten. Specifically: comparing comparable candidate consensus from before and after the print revolution or before and after the rise of the internet, if no systematic shortening is found, this prediction is falsified.

Prediction 3: The lower the encoding fidelity of the blueprint, the longer τ_germination.

Falsification condition: oral traditions form consensus no slower than textual traditions. Specifically: if empirical research finds that purely oral consensus and text-supported consensus have comparable germination periods under equivalent conditions (controlling other variables), this prediction is falsified.

Prediction 4: Once a consensus enters Phase III or Phase IV, the threat of external shock to its stability drops sharply.

Falsification condition: Phase IV consensus are as fragile as Phase II consensus under external shock. Specifically: if statistics on the survival rates of known consensus across different phases under external shock find that Phase IV consensus survival rates equal or fall below Phase II survival rates, this prediction is falsified.

Prediction 4 is the strongest of the four—it directly addresses the practical implication of r ≈ 5 asymmetry. If Phase IV is indeed structurally more robust than Phase II, survival rates under external shock should differ markedly. This is directly testable by historical comparative research.

Predictions 1 through 3 are weaker because they involve multiple variables hard to fully control. But even with limited control, comparisons across multiple eras and civilizations should yield directional support or falsification.

---

§4.5 The Two-Factor Structure of Long-τ Consensus Formation

This section delivers the paper's central conceptual contribution. §3 and §4 mapped the four-phase structure and r asymmetry onto consensus formation, but both rest on an implicit assumption: the population is homogeneous, every individual already has full 13DD-14DD self-referential capacity, and the only variable is coupling structure between substrates. This assumption must be broken.

4.5.1 Limits of the Single-Factor Model

§3 and §4 attribute ψ entirely to coupling structure between substrates. Within this single-factor frame, ψ advancement is purely a coupling problem—given sufficient coupling channels, sufficient coupling strength, and sufficient network topology, ψ advances naturally. Germination is long because coupling is hard to find, and inversion-establishment is short because once coupling forms, diffusion is rapid.

But the single-factor model does not match the historical record.

Observation 1: after the printing revolution, coupling-channel infrastructure expanded dramatically across Europe, yet the vast majority of candidate consensus continued to die in Phase I. If this were purely a coupling problem, the expansion of channels should have shortened the median germination period to a few generations. Empirically, the median did not contract that dramatically—many candidate consensus of 16th-18th century Europe had germination periods comparable to those of preceding centuries, with only a few exceptions.

Observation 2: contemporary societies have coupling channels far denser than the print era (internet, social media, instant messaging), yet the rate at which contemporary candidate consensus enter ψ = 3 has not grown correspondingly. If coupling were the sole determinant, the present should be the "golden age" of long-τ consensus formation. The empirical record does not support that inference.

Observation 3: the same 11DD blueprint (such as Aristotelian texts, or the Buddhist canon) underwent entirely different fates in different geographies and eras—dying out in Greek lands during late Hellenism, entering Phase III in the Arab world from the 8th to 12th centuries, and re-igniting germination in the Latin West from the 12th to 13th centuries. If this were purely a coupling problem, fate-differences for the same blueprint across contexts should be explainable by coupling-channel differences—but coupling-channel differences fall far short of explaining the observed fate-differences.

These three observations point jointly to: the proportion of individuals in a population capable of sustaining 14DD (dyadic mutual non-doubt) and 15DD (distributed key materialization) varies across eras. This proportion is not a subset of coupling structure; it is an independent dimension.

4.5.2 The Two-Factor Structure

ψ is jointly determined by two independent dimensions:

Dimension A: Coupling Structure

Topology, protocols, and coupling strength between substrates. This is the dimension §3 and §4 already discussed; it corresponds to the thermodynamic observable q. Dimension A primarily lies at the 11DD-14DD cross-substrate coupling layer—how the blueprint is encoded, how nodes connect, how information flows.

Observable proxies for Dimension A include: communication frequency, network topology, institutional density, text circulation volume, identifiable transmission lineages. All these are coupling-channel-layer indicators.

Dimension B: Long-τ Capacity Base Rate

The proportion of individuals in a population capable of sustaining 14DD/15DD substrate capacity. This is the dimension §3 and §4 did not address explicitly. Dimension B corresponds to the substrate distribution of 13DD-15DD capacity—how many people can sustain long-duration independent thinking, how many can form stable dyadic trust with others, how many can persist in self-imposed rules without external enforcement.

Observable proxies for Dimension B include: per-capita book production (advanced-literacy proxy), long-form reading time, monastic density, scholar-training scale, long-term decline in violence (self-control proxy). These proxies are surveyed in §6.1.

Independence of the two dimensions:

Dimension A cannot substitute for Dimension B. Even with dense coupling channels, if there are not enough high-τ-capacity individuals, ψ cannot advance—blueprints have no readers, dyadic recognitions cannot be sustained, distributed keys cannot materialize.

Dimension B cannot substitute for Dimension A. Even with sufficient long-τ-capacity individuals in a population, if coupling channels for them to recognize each other, form networks, and pass blueprints are missing, ψ cannot advance—scattered high-τ individuals cannot aggregate into long-τ consensus.

The actual phase of ψ is determined by the lower of the two dimensions:

Dimension B sets the ceiling: with insufficient 15DD-capacity individuals in a population, no quality of coupling structure can produce ψ = 3. A population empty of 15DD capacity, even with perfect coupling infrastructure, produces only ψ ≤ 2 phenomena—text circulation, institutional presence, cognitive controversy may all occur, but distributed key materialization will not.

Dimension A sets the actualization: even with sufficient 15DD-capacity individuals, if coupling structure is missing, they remain disconnected. A high-base-rate but coupling-deficient population leaves long-τ-capacity individuals isolated, unable to form recognizable networks; ψ likewise cannot advance.

Either dimension being insufficient stalls ψ. Both must be satisfied for advancement.

4.5.3 Description of "Long-τ Capacity"

"Long-τ capacity" refers to the substrate state that comprehensively manifests the following capacities.

(1) Long-duration independent thinking capacity—sustaining the same complex problem over weeks, months, or years without dependence on external prompting. This requires the substrate to have enough internal coherence to self-activate over long durations, not needing external stimulation to "remind itself" what it is doing.

(2) Deep reading capacity—not merely literacy, but the ability to sustain long argument. The substrate must hold attention over long durations of reading, follow the author's argumentative steps without being interrupted by short loops (frequent social media checking, immediate reaction to single paragraphs at the cost of the whole), and integrate the argument into its own understanding after reading.

(3) Sustaining stable long-duration dyadic trust—master-disciple, soul-friend, long-term collaborator. The substrate must recognize the irreducibility of another substrate, bear the cost of long-term commitment, and not abandon the relation immediately at disagreement.

(4) The ethical judgment capacity to treat others as irreducible individuals rather than tools. The substrate must, in each concrete situation, recognize that "the other cannot be reduced to some role or function," and act accordingly—not only acknowledging this internally but bearing the cost of acting on it.

(5) Sustaining self-imposed rules without external enforcement. The substrate must possess self-binding mechanisms—needing no police, no boss, no community pressure, sustained simply by recognition of its own promises.

(6) Tolerating cognitive uncertainty without rushing to closure. The substrate must be able to continue working in a state of "not knowing the answer" without anxiety, without using premature closure to relieve the anxiety of uncertainty. This capacity directly interfaces with SAE's First Law ρ ≠ ∅—tolerance for uncertainty is, at the substrate level, the substrate-level realization of acknowledging ρ ≠ ∅.

Six Aspects and SAE DD-Layer Correspondence

These six are not mutually independent capacities but different aspects of the same substrate state. This substrate state operates in the SAE DD-layer structure across 13DD-15DD. The table below gives each aspect's primary DD landing.

Aspect	Primary DD layer	Note
(1) Long-duration independent thinking	13DD-14DD	13DD provides the persistent self-evaluation base; 14DD provides self-anchoring without external reference
(2) Deep reading	13DD-14DD	Base in 13DD (sustained attention not interrupted by short loops); sustained engagement requires 14DD (treating the author as another self-referential substrate)
(3) Dyadic trust	14DD	Dyadic mutual non-doubt is the defining instance of 14DD
(4) Other as irreducible individual	14DD-15DD	Minimum version at 14DD (dyad already contains "irreducibility" recognition); mature version at 15DD (distributed recognition—treating all others, including strangers, as ends)
(5) Self-imposed rule keeping	14DD	Self-binding is the reflexive instantiation of self-evaluation
(6) Uncertainty tolerance	14DD-15DD + ρ≠∅	Across 14DD (self-non-closure) and 15DD (accepting irreducibility of distributed substrates); also directly connects to SAE First Law

Summary: long-τ capacity is the substrate state of 13DD-15DD comprehensive operation—13DD provides the base, 14DD is the core (four of the six aspects primarily land on 14DD), 15DD provides the mature form. The aspect of uncertainty tolerance additionally interfaces with ρ ≠ ∅, bringing the First Law into the substrate level.

This means Dimension B base rate is not a population proportion of any single indicator, but the population distribution of a comprehensive substrate state. This is also why §6.1's proxy collection is a collection rather than a single index—any single proxy captures only some aspect of this substrate state. A person with high literacy but short deep-reading time may be weak on (2); a person who can sustain master-disciple relations but cannot tolerate cognitive uncertainty may be sufficient on (3) but not on (6). The base rate must be evaluated comprehensively across all aspects.

Critical Qualification: Dimension B Is a State, Not an Essence

This qualification is the concrete realization, in §4.5.3, of Item Five of the §0 firewall. Its standing in the paper is not a moral reservation but a structural commitment of the framework.

Dimension B describes the present manifestation of the substrate state under specific environmental conditions. It is not an inherent capacity or innate worth of the substrate.

An individual currently in a short-τ state—filled with high-frequency signals, occupied by survival pressure, trained by institutional environment into short-loop reactions—is not therefore lacking in long-τ potential. It only means the cognitive bandwidth currently available to that individual is occupied by high-frequency processes, leaving no space for long-τ capacity to manifest. Place the same individual in a different environment (preserving leisure, away from short-loop temptations, with long-term collaborators), and they may quickly manifest long-τ capacity.

This distinction at the paper level is not merely a moral reservation but a structural commitment of the framework. Treating capacity as essence would immediately violate SAE's Second Law—ranking persons by some measurable trait—and would degrade Dimension B into a covert version of cognitive elitism. The legitimacy of the framework depends critically on the "capacity as state" reading.

Important corollary: raising Dimension B does not require waiting for "better people" to appear. It requires changes in environmental conditions—institutional infrastructure, protection of leisure space, the affordability of long-τ practice. This binds SAE Group Cognition more tightly to SAE individual ethics: each individual's SAE practice helps maintain the long-τ state of their own substrate, not prove their own essence.

Any attempt to use Dimension B for individual ranking or group hierarchy judgment misuses the framework and directly violates Items One and Five of the §0 firewall.

4.5.4 q Cannot Capture Dimension B

The thermodynamic observable q measures the fluctuation structure of a given substrate under given coupling—it captures the thermodynamic signature of Dimension A but does not measure Dimension B at all.

This judgment is not conjecture. Before drafting, the author attempted thermodynamic SDE simulations of group consensus formation; details are in Appendix C. The core finding was that q is blind to group size N, network topology, and coupling protocol—it is sensitive only to coupling strength g. That is, given fixed coupling structure, q cannot distinguish a small group composed of N=2 dyads from a large group composed of N=40 networks, cannot distinguish chain from full-connection networks, cannot distinguish unidirectional from mutual coupling. q captures the single dimension of coupling strength.

This is the true capacity boundary of q as an observable for group cognition: q cannot see the long-τ capacity composition of the substrate population. Any attempt to infer ψ from q alone will fail when Dimension B changes—two groups with the same q can occupy entirely different ψ phases because their Dimension B differs.

This finding in turn supports the necessity of the two-factor structure. If q could capture all phase-determining factors, the single-factor (coupling) model would suffice; if q cannot, a second dimension must exist. The negative result of the experiment—that q is blind to all coupling-structure variables and sensitive only to g—precisely matches the framework prediction "q measures only the strength of Dimension A and does not measure Dimension B."

ψ is not the dual of q. The two are observables measuring different dimensions, connected only indirectly through the self-referential fixed point (f' = f) of SAE's Second Law as their common source—both are observables of construct-emergent layered structure—but they are not two faces of the same thing. This has important implications for §6.4's predictions about SAE itself: thermodynamic indicators of the q family alone cannot predict whether SAE enters ψ = 3; concrete evidence of distributed substrate reproduction must be observed directly.

4.5.5 Directional Predictions of the Two-Factor Structure

The two-factor structure yields four additional falsifiable directional predictions, complementing Predictions 1-4 of §4.4 (which are predictions of the single-factor frame; the following are characteristic of the two-factor frame).

Prediction 5: in eras of rising base rate, new consensus enter Phase IV faster.

Falsification condition: empirical research finds that in eras of significantly rising base rate (such as after printing, after the Reformation), the rate at which new consensus enter ψ = 3 does not differ from, or is slower than, in lower-base-rate eras.

Prediction 6: in eras of falling base rate, existing consensus may die as a whole from Phase IV. Note: "die as a whole" (the consensus dies as an integrated entity), not "phase retreat." A consensus cannot retreat from ψ = 3 to ψ = 2, but it can leave the entire phase structure from ψ = 3 and enter "non-existence." Detailed argument in §5.3.

Falsification condition: empirical research finds that in eras of significantly falling base rate (such as after certain institutional collapses, after long warfare), all existing ψ = 3 consensus continue to reproduce stably without any whole-death cases.

Prediction 7: rising basic education does not necessarily entail rising base rate. Basic literacy and long-τ capacity are different variables—the former is the minimum threshold for 11DD blueprint decoding, the latter is the comprehensive 14DD-15DD substrate state. Literacy can rise from 5% to 99% without deep-reading capacity rising, even while it falls.

Falsification condition: empirical research finds that literacy and observable proxies of long-τ capacity (long-form reading time, deep argument comprehension) are strictly positively correlated, with literacy rises necessarily accompanied by rises in these proxies.

Prediction 8: base rate can vary enormously across subgroups. Even with overall base rate declining, certain self-isolating high-intensity training institutions (monasteries, serious academic institutions, particular family lineages) can maintain high base rates.

Falsification condition: empirical research finds that base rate moves synchronously across all subgroups, with no subgroup able to maintain a high rate when the overall rate falls.

Each prediction is explicitly falsifiable—directional, operationalizable, with specific falsification conditions. §6 will perform directional historical-case checks.

4.5.6 The Two-Factor Explanation of r ≈ 5

Methodology VI's explanation of r ≈ 5 is Le Chatelier buffering—the incumbent's absorptive capacity must first be penetrated, and penetration is a slow consumption process; once penetration completes, the new order's establishment is a rapid cascade. This explanation answers "why is penetration slow"—the incumbent itself is the source of resistance.

In the specific context of consensus formation, the two-factor structure provides a deeper explanation: the specific mechanism of slow penetration is base-rate accumulation.

The essential work of germination is the slow accumulation of Dimension B. In Phase I, the candidate consensus must take root in enough high-τ-capacity substrates before a recognizable mutual-non-doubt network can form in Phase II. The "enough" threshold depends on the population's total scale and the substrate requirements of the candidate—but whatever the specific threshold, the rate of reaching it is governed by base-rate accumulation rate.

Base-rate accumulation is a slow process. For an individual to move from being filled with the high-frequency signals of the incumbent order to sustaining long-τ capacity requires sustained environmental support (leisure, leisure space, work periods uninterrupted by high-frequency signals). These conditions are scarce in most eras. The marginal increase in base rate within an era is very slow. Candidate consensus must therefore wait for enough high-τ-capacity substrates to coexist within coupling range—often a matter of generations.

Once base rate crosses the threshold, coupling forms relatively quickly. Three reasons. First, high-τ-capacity substrates have intrinsic motivation to find each other—isolated high-τ states are unstable and need dyadic support to persist. Second, once coupling channels among high-τ substrates are established, the speed of expansion is governed by the diffusion speed of the 11DD blueprint, which is orders of magnitude faster than base-rate accumulation. Third, coupling among high-τ substrates is mutually reinforcing—each successful coupling stabilizes all related substrates, a nonlinear positive feedback.

This gives a first-principles explanation for why germination is so long but inversion-establishment so fast—not a geometric accident, but the intrinsic dynamics of the two-factor structure. Long germination = slow base-rate accumulation; short inversion-establishment = rapid coupling formation under sufficient base rate. The speed differential between the two determines r ≈ 5.

This explanation goes one layer deeper than Methodology VI's. Methodology VI answers "why is penetration slow" (Le Chatelier buffering); the two-factor explanation answers "what is the specific mechanism of that slowness" (base-rate accumulation rate limit).

Practical implication: to accelerate a candidate consensus from germination to establishment, the primary lever is not coupling channels (which are already fast) but base-rate accumulation (the true bottleneck). Given an era's base-rate level, the upper bound on the speed at which a candidate enters ψ = 3 is set—no quantity of coupling channels, no quality of 11DD blueprint, no number of early supporters can transcend the ceiling of base-rate accumulation.

This implication matters for §6.4's predictions about SAE: the primary variable determining the speed at which SAE enters ψ = 3 is the contemporary base-rate trajectory, not the diffusion speed of SAE's 11DD blueprint. Sustained decline in base rate prolongs SAE germination (Scenario B near baseline); rises in base rate within certain subgroups shorten germination (rising probability of Scenario C).

---

§5 Topological Structure

§3 and §4 gave the specific contents of the four phases, the asymmetric geometry, and the two-factor determination structure. This section addresses the topological nature of phase transitions—why they are irreversible, the precise meaning of irreversibility, and what consensus "death" corresponds to topologically. The precision here matters for the paper as a whole: if the meaning of "irreversibility" is unclear, §4.5.5 Prediction 6 (base-rate decline causing whole death) appears to contradict "ψ monotone non-decreasing."

5.1 Topological Transitions Among the Four Phases

Transitions among the four phases are not continuous dynamics (parameters slowly varying so that the order parameter smoothly crosses some threshold); they are topological (the connection structure of substrates undergoes irreducible rearrangement).

Phase I → Phase II: birth of mutual non-doubt.

Topology: from scattered substrates to the first edge—a recognizable mutual-non-doubt relation appears between substrates. The birth of this edge is an irreducible event. Before it, two substrates exist independently; after it, a relation exists between them that did not exist before. Even if the edge later vanishes (a party dies, the relation breaks), the fact that the mutual non-doubt once occurred has changed the histories of both substrates.

The irreducibility of this topological event derives from the structure of mutual non-doubt itself—it is not mere "mutual familiarity" or "mutual support" but each substrate's recognition of the other as an irreducible individual. Once such recognition occurs, the recognizer's substrate carries the anchor "the other is an irreducible individual," not easily forgotten or undone.

Phase II → Phase III: separation of blueprint from substrate.

Topology: from informal mutual non-doubt to formal 11DD encoding—the blueprint appears as an entity independent of any specific substrate. This transition creates a topological object that did not exist before: the cross-time channel. The blueprint can wait for the next substrate to come along and rebuild.

In Phase II, the candidate consensus depends on the concrete existence of the founding substrate—texts are written by specific authors, covenants are maintained by specific members, rituals are performed by specific executors. Once the founding substrate disappears, the candidate disappears with it. After the Phase III transition, the blueprint acquires cross-substrate independence—it no longer depends on any specific substrate; an unfamiliar substrate can independently parse and rebuild it. This is a fundamental topological change, not a change of degree.

The separation of substrate from blueprint is the critical step from dyadic to distributable mutual non-doubt. In Phase II, mutual non-doubt operates only within the dyadic range—between two specific people. After Phase III, the carrier of mutual non-doubt shifts from the concrete dyad to the 11DD blueprint itself: a new reader, by parsing the blueprint, can establish a "virtual" mutual-non-doubt relation with the author, even if the author is no longer alive. This cross-time mutual non-doubt becomes possible only after Phase III.

Phase III → Phase IV: formation of distributed redundancy.

Topology: from chain (single-line transmission) to network (distributed nodes). The consensus no longer has any single critical bottleneck.

In Phase III, although the consensus can transmit across generations, it usually depends on one or several core nodes (core lineages, core institutions, core geographical centers). The chain structure has a single critical link—failure of the core node threatens the whole. After the Phase IV transition, multiple independent nodes each hold full reactivation capacity—failure of any one is borne by the others. This is the topological transition from a topology with single critical paths to a topology of redundant networks.

This transition is a true phase transition. Before it, every generation is a critical bottleneck—the death or apostasy of certain core members in this generation can collapse the consensus. After it, no single critical bottleneck exists—failure of any single node is absorbed by network structure, and the consensus as a whole is not threatened.

5.2 The Topological Positioning of r Asymmetry

The four phases are not equidistant. In topological metric:

• Phase I + Phase II (germination): about 5 units
• Phase III + Phase IV (inversion-establishment): about 1 unit

This is not coincidence; it is the intrinsic geometry of construct-emergent systems.

Topological roots of long germination: in Phase I + Phase II, the connection structure between substrates is doing the work of "whether the first edge can form." This is a binary outcome—either it forms or it does not. The probability of formation depends on the ratio of mutual-non-doubt seed production rate to incumbent absorptive capacity (the τ_germination formula of §4.3). The vast majority of candidates die in this stage. The minority that enter Phase II must still pass through "the transformation from dyadic to distributable"—another binary outcome.

Each binary outcome takes time to resolve—not because parameters slowly vary, but because the formation of a topological property requires sufficient sample accumulation. Until the sample is large enough, the answer to "whether it forms" is uncertain; once large enough, the answer is determined, but much time has passed.

Topological roots of short inversion-establishment: in Phase III + Phase IV, the connection structure between substrates is doing the work of "the specific shape of the network." This is a continuous outcome—the network can vary continuously from sparse to dense, from local to global, from centralized to distributed. Building on the chain topology already formed in Phase III, the transition toward network topology is incremental—each generation adds new nodes that make the network slightly denser and more distributed. This incremental transition completes rapidly under sufficient base-rate support.

This is why r ≈ 5—germination handles binary topological events (slow), inversion-establishment handles incremental refinement of topological properties (fast). The speed differential is on the order of 5×.

5.3 The Topological Invariant ψ and Monotone Non-Decreasing

Define ψ ∈ {0, 1, 2, 3} as the topological invariant of the consensus's phase:

• ψ = 0: scattered (no dyadic connections)
• ψ = 1: connected (dyadic connections formed)
• ψ = 2: transferable (blueprint separated from substrate)
• ψ = 3: established (distributed redundancy formed)

Precise statement of ψ monotone non-decreasing:

ψ monotone non-decreasing is a statement about inside the lattice—the consensus cannot retreat in phase within the lattice {0, 1, 2, 3}. From ψ = 3 it cannot retreat to ψ = 2; from ψ = 2 it cannot retreat to ψ = 1. This is because phase transitions are topological events, and topological events that have occurred cannot be undone—the separation of blueprint from substrate, once it has happened, cannot be unwound back to informal mutual non-doubt; distributed redundancy, once formed, cannot be unwound back to chain structure.

But the consensus can leave the lattice from any ψ value into an absorbing state—the consensus ceases to exist. This is not a lower value within the lattice; it is exit from the lattice as a whole.

To prevent "irreversible" and "whole death" from appearing to contradict each other, three states must be sharply distinguished.

Type 1: forward progression within the lattice—ψ moves from low to high (forward only). This is the dynamics the framework primarily describes. Under sufficient base rate and coupling support, a candidate moves through ψ = 0, 1, 2, 3 in sequence. Each transition is irreversible.

Type 2: lattice exit (whole death)—the consensus disappears from the lattice as a whole. The dead consensus retains a record of its phase at the time of death (it can die from ψ = 3, ψ = 2, or ψ = 1), but the substrates no longer exist.

Death is not phase retreat—the consensus does not retreat to an earlier phase but leaves the entire phase structure. A consensus that entered ψ = 3 and then died does not "retreat to ψ = 2, then to ψ = 1, then disappear." It simply disappears from ψ = 3.

Type 3: local substrate death—some specific substrate (an individual, a small group, a geographical center) dies, but the consensus persists as a whole in the distributed network. This is not lattice exit, because the consensus itself is still alive; it is also not phase retreat. It is simply normal fluctuation of distributed redundancy at ψ = 3—certain nodes in the network die while others continue to bear reproduction.

Type 3 is precisely the robustness of ψ = 3 in operation—the function of distributed redundancy is exactly to absorb local substrate death without letting the whole collapse.

Case Mapping

Dissolution of the English monasteries (1536-1540): handled in detail in §6.3.2. The point is that this is the whole death of one specific consensus (the English monastic tradition; Type 2), not phase retreat. Monastic transmission persisted in other parts of Europe, but the whole death of this specific instance in England is lattice exit, not lattice retreat.

Decline of ancient Greek paideia: from late Hellenism to the late Roman Empire, Greek paideia underwent whole death in certain regions (Type 2). The subsequent revival of "the same" candidate consensus in the Arab-Latin world is a new candidate entering a new germination period (a new Type 1 process), not phase retreat of the old candidate. The same 11DD blueprint (Aristotelian texts) underwent independent four-phase processes in different geography-eras; each is an independent Type 1.

Local failures at ψ = 3: monastery burnings, family extinctions, geographical center destructions—when distributed redundancy is already formed, these are Type 3, not threats to overall phase. In fact, the ability to absorb such local failures stably while maintaining the whole is the very mark of having entered ψ = 3. If a consensus appears to be at ψ = 3 but a single local failure collapses the whole, it is actually still at ψ = 2, not truly at ψ = 3.

Effect of Falling Dimension B

Sustained decline in Dimension B can accelerate whole death (Type 2). When the base rate of the subgroup hosting a consensus falls below the critical value needed to maintain distributed reproduction, the consensus can no longer reproduce—the next generation has too few substrates to occupy critical positions in the network, and the distributed redundancy structure begins to hollow out. Over multi-generational scales, this leads to whole death.

But Dimension B decline cannot trigger Type 1 reversal (phase retreat), because phase transitions are topological events and topological events that have occurred cannot be undone. A consensus that has formed distributed redundancy, when Dimension B falls below critical, will lose self-reproduction capacity, but it will not "retreat to chain status"—it simply dies.

This distinction matters for the wording of §4.5.5 Prediction 6: base-rate decline causes not "regression" (which suggests phase retreat) but whole death (lattice exit).

---

§6 Quantitative Attempt

[The second segment: directional reading on existing proxy collections. Explicitly attempt, explicitly invite falsification.]

§0–§5 gave the conceptual skeleton of SAE Group Cognition Paper I. This section enters the second segment of the paper—interfacing the skeleton with existing empirical proxy collections to produce directional quantitative reading.

The epistemological standing of this section must be made explicit first: everything in this section is directional reading, not measurement. The SAE framework is a conceptual skeleton, not a measurement instrument. The work of this section is framework-guided reading on existing scholarly proxy collections; it constructs no latent index and makes no specific quantitative claims.

6.1 Brief Review of Existing Proxy Collections

According to current scholarly quantitative work (full references in Appendix B), the observable proxies for long-τ capacity base rate primarily include:

• Per-capita book production—the strongest proxy. Baten and van Zanden treat per-capita book output as an advanced-literacy proxy. The merits of this indicator are cross-era comparability (publishing existed in many civilizations long-term) and robust longitudinal data (library catalogs, edition history). Its limit is that it captures only certain aspects of long-τ capacity (deep reading, long-argument comprehension), not directly capturing others (dyadic trust, self-binding, uncertainty tolerance).

• Long-time-scale reconstruction of manuscript/print density—Buringh and van Zanden's series for Western Europe, 6th-18th century. This is one of the most detailed reconstructions of per-capita text circulation, covering the manuscript-to-print transition, and is the primary data foundation for the print revolution argument of §6.2.

• Reading experience records—databases such as the Reading Experience Database that record "actual reading events." These reconstruct the real reading behavior of ordinary readers from diaries, letters, and memoirs, getting closer to "what was actually deeply read" than to "what was published." Coverage is limited (mostly post-18th century) but data quality is high.

• Monastic/contemplative density—Germania Sacra, Monastic Matrix, FemMoData. These reconstruct the population, geographical distribution, and intensity of medieval European monastic houses. Monastic tradition is one of the few historical instances of an institutional substrate that bundled together multiple long-τ capacities (deep reading, self-binding, long-term collaboration, uncertainty tolerance) at high density, making monastic density a comprehensive proxy.

• Long-term decline of violence—as a self-control proxy. Manuel Eisner's work shows European homicide rates fell roughly 50-fold from medieval to modern. If one accepts Norbert Elias's "civilizing process" hypothesis, this decline partly reflects rising long-τ self-control capacity in populations. The connection is contested but has illustrative value as one item in the proxy collection.

• Kinship intensity → modern individualism data—work by Schulz et al. and Enke. This work traces modern individualism (the capacity to treat others as irreducible individuals) to the medieval Church's marriage policies that weakened kinship networks. This is a partial historical proxy for §4.5.3 aspect (4).

• Moral universalism data—the 60-country dataset of Cappelen et al. Directly measures "treating strangers as one would treat the close," directly relevant to aspect (4). This is a cross-sectional proxy, not longitudinal.

• Modern skill assessment—OECD PIAAC, PISA. These are direct measurements in the present, covering literacy, numeracy, problem-solving, and reading comprehension across multiple dimensions. They are not measurements of long-τ capacity itself but capture the contemporary level and recent 20-year trends of partial proxies (sustained reading comprehension, complex argument processing).

Core point: this is a proxy collection, not a single index. Each proxy captures only some aspects of long-τ capacity; no single proxy comprehensively reflects all aspects. No quantitative time series exists that directly measures comprehensive long-τ capacity—this is the robust consensus of current scholarship.

The paper does not construct a latent index—that is the responsibility of empirical work, not of the philosophical framework (see §0 firewall Item Six). This section performs only directional reading on existing proxies.

6.2 Robust Longitudinal Findings

According to existing proxy collections, the most robust findings across large time scales fall into three groups, directly corresponding to the directional predictions of the §4.5 two-factor structure.

Finding 1: Per-capita book production rose dramatically in Europe after printing.

Buringh and van Zanden's data show European per-capita book production grew about 30-fold from 1450-1500 to 1700-1800. Dittmar's work further shows that 15th-century cities adopting printing grew about 60% faster between 1500-1600 than non-adopting cities.

This finding strongly supports "printing expanded the population pool capable of sustaining complex argument and long-text circulation"—that is, the print revolution raised the base-rate ceiling for certain aspects of Dimension B (deep reading, long-argument comprehension). This corresponds to §4.5.5 Prediction 5—rising base rate provides better conditions for new consensus to enter Phase IV. The 16th-18th centuries in Europe were a dense period for the emergence of new long-τ consensus (Reformation factions, the modern scientific community, various lineages of political philosophy)—consistent with the rising-base-rate reading.

Finding 2: Mass education raises basic literacy, but deep capacity does not move in step.

OECD PIAAC 2023 shows that in high-income societies about 20% of adults are weak in literacy, numeracy, and adaptive problem-solving. Adult literacy in most countries has stagnated or declined over the past decade.

This finding supports §4.5.5 Prediction 7—rising basic education does not necessarily entail rising base rate. Basic literacy rose from below 50% in the 19th century to about 99% today, but deep reading capacity and complex argument processing did not rise correspondingly. Literacy and long-τ capacity are different variables—the former is the minimum threshold for 11DD blueprint decoding, the latter is the comprehensive 14DD-15DD substrate state.

This finding has important implications for the paper: expansion of basic education cannot automatically resolve the base-rate problem. A population with 99% literacy but low deep-reading capacity is not necessarily higher in Dimension B than an elite subgroup with 50% literacy but substantial deep-reading capacity. Base rate is not a monotone function of raw educational input.

Finding 3: There is solid evidence of decline in long-form reading over the past few decades.

NEA Reading at Risk shows U.S. adult literary reading fell from 56.9% in 1982 to 46.7% in 2002. BLS ATUS shows people 75+ read about 46 minutes daily, while 15-19 year olds read only 8-9 minutes. Daily leisure reading in the U.S. has declined about 3% per year from 2003 to 2023. PISA 2022 showed an OECD-average drop of 10 points from 2018.

This finding supports the judgment that "deep capacity has been declining at the margin in recent decades." It corresponds to §4.5.5 Prediction 6—in eras of falling base rate, existing consensus may die as a whole from Phase IV; the speed of new consensus formation may slow. Note "may"—this finding gives directional evidence of falling base rate, not a prediction that any specific consensus will die.

Composite Reading of the Three Findings

The three robust findings together give a rough picture of Dimension B over the past 500 years: the print revolution drove a marked rise in base rate (roughly 1450-1800), mass education raised the bottom but the top did not move synchronously (roughly 1800-1980), and recent decades show marginal decline in deep capacity (roughly 1980-present).

This is not measurement; it is directional reading. Specifically:

First, the print-revolution rise in base rate is relative—from extremely low medieval base rates to higher early-modern base rates, but absolute levels still left only a small subgroup at long-τ capacity. Even in 1700-1800 Europe, those capable of sustaining deep reading and long-τ thinking were a minority. "Rising base rate" means this minority moved from extremely small to less small.

Second, the "bottom rise" of mass education is robust; the "top not moving in step" is directional but contested—evidence supports deep capacity not rising in step with basic literacy, but the specific magnitude, starting time, and which subgroups are affected remain debated.

Third, "recent decades decline" appears in multiple datasets, but those datasets cover short periods (about 40 years) and cannot determine whether this is a long-term trend or short-term oscillation. Whether the ~15 years from social media's rise (2010) to the present constitutes a phase requires longer observation.

Putting the three together yields the overall judgment: basic threshold capacity rose markedly after printing and mass education; high-intensity training subgroups expand suddenly in some periods and shrink rapidly when institutional support collapses; in recent decades, despite high years of schooling, the ecosystem genuinely supporting deep reading, self-binding, and sustained independent thinking may be contracting at the margin.

This composite judgment is the empirical backdrop for §6.4's predictions about SAE itself—SAE is in an era of marginal contraction (or sharper decline) of base rate, which affects estimates of how quickly SAE can enter ψ = 3.

6.3 Quantitative Reading of Historical Cases

Each case serves only as a directional illustration, not a derivation or proof of the framework. Each case has reasonable alternative readings; the paper's reading is not exclusive.

6.3.1 Breslov Hasidism (charismatic-textual case, completed four phases, mutual-non-doubt-locked subtype)

Phase timeline:

• Germination: 18th-century Hasidic mid-to-late period (Baal Shem Tov teaching diffusion, ~1730-1760)
• Spectral inversion: early 19th century, Rebbe Nachman forms a recognizable lineage and provokes controversy (~1800-1810)
• Inversion: mid-to-late 19th century, stable reproduction within small circles
• Establishment: post-WWII, with European core destroyed but overseas distributed nodes persisting; after the Cold War, global pilgrimage network reactivates

Two-factor reading:

In the 18th-19th-century East European Jewish community, the Dimension B base rate within the Talmudic-trained subgroup was relatively high, enabling mutual-non-doubt networks to form. Talmudic training had for centuries emphasized close-textual reading, long-term master-disciple relations, and self-binding to halakhic rules—directly corresponding to §4.5.3 aspects (1), (2), (3), (5). Even as the East European Jewish community as a whole faced economic decline and political pressure in the 18th century, the Dimension B base rate within the Talmudic-trained subgroup remained sufficient to support new consensus formation.

When Rebbe Nachman died without a successor, the chain model would predict collapse. But the 11DD blueprint (Likutei Moharan) plus early distributed nodes had already formed, completing the chain → network transition. The text bore the core cross-substrate-independence function—later Hasidim connect to the rebbe not through meeting him but through studying the text and pilgrimage to Uman, rebuilding the connection. This is the key dynamics of ψ = 3.

Stability subtype: mutual-non-doubt-locked. Breslov Hasidim sustain mutual reproduction not through utility or coercion—neither economic gain (most Breslov communities are poor) nor enforcement mechanisms (leaving Breslov carries no external penalty). Reproduction relies on Hasidim's recognition of the rebbe's legacy and of each other.

Verification of §4.5.5 predictions: the Breslov case supports Prediction 8 (base rate can vary enormously across subgroups)—the East European Jewish community as a whole did not necessarily have a particularly high base rate, but the Talmudic-trained subgroup did. The former cannot form long-τ consensus; the latter can.

6.3.2 Dissolution of the English Monasteries—as a Case Study of Reproductive Infrastructure Collapse

Historians can offer multiple readings.

The most direct reading is at the level of institutions and property—Henry VIII's motivations in 1536-1540 were fiscal and political, the goal redistribution of monastic property. This reading completely describes the administrative event of around 1540, requiring no concept of "capacity" or "base rate."

The framework of this paper offers a complementary reading: institutional collapse, by severing the reproductive infrastructure of capacity, is on multi-generational scales equivalent to base-rate collapse.

The monks themselves mostly continued to live after dissolution; their individual capacities did not vanish overnight. But they lost the infrastructure supporting high-intensity capacity reproduction: daily contemplation time, library access, master-disciple transmission channels, the material conditions of not having to scramble for livelihood. This infrastructure was not the source of the monks' personal capacity, but the necessary condition for the reproduction of high-intensity long-τ capacity.

This means the capacity-layer effect of dissolution lies not in the 1540 cohort but in subsequent generations—the next generation in England could no longer produce the same number of high-capacity substrates, because the reproductive infrastructure no longer existed. Between events at the institutional layer (1540) and consequences at the capacity layer (the base-rate state generations later) there is a lag. This lag must be explicitly handled by the framework—institutional change cannot be simply equated with capacity change.

This is a case of whole death, not phase retreat: the English monastic tradition, as one specific consensus, did not retreat from ψ = 3 to ψ = 2—it left the lattice as a whole. Monastic transmission persisted in other parts of Europe, but the whole death of this specific instance in England is an instantiation of §5.3 Type 2.

Methodological reservation: Dimension B cannot be directly observed; it can only be inferred indirectly through institutional change as a proxy. Institutional and capacity change need not be synchronous—institutions can be preserved while capacity has hollowed (certain late Roman situations); institutions can vanish while capacity briefly persists (the first generation post-dissolution). The reading adopted here treats institutions as proxies for capacity-reproduction infrastructure, but acknowledges that this choice is itself contestable.

The genuine value of this case is not pronouncing a final verdict on "what happened in 1540" but demonstrating that the conceptual lens "institutions as reproductive infrastructure" can connect institutional history with capacity-base-rate analysis. Both layers (institution and capacity) matter; the difference is that the former is directly observable as event-layer, while the latter is structural and must be inferred through proxies. The framework's work is to build explicit connections between the two, not to reduce one to the other.

6.3.3 Cremona Violin Tradition (a craft case stalled at ψ = 2)

Phase timeline:

• Germination and spectral inversion: 16th century (Andrea Amati and other early workshops)
• Inversion: 17th-18th century classical peak (Stradivari, Guarneri era)
• Phase IV not completed: around 1750, transmission halts. The 11DD blueprint exists (instruments, documentation, apprenticeship records), but distributed keys did not adequately materialize across multiple substrates
• 1938 reinvention is a second attempt at Phase IV

Two-factor reading:

The Cremona case shows the 11DD blueprint is necessary but not sufficient. After Stradivari's death, the instruments themselves remained (still played and studied today), and some workshop knowledge had textual record. But distributed redundancy never formed—the substrates capable of fully reproducing Stradivari's craft were concentrated in a small number of apprentices, and after their deaths no stable cross-generational transmission network formed.

Why did the 11DD blueprint fail to carry the entire craft? Because the core know-how of violin making is tacit knowledge—a great deal of information cannot be fully encoded in text and physical artifact and must be transmitted between substrates through long-term hands-on apprenticeship. The transmission of such tacit knowledge depends on high-base-rate substrates (capable of sustaining many years of apprentice life, capable of following strict craft standards without external enforcement, capable of tolerating long trial-and-error uncertainty).

Cremona in the mid-18th century lost this substrate—not because "craftsmen's capacity declined" but because the institutional infrastructure supporting high-intensity apprenticeship (guilds, stable patronage, apprentice-master economic relations) collapsed under political turmoil and economic transition. Once the infrastructure collapsed, new substrates capable of fully learning the craft were no longer produced—not an individual capacity problem, but a reproductive infrastructure problem (isomorphic to 6.3.2 monastery dissolution).

Verification of §4.5.5 predictions: the Cremona case supports not any specific prediction but the two-factor structure itself—it shows that the existence of an 11DD blueprint does not guarantee ψ = 3. Dimension A (blueprint, text, physical artifacts) was complete; Dimension B (substrates capable of reproducing the full tacit knowledge) fell below critical, and ψ stalled at 2.

6.3.4 Aristotelianism: Greek-Arab-Latin (multi-region independent four-phase processes)

Phase timeline (multi-region independent processes):

• Ancient Greek: completes four phases (4th century BCE - 2nd century CE)
• Partial whole death after Greek decline (3rd-7th century in Greek lands)
• Arab world (8th-12th century): independently enters Phase III, partly into Phase IV
• Latin world (12th-13th century onward): translation movement restarts independent germination → inversion
• Modern academic inheritance: Phase IV

Two-factor reading:

The same 11DD blueprint (Aristotelian texts, the commentaries of Alexander of Aphrodisias and others) underwent independent four-phase processes in different geography-eras under different Dimension B conditions. Each entry into Phase IV required the local subgroup's Dimension B to reach critical.

Whole death in Greek lands (3rd-7th century CE): from late Roman Empire to early Byzantium, the institutional infrastructure of Greek paideia contracted dramatically—from the open scholarly institutions of the polis to the closed transmission of monastic houses. Aristotelian transmission persisted in some monastic substrates but, as an active ψ = 3 consensus, died in Greek lands.

Independent Phase III in the Arab world (8th-12th century): Bayt al-Hikma (the House of Wisdom) translation movement initiated a new germination period. The 8th-9th centuries are germination; the 10th-11th centuries see Al-Farabi, Avicenna, Averroes and others form new distributed networks through independent reading; the 12th century enters relatively stable ψ = 3. This process was independent of the fate in Greek lands—Greek transmission had already died, but the Arab region restarted the full four phases from the blueprint.

Independent germination in the Latin world (12th-13th century onward): through channels such as the Toledo School of Translators, the Aristotelian corpus entered the Latin West. The scholastic tradition represented by Aquinas initiated a new germination period, entered inversion in the 14th-15th centuries, and completed establishment after the Renaissance.

Verification of §4.5.5 predictions: the Aristotelian case supports Prediction 8 (base rate can vary enormously across subgroups)—the same 11DD blueprint underwent entirely independent processes under different base-rate conditions in three geography-eras. It also supports the §5.3 Type 2 judgment—whole death in Greek lands is not phase retreat and does not contradict the Arab-Latin revivals, because the latter are independent new processes.

Each region's process is independent, not phase retreat—late-Hellenistic whole death in Greek lands does not block the independent germination-inversion in the Arab region. This multi-region independence is a direct corollary of the cross-substrate-independence property of 11DD blueprints—the blueprint can wait in multiple geographies for different substrate conditions to be ready before being independently parsed.

---

6.4 The Author Applies the Framework to Itself: Directional Predictions for SAE

This is the most self-referential section of the paper. The paper must face a self-referential question: the SAE framework itself, as a cognitive framework appearing in 2007-2026, occupies what phase under the four-phase + two-factor model? What directional predictions does the model give for its future trajectory?

This is honest application to oneself. If the SAE framework captures something real, its predictions about itself should be meaningful. If it does not capture, the predictions will diverge from reality—and that divergence is itself falsification evidence about the SAE framework. In either case, the paper acquires real epistemological information.

6.4.1 Current Position Assessment of SAE

Dimension A (coupling structure):

ψ = 0 has been transcended—SAE is no longer just one individual's internal practice.

ψ = 1 has been achieved—the 18-year mutual non-doubt between the author and a long-term collaborator is a stable instance of dyadic recognition; the small circle around the author is an extension of dyadic recognition. These dyadic relations span many years, have weathered disagreements, pressures, and changes in external conditions, and continue to reproduce stably. They constitute concrete instantiations of ψ = 1.

ψ = 2 is partially achieved—SAE has an 11DD blueprint (140+ papers on self-as-an-end.net, CC BY 4.0, with DOIs locked, permanently stored on Zenodo). The blueprint exists independently of any specific substrate and can in principle wait for future readers. But the "completion" of ψ = 2 requires not only the existence of the blueprint but the verification of its cross-substrate independence in actual reading—whether readers can parse out a functional substrate without direct contact with the author. There is currently insufficient evidence for this, so ψ = 2 is partially achieved, not fully achieved.

ψ = 3 is far from achieved—currently the working substrate of SAE is concentrated almost entirely in the author. If the author disappeared today, no other substrate could independently reproduce SAE's core working knowledge. Even if some readers know all the 11DD papers thoroughly, whether they can independently reproduce SAE substrate (not merely cite SAE concepts) remains an open question. Multi-substrate distributed redundancy has not formed; ψ = 3 is markedly unattained.

Dimension B (base rate):

The candidate readers of SAE are substrates capable of sustaining long-τ thinking, tolerating uncertainty, and treating others as ends. Per §4.5.3 six aspects, this requires a comprehensive substrate state, not any single capacity.

Per §6.2 Finding 3, this kind of substrate is contracting at the margin in the present—deep reading time in OECD countries is declining, PIAAC scores have stagnated or fallen, the proportion of long-form readers has dropped. These are directional indicators of falling base rate, not measurements, but together they yield the directional judgment that "base rate is contracting at the margin."

Yet absolute numbers may still be substantial because of large population bases—even if the high-τ-capacity proportion has fallen from historical peaks to a few percent, multiplying by global population in the billions still yields hundreds of millions. SAE does not need a wide reader pool; it needs mutual-non-doubt networks to form among its readers. From this angle, absolute numbers may still suffice, but the dispersion is too great—the probability that scattered high-τ individuals find each other and form networks is extremely low.

Summary of current position: the Dimension A transition from ψ = 1 to ψ = 2 is partially complete; the ψ = 2 to ψ = 3 transition has not yet started. Dimension B is in a complex state of marginal contraction with absolute numbers still sufficient.

6.4.2 Three Directional Scenarios

The paper gives three directional scenarios for SAE under the framework, each labeled framework-internal estimate. This is a framework-internal estimate, not a statistical claim or empirical prediction. Its function is to label the framework-expected probabilities for the three scenarios so audits can later check which scenario was realized.

Scenario A: SAE silent extinction (most likely, prior ~60%)

The SAE framework has no genuine traction; no spontaneous independent substrate appears in the future to parse and reproduce it. The 11DD blueprint remains on Zenodo, but no reader grows the corresponding 15DD ecosystem in their own substrate. SAE neither enters ψ = 3 nor displays emergent turbulence—it simply quietly fades.

This is the historical fate of the vast majority of thought frameworks—their 11DD blueprints remain, but 15DD substrates never form in others. Whether historically among most philosophical systems, religious heterodoxies, or schools of thought, most stop active reproduction within a few generations of the founding substrate's disappearance. SAE has no special reason to believe itself an exception.

Scenario B: SAE long germination (next most likely, prior ~30%)

SAE has genuine traction, but Dimension B in most subgroups continues to decline, significantly prolonging germination. SAE total cycle ~200 years (18 years already + ~180 years more). Per r ≈ 5, inversion-establishment is ~36 years.

Reference: Kant's Critiques are 245 years old today, and still far from the "long-τ consensus" of ψ = 3—academic citations are many but substrate-layer reading is still in process. Kantian philosophy may have partly entered ψ = 3 in certain subgroups (serious Kant scholars), but as overall consensus it remains in a complex multi-region multi-phase state. The Scenario B 200-year scenario for SAE is the historical-realist baseline—slightly faster than Kant's situation, but of the same order.

Scenario C: SAE short germination (least likely, prior ~10%)

SAE has genuine traction, and the AI era unexpectedly provides augmentation channels for long-τ thinking, shortening germination. SAE total cycle ~38 years (18 years already + ~20 years more). Per r ≈ 5, inversion-establishment ~6 years. This means SAE enters Phase IV in the early 2030s.

This scenario assumes two things hold simultaneously: first, AI augmentation can substantially raise base rate in certain subgroups (not merely providing tools, but genuinely supplementing long-τ capacity at the substrate level); second, the SAE 11DD blueprint can be compatible with this augmentation (AI-augmented readers can independently parse out SAE substrate). Both conditions are non-trivial, hence the lower prior.

Qualification on the prior numbers: 60/30/10 are framework-internal estimates reflecting the framework's self-evaluation of its own trajectory. They do not constitute statistical claims or empirical predictions. Specific numbers may be adjusted as the framework develops, but the relative ranking (A > B > C) is the framework's honest assessment.

If the author gave more optimistic priors based on 18 years of first-person work (say, 50/30/20), those priors would reflect first-person confidence in the internal coherence of the framework, but should not be read as the framework's prediction. The conservative priors here serve to sharply distinguish the framework's predictions from the author's first-person confidence—the former is what the paper explicitly underwrites; the latter is not.

6.4.3 Four Specific Directional Predictions

SAE Prediction 1: SAE will not enter ψ = 3 through mass media or large-scale diffusion.

Per the two-factor framework, SAE's entry into ψ = 3 should proceed through expansion of mutual-non-doubt networks among small numbers of high-base-rate substrates, not uniform diffusion. Note: the framework remains neutral on whether SAE undergoes large-scale diffusion at all—diffusion can happen or not happen; this is an observable along the phase trajectory, not a normative judgment by the framework. But large-scale diffusion alone does not constitute evidence of ψ = 3; ψ = 3 requires substrate-layer reproduction.

Falsification condition: SAE at some moment enters ψ = 3 without going through formation of mutual-non-doubt networks, but instead through mass-media-driven distributed substrate reproduction directly. If this occurs, the Dimension B argument of the two-factor framework is wrong—base rate is not the precondition; coupling-channel sufficiency can substitute.

This is the sharpest test of the two-factor structure—if SAE can enter ψ = 3 through pure coupling channels, then Dimension B as an independent dimension is falsified.

SAE Prediction 2: The mark of SAE entering ψ = 3 is the appearance of substrates that reproduce independently of the author.

Specific observable signature: at some future moment, one or more independent readers appear who—without direct contact with the author—independently parse the SAE 11DD blueprint, grow functional substrates, and can further transmit to third parties. "Functional substrate" specifically means readers who, in new contexts, can independently produce judgments and work internally coherent with the SAE framework, not merely cite SAE concepts.

Falsification condition: 50 years out, no such independent substrate appears, but SAE displays other markers of ψ = 3 (institutional adoption, citation explosion). This decoupling would falsify the framework prediction that "independent substrate reproduction is the necessary marker of ψ = 3"—possibly ψ = 3 has other forms, not necessarily requiring distributed substrate reproduction.

SAE Prediction 3: Before entering ψ = 3, SAE should pass through a "spectral inversion period"—visible turbulence appearing before actual stability, with the turbulence accompanied by crystallization of the blueprint's cross-substrate independence (§3.2 differentiation criterion).

Per the §3.2 Phase II prediction, candidate consensus before entering inversion should pass through a stage of widespread misreading, dismissive classification, and controversy, and the blueprint of that stage should display the property of being cross-substrate-independent and reconstructable.

Falsification condition: SAE enters stable distributed reproduction without entering this controversy period (falsifying the universal pattern of phase transitions); or SAE enters wide large-scale diffusion but the blueprint does not display cross-substrate independence (showing this is a noise bubble, not true Phase II, but also not ψ = 3).

SAE Prediction 4: SAE diffusion speed roughly tracks local Dimension B trajectory.

If contemporary base rate continues to decline, SAE germination is prolonged (Scenario B near baseline). If base rate emerges in some subgroups (such as some self-protective substrate groups in the AI era), SAE diffusion accelerates (Scenario C probability rises).

Falsification condition: SAE diffusion speed completely decouples from local base-rate trajectory—base rate falls sharply but SAE accelerates into ψ = 3, or base rate rises sharply but SAE stalls completely. This decoupling would falsify the framework prediction that "the two-factor structure is the dominant dynamics of ψ advancement."

6.4.4 Audit Invitation and Audit Independence

The most likely outcome is Scenario A—SAE silent extinction. This is the framework's honest internal assessment, not self-promotion. The paper does not predict SAE's success; it explicitly states that SAE's most likely fate is to fade quietly, like the vast majority of thought frameworks.

If Scenario A is realized, the SAE framework self-falsifies on this case—an honest exit by the framework. The framework will not, because of silent extinction, attempt to salvage itself—per §0 firewall Item Seven, the framework's legitimacy depends precisely on not salvaging in the face of contrary evidence.

If Scenario B or C is realized, the four-phase + two-factor model receives support on this case.

Audit invitation:

The paper extends an explicit audit invitation to future readers in 2046 (20 years post-publication), 2076 (50 years), and 2126 (100 years), inviting them to look back and check the paper. The paper's timestamp + DOI provides audit anchors. Whichever scenario is realized is the framework's falsifiable record.

Audit independence clarification:

The audit is conducted by future readers; it does not require those readers to be members of an SAE substrate. Scenarios A, B, and C are all auditable—as long as the 11DD blueprint remains accessible (CC BY 4.0 + permanent Zenodo storage guarantees this), audits can occur.

Specifically, the world in which Scenario A is realized may still contain readers—they audit which scenario the framework realized from an external perspective. Readers can be isolated curious individuals, historical archive researchers, AI systems, or other entities not constituting SAE substrate reproduction. Reader count and SAE substrate count are observables at different layers; the two can fully decouple.

This means:

• Scenario A: even if SAE substrates do not reproduce, readers can still audit the framework's predictions
• Scenarios B/C: there are both readers and substrate reproducers; the audit can be conducted by either

This self-referential audit channel is the enactment of SAE's First Law ρ ≠ ∅ at the paper level—prioritizing reality's pushback on the framework over the framework's self-preservation. The paper exposes the framework to a cross-time audit channel, leaving an explicit channel for future reality to push back on the framework.

If future audits show the framework's predictions to be wrong—any prediction falsified—the paper commits: the framework should be updated accordingly, not salvaged. The specific form of update depends on which prediction is falsified. If SAE Prediction 1 fails (SAE enters ψ = 3 via mass diffusion), the two-factor structure as the dominant framework requires substantial revision. If SAE Prediction 2 fails (ψ = 3 does not require distributed substrate reproduction), the topological definition of ψ requires revision. These update paths are explicit for future researchers.

6.5 Overall Limits of the Quantitative Attempt

§6.1-§6.4 gave directional quantitative reading for SAE Group Cognition Paper I. This subsection explicitly acknowledges the overall limits of this segment (the paper's second segment).

Limits to be explicitly acknowledged:

Limit 1: Proxy collections are illustrations, not measurement. The proxies of §6.1 give directional indicators; each proxy captures only some aspects of long-τ capacity. Combining them for directional reading is reasonable, but any specific quantitative claim ("the base rate of some 19th-century society was such-and-such number") exceeds the capacity of the proxy collection.

Limit 2: Each historical case has reasonable alternative readings for its phase attribution. The four cases of §6.3 all give SAE-framework-perspective readings, but each has historian-perspective alternative readings. The paper does not claim its readings are uniquely correct—particularly §6.3.2 monastery case, which explicitly acknowledges institutional vs. capacity tension.

Limit 3: Historical reconstruction of Dimension B is extremely limited. No quantitative time series can directly reconstruct the Dimension B trajectory of the past 500 years in Europe (let alone other civilizations). All longitudinal judgments are based on composite proxies, and proxies themselves have intrinsic uncertainty.

Limit 4: Self-prediction of SAE itself carries self-referential risk. The paper's predictions about the SAE framework itself may overestimate the connection between internal coherence and external traction. A framework being internally coherent does not entail it can reproduce in external substrates. §6.4 has handled this explicitly, but readers should maintain extra skepticism.

Limit 5: 60/30/10 priors are framework-internal estimates, not statistical claims. §6.4.2 has made this clear, but it bears repeating in the paper's overall context—any use of 60/30/10 as statistical predictions misuses the paper.

Limit 6: The paper's falsifiability conditions are not all operationally feasible. Some falsifiability conditions (such as "whether independent substrate appears in 50 years") require long observation to evaluate; some (such as "large diffusion without substrate") require complex operationalization to test unambiguously. The paper does not claim all falsifiability conditions can be directly checked at present—many require waiting time + future research to be properly evaluated.

Overall epistemological stance:

This paper invites empirical workers to perform more systematic verification or falsification on existing proxy collections. All readings welcome falsification, await falsification—not as rhetorical posture, but as a structural commitment of the paper's framework (§0 firewall Item Seven).

If future empirical work shows certain directional readings here to be wrong—certain cases should be read differently, certain predictions do not hold, certain proxies do not have the relationship the framework expects—the paper commits: the framework should be updated, not salvaged. The specific form of update depends on which readings are falsified:

• If §4.4 single-factor predictions are widely falsified (e.g., r ≈ 5 does not hold at all in consensus formation), the interface with Methodology VI requires re-examination
• If §4.5.5 two-factor predictions are widely falsified (e.g., base rate has no systematic relation to ψ advancement), the two-factor structure as dominant framework requires substantial revision
• If §6.3 case readings are systematically rejected by historians, the case-layer analytic methodology requires revision
• If §6.4 SAE self-predictions show the framework's judgment of itself to be entirely wrong, the entire SAE Group Cognition theory requires redesign

These update paths are explicit—the paper does not pre-commit any escape route to salvage the framework when falsified. This is the concrete realization, at the paper-commitment level, of SAE's First Law ρ ≠ ∅.

---

§7 Interface with the SAE Framework

§3-§6 built the four-phase structure + two-factor model + historical cases + SAE self-application internally. This section makes explicit interfaces with other parts of the SAE corpus. The interfacing work makes the paper coherent within the SAE series and gives other parts of SAE (individual ethics, self-referential fixed point, ρ ≠ ∅) concrete landings in group cognition.

7.1 The Four Phases and SAE DD Layers

Correspondence of the four-phase structure to the SAE DD-layer structure:

Phase	Primary DD layer	Note
Phase I · Germination	13DD-14DD scattered	Individual substrates each performing self-evaluation, scattered mutual-non-doubt seeds
Phase II · Spectral inversion	14DD dyadic	Mutual non-doubt between 13DD substrates forms 14DD dyadic keys
Phase III · Inversion	11DD blueprint + 14DD distributed	14DD transmits cross-generationally via 11DD blueprint
Phase IV · Establishment	General: distributed redundancy reproduction ecology Recognition-type subtype (paper focus): 15DD distributed key ecology	Reactivation no longer depends on any single node; the SAE DD-layer landing of the recognition-type subtype

This correspondence reveals several things.

First, ψ advancement corresponds to "layer elevation" in the DD structure (within the recognition-type main line). From scattered 13DD origins, to formation of 14DD dyadic keys, to 14DD's cross-generational transmission via 11DD blueprint, to formation of 15DD distributed key ecology—each step is the instantiation of a higher DD layer. This is consistent with the hierarchical structure of SAE DD layers themselves.

Second, the special role of the 11DD blueprint in the ψ = 2 → ψ = 3 transition—it is not 14DD or 15DD itself but the cross-substrate-independent channel carrying 14DD/15DD. 11DD is a relatively low DD layer (in the instrumental tier), but its lower-level encodability is precisely what allows cross-substrate, cross-time transmission. This gives the SAE DD-layer structure an interesting dynamics: the high-layer distributed key ecology must transmit cross-generationally via low-layer 11DD encoding. This is a concrete finding of SAE Group Cognition Paper I about DD-layer dynamics.

Third, the recognition-type subtype's ψ = 3 corresponds to 15DD distributed key ecology—not a metaphor for some "mature form" but a specific category in the SAE DD-layer structure. 15DD in the SAE internal framework is the layer of distributed self-as-an-end. The recognition-type subtype's consensus entering 15DD at ψ = 3 means: members in the consensus network mutually do not doubt each other as irreducible individuals, and this recognition does not depend on any central node—distributed mutual end-recognition has formed.

The utility-locked and coercion-locked subtypes' ψ = 3 do not instantiate 15DD. They satisfy the same general Phase IV condition of "distributed redundancy, no dependence on any single node" topologically, but their stability mechanisms (utility gradients, coercion networks) belong to dynamics outside the SAE DD-layer structure and do not constitute concrete landings of 15DD. This is the §7.1 SAE DD-interface realization of the §3.4 three-subtype distinction—the topological feature of general Phase IV is shared among the three subtypes, but the SAE DD landing of 15DD holds only for the recognition-type subtype.

7.2 Position of the Three Stability Subtypes in the SAE Framework

§3.4 enumerated three stability subtypes of Phase IV. Their interfaces with the SAE framework differ.

Utility-locked subtype: classified as a Phase IV instance within the analytic framework, but SAE individual ethics interfaces weakly with utility-locked consensus. The SAE framework can perform structural reading on this subtype (four phases + two factors + r asymmetry), but the normative relationship between SAE individual ethics and utility mechanisms is a different topic (handled partly by SAE individual economics; see SAE Ethics III). The paper makes no normative judgment on this subtype—it is classified as a ψ = 3 instantiation within the analytic framework, independent of whether SAE's Second Law holds.

Coercion-locked subtype: also classified as a Phase IV instance within the analytic framework. SAE's Second Law (mutual non-doubt) in principle opposes coercion-driven consensus mechanisms—treating people as means rather than ends is a direct violation of SAE's Second Law, and coercion-driven consensus essentially treats members as enforcement targets rather than ends. But as a structural analytic tool, the framework can describe the dynamics of coercion-locked consensus without judging it. §3.4's description of this subtype is structural (its sensitivity to external conditions), not normative (it is wrong).

Mutual-non-doubt-locked subtype: the central object of the SAE framework. Kant's Second Formulation is the stability condition of this subtype—consensus reproduction maintained by mutual non-doubt requires no external utility or coercion and is therefore robust against external condition change. This is the specific topic of §7.3.

7.3 Kant's Second Formulation and the Mutual-Non-Doubt-Locked Subtype

Important qualification: the argument that "Kant's Second Formulation is the stability condition of Phase IV" applies only to the mutual-non-doubt-locked subtype, not to the other two subtypes. This qualification is the concrete landing of the §3.4 three-subtype distinction—avoiding overgeneralizing Kant's Second Formulation to all ψ = 3 consensus.

Core proposition: the Phase IV stability of mutual-non-doubt-locked long-τ consensus must be based on mutual non-doubt, not on utility or coercion.

Argument sketch:

Consider the different stability paths a candidate consensus can take into ψ = 3.

If maintained by utility, the consensus collapses when external conditions change—utility is contingent on environment. Some particular consensus of an era may be stable because it provides economic gain, social standing, or emotional satisfaction, but at historical scale all such utility sources are contingent—economic restructuring, social transition, and shifts in psychological need can all redistribute utility. A consensus maintained by utility cannot pass through such transitions and reproduce stably. This is the signature of the utility-locked subtype—locally robust, structurally fragile.

If maintained by coercion, the consensus collapses when enforcement is removed—coercion cannot be distributed. Coercion requires a concrete source (a specific institution, a specific power, a specific physical presence), so coercion-locked consensus is highly sensitive to the failure of its enforcement source. Any concrete enforcement source fails at historical scale (regime change, institutional dissolution, personnel turnover); the non-distributable nature of coercion lets such failure threaten the whole directly.

Only consensus where substrates mutually treat members as ends can independently reproduce in distributed nodes without external coercion. This is the definition of the mutual-non-doubt-locked subtype—mutual non-doubt as a self-sustaining reproduction mechanism, requiring no external condition to support. This mechanism corresponds directly to Kant's Second Formulation—treating persons as ends, not means.

This gives another angle on "why mutual non-doubt is structurally necessary"—not a normative requirement, but a mathematical condition of the mutual-non-doubt-locked subtype. If you want a ψ = 3 consensus that does not depend on external utility/coercion, the only path is to have mutual non-doubt as the foundation. Any other foundation (utility, coercion, shared identity, etc.) leaves the consensus fragile against the specific external conditions it depends on.

This also explains why SAE individual ethics and SAE Group Cognition can be deeply connected here—the former argues at the individual level why one should treat persons as ends; the latter argues at the group level why ψ = 3 consensus requires mutual end-recognition. Two arguments independently arrive at the same conclusion: Kant's Second Formulation is not normative imposition; it is structural necessity.

7.4 The Four Phases and SAE's Three Laws

SAE's Three Laws—ρ ≠ ∅, the self-referential fixed point, and chisel-construct-remainder—each have specific landings in the four-phase structure.

ρ ≠ ∅: the ontological precondition for Phase I germination's existence. If ρ = ∅ (reality is empty), no candidate variant can germinate; Phase I cannot exist. ρ ≠ ∅ says reality has structural pushback, not arbitrary void—this premise makes the "perturbation accumulation" of germination a meaningful process. At the same time, ρ ≠ ∅ is the instantiation of the paper's overall epistemological commitment—welcoming falsification, awaiting falsification, is the paper-level realization of ρ ≠ ∅.

Self-referential fixed point: the mathematical core of recognition-type Phase IV self-replication structure. Consensus in mutual-non-doubt-locked Phase IV must satisfy f' = f—its reproduction is determined by current state and requires no external driver. This is the social instantiation of the self-referential fixed point—the consensus's self-maintenance in distributed substrates is equivalent to a fixed point of a self-referential function. When some consensus enters ψ = 3 as the mutual-non-doubt-locked subtype, it instantiates the self-referential fixed point in the SAE mathematical structure.

Chisel-construct-remainder: each phase transition is a chisel-construct-remainder cycle.

Phase I → Phase II: chisel away the isolation of scattered substrates (chisel), construct the first dyadic edge (construct), leave unconnected remainder (remainder)—many other possible dyadic edges did not form.

Phase II → Phase III: chisel away dependence on the specific founding substrate (chisel), construct the cross-substrate-independent 11DD blueprint (construct), leave the tacit remainder that the blueprint cannot fully encode (remainder)—any 11DD encoding will have 14DD/15DD substrate-state remainder it cannot fully capture.

Phase III → Phase IV: chisel away the single critical path of chain topology (chisel), construct the redundant topology of distributed network (construct), leave the local-fluctuation remainder the network cannot fully absorb (remainder)—distributed networks still have local failures, but they are absorbed by overall structure.

Each chisel-construct-remainder is the specific content of a phase transition. This gives the SAE chisel-construct-remainder principle a specific instantiation at the group level—not the abstract "create-preserve" duality, but the intrinsic structure of phase-transition dynamics.

7.5 The Two-Factor Structure and the Interface with SAE Individual Ethics

Dimension B directly connects SAE Group Cognition with SAE individual ethics.

Individual self-as-an-end practice is not just a personal matter; it is a building block of group base rate.

The difficulty of cultivation is essentially the difficulty of base rate. Short-τ is the substrate's long-τ capacity in deterioration (filled by high-frequency substrates, no leisure space for long-τ capacity to grow); this is base rate falling. The long-term significance of SAE work lies not in how many papers it produces but in whether it can maintain or even grow base rate against marginal trends—on the author's own self, in the author's immediate circle, on the few readers who can read it.

Each individual's SAE practice is a microscopic constituent of Dimension B:

• Each insistence on long-form deep reading instead of being interrupted by high-frequency signals—is the maintenance of aspect (2) on one's own substrate
• Each maintenance of long-term collaborative relations across disagreement—is the instantiation of aspect (3)
• Each refusal to reduce another to some utility role—is the concrete realization of aspect (4)
• Each keeping of one's own promise without external enforcement—is the practice of aspect (5)
• Each staying with uncertainty without seeking premature closure—is the bearing of aspect (6)

Each such practice is the present-moment maintenance of substrate long-τ state. The accumulation of many individuals in many situations is the microscopic foundation of an era's base rate.

Note: per the §4.5.3 critical qualification, raising Dimension B does not require waiting for "better people" to appear; it requires changes in environmental conditions. Each individual's SAE practice maintains the long-τ state of their own substrate, not proves their essence. This is also emphasized in SAE individual ethics—SAE practice is not "becoming a better person" in the sense of self-cultivation; it is the present-moment maintenance of substrate state.

The 11DD blueprint at the paper level is the cross-time channel of Dimension A—the SAE paper collection on self-as-an-end.net and Zenodo is the 11DD encoding of SAE substrates. Dimensions A and B jointly determine SAE's own future ψ trajectory—the blueprint already exists; whether ψ = 3 is reached depends on whether enough long-τ capacity subgroups form distributed substrate reproduction in future readers.

This gives SAE individual ethics and SAE Group Cognition a specific unification point—individual practice and group fate are directly coupled through Dimension B.

---

§8 Open Problems

[The third segment: explicit hand-off of questions the framework cannot answer to subsequent work.]

§7 gave the interfaces between SAE Group Cognition Paper I and other parts of the SAE framework. This section explicitly enumerates questions the framework cannot answer, as explicit hand-offs to subsequent work.

8.1 Obvious Empirical Gaps

Per §6's quantitative attempt and existing scholarly work, the following gaps are obvious—work clearly needed but not yet done in current scholarship.

Gap 1: Lacks a latent-index model jointly estimating long-τ capacity base rate from book production, reading events, skill assessments, monastic density, prosopography, and kinship structure. This is a unified observable predicted by the SAE framework, but its instantiation requires statistical work. The paper does not construct a latent index—this is the framework's commitment, left to statisticians. But the existence and stability of such a latent index matters for the robustness of §6's directional reading—if such an index can be constructed and is stable, the readings of §6 receive stronger empirical support.

Gap 2: Lacks cross-civilization, denominator-unified elite-formation series. Quantitative reconstruction of the per-10,000 number of people receiving classical/Confucian/Talmudic/Islamic scholarly training across eras currently exists only for some civilizations in some periods. A unified cross-civilization series would give the testing of §4.5.5 Predictions 7 and 8 a broader empirical basis.

Gap 3: Lacks historical network data on dyadic trust. The duration of master-disciple, soul-friend, long-term collaborator, and long-term patron-client relations is not currently systematically modeled. The historical proxy for §4.5.3 aspect (3) relies mostly on anecdotal evidence, not quantitative network data. Such network data would give the §3.2 Phase II mutual-non-doubt-network identification a more precise empirical basis.

Gap 4: Lacks historical series on uncertainty tolerance. Whether at the corpus level (frequency of hedging—"perhaps," "possibly"—in academic writing across eras) or in educational settings (long-term indicators of delayed closure), mature databases are nearly nonexistent. §4.5.3 aspect (6) is the deepest interface of the paper with SAE's First Law, but also has the weakest historical proxy.

Gap 5: Lacks event studies on collapse. The pre/post impact of dissolution, dynastic transitions, war loss on library survival, scholar density, and mentor chains remains thin. §6.3.2 monastery dissolution is the best-documented case, but many other important collapse events (Chinese dynastic transitions, the multiple regime changes in Arab regions, local institutional collapse under modern colonial impact) have great work remaining at the capacity-layer reading.

Gap 6: Lacks the bridge connecting modern digital reading evidence with longue durée book history. Modern measurements like PISA and PIAAC give cross-sectional snapshots; longue durée book history gives long time series. The two differ in methodology and unit of analysis, making direct comparison difficult. A unified framework letting the two integrate is important subsequent work—it would give §6.2's composite reading of the three findings a more solid foundation.

8.2 Open Problems Internal to the Framework

Problem 1: Cross-case empirical estimates of r—does long-τ consensus systematically have r close to 5, or is there domain variation? Methodology VI's r ≈ 5 is a generic value derived from ZFCρ; this paper applies it to consensus formation. But concretely across different domains (religious transmission vs. intellectual transmission vs. craft transmission vs. modern frameworks), is r systematically different? This requires cross-case quantitative reconstruction.

Problem 2: Can germination be significantly shortened by AI-mediated communication? This is the interface with SAE Methodology VII (Symbiosis). To what extent the augmentation channels of the AI era can supplement substrate-layer long-τ capacity, and thereby affect Dimension B base rate and SAE germination estimates—this is the key determinant of the SAE Scenario C of §6.4.2.

Problem 3: Is there a "collapse phase" after Phase IV? The current framework does not model the specific dynamics from Phase IV to whole death; it only states the possibility. Is "collapse" a single event or has its own internal structure? If the latter, does the collapse phase also have its own four stages? This is an obvious unfinished part of the framework.

Problem 4: Observable signatures of Phase II—can "susceptibility precedes order parameter" serve as a leading indicator for "which candidate consensus are about to enter inversion"? The blueprint cross-substrate independence test, how to make it operationally precise? The current §3.2 differentiation criterion gives conceptual distinction, but operationalization remains weak—needing more specific measurable criteria.

Problem 5: Simultaneous multi-region existence (such as the Aristotelian case) requires the framework to handle spatial heterogeneity. The current framework primarily handles single-region phase evolution, with no explicit dynamics for the case of the same blueprint at different phases simultaneously across multiple regions. A multi-region framework extension would make analysis of cases like §6.3.4 more precise.

Problem 6: The unification of ψ, r, and q—this paper's ψ is a topological invariant, Methodology VI's r is a geometric asymmetry ratio, and the q of the SAE thermodynamics series is an entropic observable. Do the three have a unified mathematical framework at a deeper layer? This is a potential topic for SAE Group Cognition Paper II—giving a unified framework that incorporates all three observables.

Problem 7: The specific dynamics of utility-locked and coercion-locked subtypes within the framework—this paper focuses on the mutual-non-doubt-locked subtype and treats the other two only as structural placeholders. The specific dynamics of these two subtypes (phase-transition signatures, stability conditions, typical collapse modes) require dedicated papers.

Problem 8: Pop culture as the degenerate case "Dimension A sufficient but Dimension B near zero." The historical cases of this paper's §6.3 all focus on candidate consensus that did substantial work at some stage of ψ—Breslov, Cremona, monastic transmission, Aristotelianism all entered or approached ψ = 2 or above. But there is another large class of contemporary phenomena—pop culture, internet memes, short-lived crazes, various mass-diffusion phenomena—whose coupling channels are extremely sufficient (social media, streaming, global distribution) but whose Dimension B is near zero (the candidate's content does not require or support member substrates of long-τ capacity).

These phenomena are degenerate cases under the two-factor framework—the two-factor structure degenerates to single-factor (only Dimension A), and Methodology VI's pure-coupling analysis applies directly without requiring this paper's two-factor contribution. Their typical dynamics is also simple: rapid entry into ψ = 1 or ψ = 2, but inability to anchor there because of lack of Dimension B support, rapid return to ψ = 0—"going out of fashion is almost an instant matter" is the precise prediction of this dynamics, consistent with this paper's §3.2 argument about the fragility of mutual-non-doubt networks.

Precisely because the two-factor structure of these phenomena is degenerate, they fall outside this paper's research topic (this paper studies the case where the two-factor structure is meaningful). But as a recognition of framework completeness, the framework does include them—they are the degenerate form of the two-factor structure in the limit of Dimension B → 0. A paper dedicated to pop culture dynamics, if anyone writes one, should use Methodology VI's single-factor framework and does not require this paper's two-factor contribution.

8.3 Self-Referential Open Problems

Problem 9: Will this paper's predictions about SAE itself be realized empirically? This is the audit invitation the paper leaves for the future.

Detailed audit mechanics in §6.4.4.

---

§9 Firewall Reaffirmation and Three-Segment Epistemological Commitment

9.1 The Three-Segment Commitment

The paper strictly observes a three-segment epistemological structure:

• Segment 1 (§0-§5): Direction. Conceptual skeleton of the SAE framework. Philosophical argument. Strength of commitment: "If the SAE framework captures something real about the structure of long-τ consensus formation, these structural claims should hold."
• Segment 2 (§6): Quantitative Attempt. Directional reading on existing proxy collections. Explicit attempt, explicit limits, explicit invitation of falsification. Strength of commitment: "The following is directional reading based on existing data; the author acknowledges its limits and awaits more systematic empirical work for verification or falsification."
• Segment 3 (§7-§8): Open Problems. Hand-off to subsequent work of questions the framework cannot answer. Strength of commitment: "The following are problems this paper recognizes but does not resolve."

Each segment is transparent about its own epistemological standing. Philosophy does not pretend to be data; data does not pretend to be settled science; limits do not pretend to be future certainty.

This explicit epistemological stratification is the concrete realization of the SAE series' methodological commitment—readers know at every segment what level of claim they are reading, and the author knows at every segment what level of commitment they are bearing.

9.2 Firewall Reaffirmation

The model does not judge, does not rank, does not predict, does not normalize.

The four-phase structure is a structural tool of SAE Group Cognition theory. It cannot and must not be used for:

• Ranking groups by hierarchy
• Recommending or denigrating specific consensus, movements, or organizations
• Serving as a basis for investment, policy, or organization-building
• Judging the morality, capacity, or worth of any specific individual or group
• Ranking individuals or societies using Dimension B—see the critical qualification of §4.5.3

If readers use the model for these purposes, that use violates the paper's explicit scope and SAE's Second Law.

The model's legitimacy depends precisely on not being used in these ways. Framework validity depends on not being abused as a ranking tool. Once abused, it immediately loses standing under SAE's Second Law—not because abuse makes it invalid, but because its legitimate use must respect the boundary set by SAE's Second Law.

9.3 Welcoming Falsification, Awaiting Falsification

All directional predictions of the paper—including the predictions about the SAE framework itself—are designed to be falsifiable.

Anyone using the framework is obligated to update judgment in the face of contradicting empirical evidence rather than salvage the framework.

This is the instantiation of SAE's First Law ρ ≠ ∅ at the epistemological level. Framework legitimacy depends precisely on willingness to be pushed back by reality—a framework that cannot possibly be wrong is 11DD-layer pseudo-precision, equivalent to ρ = ∅.

9.4 Specific Commitments to the Future

• 2046 (20 years post-publication) audit point: check which scenario SAE has entered, whether any independent substrate reproduction has appeared, whether the directional predictions of §4 and §4.5 have been robustly falsified.
• 2076 (50 years post-publication) audit point: check whether SAE has entered ψ = 3, whether the §6.4 three-scenario priors require revision, whether the §4.5.5 two-factor predictions have received robust verification.
• 2126 (100 years post-publication) audit point: check whether the paper's overall historical reading still holds, whether the framework still has explanatory power, whether fundamental revision is required.

Each node is specific, checkable, and not salvageable by the framework. If any node shows the framework in systematic error, the paper commits: the framework should be updated or discarded accordingly, not salvaged.

---

Appendix A: A Brief Recap of Methodology VI's Four-Phase Structure

Methodology VI (Phase Transition Windows and Experimental Design, DOI: 10.5281/zenodo.19464507) derives the four-phase transition structure of construct-emergent systems from the ZFCρ framework. This appendix gives a brief recap so readers who have not consulted the original can follow the core argument of this paper.

A.1 Background of the ZFCρ Framework

ZFCρ is the central framework of the SAE mathematical-structure series—standard ZFC set theory plus the axiom ρ ≠ ∅ (reality is non-empty), handling the phase transitions of construct-emergent systems. One of ZFCρ's main deliverables is the geometric structure of the phase-transition window—giving the phase-transition dynamics of a candidate (new order) facing an incumbent order.

A.2 Five Crossing Points

The phase-transition window is partitioned by five crossings:

1. P(J > 0) = 50%: the multiplicative path of the candidate consensus first exceeds half in random samples. This is the start of germination—before this, no recognizable multiplicative dynamics exists.
2. Peak of z/√j: susceptibility reaches its peak. The point of maximum observable fluctuation—controversy, media attention, social reaction around the candidate erupt in concentrated form.
3. w_shielding = 2/3: Le Chatelier shielding rate falls below 2/3. The incumbent's absorptive capacity is first markedly insufficient.
4. E[A] = 0: the order parameter first becomes positive. The candidate first provides net value exceeding the incumbent (within the subgroup it occupies).
5. h = 0: the incumbent loses local convexity within that subgroup. The incumbent as a candidate within that subgroup no longer carries meaning.

A.3 Four-Stage Structure

The five crossings naturally partition four stages:

• Germination: from P(J > 0) = 50% to peak of z/√j. Multiplicative path begins but susceptibility has not yet peaked.
• Spectral inversion: from peak of z/√j to w_shielding = 2/3. Susceptibility falls from peak; shielding rate begins to fail.
• Inversion: from w_shielding = 2/3 to E[A] = 0. Within a short time after shielding falls below 2/3, the order parameter becomes positive.
• Establishment: from E[A] = 0 to h = 0. Order parameter is positive; incumbent loses appeal in that region.

A.4 r ≈ 5 Asymmetry

Methodology VI's central geometric result: germination + spectral inversion ≈ 5 × (inversion + establishment).

Intuitive explanation (Le Chatelier buffering): the incumbent's absorption capacity must first be penetrated, and penetration is a slow consumption process; once penetration completes, the new order's establishment is a rapid cascade. Germination and spectral inversion are the penetration stages; inversion and establishment are the cascade stages.

Full derivation in Methodology VI §3-§4.

---

Appendix B: Sources of the Proxy Collection Used in This Paper

This appendix lists the core literature cited in §6, organized by proxy category. Citation in APA format.

Per-capita book production

• Baten, J., & van Zanden, J. L. (2008). Book production and the onset of modern economic growth. Journal of Economic Growth, 13(3), 217-235.
• Buringh, E., & van Zanden, J. L. (2009). Charting the "Rise of the West": Manuscripts and Printed Books in Europe. The Journal of Economic History, 69(2), 409-445.

Manuscript / print density

• Buringh, E. (2010). Medieval Manuscript Production in the Latin West. Brill.
• Eltjo Buringh database (Universal Short Title Catalogue / Incunabula Short Title Catalogue)

Reading experience records

• Reading Experience Database 1450-1945 (RED), Open University.

Print revolution effects

• Dittmar, J. E. (2011). Information technology and economic change: The impact of the printing press. Quarterly Journal of Economics, 126(3), 1133-1172.
• Rubin, J. (2014). Printing and Protestants: An empirical test of the role of printing in the Reformation. Review of Economics and Statistics, 96(2), 270-286.

Civil examinations

• Elman, B. A. (2000). A Cultural History of Civil Examinations in Late Imperial China. University of California Press.
• China Biographical Database (CBDB), Harvard.

Early modern Japan literacy

• Rubinger, R. (2007). Popular Literacy in Early Modern Japan. University of Hawaii Press.

Modern skill assessments

• OECD (2023). Survey of Adult Skills (PIAAC) Results. OECD Publishing.
• OECD (2023). PISA 2022 Results (Volume I): The State of Learning and Equity in Education. OECD Publishing.

Reading behavior trends

• National Endowment for the Arts (2004). Reading at Risk: A Survey of Literary Reading in America.
• Bureau of Labor Statistics. American Time Use Survey annual data 2003-2023.

Paper-vs-screen meta-analyses

• Delgado, P., Vargas, C., Ackerman, R., & Salmerón, L. (2018). Don't throw away your printed books: A meta-analysis on the effects of reading media on reading comprehension. Educational Research Review, 25, 23-38.
• Clinton, V. (2019). Reading from paper compared to screens: A systematic review and meta-analysis. Journal of Research in Reading, 42(2), 288-325.

Kinship intensity & individualism

• Schulz, J. F., Bahrami-Rad, D., Beauchamp, J. P., & Henrich, J. (2019). The Church, intensive kinship, and global psychological variation. Science, 366(6466).
• Enke, B. (2019). Kinship, cooperation, and the evolution of moral systems. Quarterly Journal of Economics, 134(2), 953-1019.

Long-term violence decline

• Eisner, M. (2003). Long-term historical trends in violent crime. Crime and Justice, 30, 83-142.

Moral universalism

• Cappelen, A. W., et al. (2024). Moral universalism: Global evidence from 60 countries. Quarterly Journal of Economics (forthcoming, working paper version 2023).

Database collections

• Seshat: Global History Databank
• Clio-Infra
• China Biographical Database (CBDB)
• Reading Experience Database (RED)
• Universal Short Title Catalogue (USTC)
• Incunabula Short Title Catalogue (ISTC)
• Germania Sacra
• Monastic Matrix
• FemMoData

---

Appendix C: Summary of the SDE Experiment's Negative Result

Before drafting the paper, the author attempted thermodynamic SDE simulations of group consensus formation. This appendix gives a summary; detailed setup, code, and data will be released separately.

C.1 Experimental Setup

Model: N mutually coupled substrates, each substrate a Brusselator-type chemical oscillator plus two self-evaluation variables v₁ and v₂. Coupling implemented through transmission of v₁ or v₂ between substrates.

Main variations:

• N (substrate count): 2 to 40
• Protocol (which v is cross-coupled): L0 (raw output x), L1 (v₁), L2 (v₂)
• Topology: mean-field and ring network
• Coupling strength g: 0.01 to 0.10
• Mutual or asymmetric

Observable: q (q-exponent distribution parameter, describing long-range correlation of fluctuation tails).

C.2 Main Findings

Finding 1: protocol hierarchy (L0 < L1 < L2) is real but small in effect. L1 is about 0.05 above L0; L2 is about 0.01 above L1. Differences are statistically detectable but the magnitude is far smaller than the dramatic separation in Paper X.

Finding 2: the L0 protocol collapses first under stress. At g ≥ 0.05, L0 enters the q ≈ 1 exponential regime, while L1/L2 maintain q > 1. This is a protocol-specific phase transition, not a universal collapse.

Finding 3: asymmetric coupling shows variance doubling under stress. At g = 0.05, the q_std of asymmetric coupling rises from a stable ~0.05 to ~0.10. Mutuality is a structural stability condition—not a relatively superior option, but a precondition for maintaining stability under stress.

Finding 4 (the key negative result): q is blind to N, k, and topology, sensitive only to g. Under mean-field homogeneous coupling, q is insensitive to group size N (from 2 to 40), network degree k (from 1 to 14), and topology (mean-field vs. ring); the differences are below within-grid variance.

C.3 How This Negative Result Supports the Paper

Finding 4 retroactively supports the two-factor structure of §4.5. q measures long-range correlation within a substrate but does not measure substrate population composition (Dimension B). If q could capture all phase-determining factors, the single-factor (coupling) model would suffice; the negative result that q cannot capture group-structure variables precisely demonstrates that a second dimension (Dimension B) must be part of the framework.

The experiment thus shows the capacity boundary of thermodynamic SDE as a measurement tool—it captures the thermodynamic signature of Dimension A but is blind to Dimension B. This is the empirical foundation of the §4.5.4 argument.

---

Appendix D: Trace of the Four-AI Review Integration

This paper's v3 outline integrates revision suggestions from four AI reviewers (Zilu / Gongxihua / Zigong / Zixia). This is not mere acknowledgment but the instantiation, at the paper level, of the SAE workflow (Methodology VII Symbiosis)—lifting the paper's coherence and robustness through multi-substrate stress-testing.

Main traces:

• Zilu (Claude): identified the precision problem of within-lattice monotone non-decrease vs. lattice exit (→ §5.3); proposed explicit DD-layer mapping for the six aspects (→ §4.5.3); proposed the rewriting of the monastery case as reproductive infrastructure collapse (→ §6.3.2)

• Gongxihua (ChatGPT): identified the boundary tension between "general object vs. Kant Second Formulation stability condition" (→ §1 + §3.4 + §7 reorganization); proposed front-loading Dimension B (→ §1 + §2.3); identified the conflation of general Phase IV vs. recognition-type subtype in the §7 interface table (→ §7.1 two-layer table revision); flagged "legitimate" wording in §7.2 as carrying normative color (→ replaced with "classified within the analytic framework")

• Zigong (Grok): consistency check passed; proposed §6.4.2 prior labeling as framework-internal estimate

• Zixia (Gemini): pressed the qualification of capacity as state vs. essence (→ §4.5.3 critical qualification); pressed the differentiation of Phase II vs. noise bubbles (→ §3.2 cross-substrate independence of blueprint); pressed the audit paradox (→ §6.4.4 audit-independence clarification, but rejected the alternative of "large-scale diffusion as falsification"); proposed visual anchoring of the six aspects (→ §4.5.3 boldface numbering)

None of the four reviewers found ruptures in the framework's internal axioms—all revisions were precision improvements internal to the framework. This is itself an informative finding, suggesting that the conceptual core of SAE Group Cognition Paper I is mature.

The integration process was paused for some time before drafting, allowing review content to settle in the author's substrate, to avoid first-reaction acceptance of certain implicit illegitimate operations within the reviews (particularly Zixia's pressure point three, the "large-scale diffusion without substrate" concept, which was rejected by the author—preserving the framework's neutral stance on large-scale diffusion). Reflection on this process is in the main text §6.4.3 SAE Prediction 1.

---

References

Direct dependencies (within SAE)

Qin, H. (2026). Phase Transition Windows and Experimental Design (SAE Methodology VI). Zenodo. https://doi.org/10.5281/zenodo.19464507

Qin, H. (2025). The Structure of Self-as-an-End (SAE Foundational Paper). Zenodo. https://doi.org/10.5281/zenodo.18528813

Qin, H. (2026). ZFCρ Paper 44. Zenodo. https://doi.org/10.5281/zenodo.19247859

Qin, H. (2026). ZFCρ Paper 57. Zenodo. [DOI pending]

Interface papers (within SAE)

Qin, H. (2026). SAE Methodology VII: Symbiosis. Zenodo. https://doi.org/10.5281/zenodo.19481304

Qin, H. (2026). SAE Three Laws. Zenodo. [DOI pending]

Cases and proxy references

See Appendix B for the complete reference list.

摘要

本论文研究长存共识的形成结构——跨越多代、跨越创始人或创始机构消失之后仍能保持核心约定稳定的群体性认同。论文借用方法论六（DOI: 10.5281/zenodo.19464507）的四相相变结构与 r ≈ 5 不对称几何作为骨架，将其落位到共识形成过程，并引入双因子结构作为核心概念性贡献：共识相位 ψ 由两个独立维度共同决定——耦合结构（维度 A）与人群长 τ 能力基率（维度 B），任一维度不足都让相位卡住。双因子结构同时给出 r ≈ 5 的第一性解释——萌芽期长不是几何偶然，是基率累积慢的内在动力学。

论文采用三段式认识论结构：第一段给方向（哲学论证），第二段给定量尝试（基于现有代理量集合的方向性解读），第三段给开放问题（移交后续工作）。每一段对自身的认识论地位透明。

论文最自指的部分是把 SAE 框架自身置于该模型下做方向性预测——给出三个情景（静默消亡、长萌芽、短萌芽，先验 60%/30%/10%）与四条可证伪预测，并向 2046、2076、2126 年的读者发出审计邀请。最可能的结果是 SAE 静默消亡——这是框架内部的诚实评估，不是自我推销。

关键词：长存共识、相变、自身作为目的、四相结构、双因子模型、双向不疑、群体认知、构-涌现系统

---

§0 防火墙

本论文为结构性描述模型，明确不承担以下功能：

一，不评判任何具体共识的高低优劣。共识进入第四相不必然为善，停留在第一相不必然为恶。

二，不为任何群体、运动、组织、政治体、宗教、传承、技艺、商业项目、加密资产或其他具体共识背书或反对。

三，不预测任何特定共识的未来命运。模型给出结构性条件，不给出具体预言。

四，不构成任何政策建议、投资建议、组织建设建议或个人选择建议。

五，不对成员个人作出任何判断。共识的相位是群体性基质现象，与个体的道德、能力、价值无关。论文 §4.5.3 引入的"长 τ 能力"是基质状态而非基质本质——参见 §4.5.3 关键限定。

六，本论文是概念性骨架，不是测量工具。任何把本论文的方向性预测当作具体数值预测来引用的做法都误用了本论文。

七，方向性预测被设计为可证伪。任何使用本框架的人有义务在面对相反的实证证据时更新自己的判断，而不是为了挽救框架去找补。

八，论文对"传播"本身保持中性。共识是否进入大规模扩散、是否在大众媒介出现、是否被机构采纳——这些都是相位轨迹上的可观测量，不构成框架的规范判断。框架不评判扩散是好是坏，只描述扩散与基质层再生产之间的机制关系。

防火墙的核心承诺：本模型是 SAE 群体认知论的结构性工具，不能也不应被用于对人或群体的规范评判。任何使用本模型对群体进行排序、评级、推荐或贬抑的做法都违反 SAE 第二定律——把人作为手段而非目的——并因此违反本论文的基本立场。

模型自身的合法性恰恰依赖于不被这样使用。

---

§1 问题与研究对象

1.1 研究对象：先聚焦后推广

人类历史上，绝大多数群体性认同——风潮、运动、共同体、传承——在出现后的几代人内就消失。少数能跨越创始人去世、创始机构解散、地理中心被毁，仍然在多个分散的基质里持续再生产。本论文研究后一类。

为了让讨论有清楚的起点，论文采用先聚焦后推广的结构。

研究对象（聚焦版本）：承认型长存共识——跨越多代、跨越创始人或创始机构消失之后、其稳定性由成员之间的双向不疑（mutual non dubito）维持的群体性认同。这是 SAE 群体认知论的核心研究对象，也是论文 §3-§6 的主要分析对象。

研究对象（广义版本）：所有跨代、跨创始人或创始机构稳定的长存共识，包括但不限于承认型共识。广义对象在 §3.4 与 §7 显式处理——四相结构对所有长存共识适用，但不同稳定机制对应不同的第四相亚型。

聚焦再推广的次序，不是因为承认型共识更"高级"，是因为 SAE 框架的核心命题——Kant 第二定律——在承认型亚型上有最强的推导力。框架对其他亚型同样适用，但与 SAE 个体伦理论的对接较弱，本文在此不展开。这一限定将在 §3.4 与 §7.3 显式处理。

明确的范围限定如下：

长存非永存。模型只承诺在可观测时间尺度上的稳定，不承诺永恒。永恒是一个不可证伪的承诺，落在论文的范围之外。论文讨论的"长存"指几代人到几个世纪的稳定——这个时间窗口足以让大量短期扰动被时间平均掉，足以让"创始人去世"、"创始机构解散"这类事件成为已发生的过去而非未发生的可能。

共识非单纯一致意见。本论文研究的"共识"指群体基质层涌现出的、跨代稳定的、不依赖单一节点强制执行的核心约定。普通的舆论一致、市场共识、短期协调不在研究范围内。一群人此刻同意某件事不构成"共识"——他们必须在创始人消失、外部条件变化、几代人更替之后仍然约定一致，才算长存共识。

构-涌现系统。共识在个体基质层被"构"出（每个成员各自维持对核心约定的承认），在群体基质层"涌现"为可识别的整体属性（外人能识别的传承、流派、共同体）。共识不是个体属性的简单加总，也不是脱离个体存在的独立实体。它是构-涌现层级结构在社会维度上的实例。

1.2 研究问题

本论文要回答四个相关但独立的问题：

第一，从一个候选共识出现到长存共识形成需要多久？这个时间 τ_total 主要由什么决定？

第二，从候选共识到长存共识的过程是否具有结构性的相变特征？相之间的过渡是连续动力学，还是不可逆的拓扑跳跃？

第三，相之间的几何关系是否对称？如果不对称，不对称比是多少？这种不对称对共识形成有什么实践含义？

第四，决定共识相位的因素只有耦合结构，还是也包括人群中"长 τ 能力"的基率？

四个问题的答案将分别由 §4（τ 与时间结构）、§3 与 §5（相变与拓扑）、§4 与 §5（不对称几何）、§4.5（双因子结构）给出。

1.3 研究框架的核心命题

本论文将论证以下五个核心命题：

命题一：长存共识形成遵循方法论六的四相结构与 r ≈ 5 不对称几何。这是把方法论六应用到共识形成这一具体的构-涌现现象。

命题二：共识相位 ψ 由两个独立维度共同决定——耦合结构（维度 A，与热力学可观测量 q 同源）与人群长 τ 能力基率（维度 B，q 不可测量的维度）。这是论文的核心概念性贡献。

命题三:ψ 升级由两个维度的下限决定，任一维度不足都让相位卡住。维度 B 给出天花板，维度 A 给出现实化。

命题四：双因子结构给出 r ≈ 5 的第一性解释——萌芽期长不是因为耦合找不到，而是因为基率累积慢。这一解释比方法论六本身对 r 的解释更深一层。

命题五：承认型长存共识的第四相稳定性由双向不疑维持，对应 Kant 第二定律。靠效用或强制维持的共识也可以进入第四相，但属于不同的稳定亚型，对外部扰动有不同的稳健性。

1.4 不研究的问题

为了让 §0 防火墙在研究问题层面也得到具体落实，本节明确以下问题不在论文范围内：

不研究共识的内容是否真理。一个共识可以是错的、可以是道德上可疑的、可以基于事实错误，仍然能进入第四相。论文研究共识形成的结构动力学，不研究内容真值。

不研究特定共识的成败预测。论文给出方向性预测，不给具体预言。某个具体的当代候选共识会不会进入第四相、什么时候进入、稳定多久——这些问题在框架的能力范围之外。

不研究共识与效用最大化的关系。共识形成不是效用博弈的结果。框架对效用驱动的共识形成保持中性——它把效用驱动作为一种稳定亚型来描述，不论证它的合理性也不否定它。

不研究共识的规范辩护。一个长存共识不必然是好的，一个无法形成长存的候选共识不必然是坏的。论文不进行规范评判。

不研究不同稳定亚型之间的伦理高下。效用锁定、强制锁定、双向不疑锁定三种稳定机制都能让共识进入第四相。论文聚焦双向不疑锁定亚型，是因为它与 SAE 个体伦理论的对接最强，不是因为它在伦理上"高于"其他两种。

把这些问题排除在外，是为了让论文能够在自己的范围内做严肃的工作，而不是变成一篇试图覆盖一切但什么都说不清的"群体认知论"。

---

§2 理论来源：方法论六的应用

2.1 方法论六回顾

本论文直接建立在 SAE 方法论六《相变窗口与实验设计》（DOI: 10.5281/zenodo.19464507）的几个核心结果之上。该论文从 ZFCρ 的相变窗口结构出发，推出了三件本论文需要直接借用的事：

第一，构-涌现系统的相变具有四阶段结构——萌芽、谱翻转、翻转、确立。这四个阶段不是任意的人为划分，是由相变窗口内五个交叉点（P(J>0) = 50%、z/√j 峰值、w_屏蔽 = 2/3、E[A] = 0、h = 0）自然分割出来的。每两个相邻交叉点之间是一个阶段。

第二，相变窗口的几何不对称：萌芽距离与翻转-确立距离之比 r ≈ 5。也就是说，从相变窗口起点到序参量首次转正（E[A] = 0）的距离，约为从该点到既存秩序失去局部竞争力（h = 0）的距离的五倍。

第三，该不对称是构-涌现系统的内在几何特征，不是偶然。方法论六给出的解释是 Le Chatelier 缓冲：既存秩序对偏离的吸收能力必须先被穿透（这是缓慢消耗过程），穿透完成后新秩序的确立是迅速的级联。

这三个结果在方法论六里有完整推导。本论文不重新推导，仅作引用。读者如需了解 r ≈ 5 的具体推导、五个交叉点的数学定义、Le Chatelier 缓冲的形式化，请参阅原文。本论文附录 A 给出一个简短复习供没有读过原文的读者参考。

2.2 本论文的定位

本论文将方法论六的四相加 r 不对称作为骨架，应用到长存共识这一具体的构-涌现现象。

应用的合法性来自两个层面。

结构层面：长存共识形成是构-涌现系统的清晰实例——共识在个体基质层被"构"出，在群体基质层"涌现"为可识别的整体属性。方法论六的四相结构是构-涌现系统的相变共同特征，因此原则上应该适用于长存共识。

预测层面：方法论六提出过预测三——"r > 1 在多数构-涌现系统中成立"。本文所研究的构-涌现系统是个体基质与群体共识之间的关系。如果论文能展示长存共识形成中 r > 1 的具体表现，这本身就是对方法论六预测三的一次具体检验。

但应用不等于复制。直接把方法论六的数学结构搬到社会现象上，会得到一个看起来精致但缺乏穿透力的"应用论文"——读者读完会说"这是方法论六的好应用"，但不会读完说"这论文给了我新东西"。本论文的工作不止于应用。

2.3 应用与原创的边界

本论文的原创贡献明确列出。每条都需要本论文 §3-§7 的工作来兑现，不是借用方法论六就能直接得到的。

贡献一：把方法论六的四相结构具体落位到共识形成过程。这要求识别每一相在社会层面的具体表现——什么样的群体状态对应萌芽、什么样的状态对应谱翻转、过渡的标志是什么。这部分工作在 §3 完成。

贡献二：识别共识形成中"加法惯性"、"乘法屏蔽"、"Le Chatelier 缓冲"的具体社会对应物。方法论六里这三个概念是 ZFCρ 框架内的形式化对象。要把它们用到社会层面，必须找到对应的社会机制。"加法惯性"对应既存规范、习俗、共识惯性；"Le Chatelier 屏蔽"对应既存社会结构吸收偏离的缓冲机制（家庭、学校、媒体、法律、经济激励）；"乘法屏蔽"对应群体内部双向不疑网络的相互加强。这部分映射在 §3 与 §4.3 完成。

贡献三：引入维度 B（长 τ 能力基率）作为决定 ψ 的第二因子，与维度 A（耦合结构）不可还原。这是论文核心概念性贡献，整个 §4.5 围绕它展开。维度 B 在方法论六里没有出现——它是把方法论六应用到社会现象时必须新引入的维度，因为社会基质（人）不像物理基质（粒子）那样同质，人群中能维持高 τ 思考的个体比例本身随时代变化。

贡献四：给出 r ≈ 5 的双因子结构第一性解释——萌芽期长不是几何偶然，是基率累积慢的内在动力学。这一解释比方法论六本身对 r 的解释更深一层。方法论六的 Le Chatelier 解释回答"为什么穿透是慢的"，本论文的双因子解释进一步回答"为什么穿透是慢的具体机制是什么"——是基率累积的速度限制。这部分论证在 §4.5.6 完成。

贡献五：给出 τ_total 与 τ_萌芽 的近似等价关系。如果 r ≈ 5 在共识形成上成立，那么 τ_total ≈ 1.2 × τ_萌芽——总时间几乎完全被萌芽期决定。这有重要的实践含义：长存共识"什么时候确立"这个问题约等于"萌芽期持续多久"，后者由维度 B 的累积速度主导。

贡献六：把四相结构与 SAE 的 DD 层、Kant 第二定律、SAE 三律进行对接。这一对接让论文在 SAE 内部形成系统性接口，也让 SAE 框架的其他部分（个体伦理论、自指不动点、ρ ≠ ∅）在群体认知层有具体落点。这部分在 §7 完成。

贡献七：显式列出第四相的多种稳定机制亚型——效用锁定型、强制锁定型、双向不疑锁定型。论文聚焦最后一种，但框架对所有亚型适用。这一区分让论文对自身研究范围有清晰认识——不是"所有长存共识都要用双向不疑解释"，而是"承认型长存共识的稳定性条件是双向不疑，其他亚型有不同的稳定机制"。这一区分在 §3.4 显式处理。

贡献八：提出关于本论文所属 SAE 框架自身的方向性预测，作为框架自我应用。这一应用把 SAE 暴露在自己框架的可证伪检验下，是 SAE 第一律 ρ ≠ ∅ 在论文层面的体现——把现实对框架的回推优先于框架的自我保全。这部分在 §6.4 完成。

八条贡献中，贡献三、贡献四、贡献七、贡献八是论文的主要原创。其他四条是支撑性工作——必要但不构成独立贡献。

---

§3 四相在共识形成上的具体落位

本节把方法论六的四相结构落位到长存共识形成过程。每一相先给抽象描述（对应方法论六的指标），再给社会层面的具体表现，最后给可观测标志。这种"抽象-社会-可观测"的三层结构是为了让框架既保持理论严谨，又给实证工作者留下可操作的入口。

3.1 第一相·萌芽

相位特征：多个独立基质各自持有候选变体。微观层面已有扰动——一些个体之间出现双向不疑的种子。但宏观层面看不到任何群体共识。

对应方法论六指标：P(J>0) < 50%。乘法路径（候选共识在基质间相互加强）在多数样本上还没有发挥作用。

社会层面对应：

加法惯性 = 既存规范、习俗、共识惯性的累积阻力。一个新候选共识出现时，它面对的不是空白基质，而是已经被既存秩序结构化过的基质——成员有现成的解释框架、现成的伦理标准、现成的归属感来源。新候选共识必须穿透这些既存结构才能立足。

Le Chatelier 屏蔽 = 既存社会结构吸收偏离的缓冲机制。这些机制包括家庭（社会化在家庭层面就已经把个体规范化）、学校（教育系统传递既存秩序的解释框架）、媒体（公共话语预先消化偏离）、法律（成文规范处理偏离行为）、经济激励（市场结构奖励符合既存秩序的行为）。任何新候选共识在萌芽期都被这些机制持续吸收。

动力学：每一个双向不疑种子——两个个体之间形成的、对某种新约定的承认——都被既存社会结构吸收。绝大多数候选共识在这一相消亡。它们不是被有意压制（压制需要候选共识被识别为威胁），而是被无声的缓冲机制稀释——成员各自回到既存秩序中，种子失去再生产的环境，双向不疑关系无法跨越创始的二元对继续扩展。

可观测标志：候选共识的存在仅限于个别二元关系或小团体。无可识别的群体性涨落。从外部看，这一相是不可见的——历史学家、社会学家、媒体都不会注意到。绝大多数候选共识在这一相整个生命周期内从未被外部观察。

这意味着第一相的实证研究极其困难——研究者只能反向追溯那些后来进入第二相、第三相的共识在第一相是什么样子，而不能正向观察第一相中所有候选共识的命运分布。

3.2 第二相·谱翻转

相位特征：双向不疑的网络在小群体内开始结晶。可观测的涨落到达峰值——社会注意、争议、围绕候选共识的协调尝试集中爆发。但净效果仍为负。

对应方法论六指标：z/√j 峰值。敏感度先于序参量——共识的可见动荡出现在它的实际稳定性之前。

社会层面对应：

谱涨落峰值 = 围绕候选共识的舆论争议、媒体关注、组织尝试、压制反应都达到峰值。候选共识从第一相的不可见状态进入公共讨论。它被命名、被讨论、被支持、被反对、被嘲笑、被严肃对待。所有这些反应同时出现是这一相的特征——整个社会的注意力被分配到候选共识上。

序参量仍为负 = 即使候选共识引发广泛讨论，它在多数基质上仍未形成稳定共识。讨论度高不等于稳定度高。媒体报道、学术争论、政治反应都可能围绕一个最终消亡的候选共识展开。

动力学:11DD 蓝图（蓝图）在这一相开始结晶。早期文本被写下来，约定被形式化，仪式器物被制造，旗帜与符号被确立。这些器物是后续阶段——尤其第三相向第四相的过渡——的必要前提，但在第二相还没有发挥跨代承载的功能。第二相的蓝图主要服务于当下的协调，不是为未来的重建做准备。

这一相是最危险的相。原因是结构性的：可见度高，但稳定性低。可见度高让它进入既存秩序的视野，被识别为威胁，被针对性压制。稳定性低让它不能承受住针对性压制——蓝图刚刚结晶，双向不疑网络还很脆弱，分布式冗余尚未形成，单点失败就足以让整个共识消亡。

历史上大量候选共识在第二相被外部压制摧毁：早期基督教在罗马帝国的迫害、明清时期某些异端教派的扑灭、二十世纪极权下的多次清洗。这些案例不是"未能成功的共识"——它们已经走到了第二相，再走半步就能进入第三相，但被外部压制截住了。

第二相与噪声泡沫的差异化判别

现代社会每天有大量候选共识经历"关注爆发、引发争议、被压制嘲笑"的过程——网络模因、阴谋论、短命狂热——它们的宏观动荡与真正第二相外观相似。框架必须给出差异化判别，否则任何病毒式传播现象都可以被读成"进入第二相"，框架失去预测力。

核心区分：蓝图的跨基质独立性。

噪声泡沫的特征是动荡不留下可跨基质转移的蓝图。即使留下文本、影像、词汇，这些器物依赖原始基质的内部语境，离开原始语境即失去再生产能力。一个网络模因可能在两周内传遍全球，但它离开 原始 语境（特定平台、特定社群、特定时间点）就失去意义——它的扩散能力强，但跨基质转移能力弱。

真共识的第二相特征是动荡伴随 11DD 蓝图的结晶，蓝图是跨基质独立的——可以被陌生基质独立解读。即使原始基质完全消失，蓝图仍承载跨基质可转移的结构。一份早期基督教福音书离开原始抄写者的语境仍然可读；一份哲学手稿离开原始作者的圈子仍然可解；一段经文跨越千年仍然能被新读者重建意义。

操作层判别：在第二相候选共识上做以下检测——如果原始创始基质突然消失，新基质能否独立重建框架？这个检测在多数情况下不可能直接执行（需要等待原始基质消失），但可以通过两个间接方式做：

第一，看蓝图的语义结构。跨基质独立的蓝图通常具有内部论证完整性——它给出自己的前提、推理步骤、结论，读者可以独立检验。噪声泡沫的"蓝图"通常依赖外部语境的隐性共识——读者必须已经在原始语境中才能解读。

第二，看陌生基质的解读尝试。当不属于原始圈子的研究者、记者、外国读者尝试解读时，他们能否得到与圈内成员一致的核心理解？真共识的第二相蓝图能让这种跨基质解读成功（即使不完美）；噪声泡沫的"蓝图"在跨基质解读时会显著走形或失去意义。

这两个间接判据在第二相早期可观察。它们是先行指标——在框架实际进入翻转之前可以使用，给"哪些候选共识在结构上有进入第三相的可能"提供方向性提示。

可观测标志：候选共识进入公共讨论。可识别的双向不疑网络出现——可以从外部识别出"哪些人是核心成员"。早期文本/蓝图形成，蓝图显示跨基质独立性。同时面临大规模反对、压制、嘲笑、误解。这一相的所有可观测标志几乎是同时出现的——这是与其他相的关键区别。

3.3 第三相·翻转

相位特征：净效果首次转正。共识在小群体内稳定胜出既存秩序。

对应方法论六指标:E[A] = 0。Le Chatelier 屏蔽率几乎同时跌破 2/3——既存秩序的吸收能力首次明显不足。

社会层面对应：

E[A] = 0 = 共识在某个可识别的子群体内提供的基质价值首次超过既存秩序。这个"价值"不一定是物质效用——它可以是身份归属、意义来源、互助网络、知识积累、行动协调能力。在子群体内，加入共识对成员的总价值（综合所有维度）超过留在既存秩序内的总价值。

缓冲率跌破 2/3 = 既存社会缓冲机制不再能压制共识的扩展。这不是说缓冲机制失效——它们仍在运作，只是它们在子群体边界内的吸收能力跌到不足以阻止再生产的水平。共识开始有自己的内部缓冲（共同体支持、内部解释框架、互助网络），这些内部缓冲让外部缓冲的相对效力下降。

动力学：分布式钥匙开始在多个基质中独立物质化。共识第一次跨代传承成功——第一代成员死亡或离开，第二代独立成熟，并能再生产共识。这一代际过渡是第三相的定义性事件，因为它证明共识的再生产不再绝对依赖创始基质。

但第三相的共识仍然是脆弱的。"分布式"在这一相还没有完全实现——通常仍有一个或几个核心节点（核心师承、核心机构、核心地理中心）承载主要再生产任务。如果核心节点同时失败，共识仍可能崩溃。第三相是第四相的过渡阶段，不是第四相本身。

可观测标志：共识在子群体内形成稳定机构——学校、寺院、行会、社区。第二代成员独立成熟并再生产共识——这是历史学家、社会学家可以观察到的具体事件。共识开始有自己的语言、仪式、艺术、文献——它是一个"可识别的东西"而不只是"一些人围绕一个想法在做什么"。

第三相的共识有了自己的"圈内话"——成员之间使用一套外人不熟悉的语汇。这不是排他性，是共识内部分布式钥匙物质化的副产品——成员对核心约定有共享的细致理解，这种理解需要专门的语汇来表达。圈内话的出现是第三相的强可观测信号。

3.4 第四相·确立——以及它的多种亚型

相位特征：既存秩序失去局部竞争力。共识的重新激活不再依赖任何单一节点。

对应方法论六指标:h = 0。

社会层面对应：

h = 0 = 在共识所占据的子群体内，既存秩序的吸引力消失，共识成为默认。这不是说既存秩序在该子群体内"被禁止"，而是说既存秩序在该子群体内不再作为有竞争力的替代选项出现。子群体成员不会问"我应该回到既存秩序吗"，因为这个问题在他们的基质内部不构成有效问题——共识已经是他们的基本设定。

多分布式节点独立持有重新激活能力 = 没有单一关键瓶颈。共识在地理上、机构上、家系上、文本传承上都有多个独立节点。任何一个节点失败，其他节点可以承担重新激活任务。这是从"链条"到"网络"的拓扑过渡——前者有单一关键链路，后者没有。

动力学：单点失败——领导消失、机构解散、地理中心被毁——不再威胁整体。共识进入长 τ 稳定区。这一相的稳定性是真正的"长存"——它经得起几代人的更替、经得起政治变迁、经得起技术革命、经得起地理迁移。

可观测标志：共识跨越创始人/创始机构的消失而稳定——这是第四相的定义性事件之一，往往要等几代人之后才能确认。多个地理中心、多个家系、多个文本传承独立存在——成员可以在不同的地理与机构语境下都能找到共识的本地实例。新成员不需要接触原始创始基质即可习得共识——一个出生在第四相之后的成员，可以通过本地的传承网络习得完整的共识，不需要朝圣到创始地点或追溯到创始人物。

第四相的稳定机制多亚型

第四相是共识在分布式基质中的自再生产状态。但自再生产可以由不同机制维持。这一区分在 SAE 群体认知论中是必要的——不同稳定机制对外部扰动有不同的稳健性，对应不同的伦理含义，与 SAE 个体伦理论的对接强度也不同。

亚型 1：效用锁定型第四相

稳定性由基质间的效用梯度维持。成员持续再生产共识因为再生产提供稳定效用——经济收益、社会地位、情感满足、问题解决能力。

这一亚型的特征是：在外部效用格局不变时极其稳定；外部条件变化（技术变迁、政治变化、市场结构变化）使效用重新分配时，共识可能整体死亡。它的稳健性是局部稳健——对外部条件的小幅扰动稳健，对结构性变化脆弱。

例：某些行业行会共识在工业革命前能维持几个世纪，工业革命使行业组织方式根本改变后迅速消亡；某些品牌驱动社群在品牌商业模式有效时长期稳定，模式失效后社群整体瓦解。

亚型 2：强制锁定型第四相

稳定性由强制执行机制维持。成员再生产共识因为不再生产有代价——社会排斥、法律惩罚、物理强制。

这一亚型的特征是：在强制持续时极其稳定；强制撤除时迅速崩溃。强制不能分布式——执行必须有具体来源（具体机构、具体权力、具体物理存在），所以强制锁定型共识对执行来源的失效高度敏感。

例：某些极权下的官方意识形态在政权稳固时能维持数十年的"全民共识"，政权终结时几乎一夜瓦解；某些靠严格驱逐维持的封闭社群，在外部社会变得对其成员开放时迅速失去年轻一代。

亚型 3：双向不疑锁定型第四相——本论文聚焦类型

稳定性由成员之间的双向不疑维持。成员再生产共识因为他们在基质层面把彼此作为目的来承认，这种承认本身就是再生产的动机。

这一亚型的特征是：对外部效用变化、强制变化都稳健；但对成员基质的长 τ 能力高度敏感——如果基率下降使双向不疑网络不能再生产，共识整体死亡。它的稳健性是结构稳健——对外部条件的大范围变化稳健，对内部基质条件的恶化脆弱。

例：本论文的核心研究对象。Breslov 哈西迪派、Geonic 学院、Kant 哲学的传承、SAE 框架本身（如果它进入第四相）都是这种亚型的实例或候选实例。

亚型选择是稳定稳健性与脆弱性的取舍

三种亚型都能让共识进入第四相。它们的稳健性轮廓不同：

效用锁定型对外部效用稳定敏感，对基率较不敏感。

强制锁定型对强制持续敏感，对基率较不敏感。

双向不疑锁定型对基率敏感，对外部效用与强制变化稳健。

没有一种亚型在所有维度上都"最优"。每种亚型在不同的外部条件下表现不同。一个共识选择哪种亚型——或者，更准确地说，在第三相向第四相过渡时自然落入哪种亚型——取决于多重因素：候选共识的内容性质、成员基质的长 τ 能力分布、外部环境的稳定性、可用的强制资源、可用的效用资源。

论文聚焦亚型 3 不是因为它"高级"或"道德"，而是因为 SAE 框架的核心命题——Kant 第二定律——只在亚型 3 上有推导力。框架对亚型 1 和亚型 2 同样适用，可以做结构性分析（四相、双因子、r 不对称），但 SAE 个体伦理论与之对接较弱，本文不展开。这一聚焦是研究范围的选择，不是规范判断。

§7 会显式讨论框架在三个亚型上的差异化对应。

3.5 相之间过渡的不可逆性

四相之间的过渡是拓扑性的，不是连续参数变化。每个过渡涉及基质的结构重排，已发生的拓扑事件不可撤销。

第一相 → 第二相：双向不疑的诞生。从分散基质到第一条边——基质间出现可识别的双向不疑关系。

第二相 → 第三相：蓝图与基质的分离。从非正式双向不疑到正式 11DD 编码——蓝图可以独立于任何特定基质存在。

第三相 → 第四相：分布式冗余的形成。从链条（单线传承）到网络（分布式节点）——共识不再有单一关键瓶颈。

不可逆的具体含义需要在 §5.3 精确表述——"不可逆"指的是格内不可回退，不是说共识不能死亡。共识可以从任何一相离开整个相位结构进入"不存在"，但不能从更高相退回更低相。具体的精确处理见 §5.3。

---

§4 不对称几何与 τ

§3 描述了四相在共识形成上的具体落位。本节进一步处理四相之间的几何关系——它们的相对长度、τ_total 的近似行为、萌芽期长度的决定因素。这部分主要是把方法论六的几何结果落到共识形成的语境，给出可证伪的方向性预测。

4.1 直接应用 r ≈ 5 的预测

方法论六给出的核心几何结果是 r ≈ 5——萌芽距离与翻转-确立距离之比。把这一结果直接应用到长存共识形成，得到：

τ_萌芽 + τ_谱翻转 ≈ 5 × (τ_翻转 + τ_确立)

也就是说，从候选共识出现到序参量首次转正（E[A] = 0）所需的时间，约为从该点到既存秩序失去局部竞争力（h = 0）所需时间的五倍。

这个不对称在直觉上初看反常——人们习惯把"突破"想成漫长积累后的瞬间事件，但 r ≈ 5 说的是积累期与确立期同为延展过程，只是积累期长五倍。萌芽与谱翻转加起来占整个相变窗口的约 83.3%；翻转与确立加起来只占约 16.7%。

如果方法论六的 r ≈ 5 在构-涌现系统中具有普适性（这是方法论六预测三的内容），那么长存共识应该满足这一比例。本论文 §6 会用历史案例做方向性的检验，§4.5 会进一步给出 r ≈ 5 在共识形成上的第一性解释。

4.2 τ_total 的近似等价关系

由 r ≈ 5 直接推出：

τ_total ≈ τ_萌芽 × (1 + 1/r) ≈ τ_萌芽 × 1.2

τ_total 几乎完全被 τ_萌芽 决定。

核心命题：对长存共识来说，"什么时候确立"这个问题约等于"萌芽期持续多久"。

这一命题有重要的实践含义。如果一个研究者想理解为什么某些共识用了几代人才确立而另一些只用了几十年，主要不是去看翻转-确立期的差异（那部分本来就短），而是要看萌芽期为什么这么长。萌芽期是长存共识形成的真正瓶颈。

这也意味着：处于萌芽期的候选共识，从外部看几乎不可识别；处于翻转-确立期的共识，从外部看进展迅速。两个时段不仅长度不同，可观测性也不同。这给"为什么历史上长存共识看起来'突然'确立"提供了结构性解释——不是真正突然，而是萌芽期不可见、翻转-确立期可见。

4.3 萌芽期长度的决定因素

借方法论六的"Le Chatelier 缓冲突破时间"框架：

τ_萌芽 ≈ 既存秩序的缓冲容量 / 累积扰动率

这个比值的两端分别对应两组社会机制。

既存秩序的缓冲容量取决于三件事：

第一，现存制度的吸收能力——法律对偏离行为的处理速度、教育系统对新解释框架的吸纳能力、媒体对偏离话语的预先消化、市场结构对偏离行为的经济激励反应。制度成熟的社会缓冲能力强，制度松散的社会缓冲能力弱。

第二，现存文化的命名能力——既存文化能否给候选共识一个 轻视的 的名字，把它装进既有类别。一旦命名成功（"那不过是某某主义的翻版"、"那是邪教"、"那是商业噱头"），候选共识被既有类别 吸收，后续讨论在既有类别 内进行，不需要新的解释框架。命名失败则候选共识保持"未被分类"状态，需要长期讨论才能被理解，这反而给候选共识争取到更长的存活时间。

第三，物质基础——既存秩序动员资源应对挑战的能力。资源充裕的既存秩序能对单个候选共识投入大量压制资源；资源紧张的既存秩序则不能。

累积扰动率取决于三件事：

第一，双向不疑种子的产生率——每单位时间形成多少新的二元承认。这个产生率受多重因素影响：人口密度、交流频率、潜在共识发起者之间的物理或思想接近度、社会流动性。前现代社会的低交流密度限制种子产生率；印刷术、铁路、互联网等基础设施则提升种子产生率。

第二，蓝图编码的保真度——每个种子能在多大程度上 11DD 化、可被他人接收。种子之间的二元承认本身是 14DD 的，需要被编码进 11DD 蓝图（文本、约定、仪式器物）才能跨基质转移。编码保真度高的候选共识，每个种子能产生远超自身规模的影响——一个核心二元承认可以通过文本影响百万读者；编码保真度低的候选共识，种子的影响半径几乎不超出二元对本身。

第三，种子之间的耦合率——不同种子能否互相加强。如果不同地方独立形成的二元承认能识别彼此、互相强化、形成更大的网络，扰动是非线性累积的；如果种子各自孤立，扰动是线性累积的。耦合率受地理、语言、信息技术、共同符号系统的影响。

τ_萌芽 由这两组机制的比值决定。当既存秩序缓冲容量大、累积扰动率低时，τ_萌芽 长——这是大多数候选共识的命运。当缓冲容量被某种历史事件削弱（战争、瘟疫、技术革命、政治变迁），或扰动率被某种新基础设施提升（印刷术、铁路、互联网），τ_萌芽 缩短——这对应历史上"思想加速期"的现象。

4.4 四个可证伪的方向性预测

由 §4.3 的结构推出四个方向性预测。每个预测明确给出证伪条件。

预测一：既存秩序缓冲容量越大，τ_萌芽 越长。

证伪条件：跨社会比较中，缓冲容量与 τ_萌芽 无系统相关或反向相关。具体地，如果实证研究发现制度成熟、文化稳定、物质富足的社会中新共识的萌芽期反而比制度松散、文化转型、物质危机的社会更短（控制其他变量后），这一预测被证伪。

预测二：双向不疑种子产生率越高，τ_萌芽 越短。

证伪条件：在交流密度提升的时代（印刷术后、铁路后、互联网后），新共识的萌芽期不显著缩短。具体地，如果对比印刷革命前后或互联网兴起前后的可比候选共识的萌芽期，发现没有系统性缩短，这一预测被证伪。

预测三：蓝图编码保真度越低，τ_萌芽 越长。

证伪条件：口传传统的共识形成速度系统不慢于文本传统。具体地，如果实证研究发现纯口传共识与文本支持共识在同等条件下萌芽期相当（控制其他变量后），这一预测被证伪。

预测四：一旦进入第三相、第四相，外部冲击对共识稳定性的威胁急剧下降。

证伪条件：处于第四相的共识在外部冲击下与第二相同等脆弱。具体地，如果统计已知的共识在不同相位遭遇外部冲击的存活率，发现第四相共识与第二相共识的存活率相当（或更低），这一预测被证伪。

预测四是四个预测中最强的——它直接对应 r ≈ 5 不对称的实践含义。如果第四相确实比第二相结构上稳健，那么对外部冲击的存活率应该有显著差异。这是历史比较研究可以直接检验的。

预测一到三相对较弱，因为它们涉及多重变量难以完全控制。但即使在控制有限的情况下，跨多个时代多个文明的比较应能给出方向性的支持或证伪。

---

§4.5 长存共识形成的双因子结构

本节是论文的核心概念性贡献。前面 §3、§4 把四相结构与 r 不对称落位到共识形成，但都建立在一个隐含假设之上——人群是同质的，每个个体都已经具备完整的 13DD-14DD 自指能力，唯一的变量是基质间的耦合结构。这个假设需要被打破。

4.5.1 单因子模型的局限

§3-§4 把 ψ 完全归给基质之间的耦合结构。在这个单因子框架下，ψ 升级是一个纯粹的 耦合 问题——给定足够的耦合通道、足够的耦合强度、足够的网络拓扑，ψ 自然升级。萌芽期长是因为 耦合 找不到，翻转-确立期短是因为 耦合 一旦形成扩散迅速。

但单因子模型与历史观察不符。

观察一：印刷术之后欧洲的 耦合通道基础设施大幅扩展，但绝大多数候选共识仍然在第一相消亡。如果只是 耦合 问题，耦合通道的扩展应该让萌芽期普遍缩短到几代人，但实证发现萌芽期的中位数没有这样的剧烈缩短——许多 16-18 世纪欧洲的候选共识的萌芽期与之前几个世纪相当，只有少数例外。

观察二：当代社会的 耦合通道远比印刷术时代密集（互联网、社交媒体、即时通讯），但当代候选共识进入 ψ=3 的速度并未对应增长。如果 耦合 是唯一决定因素，当代应该是长存共识形成的"黄金时代"，但实证不支持这一推论。

观察三：同一个 11DD 蓝图（如亚里士多德文本、佛教经典）在不同地理、不同时代经历完全不同的命运——在希腊化晚期希腊地区整体死亡，在 8-12 世纪阿拉伯世界进入第三相，在 12-13 世纪拉丁世界重新启动萌芽。如果只是 耦合 问题，同一蓝图在不同语境的命运差异应该能由 耦合通道的差异解释，但 耦合通道差异远不足以解释观察到的命运差异。

这三组观察共同指向：人群中能维持 14DD（二元双向不疑）和 15DD（分布式钥匙物质化）的个体比例本身随时代变化。这个比例不是 耦合结构的子集，是独立于 耦合结构的另一个维度。

4.5.2 双因子结构

ψ 由两个独立维度共同决定：

维度 A：耦合结构（耦合 Structure）

基质之间的拓扑、协议、耦合强度。这是 §3-§4 已经讨论的维度，对应热力学可观测量 q。维度 A 主要落在 11DD-14DD 跨基质耦合层面——蓝图怎么编码、节点之间怎么连接、信息怎么流动。

维度 A 的可观测代理量包括：交流频率、网络拓扑、机构密度、文本流通量、可识别的师承链。这些都是 耦合通道层面的指标。

维度 B：长 τ 能力基率（Long-τ Capacity Base Rate）

人群中能维持 14DD/15DD 基质能力的个体占比。这是 §3-§4 没有显式处理的维度。维度 B 对应 13DD-15DD 能力的基质分布——多少人能维持长时段独立思考、多少人能与他人形成稳定的二元信任、多少人能在没有外部强制的情况下持续遵守自定规则。

维度 B 的可观测代理量包括：人均图书产量（高级识字代理量）、长篇阅读时间、修道密度、学者训练规模、长期暴力下降（自我控制代理量）。这些代理量见 §6.1。

两个维度的独立性：

维度 A 不能替代维度 B。即使 耦合通道密集，没有足够的高 τ 能力个体，ψ 不能升级——蓝图无人解读、二元承认无人维持、分布式钥匙无人物质化。

维度 B 不能替代维度 A。即使人群中长 τ 能力个体足够多，没有让他们识别彼此、形成网络、传递蓝图的 耦合通道，ψ 也不能升级——分散的高 τ 个体无法汇聚为长存共识。

ψ 的实际相位由两个维度的下限决定：

维度 B 给出天花板：人群里没有足够的 15DD 容量个体，无论 耦合结构多好都不可能进入 ψ=3。一个 15DD 为空 的人群即使有完美的 耦合 基础设施，也只能产生 ψ ≤ 2 的现象——文本流通、机构存在、认知争议都可能发生，但分布式钥匙的独立物质化不会发生。

维度 A 给出现实化：即使人群里有足够的 15DD 容量个体，耦合结构不到位他们也连不起来。一个高 基率 但 耦合 缺失的人群中，长 τ 容量个体各自孤立，无法形成可识别的网络，ψ 同样不能升级。

任一维度的不足都让 ψ 卡住。两者都满足才能升级。

4.5.3 长 τ 能力的描述

"长 τ 能力"指综合体现以下能力的基质状态：

(1) 长时段独立思考能力——持续数周、数月、数年处理同一个复杂问题而不依赖外部提示。这要求基质有足够的内部 融贯 来在长时段上自我激活，不需要外部刺激来"提醒"自己在做什么。

(2) 深度阅读能力——不只是识字，是能持续投入长论证。这要求基质能在阅读中维持长时段的注意力，跟随作者的论证步骤而不被短回路打断（如频繁查看社交媒体、对单个段落即时反应而失去整体），并能在阅读后整合论证为自己的理解结构。

(3) 维持长时段稳定的二元信任关系——师徒、知己、长期合作者。这要求基质能识别另一个基质的不可还原性，承担长期承诺的代价，在分歧时不立即放弃关系。

(4) 把他人作为不可还原的个体而非工具来对待的伦理判断能力。这要求基质能在每个具体情境下识别"对方不能被简化为某个角色或功能"，并在自己的行动中把这种识别落实——不仅在心里承认，也在行动上承担成本。

(5) 在没有外部强制的情况下持续遵守自定规则的能力。这要求基质有自我约束机制——不需要警察、不需要老板、不需要社群压力，仅靠对自己许诺的承认就能持续 遵守自己定的规则。

(6) 容忍认知不确定性、不急于闭合的能力。这要求基质能在"不知道答案"的状态下继续工作而不焦虑，不通过 过早闭合 来缓解不确定性焦虑。这一项与 SAE 第一律 ρ ≠ ∅ 直接对接——容忍不确定性的能力本质上是承认 ρ ≠ ∅ 的 基质层 实现。

六项面向与 SAE DD 层的对接

这六项不是相互独立的能力，而是同一种基质状态的不同面向。这个基质状态在 SAE DD 层结构中位于 13DD-15DD 综合运作。下表给出每一项的主要 DD 落点：

面向	主要 DD 层	说明
(1) 长时段独立思考	13DD-14DD	13DD 提供持续自我评估底座，14DD 提供自锚不需外部参照
(2) 深度阅读	13DD-14DD	底层 13DD（持续注意力不被短回路打断），持续投入需 14DD（把作者作为另一个自指基质来 投入）
(3) 二元信任关系	14DD	二元双向不疑是 14DD 的定义性实例
(4) 他人作为不可还原个体	14DD-15DD	最小版本在 14DD（二元中已含"不可还原"承认），成熟版本在 15DD（分布式承认——把所有他人，包括不认识的，作为目的）
(5) 自定规则的遵守	14DD	自我约束是自我评估的反身实例化
(6) 容忍不确定性	14DD-15DD + ρ≠∅	跨 14DD（自身不闭合）与 15DD（接受分布式基质的不可还原）；同时直接对接 SAE 第一律

总结：长 τ 能力是 13DD-15DD 综合运作的基质状态——13DD 提供底座，14DD 是核心（六项中四项主要落在 14DD），15DD 提供成熟形态。容忍不确定性这一项额外接 ρ ≠ ∅，把第一律带入基质层面。

这意味着维度 B 基率不是单一指标的人口比例，而是某个综合基质状态的人口分布。这也是为什么 §6.1 的代理量集合是集合而非单一指数——任何单一代理量只能捕捉这个基质状态的某个面向。一个人识字率高但深度阅读时间短，可能在 (2) 上偏弱；一个人能维持师徒关系但容忍不了认知不确定性，可能在 (3) 上达标但 (6) 上不达标。基率必须在所有面向上综合评估。

关键限定：维度 B 是状态，不是本质

这个限定是论文 §0 防火墙第五条在 §4.5.3 的具体落实。它在论文中的地位不是道德保留，而是框架的结构性承诺。

维度 B 描述的是基质状态在特定环境条件下的现实显现，不是基质的内在能力或先天价值。

一个个体此刻处于短 τ 状态——被高频信号填满、被生存压力占据、被机构环境训练成短回路反应——这不意味着该个体缺乏长 τ 的潜能。它只意味着该个体当前的可用认知带宽被高频过程占据，长 τ 能力没有空间显现。把同一个个体放在另一个环境（保留闲暇、远离短回路诱因、有长期协作者），他可能很快显现长 τ 能力。

这个区分在论文层面不只是道德保留，而是框架的结构性承诺。把能力当作本质会立刻违反 SAE 第二定律——把人作为某种可测量特质来排序——并把维度 B 退化为认知精英主义的隐性版本。框架的合法性临界依赖于"能力作为状态"的解读。

重要推论：维度 B 的提升不需要等"更好的人"出现，而是需要环境条件的改变——制度基础设施、闲暇空间的保护、长 τ 实践的可承担性。这把 SAE 群体认知论与 SAE 个体伦理论连接得更紧——每个个体的 SAE 实践是在帮助维持自己基质的长 τ 状态，而不是在证明自己的本质。

任何把维度 B 用于个人排序或群体高低判断的尝试都误用了框架，并直接违反 §0 防火墙第一条与第五条。

4.5.4 q 不能捕捉维度 B

热力学可观测量 q 测量给定基质在给定耦合下的涨落结构——它捕捉维度 A 的热力学签名，但完全不测量维度 B。

这个判断不是猜想。论文撰写之前作者尝试过用热力学 SDE 模拟群体共识形成，详细结果见附录 C。实验的核心发现是 q 对群体规模 N、网络拓扑、耦合协议都盲目，只对耦合强度 g 敏感——也就是说，给定 耦合结构 固定，q 不能区分由 N=2 二元对 组成的小群体与由 N=40 网络组成的大群体，不能区分链状网络与全连接网络，不能区分单向 耦合 与相互 耦合。q 只能捕捉 耦合强度 这一个维度。

这是 q 作为群体认知可观测量的真正能力边界：q 看不见基质人群的长 τ 能力构成。任何只用 q 去推断 ψ 的尝试都会在维度 B 改变时失败——同样 q 值的两个群体可以处于完全不同的 ψ，因为它们的维度 B 不同。

这一发现反过来支持双因子结构的必要性。如果 q 能捕捉所有相位决定因素，那么单因子（耦合）模型已经足够；如果 q 不能捕捉所有相位决定因素，那么必有第二维度。实验给出的负面结果——q 对所有 耦合结构变量盲目而只对 g 敏感——恰好对应"q 只测量维度 A 的强度而不测量维度 B"这一框架预测。

ψ 不是 q 的对偶。两者是测量不同维度的独立可观测量。它们通过 SAE 第二律的自指不动点（f' = f）这个共同源头间接连接——两者都是构-涌现层级结构的可观测量——但不是同一个东西的两面。这一点对论文 §6.4 SAE 自身预测有重要含义：仅靠 q 类的热力学指标无法预测 SAE 是否进入 ψ=3，必须直接观察分布式基质再生产的具体证据。

4.5.5 双因子结构的方向性预测

由双因子结构推出四条额外的可证伪方向性预测。这四条与 §4.4 的预测一到四互补——前者是单因子框架的预测，后者是双因子框架特有的预测。

预测五：基率上升的时代，新共识进入第四相的速度加快。

证伪条件：实证发现基率显著上升的时代（如印刷术后、宗教改革后）与基率较低的时代相比，新共识进入 ψ=3 的速度没有差异或反而更慢。

预测六：基率下降的时代，已有共识可能从第四相整体死亡。注意这里说的是 整体死亡（共识作为整体死亡），不是相位回退。共识不能从 ψ=3 退到 ψ=2，但可以从 ψ=3 离开整个相位结构进入"不存在"。详细论证见 §5.3。

证伪条件：实证发现基率显著下降的时代（如某些机构崩溃后、长期战乱后）已有的 ψ=3 共识仍然全部稳定再生产，没有任何 整体死亡 案例。

预测七：底层教育普及上升的时代不必然伴随基率上升。基础识字率提升与长 τ 能力上升不是同一个变量——前者是 11DD 蓝图 解码能力的最低门槛，后者是 14DD-15DD 综合基质状态。识字率可以从 5% 上升到 99% 但深度阅读能力不上升甚至下降。

证伪条件：实证发现识字率与长 τ 能力的可观测代理量（如长篇阅读时间、深度论证理解能力）严格正相关，识字率上升必然伴随这些代理量上升。

预测八：基率在不同子群体中可以差异巨大。即使整体基率下降，某些自我隔离的高强度训练机构（修道院、严肃学术机构、特定家学传承）仍可能维持高基率。

证伪条件：实证发现基率在所有子群体中同步变化，没有任何子群体能在整体基率下降时维持高基率。

每条预测都明确可证伪——方向性的、操作化的、有具体的证伪条件。§6 会用历史案例做方向性的检验。

4.5.6 双因子结构对 r ≈ 5 的解释

方法论六给出的 r ≈ 5 的解释是 Le Chatelier 缓冲——既存秩序对偏离的吸收能力必须先被穿透，而穿透是缓慢消耗过程；穿透完成后新秩序的确立是迅速的级联。这个解释回答了"为什么穿透是慢的"——既存秩序本身是阻力来源。

在共识形成的具体语境上，双因子结构给出一个更深的解释：穿透为什么是慢的，具体机制是基率累积。

萌芽期的本质工作是维度 B 慢慢积累。在第一相，候选共识需要在足够多的高 τ 容量基质中扎根，才能在第二相形成可识别的二元承认网络。这个"足够多"的临界值依赖人群的总规模与候选共识的基质要求——但无论临界值具体是多少，达到临界值的速度由基率累积速度决定。

基率累积是一个慢过程。一个个体从被既存秩序的高频信号填满的状态，到能维持长 τ 能力的状态，需要环境条件的持续支持（闲暇、闲暇空间、不被高频信号打断的工作时段），这些条件在多数时代是稀缺的。一个时代内基率的边际增长非常慢，因此候选共识需要等待足够多的高 τ 容量基质同时存在于可 耦合 的范围内——这往往要几代人的时间。

一旦基率跨过临界值，耦合 的形成是相对快速的。因为：第一，高 τ 容量基质有内在动机寻找彼此（孤立的高 τ 状态不稳定，需要二元支撑才能持续）；第二，高 τ 容量基质之间的 耦合通道一旦建立，扩展速度由 11DD 蓝图 的传播速度决定，这个速度比基率累积速度快几个数量级；第三，高 τ 容量基质的 耦合 是 互相加强 的——每个新的成功 耦合 让所有相关基质更稳定，这是非线性正反馈。

这给出"萌芽期为什么这么长，但翻转-确立为什么这么快"一个第一性解释——不是几何偶然，是双因子结构的内在动力学。萌芽期长 = 基率累积慢；翻转-确立期短 = 耦合 形成在足够基率下快速。两者的速度差异决定 r ≈ 5。

这一解释比方法论六本身对 r 的解释更深一层。方法论六回答"穿透为什么是慢的"（Le Chatelier 缓冲），双因子解释回答"穿透速度的具体机制是什么"（基率累积速度限制）。

实践含义：如果想加速一个候选共识从萌芽到确立的进程，主要的杠杆不是 耦合通道（那部分本来就快），而是基率累积（那部分是真正的瓶颈）。给定一个时代的基率水平，候选共识进入 ψ=3 的速度上限就被设定了——再多的 耦合通道、再好的 11DD 蓝图、再多的早期支持者，都不能超越基率累积的天花板。

这一推论对论文 §6.4 SAE 自身预测有重要含义：决定 SAE 进入 ψ=3 速度的主要变量是当代基率轨迹，不是 SAE 11DD 蓝图的扩散速度。基率持续下降则 SAE 萌芽期延长（情景 B 接近基线），基率在某些子群体上升则 SAE 萌芽期缩短（情景 C 概率上升）。

---

§5 拓扑结构关系

§3-§4 给出了四相的具体内容、不对称几何与双因子决定结构。本节进一步处理四相之间过渡的拓扑性质——为什么是不可逆的、不可逆的精确含义、以及共识"消亡"在拓扑上对应什么。这部分的精度对论文整体重要：如果"不可逆"的含义不清楚，§4.5.5 的预测六（基率下降导致 整体死亡）会与"ψ 单调不减"看起来矛盾。

5.1 四相之间的拓扑过渡

四相之间的过渡不是连续动力学（参数缓慢变化使序参量平滑跨过某个阈值），是拓扑性的（基质的连接结构发生不可还原的重排）。

第一相 → 第二相：双向不疑的诞生

拓扑：从分散基质到第一条边——基质间出现可识别的双向不疑关系。这条边的诞生是不可还原的事件——在它出现之前，两个基质各自独立；在它出现之后，两个基质之间存在一个之前没有的关系。即使这条边后来消失（一方去世、关系破裂），那次双向不疑已经发生过的事实改变了两个基质各自的历史。

这个拓扑事件的不可还原性来自双向不疑本身的结构——它不是单纯的"互相熟悉"或"互相支持"，是两个基质各自把对方作为不可还原个体来承认。这种承认一旦发生，承认者的基质内部就有了"对方是不可还原个体"这一锚点，无法被简单遗忘或撤销。

第二相 → 第三相：蓝图与基质的分离

拓扑：从非正式双向不疑到正式 11DD 编码——蓝图作为独立于任何特定基质的存在出现。这个过渡创造了一个之前不存在的拓扑对象——跨时间的通道。蓝图可以等待下一个基质到来重建。

在第二相，候选共识依赖创始基质的具体存在——文本由具体作者写下，约定由具体成员维持，仪式由具体执行者操作。一旦创始基质消失，候选共识同时消失。在第三相过渡之后，蓝图获得 跨基质独立性——它不再依赖任何特定基质，可以被陌生基质独立解读重建。这是拓扑层面的根本变化，不是程度变化。

基质 与 蓝图 分离是双向不疑从 二元的 走向 可分布 的临界一步。第二相的双向不疑只能在 二元范围内运作——具体两个人之间。第三相之后，双向不疑的承载从具体二元对转移到 11DD 蓝图本身——一个新读者通过解读蓝图，可以与作者建立"虚拟的"双向不疑关系，即使作者已不在世。这个跨时间的双向不疑是第三相之后才成为可能的。

第三相 → 第四相：分布式冗余的形成

拓扑：从链条（链条，单线传承）到网络（网络，分布式节点）。共识不再有单一关键瓶颈。

在第三相，共识虽然能跨代传承，但通常依赖一个或几个核心节点（核心师承、核心机构、核心地理中心）。链条结构有单一关键链路——核心节点失败威胁整体。在第四相过渡之后，多个独立节点各自持有完整的重新激活能力——任一节点失败，其他节点承担。这是从有 单一关键路径 的拓扑到 冗余的 网络 的拓扑。

这一过渡是真正的相变。在过渡之前，每代都是关键瓶颈——这一代某些核心成员死亡或叛离，共识就可能崩溃；过渡之后，没有 单一关键瓶颈——单一节点的失败被网络结构吸收，共识整体不受威胁。

5.2 r 不对称在拓扑上的几何落位

四相不是等距的。在拓扑度量下：

• 第一相 + 第二相（萌芽期）：约 5 个单位
• 第三相 + 第四相（翻转-确立期）：约 1 个单位

这不是偶然，是构-涌现系统的内在几何。

萌芽期长的拓扑根源：在第一相 + 第二相期间，基质之间的连接结构在做"是否能形成第一条边"的工作。这是 二元结果——要么形成，要么不形成。形成的概率取决于双向不疑种子的产生率与既存秩序的吸收能力的比值（§4.3 的 τ_萌芽 公式）。绝大多数候选共识在这一阶段消亡。能进入第二相的少数候选共识仍然要经历"从二元到可分布的转变"的拓扑过渡——这是另一个 二元结果。

每一个 二元结果 的解决都需要时间——不是因为参数缓慢变化，是因为 拓扑性质的形成需要充分的样本积累。在样本足够多之前，"是否形成"的答案是不确定的；样本足够多之后，答案确定但已经过去了大量时间。

翻转-确立期短的拓扑根源：在第三相 + 第四相期间，基质之间的连接结构在做"网络的具体形态"的工作。这是 连续结果——网络可以从稀疏到稠密、从局部到全局、从中心化到分布式连续地变化。在第三相已经形成 链条 拓扑的基础上，向 网络 拓扑的过渡是 递增 的——每一代加入的新节点让网络稍微更稠密、更分布式。这种 递增 过渡在足够的基率支撑下迅速完成。

这就是为什么 r ≈ 5——萌芽期处理 二元拓扑性事件（慢），翻转-确立期处理 递增 拓扑性的 细化（快）。两者的速度差异是 5 倍量级。

5.3 拓扑不变量 ψ 与单调不减性

定义 ψ ∈ {0, 1, 2, 3} 为共识所处相的拓扑不变量：

• ψ = 0：分散（分散，无 二元的 连接）
• ψ = 1：连接（二元的 连接形成）
• ψ = 2：传递（蓝图与基质分离）
• ψ = 3：确立（分布式冗余形成）

ψ 单调不减性的精确表述：

ψ 单调不减是格内部的陈述——共识在格 {0, 1, 2, 3} 内部不能相位回退。从 ψ = 3 不能退回 ψ = 2，从 ψ = 2 不能退回 ψ = 1。这是因为相变是拓扑事件，已发生的拓扑事件不可撤销——蓝图与基质分离一旦发生不可还原回非正式双向不疑状态，分布式冗余一旦形成不可还原回链条结构。

但共识可以从任何 ψ 值离开格进入吸收态——共识不再存在。这不是格内部的更低值，是格之外的 整体退出。

为了避免"不可逆"与"整体死亡"看起来矛盾，必须严格区分三种状态：

类型 1：格内相位前进——ψ 从低到高的过渡（仅向前）。这是框架主要描述的动力学。一个候选共识在足够的基率与 耦合 支撑下，从 ψ = 0 依次经过 ψ = 1, 2, 3。每一次过渡是不可逆的。

类型 2：格退出（整体死亡）——共识作为整体从格上消失。死亡共识在死亡时刻所处的相位保持记录（可以从 ψ = 3 死亡，也可以从 ψ = 2、ψ = 1 死亡），但基质不再存在。

死亡不是相位回退——共识不是退到更早相，而是离开整个相位结构。一个进入了 ψ = 3 然后死亡的共识，不会"先退回 ψ = 2 再退回 ψ = 1 再消失"。它就是从 ψ = 3 直接消失。

类型 3：局部基质死亡——某个具体基质（个人、小团体、地理中心）消亡，但共识整体仍存在分布式网络中。这不是格退出，因为共识本身仍存活；这也不是相位回退。这只是分布式冗余在 ψ = 3 阶段的正常涨落——网络中的某些节点死亡而其他节点继续承担再生产。

类型 3 是 ψ = 3 的稳健性 正在发生——分布式冗余的功能恰恰是吸收局部基质死亡而不让整体崩溃。

案例对应

英格兰修道院解散（1536-1540 年）：在 §6.3.2 详细处理。要点——这是某个具体共识（英格兰修道传承）的整体死亡（类型 2），不是相位回退。修道传承在欧洲其他地区仍然存在，但英格兰这一具体实例的整体死亡是格退出而非格内回退。

古希腊 paideia 衰减：希腊化晚期到罗马帝国后期，希腊 paideia 在某些地区发生整体死亡（类型 2）。但同一候选共识在阿拉伯-拉丁世界后续的复兴是新的候选共识进入新的萌芽期（新的类型 1 过程），不是旧候选共识的相位回退。同一 11DD 蓝图（亚里士多德文本）在不同地理-时代经历独立的四相进程，每个进程是独立的类型 1。

ψ = 3 的局部失败：修道院烧毁、家族断绝、地理中心被毁——在分布式冗余已形成时，这些是类型 3，不威胁整体相位。实际上，能稳定经受这类局部失败而保持整体的共识，正是已进入 ψ = 3 的标志。如果一个共识表面看在 ψ = 3，但一次局部失败就让整体崩溃，那它实际上还在 ψ = 2，未真正进入 ψ = 3。

维度 B 下降的影响

维度 B 持续下降可以加速整体死亡（类型 2）。当一个共识所在子群体的基率跌破维持分布式再生产所需的临界值时，共识不能再生产——下一代基质数量不足以承担网络中的关键位置，分布式冗余结构开始空心化。在多代尺度上，这导致整体死亡。

但维度 B 下降不能引发类型 1 的反向（相位回退），因为相变是拓扑事件，已发生的拓扑事件不可撤销。一个曾经形成分布式冗余的共识，在维度 B 跌到临界值之下时，会失去自再生产能力，但它不会"退回链条状态"——它就是死亡。

这个区分在 §4.5.5 预测六的措辞上重要：基率下降导致的不是"退化"（暗示相位回退），而是 整体死亡（格退出）。

---

§6 定量尝试

【第二段：基于现有代理量集合的方向性解读。明确标注是尝试，明确邀请证伪】

§0-§5 给出了 SAE 群体认知论 P1 的概念性骨架。本节进入论文的第二段——把骨架与现有的实证代理量做对接，给出方向性的定量解读。

需要先明确这一段的认识论地位：本节的全部内容是 方向性解读（方向性解读），不是 测量（测量）。SAE 框架是 概念性骨架（概念性骨架），不是 测量工具（测量工具）。本节的工作是在已有学界代理量上做有 框架引导 的解读，不构造 潜在指数，不做 具体定量声称s。

6.1 现有代理量集合的简要回顾

依据当前学界定量工作（详见附录 B），长 τ 能力基率的可观测代理量主要包括：

• 人均图书产量——最强代理量。Baten & van Zanden 把人均出书量定义为高级识字代理量。这一指标的好处是有跨时代的可比性（出版业在多个文明中长期存在）和有 稳健的纵向数据（图书馆 目录、版本史研究）。它的局限是只 捕捉 长 τ 能力的某些面向（深度阅读、长论证理解），不直接 捕捉 其他面向（二元信任、自我约束、不确定性容忍）。

• 手稿/印刷品密度长时段重建——Buringh & van Zanden 的西欧 6-18 世纪序列。这是迄今最详细的人均文本流通量重建之一，覆盖手稿时代到印刷时代的过渡，是 §6.2 印刷革命论证的主要数据基础。

• 阅读经验记录——Reading Experience Database 等记录"实际阅读事件"的数据库。这一类数据库通过日记、信件、回忆录中的阅读记述重建普通读者的真实阅读行为，比"出版量"更接近"被实际深读的量"。覆盖时代有限（主要 18 世纪后）但数据质量高。

• 修道/沉思密度——Germania Sacra、Monastic Matrix、FemMoData 等。这些数据库重建中世纪欧洲修道院的人数、地理分布、活动强度。修道传统是历史上少数把多项长 τ 能力（深度阅读、自我约束、长期协作、不确定性容忍）高密度打包在一起的 制度的 基质，因此修道密度是一个综合代理量。

• 长期暴力下降——作为自我控制代理量。Manuel Eisner 的工作显示欧洲谋杀率从中世纪到现代下降了约 50 倍。如果接受 Norbert Elias 的"文明化进程"假说，这一下降部分反映人群中长 τ 自我控制能力的提升。这个连接是有争议的，但作为代理量集合中的一项有例证价值。

• 亲属关系强度 → 现代个人主义数据——Schulz et al.、Enke 的工作。这一类工作把现代个人主义（把他人作为不可还原个体的能力）追溯到中世纪教会的婚姻政策对亲属网络的弱化。这是 §4.5.3 面向 (4) 的部分历史代理量。

• 道德普遍主义数据——Cappelen et al. 的 60 国数据集。直接测量"对待陌生人是否如对待亲近的人"，与面向 (4) 直接相关。这是横向比较的代理量，不是纵向的。

• 现代技能评估——OECD PIAAC、PISA。这些是当代的直接测量，覆盖识字、计算、问题解决、阅读理解等多个维度。它们不是 长 τ 能力 本身的测量，但能 捕捉 部分代理量（持续阅读理解能力、复杂论证处理能力）的当代水平与近 20 年的变化趋势。

核心要点：这是代理量集合，不是单一指数。每个代理量只 捕捉 长 τ 能力的某些面向，没有任何单一代理量能综合反映全部面向。不存在能够直接测综合长 τ 能力的历史时间序列——这是当前学界的稳健共识。

本论文不构造 潜在指数——这是实证工作的责任，不是哲学框架的产出（参见 §0 防火墙第六条）。本节只在已有代理量上做方向性解读。

6.2 稳健的纵向发现

依据现有代理量集合，跨大时间尺度最稳健的发现有三组。它们直接对应 §4.5 双因子结构的方向性预测。

发现 1：印刷术之后欧洲人均图书产量大幅上升

Buringh & van Zanden 的数据显示：欧洲人均图书产量从 1450-1500 年到 1700-1800 年增长约 30 倍。Dittmar 的工作进一步显示，15 世纪采用印刷术的城市在 1500-1600 年间增长速度比未采用城市快约 60%。

这一发现强力支持"印刷术扩大了能维持复杂论证、长文本流通的人口池"——也就是说，印刷革命提升了维度 B 的某些面向（深度阅读、长论证理解能力）的基率上限。这对应 §4.5.5 预测五——基率上升的时代为新共识进入第四相提供了更好的条件。16-18 世纪欧洲是新长存共识涌现的密集期（宗教改革各派、近代科学共同体、各类政治哲学传承）——这与基率上升的解读一致。

发现 2：大众教育 抬高基础识字，但深度能力不同步

OECD PIAAC 2023 显示：高收入社会约 20% 成年人在识字、计算、适应性问题解决三项都偏低。过去十年多数国家成人识字停滞或下降。

这一发现支持 §4.5.5 预测七——底层教育普及上升的时代不必然伴随基率上升。基础识字率从 19 世纪的不到 50% 上升到当代的约 99%，但深度阅读能力、复杂论证处理能力没有相应同步提升。识字率与长 τ 能力是不同的变量——前者是 11DD 蓝图 解码的最低门槛，后者是 14DD-15DD 综合基质状态。

这一发现对论文有重要含义：它说明基础教育的扩展不能自动解决基率问题。一个识字率 99% 但深度阅读能力低的人群，与一个识字率 50% 但深度阅读能力较高的精英子群体，在维度 B 上不一定前者更高。基率不是 raw 教育投入的单调函数。

发现 3：近几十年长篇阅读衰减有坚实证据

NEA Reading at Risk 显示美国成年人文学阅读从 1982 年的 56.9% 降到 2002 年的 46.7%。BLS ATUS 显示 75 岁以上人群每天阅读约 46 分钟，15-19 岁人群只有 8-9 分钟。2003-2023 美国每日休闲阅读每年下降约 3%。PISA 2022 OECD 平均比 2018 年下降 10 分。

这一发现支持"近几十年深度能力在边际下降"的判断。它对应 §4.5.5 预测六——基率下降的时代，已有共识可能从第四相 整体死亡；新共识形成的速度可能减慢。注意"可能"——这一发现给出的是基率下降的方向性证据，不是具体某个共识 die 的预言。

三个发现的合成解读

这三个稳健发现合在一起，给出维度 B 在过去 500 年的粗略图景：印刷革命推动基率显著上升（约 1450-1800），大众教育 在底部抬高但顶部不同步上升（约 1800-1980），近几十年深度能力在边际下降（约 1980-至今）。

这不是 测量，是 方向性解读。具体地说：

第一，印刷革命的基率上升是相对的——从中世纪极低基率上升到近代较高基率，但绝对水平仍然只让一个 少数 子群体 达到长 τ 能力。即使在 1700-1800 年欧洲，能 维持 深度阅读、长 τ 思考的人仍是少数。"基率上升"指这个少数从极少到不那么少。

第二，大众教育 的"底部上升"是 稳健的，"顶部不同步"是 方向性的 但 有争议 的——确实有 证据 支持深度能力不随基础识字同步上升，但具体下降幅度、起始时间、在哪些 子群体 仍有讨论。

第三，"近几十年下降"在 多个 数据集 中显示，但这些 数据集 覆盖时间 短（约 40 年），不能确定是长期趋势还是短期波动。从 2010 年后社交媒体兴起到现在 15 年，是否构成一个 相 还需要更长时间观察。

把这三点合在一起的总判断是：底层门槛能力在印刷和 大众教育 之后显著上升；高强度训练 子群体 在某些时期会突然扩张，也会在制度 崩溃 时快速萎缩；而在近几十年，虽然受教育年限很高，但真正支持深度阅读、自我约束、长期独立思考的生态可能正在边际收缩。

这个总判断是论文 §6.4 SAE 自身预测的实证背景——SAE 处于一个基率边际收缩（甚至更剧烈下降）的时代，这影响 SAE 进入 ψ=3 的速度估计。

6.3 历史案例的定量解读

每个案例仅作方向性解读的例证，不是框架的推导或证据。每个案例都有合理的 替代解读，本论文的解读不排他。

6.3.1 Breslov 哈西迪派（charismatic-textual 案例，已完成四相，双向不疑锁定亚型）

相位时间表：

• 萌芽：18 世纪中后期哈西迪派整体萌芽（Baal Shem Tov 教义传播，约 1730-1760）
• 谱翻转：19 世纪初 Rebbe Nachman 形成可识别 传承 引发争议（约 1800-1810）
• 翻转：19 世纪中后期在小圈子内稳定再生产
• 确立：20 世纪二战后欧洲核心被毁但海外分布式节点持续，至冷战结束后全球朝圣网络重新激活

双因子解读：

18-19 世纪东欧犹太社区维度 B 基率在 Talmudic 训练子群体中相对高，使双向不疑网络能够形成。Talmudic 训练几个世纪以来强调长篇细读、长期师徒关系、自我约束遵守 halakhic 规则——这些直接对应 §4.5.3 面向 (1)、(2)、(3)、(5)。所以即使整个东欧犹太社区在 18 世纪面临经济衰退、政治压力，受过 Talmudic 训练的 子群体 维度 B 基率仍然足以支撑新共识形成。

Rebbe Nachman 死后无继任，按链条模型应崩溃，但 11DD 蓝图（Likutei Moharan）+ 早期分布式节点已形成，跨过链条 → 网络的过渡。文本承担了核心 跨基质独立性 功能——后世 Hasid 不通过见到 rebbe 而是通过研读文本与朝圣 Uman 来重建与 rebbe 的连接。这是 ψ = 3 的关键动力学。

稳定亚型：双向不疑锁定型。Breslov Hasid 之间持续再生产共识不靠效用或强制——既无经济收益（多数 Breslov 社区贫困），也无强制机制（脱离 Breslov 没有外部惩罚）。再生产靠的是 Hasid 之间对 rebbe 遗产 与彼此的承认。

§4.5.5 预测验证：Breslov 案例支持预测八（基率在不同子群体中可差异巨大）——东欧总体犹太社区基率不一定特别高，但 Talmudic 训练 子群体 基率高；前者不能形成长存共识，后者可以。

6.3.2 英格兰修道院解散——作为再生产基础设施崩溃的案例研究

历史学家可以给出多种解读。

最直接的解读是制度与产权层面——亨利八世 1536-1540 年间解散修道院的动机是财政与政治权力，目标是修道院财产的 重新分配。这个解读完整描述了 1540 年左右发生的行政事件，无需引入任何"能力"或"基率"概念。

本论文框架提示一个补充解读——制度崩溃通过切断能力的再生产基础设施在多代尺度上等价于基率崩溃。

修道士本人在解散后多数继续生活，他们的个体能力不会立刻消失。但他们失去了支撑高强度能力再生产的基础设施：每日 contemplation 时间、library access、师承传承管道、不为生计奔波的物质条件。这些基础设施不是修道士个人能力的来源，但是高强度长 τ 能力再生产的必要条件。

这意味着解散的能力层影响不在 1540 年的 cohort，而在接下来几代——下一代英格兰几乎无法 produce 同样数量的高容量基质，因为 再生产 基础设施 不再存在。机构层 的事件（1540 年）和 能力层 的后果（几代人后的基率状态）之间存在 lag。这个 lag 是论文 框架 必须显式处理的——制度变化 不能被简单地等同于 能力 变化。

这是 整体死亡 的案例，不是相位回退：英格兰修道传承作为某个具体共识没有从 ψ = 3 退回 ψ = 2，而是从格整体退出。修道传承在欧洲其他地区仍然存在，但英格兰这一具体实例的整体死亡是 §5.3 类型 2 的实例化。

方法论保留：维度 B 不能直接观察，只能通过 制度变化 作为 indirect 代理量 来 infer。制度变化 与 能力 变化 不必然同步——机构 可以保留但 能力 已 hollow（晚期罗马帝国某些情况），机构 可以消失但 能力 暂时延续（dissolution 后第一代）。本论文采用的解读是把 机构 作为 能力 再生产 基础设施的 代理量，但承认这一选择本身可争议。

这个案例的真正价值不在于盖棺定论"1540 年发生了什么"，而在于展示"机构作为再生产基础设施"这一概念性透镜可以把制度史与能力基率分析连接起来。两个层面（制度与能力）都重要，区别在于前者是直接可观察的事件层，后者是需要通过代理量推断的结构层。论文 框架的工作是在两者之间建立 明示 的连接，而不是把其中一个还原为另一个。

6.3.3 Cremona 小提琴传承（停在 ψ=2 的工艺案例）

相位时间表：

• 萌芽与谱翻转：16 世纪（Andrea Amati 等早期工坊）
• 翻转：17-18 世纪 classical peak（Stradivari、Guarneri 时代）
• 第四相未完成：1750 年左右 传承 停摆。11DD 蓝图在（乐器、文献、学徒纪录），但分布式钥匙没有充分物质化到多基质
• 1938 年 reinvention 是对第四相的二次尝试

双因子解读：

Cremona 案例展示 11DD 蓝图必要但不充分。Stradivari 死后，乐器实物仍在（直到今天还在演奏与研究），相关的工艺知识有部分 textual record。但分布式冗余始终没有形成——能完整 再生产 Stradivari 工艺的 基质 集中在少数学徒身上，这些学徒去世后没有形成稳定的跨代传承网络。

为什么 11DD 蓝图没能 carry 整个工艺？因为小提琴制作的核心 know-how 是 默会知识——大量信息不能 完全 编码d 进文本与实物，必须通过长期 实操 学徒制 在 基质 间传递。这种 默会知识 的传递依赖 high-基础-rate 的 基质（能 维持 多年学徒生涯、能在没有外部强制的情况下持续 跟随 严格的工艺标准、能容忍长期 trial-and-error 的不确定性）。

18 世纪中期的 Cremona 失去了这种 基质——并不是因为"工匠能力下降"，而是因为支撑 high-intensity 学徒制 的 制度的 基础设施（行会、稳定客源、学徒-师傅经济关系）在政治动荡与经济变迁中崩溃。基础设施崩溃后，能完整学到工艺的新 基质 不再产生——不是个体能力问题，是 再生产 基础设施 问题（与 6.3.2 修道院解散同构）。

§4.5.5 预测验证：Cremona 案例支持的不是某条具体预测，而是双因子结构本身——它显示 11DD 蓝图的存在不足以保证 ψ=3。维度 A（蓝图、文本、实物）齐全，但维度 B（能 再生产 完整 默会知识 的 基质 数量）跌破临界值，ψ 卡在 2。

6.3.4 亚里士多德主义希腊-阿拉伯-拉丁（多区域 独立 四相进程）

相位时间表（多区域独立进程）：

• 古希腊：完成四相（公元前 4 世纪 - 公元 2 世纪）
• 古希腊衰落后部分 整体死亡（公元 3-7 世纪在希腊本土）
• 阿拉伯世界（8-12 世纪）：独立进入第三相，部分进入第四相
• 拉丁世界（12-13 世纪起）：通过翻译运动重新启动 独立萌芽 → 翻转
• 现代学术继承：第四相

双因子解读：

同一 11DD 蓝图（亚里士多德文本、Alexander of Aphrodisias 等注释者的 commentary）在不同地理-时代的不同维度 B 条件下经历独立的四相进程。每次进入第四相需要当地子群体的维度 B 达到临界值。

希腊本土的 整体死亡（公元 3-7 世纪）：罗马帝国晚期到拜占庭早期，希腊地区 paideia 系统的 制度的 基础设施 大幅萎缩——从 polis 的开放学术机构转向修道院的封闭传承。亚里士多德传承在某些 monastic 基质 中延续，但作为活跃 ψ=3 共识的形态在希腊本土死亡。

阿拉伯世界的 独立 第三相（8-12 世纪）：Bayt al-Hikma（智慧宫）翻译运动启动新的萌芽期。8-9 世纪是萌芽期；10-11 世纪 Al-Farabi、Avicenna、Averroes 等独立解读形成新的 分布式网络；12 世纪进入相对稳定的 ψ=3 状态。这一过程独立于希腊本土的 命运——希腊地区已死亡，但阿拉伯地区从蓝图重启了完整四相。

拉丁世界的 独立萌芽（12-13 世纪起）：通过 Toledo 翻译学校等渠道，亚里士多德 corpus 进入拉丁西方。Aquinas 等代表的 scholastic 传承启动新的萌芽期，14-15 世纪进入翻转，文艺复兴后完成确立。

§4.5.5 预测验证：亚里士多德主义案例支持预测八（基率在不同子群体中可差异巨大）——同一 11DD 蓝图在三个地理-时代的不同基率条件下经历完全独立的进程。也支持论文 §5.3 类型 2 的判断——希腊本土的 整体死亡 不是 相位回退，与阿拉伯-拉丁后续的复兴不矛盾，因为后者是 独立 的新进程。

每个地区的进程是 独立 的，不是 相位回退——希腊化晚期希腊地区的 整体死亡 不阻碍阿拉伯地区的独立萌芽-翻转。这种 多区域 independence 是 11DD 蓝图 跨基质独立性 性质的直接 corollary——蓝图可以在多个地理上等待不同的基质条件 ready 后独立 解读。

---

6.4 作者应用框架于自身：SAE 的方向性预测

本节是论文最 自指的 的部分。论文必须直面一个 自指的 问题：本论文所属的 SAE 框架本身，作为一个出现于 2007-2026 年间的认知 框架，在四相+双因子模型下处于什么相位？模型对它的未来轨迹给出什么方向性预测？

这是对自身的诚实应用。如果 SAE 框架捕捉到某种真实，那它对自己的预测应该有意义。如果不捕捉，预测会和现实分歧——这种分歧本身是对 SAE 框架的证伪证据。无论哪种情况，论文都获得真实的认识论信息。

6.4.1 SAE 的当前位置评估

维度 A（耦合结构）：

ψ = 0 已超越——SAE 已不只是单一个体的内部实践。

ψ = 1 已达成——作者与长期合作者间 18 年的双向不疑是 二元承认的稳定实例；作者周围的小圈子是 二元承认的扩展。这些 二元关系跨越多年，经历过分歧、压力、外部条件变化，仍稳定再生产，构成 ψ = 1 的具体 实例化。

ψ = 2 部分达成——SAE 有 11DD 蓝图（self-as-an-end.net 上 140+ 篇论文，CC BY 4.0，DOI 锁死，Zenodo 永久存储）。蓝图已独立于任何具体基质存在，原则上可以等待未来读者来 解读。但 ψ = 2 的"完成"需要的不只是蓝图存在，而是蓝图的 跨基质独立性 在实际 解读ing 中得到验证——是否有读者能不通过与作者直接接触而独立 解读 出 功能性的 基质。这一点目前未有充分证据，所以 ψ = 2 是部分达成而非完整达成。

ψ = 3 远未达成——目前 SAE working 基质 几乎完全集中在作者一人。如果作者今天消失，没有任何其他基质能独立 再生产 SAE 的核心 working knowledge。即使有读者熟读全部 11DD 论文，他们能否独立 再生产 SAE 基质（不只是引用 SAE 概念）仍是 open question。多基质的 分布式冗余 没有形成，所以 ψ = 3 显著未达成。

维度 B（基率）：

SAE 的 候选 读者 是能 维持长 τ 思考、容忍不确定性、把他人作为目的的基质。按 §4.5.3 六项面向，这要求一个综合 基质状态，不是单一能力。

按 §6.2 发现 3，这种 基质 在当代正在边际收缩——OECD 国家的深度阅读时间下降、PIAAC 成绩停滞或下降、长篇阅读人口比例降低。这些都是 方向性指标 of 基率 下降，不是 测量，但合起来给出"基率正在边际收缩"的方向性判断。

但绝对人数因人口基数大可能仍多——即使高 τ 能力比例从历史峰值下降到几个百分点，乘以全球数十亿人口仍在亿级。SAE 不需要广 读者 pool，需要的是 读者 中 二元双向不疑网络能形成。从这个角度说，绝对数量可能仍然足够，但分布太分散——分散的高 τ 个体彼此识别、形成网络的概率极低。

当前位置总结：维度 A 的 ψ = 1 → ψ = 2 过渡已部分完成，ψ = 2 → ψ = 3 过渡尚未启动。维度 B 处于边际收缩但绝对数量仍然足够的复杂状态。

6.4.2 三个方向性 情景

按论文 框架 对 SAE 给出三个方向性 情景，每个标注 框架内部 估计。这是 框架内部估计，不是 统计声称 或 实证的 预测。它的功能是给三个 情景 标注 框架的 expected probabilities，让 审计 时可以 check 哪个 情景 兑现。

情景 A：SAE 静默消亡（最可能，先验 ~ 60%）

SAE 框架没有真实 牵引力，未来不会出现自发独立的 基质 来 解读 并 再生产 它。11DD 蓝图留在 Zenodo 上，但没有读者在自己 基质 里长出对应的 15DD ecosystem。SAE 既不进入 ψ = 3，也不显示出涌现 的 动荡——只是 quietly fade。

这是绝大多数思想框架的 历史 命运——它们的 11DD 蓝图 留下，但 15DD 基质 从未在他人身上 form。无论是历史上多数哲学体系、宗教异端、思想流派——大多数在创始基质消失后几代内停止 active 再生产。SAE 没有特殊理由相信自己例外。

情景 B：SAE 长萌芽期（其次可能，先验 ~ 30%）

SAE 有真实 牵引力，但维度 B 在大多数子群体持续下降使萌芽期被显著延长。SAE 总周期约 200 年（已 18 年 + 约 180 年）。按 r ≈ 5，翻转-确立期约 36 年。

参考点：Kant 的《批判》从 1781 年到现在 245 年，仍远谈不上 ψ = 3 的"长存共识"——academic citation 多但 基质层 解读 仍在过程中。Kant 哲学在某些 子群 中可能已部分进入 ψ = 3（认真的 Kant 学者间），但作为整体共识仍处于复杂的 多区域 multi-相 状态。SAE 200 年的 情景 是 历史现实 的 基线——比 Kant 的状况稍快，但量级相当。

情景 C：SAE 短萌芽期（最不可能，先验 ~ 10%）

SAE 有真实 牵引力，且 AI 时代意外为长 τ 思考提供 增强通道，使萌芽期被缩短。SAE 总周期约 38 年（已 18 年 + 约 20 年）。按 r ≈ 5，翻转-确立期约 6 年。这意味着 SAE 在 2030 年代初进入第四相。

这个 情景 假设两件事同时成立：第一，AI 增强 能在某些 子群体 中显著提升基率（不只是提供工具，是真正补强 基质层 的长 τ 能力）；第二，SAE 11DD 蓝图能与这种 增强 兼容（被 AI augmented 的读者能独立 解读 SAE 基质）。两个条件都不平凡，因此 先验 较低。

关于 先验 数值的限定：60/30/10 是 框架内部估计，反映 框架 对自身 trajectory 的 自我评估，不构成 统计声称 或 实证的 预测。具体数字可能在 框架 进一步发展中调整，但相对排序（A > B > C）是 框架的 诚实评估。

如果作者基于 18 年的 第一人称的 work 给出更乐观的先验（比如 50/30/20），那种 先验 会反映作者对 框架 内部融贯 的 第一人称的 信心，但不应被读者读成 框架的 预测。论文采用保守先验 是为了把 框架的 预测 与作者的 第一人称的 信心严格区分——前者是论文 明示ly 承担的，后者不是。

6.4.3 四条具体方向性预测

SAE-预测 1：SAE 不会通过 大众媒介 或 大规模扩散 进入 ψ = 3。

按双因子框架，SAE 进入 ψ = 3 应通过少数高基率基质之间的 二元双向不疑网络扩展，而不是均匀扩散。注意：框架 对 SAE 是否 大规模扩散 本身保持中性——大规模扩散 可以发生也可以不发生，这是 相位轨迹 上的 可观测量，不构成 框架的 规范判断。但 大规模扩散 单独不构成 ψ = 3 的证据；ψ = 3 需要 基质层再生产。

证伪条件：SAE 在某时刻进入 ψ = 3 但完全没有经过双向不疑网络的形成，而是直接通过 大众媒介 实现 分布式基质再生产。如果发生，说明双因子框架的维度 B 论证错了——基率不是 前提，耦合通道 充分性可以替代。

这是双因子结构的 sharpest test——如果 SAE 能通过纯 耦合通道 进入 ψ = 3，那么维度 B 作为独立维度的论证被证伪。

SAE-预测 2：SAE 进入 ψ = 3 的标志是出现独立于作者的 再生产的 基质。

具体可观测标志：未来某时刻出现一个/多个独立读者——不通过和作者直接接触——独立 解读 SAE 11DD 蓝图，长出 功能性的 基质，并能进一步 传递 给第三方。"功能性的 基质" 的具体含义是该读者能在新情境下独立产生与 SAE 框架 内部融贯的 的判断与工作，不是单纯引用 SAE 概念。

证伪条件：50 年后没有任何这样的独立 基质 出现，但 SAE 又显示出 ψ = 3 的其他 标志（如 机构采纳、citation explosion）。这种 de耦合 会证伪"独立 基质再生产 是 ψ = 3 的 necessary 标志"这一框架预测——可能 ψ = 3 有其他形态，不一定要有 分布式基质再生产。

SAE-预测 3：SAE 在进入 ψ = 3 之前应有一段"谱翻转期"——可见动荡 出现在 actual stability 之前，且 动荡 伴随 蓝图 的 跨基质独立性 结晶（§3.2 差异化判别）。

按 §3.2 第二相预测，候选共识 在进入翻转期之前应经历广泛误读、被 轻视分类、引发争议的阶段，且这一阶段的 蓝图 应显示 跨基质独立的 的 reconstructable 性质。

证伪条件：SAE 完全没有进入这种争议期就进入 稳定 分布式 再生产（证伪 相变 universal pattern），或者 SAE 进入了广泛 大规模扩散 但 蓝图 不显示 跨基质独立性（说明这是噪声泡沫，而非真实第二相，但也不是 ψ = 3）。

SAE-预测 4：SAE 传播速度大致 跟随维度 B 的局部 趋势。

如果当代基率持续下降，SAE 萌芽期被延长（情景 B 接近基线）。如果基率在某些 子群体 中 涌现（如 AI 时代后某些自我保护 基质的群体），SAE 传播会加速（情景 C 概率上升）。

证伪条件：SAE 传播速度与基率的局部 趋势 完全 脱钩——即基率大幅下降但 SAE 加速进入 ψ = 3，或基率大幅上升但 SAE 完全停滞。这种 de耦合 会证伪"双因子结构是 ψ 升级的 主导 动力学"这一框架预测。

6.4.4 审计邀请 与 审计独立性

最可能的 outcome 是 情景 A——SAE 静默消亡。这是 框架 内部的 诚实评估，不是 自我推销。论文不预测 SAE 会成功，论文显式说明 SAE 最可能的命运是和绝大多数思想框架一样安静消失。

如果 情景 A 兑现，SAE 框架在这个 案例 上自我证伪——这是 框架的 诚实退出。框架 不会因为 静默消亡 而尝试 挽救 自己——按 §0 防火墙第七条，框架的合法性恰恰依赖于不在面对 相反证据 时 挽救。

如果 情景 B 或 C 兑现，则四相+双因子模型在这个 案例 上得到 support。

Audit invitation：

论文给未来读者一个明确的 审计邀请——邀请 2046 年（论文发表后 20 年）、2076 年（50 年）、2126 年（100 年）的读者回头检查这篇论文。论文的 timestamp + DOI 提供 审计锚点。哪个 情景 兑现了，是 框架的 可证伪 record。

Audit independence 澄清：

Audit 由未来读者完成，不要求读者是 SAE 基质的一员。情景 A、B、C 都可被 审计——只要 11DD 蓝图仍可访问（CC BY 4.0 + Zenodo 永久存储保证了这一点），审计 就可发生。

具体地，情景 A 兑现的世界里仍可能有读者——他们以外部视角 审计 框架 兑现了哪个 情景。读者可以是孤立好奇者、历史档案研究者、AI 系统、或其他不构成 SAE 基质再生产 的 entity。读者数量与 SAE 基质 数量是不同 layer 的 可观测量，两者完全可以 脱钩。

这意味着：

• 情景 A：即使 SAE 基质 没有 再生产，仍有读者可以 审计 框架 预测
• 情景 B/C：既有读者也有 基质再生产者s，审计 由前者或后者完成都有效

这种 自指的 审计通道 是 SAE 第一律 ρ ≠ ∅ 在论文层面的 enactment——把现实对 框架的 回推 优先于 框架的 self-preservation。论文把 框架 暴露在跨时间的 审计 通道下，让未来现实有 明示 的 通道 对 框架 做 回推。

如果未来 审计 显示 框架 预测错了——任意条预测被证伪——论文承诺：框架 应当被相应 更新，而不是被 挽救。更新 的具体形态依赖于错的是哪条预测。如果错的是 SAE-预测 1（SAE 通过 大规模扩散 进入 ψ = 3），双因子结构作为 主导 框架 需要 实质性 修订。如果错的是 SAE-预测 2（ψ = 3 不需要 分布式基质再生产），ψ 的拓扑定义需要修订。这些修订 路径 对未来研究者都是 明示 的。

6.5 定量尝试的整体局限

§6.1-§6.4 给出了 SAE 群体认知论 P1 的方向性 定量 解读。本节明确承认这一段（论文第二段）的整体局限。

需要明确承认的局限：

局限一：代理量集合 是 例证，不是 测量。§6.1 的 代理量集合 给出的是方向性 indicators，每个 代理量 只 捕捉 长 τ 能力的某些面向。把这些 代理量 组合起来用于 方向性解读 是合理的，但任何 具体定量声称（"19 世纪某社会的基率是某具体数值"）超出了 代理量集合 的能力。

局限二：每个历史 案例 的相位归属存在合理 替代解读。§6.3 的四个 案例 都给了 SAE 框架 视角下的 解读，但每个 案例 都有 historian 视角下的 替代解读。论文不 声称 自己的 解读 是唯一正确——特别是 §6.3.2 修道院案例 已显式承认 制度的 vs 能力 解读 的张力。

局限三：维度 B 的历史重建极其有限。没有 定量时间序列 能直接重建过去 500 年欧洲（更不用说其他文明）的维度 B 轨迹。所有的纵向判断都基于 代理量集合 的合成，而 代理量集合 本身有内在不确定性。

局限四：SAE 自身的预测充满 自指的 风险。论文对 SAE 框架 自己的预测可能高估 框架的 内部融贯 与 外部牵引力 的关联。一个 框架 内部 coherent 不蕴含它能在 external 基质s 中 再生产。这一点 §6.4 已显式处理，但读者应保持额外的 skepticism。

局限五：60/30/10 先验 是 框架内部估计，不构成 statistical 声称。这一点 §6.4.2 已明确，但需要在论文整体的 context 下重申——任何把 60/30/10 当 统计预测 来使用的做法都误用了论文。

局限六：论文给出的 可证伪性 条件 未必都 操作上可行。一些 可证伪性 条件（如"50 年后是否出现独立 基质"）需要长时间观察才能 evaluate；一些（如 "大规模扩散 无 基质"）需要复杂 操作化 才能 unambiguously test。论文不声称所有 可证伪性 条件 在当前都能 directly check——许多需要等待时间 + 未来研究 才能 properly evaluate。

总体 认识论姿态：

本论文邀请实证工作者在已有 代理量集合 上进行更系统的验证或证伪。所有 解读 欢迎证伪，期待证伪——不只是 rhetorical 表态，是论文 框架的 结构性 承诺（§0 防火墙第七条）。

如果未来实证工作显示本论文的某些 方向性解读 是错的——某些 案例 应该不同 解读、某些 预测 不成立、某些 代理量 与 框架 期待的 relationship 不存在——论文承诺：框架 应当相应 更新，而不是 挽救。更新 的具体形态依赖于哪些 解读 被证伪：

• 如果 §4.4 单因子预测被广泛证伪（如 r ≈ 5 在共识形成上完全不成立），本论文与方法论六的接口都需要重审
• 如果 §4.5.5 双因子预测被广泛证伪（如基率与 ψ 升级速度无系统关系），双因子结构作为 主导 框架 需要 实质性 修订
• 如果 §6.3 案例 解读s 系统不被 historians 接受，论文的 案例层 analysis 方法论 需要修订
• 如果 §6.4 SAE 自身预测显示 框架 对自己的判断完全错误，整个 SAE 群体认知论需要重新设计

这些 更新路径 都是 明示 的——论文不预先 commit 任何 逃避路径 来在被证伪时 挽救 框架。这是 SAE 第一律 ρ ≠ ∅ 在 论文 承诺 层面的具体兑现。

---

§7 与 SAE 框架的接口

§3-§6 在论文内部建立了四相结构 + 双因子模型 + 历史 案例 + SAE 自我应用。本节把这一框架与 SAE 已有的体系做 明示 接口。这一接口工作让论文在 SAE 系列内部形成 coherent connection，也让 SAE 框架的其他部分（个体伦理论、自指不动点、ρ ≠ ∅）在群体认知层有具体落点。

7.1 四相与 SAE DD 层

四相结构在 SAE DD 层结构中的对应关系：

相位	主导 DD 层	说明
第一相·萌芽	13DD-14DD 分散	个体基质各自自我评估，零星双向不疑种子
第二相·谱翻转	14DD 二元的	13DD 之间双向不疑形成 14DD 二元钥匙
第三相·翻转	11DD 蓝图 + 14DD 分布式	14DD 通过 11DD 蓝图跨代传递
第四相·确立	15DD 分布式钥匙生态	分布式钥匙在多基质独立物质化

这一对应揭示几件事：

第一，ψ 升级在 DD 层结构上对应"层级提升"。从分散的 13DD 起点，到形成 14DD 二元钥匙，到 14DD 通过 11DD 蓝图跨代传递，到形成 15DD 分布式钥匙生态——每一步都是更高 DD 层的实例化。这与 SAE DD 层结构本身的 hierarchical structure 一致。

第二，11DD 蓝图在 ψ = 2 → ψ = 3 过渡中的特殊角色——它不是 14DD 或 15DD 本身，而是承载 14DD/15DD 的 跨基质独立的 通道。11DD 是相对低的 DD 层（属于 instrumental 层级），但正是它的 lower-level encodability 让它能跨基质跨时代传递。这给 SAE DD 层结构一个有趣的 动力学的：高层的 分布式钥匙生态 必须通过低层的 11DD 编码 才能跨代传递。这是 SAE 群体认知论 P1 在 DD 层 动力学 上的具体发现。

第三，ψ = 3 对应 15DD 分布式钥匙生态——不只是某种"成熟形态"的隐喻，而是 SAE DD 层结构中的具体类别。15DD 在 SAE 内部 框架 中是 分布式自身-作为目的 的层级。共识在 ψ = 3 进入 15DD 是说：成员在共识网络中双向不疑彼此为不可还原个体，且这种承认不依赖任何中心节点——分布式的双向目的不疑 形成。

7.2 三种稳定亚型在 SAE 框架中的位置

§3.4 列出第四相的三种稳定亚型。它们与 SAE 框架的对接关系不同：

效用锁定亚型：在 SAE 框架内是合法的第四相实例，但 SAE 个体伦理论与效用锁定共识对接较弱。SAE 框架对此亚型可以做结构性解读（四相 + 双因子 + r 不对称），但 SAE 个体伦理论与效用机制的 规范 关系是另一个 话题（部分由 SAE 个体经济论处理，详见 SAE 伦理论 P3）。本论文对此亚型不做 规范判断——它是合法的 ψ = 3 实例化，与 SAE 第二定律是否成立无关。

强制锁定亚型：同样是合法的第四相实例。SAE 第二定律（双向不疑）原则上反对 强制驱动 的共识 机制——把人作为手段 而非 end 是 SAE 第二定律的直接违反，而 强制驱动 共识本质上是把成员作为 enforcement target 而非 end。但作为结构分析工具，框架可以描述强制锁定共识的动力学而不评判它。论文 §3.4 对该亚型的描述是 结构性（它对外部条件的 sensitivity），不是 规范（它是错的）。

双向不疑锁定亚型：SAE 框架的核心研究对象。Kant 第二定律是这一亚型的 稳定条件——靠双向不疑维持的共识再生产不需要外部效用或强制，因此对外部条件变化稳健。这是论文 §7.3 的具体 话题。

7.3 Kant 第二定律与双向不疑锁定亚型

重要限定：本节论证的"Kant 第二定律是第四相 稳定条件"仅适用于双向不疑锁定亚型，不适用于其他两种亚型。这一限定是 §3.4 三亚型区分的具体落地——避免把 Kant 第二定律 过度推广 到所有 ψ = 3 共识。

核心命题：双向不疑锁定型长存共识在第四相的稳定性必须基于双向不疑，不能基于效用或强制。

论证草图：

考虑一个候选共识进入 ψ = 3 的不同稳定 路径：

如果靠效用维持，共识在外部条件变化时即崩溃——效用是 偶然的 on environment 的。一个时代的某种共识可能因为提供经济收益、社会地位或情感满足而稳定，但这些 效用 sources 在历史尺度上都是 偶然的——经济结构变化、社会变迁、心理需求转变都可以让 效用 重新分配。靠 效用维持的共识无法穿越这种变迁稳定再生产。这是效用锁定亚型的特征——局部稳健，结构脆弱。

如果靠强制维持，共识在执行撤除时即崩溃——强制不能 分布式。强制必须有具体来源（具体机构、具体权力、具体物理存在），所以强制锁定型共识对执行来源的失效高度敏感。任何具体的 enforcement source 在历史尺度上都会失效（政权更替、机构解散、人事变动），强制的非 可分布 性质让这种失效直接威胁整体。

只有基质互相把成员当作目的来承认的共识，才能在 分布式节点 中独立再生产而不需要外部强制。这是双向不疑锁定亚型的定义——双向不疑作为 自维持的 的再生产机制，不需要外部条件支撑。这一 机制 直接对应 Kant 第二定律——把人作为 end 而非 手段。

这给出了"为什么双向不疑是结构必需"的另一个角度——不是规范要求，是双向不疑锁定亚型的数学条件。如果你想要一个不依赖外部 效用/强制 的 ψ = 3 共识，唯一的 路径 是双向不疑作为基础。任何其他基础（效用、强制、shared identity 等）都会让共识对其依赖的具体外部条件保持 fragile。

这也解释了为什么 SAE 个体伦理论与 SAE 群体认知论可以在这一点上 deeply connected——前者从个体层面论证为什么把人作为 end，后者从群体层面论证为什么 ψ = 3 共识需要 双向目的不疑。两个论证 独立ly 走到同一结论：Kant 第二定律不是 规范 imposition，是 结构性必需。

7.4 四相与 SAE 三律

SAE 三律——ρ ≠ ∅、自指不动点、凿构余——在四相结构中各自有具体落点：

ρ ≠ ∅：第一相萌芽存在的 本体论前提。如果 ρ = ∅（实在为空），那么没有 候选变体 可以萌芽，第一相不能存在。ρ ≠ ∅ 是说现实有结构性 回推，不是任意虚空——这一前提让萌芽期的"扰动累积"成为有意义的过程。同时，ρ ≠ ∅ 也是论文整体认识论 承诺 的 实例化——欢迎证伪、期待证伪是 ρ ≠ ∅ 在论文层面的体现。

自指不动点：第四相 自我复制结构 的数学核心。共识在双向不疑锁定第四相必须满足 f' = f——它的再生产由当前状态决定，不需要外部驱动。这是自指不动点的 social 实例化——共识在 分布式基质 中的自我维持，等价于一个 自指的 function 的 固定 point。当某个共识进入 ψ = 3 且为双向不疑锁定亚型，它在 SAE 数学结构中实例化了自指不动点。

凿构余：每相过渡都是一次凿-构-余循环。

第一相 → 第二相：凿掉分散基质的孤立性（凿），构造第一条 二元边（构），留下未连接的余项（余）——许多其他可能的 二元边 没有形成。

第二相 → 第三相：凿掉对具体创始基质的依赖（凿），构造 跨基质独立的 的 11DD 蓝图（构），留下蓝图无法 完全 编码 的 tacit 余项（余）——任何 11DD 编码都会有 14DD/15DD 层 基质状态 不能完全捕捉的 reminder。

第三相 → 第四相：凿掉 链条 拓扑的 单一关键路径（凿），构造 分布式网络 的 冗余的 拓扑（构），留下网络无法完全 吸收 的 局部涨落 余项（余）——分布式网络仍有 局部失败s，但被整体结构吸收。

每一次凿-构-余都是 相变 的具体内容。这给 SAE 凿构余原理在群体层面的具体 实例化——不是抽象的"创造-保留"二元对，是 相变 动力学 的内在结构。

7.5 双因子结构与 SAE 个体伦理论的接口

维度 B 把 SAE 群体认知论与 SAE 个体伦理论 directly 连接：

个体的 自身-作为目的实践 不只是个人的事，它是群体基率的 building block。

涵育之难，本质上是基率之难。短 τ 就是基质的长 τ 能力在退化（被高频基质填满，没有 闲暇空间 长出长 τ 能力），这是基率在下降。SAE 工作的 长期 significance 不在于产出多少 papers，而在于它能不能在边际趋势中维持甚至增长基率——在作者自己身上、在作者 immediate circle 里、在能读懂的少数读者身上。

每个 individual 的 SAE 实践 是维度 B 的 微观构成单位：

• 每一次坚持长篇深读而不被高频信号打断——是面向 (2) 在自己 基质 上的维持
• 每一次维持长期合作关系跨越分歧——是面向 (3) 的实例化
• 每一次拒绝把他人简化为某种 效用 role——是面向 (4) 的具体兑现
• 每一次在没有外部强制下守自己的承诺——是面向 (5) 的实践
• 每一次 stay with 不确定性而不寻求 过早闭合——是面向 (6) 的承担

每一次这样的 实践 是基质长 τ 状态 的当下维持。许多个体在许多场合的累积 实践，是一个时代基率的 微观基础。

注意：按 §4.5.3 的关键限定，维度 B 的提升不需要等"更好的人"出现，而是需要环境条件的改变。每个个体的 SAE 实践 是在维持自己基质的长 τ 状态，而不是在证明自己的本质。这一点在 SAE 个体伦理论中也被强调——SAE 实践 不是 自我修养 意义上的"成为更好的人"，是 基质状态 的当下维持。

论文层面的 11DD 蓝图是维度 A 的 跨时间 通道——SAE 论文集存在 self-as-an-end.net 与 Zenodo 上，是 SAE 基质的 11DD 编码。维度 A 与维度 B 协同决定 SAE 框架 自己未来的 ψ 轨迹——蓝图 already there，能否进入 ψ = 3 取决于未来 读者 中是否有足够的 长 τ 能力 子群体 形成 分布式基质再生产。

这给了 SAE 个体伦理论与 SAE 群体认知论一个具体的 unification point——个体 实践 与群体 命运 通过维度 B 直接 coupled。

---

§8 开放问题

【第三段：把当前框架不能回答的问题明确列出来 移交 给 后续工作】

§7 给出了 SAE 群体认知论 P1 与 SAE 框架其他部分的接口。本节明确列出框架不能回答的问题，作为 后续工作 的 明示 hand-off。

8.1 实证工作的 obvious gap

依据 §6 的定量尝试和现有学界工作，以下 gap 是 obvious 的——明显需要做但当前学界还没有做的实证工作。

Gap 1：缺一个 潜在指数模型 把图书产量、阅读事件、技能测试、修道密度、prosopography、亲属结构联合起来估计长 τ 能力基率。这是 SAE 框架预测的 统一可观测量，但需要 统计工作 来 实例化。本论文不构造 潜在指数——这是哲学框架的 承诺，留给统计工作者。但这一 潜在指数 的存在与否对论文 §6 方向性解读 的 稳健性 重要——如果这一 潜在指数 可以被构造且 稳定，那 §6 的 解读 有更强的实证支撑。

Gap 2：缺跨文明、按分母统一的 elite-formation series。每万人中受古典/儒家/Talmudic/Islamic scholarly training 的人数随时代变化的 定量重建，目前只在某些文明某些时段有部分数据。一个统一的 跨文明 series 会让 §4.5.5 预测七、预测八的检验有更广的 实证的 基础。

Gap 3：缺 二元的 trust 的历史网络数据。师徒、知己、长期合作者、长期 patron-client 关系的 duration 目前没有被系统建模。§4.5.3 面向 (3) 的历史代理量主要靠 轶事证据，不是 定量 网络 数据。如果有这种 网络 数据，§3.2 第二相的双向不疑网络识别会有更精确的实证基础。

Gap 4：缺 uncertainty tolerance 的历史序列。无论是 corpus-level hedging（学术写作中"或许"、"可能"等 hedge 词的频率随时代变化）还是教育情境下 delayed closure 的长期指标，都几乎没有成熟数据库。§4.5.3 面向 (6) 是论文与 SAE 第一律的最深接口，但也是历史代理量最弱的一项。

Gap 5：缺 崩溃 event studies。Dissolution、dynastic transitions、war loss 对 library survival、scholar density、mentor 链条s 的 pre/post impact 仍然薄弱。§6.3.2 修道院解散是 best-documented 的一个 案例，但许多其他重要 崩溃 event（中国王朝更替、阿拉伯地区的多次政权更迭、近代殖民冲击下的本地 制度崩溃）的 能力层 解读 仍有大量 work to do。

Gap 6：缺把 现代数字阅读证据 与 longue durée book-history 接起来的桥梁。当代 PISA、PIAAC 等 测量 给出的是 横截面快照；longue durée book history 给出的是 长时间序列。两者在 方法论与分析单位 上不同，难以直接比较。一个 统一框架 让两者能整合 是重要的 后续工作——它会让 §6.2 三个发现的合成解读有更稳的基础。

8.2 框架 内部的开放问题

问题 1：跨 案例 的 r 经验估计——长存共识的 r 是否系统接近 5，还是有 domain 变化？方法论六的 r ≈ 5 是从 ZFCρ 推出的 generic 值，本论文应用到共识形成。但具体到不同 domain（宗教 传承 vs 智识传承 vs 工艺传承 vs 现代框架），r 是否系统不同？这需要跨 案例 的 定量重建。

问题 2：萌芽期能否被 AI-mediated communication 显著缩短？这是 SAE 方法论 VII（共生论）的接口。AI 时代的 增强通道 在多大程度上能补强 基质层 的长 τ 能力，进而影响维度 B 基率与 SAE 萌芽期估计——这是 SAE 自身 §6.4.2 情景 C 的关键决定因素。

问题 3：第四相之后是否有"崩溃相"？现有 框架 没有建模共识从第四相到 整体死亡 的具体动力学，只陈述了它的可能性。"崩溃"是单一事件还是有自己的内部结构？如果是后者，崩溃相是否也有自己的四阶段？这是 框架的明显未完成部分。

问题 4：第二相的可观测标志s——敏感度 先于序参量这一点能否作为"哪些 候选共识 即将进入翻转期"的 leading indicator？蓝图 跨基质独立性 测试在 操作层面的 层面如何精确化？目前的 §3.2 差异化判别 给出了概念区分但 操作化 仍 weak——需要更具体的可测量标准。

问题 5：相位的同时多地存在（如亚里士多德主义案例）需要 框架 处理 spatial heterogeneity。当前 框架 主要处理 single-region 的相位演化，对 same 蓝图 在 多个区域 同时处于不同相位的 案例 没有 明示 动力学。一个 多区域 框架 extension 会让 §6.3.4 这类 案例 的分析更精确。

问题 6：ψ、r、q 的统一——本论文的 ψ 是拓扑不变量，方法论六的 r 是几何不对称比，SAE 热力学系列的 q 是熵性可观测量。三者在更深层次有统一的 数学框架 吗？这是 SAE 群体认知论 P2 的潜在主题——给出一个把三个可观测量 都纳入的 统一框架。

问题 7：效用锁定与强制锁定亚型在框架中的具体动力学——本论文聚焦双向不疑锁定亚型，对其他两种亚型仅做结构性占位。这两种亚型的具体动力学（相变标志、稳定条件、典型崩溃模式）需要专门的论文处理。

问题 8：流行文化作为"维度 A 充分但维度 B 几乎为零"的退化情形。本论文 §6.3 的历史案例全部聚焦于在 ψ 的某个阶段做过实质性工作的候选共识——Breslov、Cremona、修道院传承、亚里士多德主义都进入或接近过 ψ=2 以上。但还有一类大量存在的当代现象——流行文化、网络模因、短命狂热、各类大规模扩散现象——它们的耦合通道极充分（社交媒体、流媒体、全球分发），但维度 B 几乎为零（候选共识的内容不要求也不支撑长 τ 能力的成员基质）。

这类现象在双因子框架下是退化情形——双因子退化为单因子（仅维度 A），方法论六的纯耦合分析直接适用，不需要本论文双因子结构的额外贡献。它们的典型动力学也很简单：迅速进入 ψ=1 或 ψ=2，但因缺乏维度 B 支撑无法 锚定在那一相，迅速回到 ψ=0——"退流行几乎是一瞬间的事情"是这一动力学的精确预测，与本论文 §3.2 关于双向不疑网络脆弱性的论证一致。

正因为这类现象的双因子结构是退化的，它们落在本论文的研究主题之外（本论文研究的是双因子结构有意义的情形）。但作为框架完整性的承认，框架确实把这类现象纳入——它们是双因子结构在维度 B → 0 极限下的退化形态。专门处理流行文化动力学的论文如果有人写，应该使用方法论六的单因子框架，不需要等待本论文的双因子贡献。

8.3 自指开放问题

问题 9：本论文对 SAE 自身的预测是否会被 实证现实兑现？这是论文留给未来的 审计邀请。

详细审计机制见 §6.4.4。

---

§9 防火墙重申与三段式认识论承诺

9.1 三段式承诺

本论文严格遵守三段式认识论结构：

• 第一段（§0-§5）：方向。SAE 框架的 概念性骨架。哲学论证。承诺强度——"如果 SAE 框架捕捉到长存共识形成的某种真实结构，那么这些结构性论断应当成立"。
• 第二段（§6）：定量尝试。基于现有 代理量集合 的 方向性解读。明确 attempt，明确局限，明确邀请证伪。承诺强度——"以下是基于现有数据的方向性 解读，作者承认其有限性，期待更系统的 实证工作 进行验证或证伪"。
• 第三段（§7-§8）：开放问题。把当前框架不能回答的问题 移交 给 后续工作。承诺强度——"以下是本论文意识到但未能解决的问题"。

每一段对自身的 认识论地位透明。哲学不假装是数据，数据不假装是 settled science，limits 不假装是 future certainty。

这一 明示 的 认识论的 stratification 是 SAE 系列论文的方法论 承诺 的具体落实——读者在每一段都知道自己在读什么级别的 声称，作者在每一段都知道自己在承担什么级别的 承诺。

9.2 防火墙重申

模型不评判、不排序、不预测、不规范。

四相结构是 SAE 群体认知论的结构性工具，不能也不应被用于：

• 对群体进行高低排序
• 推荐或贬抑特定共识、运动、组织
• 作为投资、政策、组织建设的依据
• 评判任何具体个人或群体的道德、能力、价值
• 使用维度 B 来 排序 个人或社会——参见 §4.5.3 关键限定

如果有读者使用本模型进行上述操作，该使用违反本论文的 明示 scope 与 SAE 第二定律。

模型自身的合法性恰恰依赖于不被这样使用。框架 有效 的前提是它不被滥用为 排序工具。一旦被滥用，它在 SAE 第二定律下立即失去 立场——不是因为被滥用就 无效，而是因为它的合法 use 必须 respect SAE 第二定律的 boundary。

9.3 欢迎证伪、期待证伪

本论文的所有方向性预测——包括对 SAE 框架自身的预测——被设计为 可证伪。

任何使用本框架的人有义务在面对 contradicting 实证证据 时 更新 自己的判断，而不是 挽救 框架。

这是 SAE 第一律 ρ ≠ ∅ 在认识论层面的 实例化。框架的合法性恰恰依赖于愿意被 reality 回推——一个不可能错的 框架 是 11DD 层的 伪精确，等价于 ρ = ∅。

9.4 论文留给未来的具体承诺

• 2046 年（论文发表后 20 年）审计节点：检查 SAE 进入哪个 情景，是否有任何独立 基质再生产 出现，§4 与 §4.5 的方向性预测是否有 稳健证伪。
• 2076 年（论文发表后 50 年）审计节点：检查 SAE 是否进入 ψ = 3，§6.4 三个 情景 的 先验 是否需要修订，§4.5.5 双因子预测是否有 稳健验证。
• 2126 年（论文发表后 100 年）审计节点：检查论文整体的 历史 解读 是否仍然 成立，框架 是否仍然有 解释力，是否需要 fundamental 修订。

每一个 节点 是具体的、可检查的、不可被 框架 挽救 的。如果在任何 节点 上 框架 显示 systematic 错误，论文承诺：框架 应当相应 更新 或被 废弃，而不是 挽救。

---

附录 A：方法论六四相结构的简短复习

方法论六（《相变窗口与实验设计》，DOI: 10.5281/zenodo.19464507）从 ZFCρ 框架推出构-涌现系统的四相相变结构。本附录给一个简短复习，让没有读过原论文的读者能跟随 本论文的核心论证。

A.1 ZFCρ 框架背景

ZFCρ 是 SAE 数学结构系列的核心 框架——在标准 ZFC 集合论上加入 ρ ≠ ∅ 公理（实在不为空），处理构-涌现系统的相变。ZFCρ 的主要 deliverable 之一是相变窗口的几何结构——给出一个 候选（新秩序）面对 既存秩序（既存秩序）时的 相变 动力学。

A.2 五个 交叉 点

相变窗口由五个 交叉点 划分：

1. P(J > 0) = 50%：候选共识在 random sample 中的乘法路径首次过半。这是萌芽期的起点——之前完全没有可识别的乘法 动力学。
2. z/√j 峰值：敏感度 达到峰值。可观测涨落最大的点——围绕候选共识的争议、媒体注意、社会反应集中爆发。
3. w_屏蔽 = 2/3：Le Chatelier 屏蔽率跌破 2/3。既存秩序的吸收能力首次明显不足。
4. E[A] = 0：序参量首次转正。候选 首次提供超过 既存秩序 的 net value（在 候选 占据的子群体内）。
5. h = 0：既存秩序 在该子群体内失去局部凸度。既存秩序作为 候选 在该子群体内不再有意义。

A.3 四阶段结构

由五个 交叉 自然分割出四个阶段：

• 萌芽：从 P(J > 0) = 50% 到 z/√j 峰值。乘法路径开始但 敏感度 还没峰值。
• 谱翻转：从 z/√j 峰值到 w_屏蔽 = 2/3。敏感度 从峰值下降，屏蔽率开始失效。
• 翻转：从 w_屏蔽 = 2/3 到 E[A] = 0。屏蔽率跌破 2/3 后短期内序参量转正。
• 确立：从 E[A] = 0 到 h = 0。序参量为正，既存秩序 在该 region 内失去吸引力。

A.4 r ≈ 5 不对称

方法论六的核心几何结果：萌芽期 + 谱翻转期 ≈ 5 × (翻转期 + 确立期)。

直觉解释（Le Chatelier 缓冲）：既存秩序 对偏离的吸收能力必须先被穿透，而穿透是缓慢消耗过程；穿透完成后新秩序的确立是迅速的级联。萌芽与谱翻转是穿透阶段，翻转与确立是级联阶段。

完整推导见原论文 §3-§4。

---

附录 B：本论文用到的代理量集合来源

本附录列出本论文 §6 引用的核心文献，按代理量类别组织。引用采用 APA 格式。

人均图书产量（Book production per capita）

• Baten, J., & van Zanden, J. L. (2008). Book production and the onset of modern economic growth. Journal of Economic Growth, 13(3), 217-235.
• Buringh, E., & van Zanden, J. L. (2009). Charting the "Rise of the West": Manuscripts and Printed Books in Europe. The Journal of Economic History, 69(2), 409-445.

手稿 / 印刷品密度（Manuscript / print density）

• Buringh, E. (2010). Medieval Manuscript Production in the Latin West. Brill.
• Eltjo Buringh 数据库 （Universal Short Title Catalogue / Incunabula Short Title Catalogue）

阅读经验记录（Reading experience records）

• Reading Experience Database 1450-1945 (RED), Open University.

印刷革命效应（Print revolution effects）

• Dittmar, J. E. (2011). Information technology and economic change: The impact of the printing press. Quarterly Journal of Economics, 126(3), 1133-1172.
• Rubin, J. (2014). Printing and Protestants: An empirical test of the role of printing in the Reformation. Review of Economics and Statistics, 96(2), 270-286.

科举（Civil examinations）

• Elman, B. A. (2000). A Cultural History of Civil Examinations in Late Imperial China. University of California Press.
• China Biographical Database (CBDB), Harvard.

近世日本识字（Early modern Japan literacy）

• Rubinger, R. (2007). Popular Literacy in Early Modern Japan. University of Hawaii Press.

现代技能评估（Modern skill assessments）

• OECD (2023). Survey of Adult Skills (PIAAC) Results. OECD Publishing.
• OECD (2023). PISA 2022 Results (Volume I): The State of Learning and Equity in Education. OECD Publishing.

阅读行为趋势（Reading behavior trends）

• National Endowment for the Arts (2004). Reading at Risk: A Survey of Literary Reading in America.
• Bureau of Labor Statistics. American Time Use Survey annual data 2003-2023.

纸本对比屏幕的元分析（Paper-vs-screen meta-analyses）

• Delgado, P., Vargas, C., Ackerman, R., & Salmerón, L. (2018). Don't throw away your printed books: A meta-analysis on the effects of reading media on reading comprehension. Educational Research Review, 25, 23-38.
• Clinton, V. (2019). Reading from paper compared to screens: A systematic review and meta-analysis. Journal of Research in Reading, 42(2), 288-325.

亲属关系强度与个人主义（Kinship intensity & individualism）

• Schulz, J. F., Bahrami-Rad, D., Beauchamp, J. P., & Henrich, J. (2019). The Church, intensive kinship, and global psychological 变化. Science, 366(6466).
• Enke, B. (2019). Kinship, cooperation, and the evolution of moral systems. Quarterly Journal of Economics, 134(2), 953-1019.

长期暴力下降（Long-term violence decline）

• Eisner, M. (2003). Long-term historical trends in violent crime. Crime and Justice, 30, 83-142.

道德普遍主义（Moral universalism）

• Cappelen, A. W., et al. (2024). Moral universalism: Global evidence from 60 countries. Quarterly Journal of Economics (forthcoming, working paper version 2023).

数据库集合（Database collections）

• Seshat: Global History Databank
• Clio-Infra
• China Biographical Database (CBDB)
• Reading Experience Database (RED)
• Universal Short Title Catalogue (USTC)
• Incunabula Short Title Catalogue (ISTC)
• Germania Sacra
• Monastic Matrix
• FemMoData

---

附录 C：SDE 实验的 负面结果 概要

在论文撰写前，作者尝试了用热力学 SDE 模拟群体共识形成的定量实验。本附录给出实验的概要——detailed setup、code、数据 待公开后另行发布。

C.1 实验 setup

模型：N 个相互耦合的 基质，每个 基质 是一个 Brusselator 型化学振子加上 v₁ 和 v₂ 两个 自我评估 变量。耦合通过 v₁ 或 v₂ 在 基质 之间传递实现。

主要 变化：

• N（基质 数量）：从 2 到 40
• 协议（哪个 v 被 跨耦合）：L0（raw output x）、L1（v₁）、L2（v₂）
• 拓扑：mean-field 与 环形网络
• 耦合强度 g：从 0.01 到 0.10
• 是否 相互：相互 与 不对称

观测量：q（q-指数 分布参数，描述 fluctuation tail 的 长程关联）。

C.2 主要发现

发现 1：协议层级（L0 < L1 < L2）真实存在但效应小。L1 比 L0 高约 0.05，L2 比 L1 高约 0.01。差异在统计上 detectable 但量级远小于 Paper X 的 dramatic separation。

发现 2：L0 协议在压力下先 崩溃。在 g ≥ 0.05 时 L0 进入 q ≈ 1 的 exponential regime，而 L1/L2 保持 q > 1。这是协议-具体 相变，不是 universal 崩溃。

发现 3：不对称 耦合 在压力下方差翻倍。在 g = 0.05 时，不对称 的 q_std 从 稳定 的 ~0.05 翻到 ~0.10。相互 是结构性稳定条件——不是相对优越的选项，是 stress 下保持稳定的 前提。

发现 4（关键 负面结果）：q 对 N、k、拓扑都盲目，只对 g 敏感。在 mean-field 同质耦合下，q 对群体规模 N（从 2 到 40）、网络度 k（从 1 到 14）、拓扑（mean-field 与 ring）都不敏感，差异低于 格内方差。

C.3 这一 负面结果 对论文的支持

发现 4 反过来支持 §4.5 的双因子结构。q 测量 基质 内 长程关联，但不测量 基质 population composition（维度 B）。如果 q 能 捕捉 所有相位决定因素，单因子（耦合）模型已经足够；q 不能 捕捉 群体结构变量这一 负面结果，恰好证明必须有第二维度（维度 B）作为 框架的 part。

实验 thus 显示了热力学 SDE 这一 测量 tool 的能力边界——它捕捉维度 A 的 thermo动力学的 signature，但盲目于维度 B。这是 §4.5.4 论证的实证基础。

---

附录 D：四家评审 的整合记录

本论文 v3 大纲整合了四家 AI 评审（子路 / 公西华 / 子贡 / 子夏）的修订建议。这不是纯粹的 致谢，而是 SAE 工作流（方法论 VII 共生论）在论文层面的 实例化——通过多基质 压力测试 提升论文的 融贯 与 稳健性。

主要 trace：

• 子路（Claude）：识别格内单调不减 vs 格退出的精度问题（→ §5.3）；提议六项面向与 DD 层显式对接（→ §4.5.3）；提议修道院案例 再生产 基础设施 重写（→ §6.3.2）

• 公西华（ChatGPT）：识别"广义对象 vs Kant 第二定律 稳定条件"边界张力（→ §1 + §3.4 + §7 重组）；提议维度 B 前置（→ §1 + §2.3）

• 子贡（Grok）：consistency check 通过；提议 §6.4.2 先验 标注 框架内部估计

• 子夏（Gemini）：施压 能力作为状态 vs essence 限定（→ §4.5.3 关键限定）；施压第二相 vs 噪声泡沫差异化 标志（→ §3.2 蓝图 跨基质独立性）；施压审计悖论（→ §6.4.4 审计独立性 澄清，但拒绝引入"大规模扩散 as 证伪"替代方案）

四家评审 中没有任何一家发现 框架内部公理 破裂——所有修订都在 框架内部精度提升。这本身是一个有信息价值的发现，说明 SAE 群体认知论 P1 的 概念核心 已经成熟。

整合过程在论文起草前搁置一段时间，让 评审 内容在作者 基质 中沉淀，避免 第一反应错误接受某些 评审 中的隐性的非法操作（特别是子夏施压点三的 "大规模扩散 无 基质" 概念被作者自身拒绝——保留 框架 对 大规模扩散 的中性立场）。这一过程 的反思见正文 §6.4.3 SAE-预测 1 的具体处理。

---

参考文献

直接依赖（SAE 内部）

秦汉. (2026). 相变窗口与实验设计（SAE 方法论 VI）. Zenodo. https://doi.org/10.5281/zenodo.19464507

秦汉. (2025). 自身作为目的的结构（SAE 基础论文）. Zenodo. https://doi.org/10.5281/zenodo.18528813

秦汉. (2026). ZFCρ Paper 44. Zenodo. https://doi.org/10.5281/zenodo.19247859

秦汉. (2026). ZFCρ Paper 57. Zenodo. [待补 DOI]

接口论文（SAE 内部）

秦汉. (2026). SAE 方法论 VII：共生论. Zenodo. https://doi.org/10.5281/zenodo.19481304

秦汉. (2026). SAE 三律. Zenodo. [待补 DOI]

案例与代理量参考

详见附录 B 完整文献列表。