The Causality of the Non-Binding Vow - On Trust, Promise, and the Generation of Expectation in the Absence of the Other's Sword 【（Preliminary）Draft】

Posted on In , Views: Disqus: Word count in article: 101k Reading time ≈ 6:08

ENGLISH

Philosophy of Intimacy and the Theory of Justice · Paper XIII

The Causality of the Non-Binding Vow

On Trust, Promise, and the Generation of Expectation in the Absence of the Other’s Sword

[ Working Draft ]

On the Mechanism by which a Promise without Force Generates Trust, the Translation of Sincerity across the Borders of Epistemology, and the Descent of a Relational Divinity

Wanhong

Working draft, not for citation or circulation

死生契阔，与子成说。
Through death and life, through parting and reunion, with you I made my word.

匪报也，永以为好也。
Not as requital—but to be, with you, in good faith for ever.

Where nothing binds two hearts, they hold each other still; and a vow is believed, if at all, in the very place where nothing binds it.

— 《诗经》Shijing (The Book of Songs), 邶风·击鼓 and 卫风·木瓜; and a remark of the author’s own.

For the forest girl,
to whom a promise was made
that no sword stands behind,
and which is, for that very reason,
the more wholly given.

本乎此心，自成永恒。

— To the one I love most.

Abstract

This paper asks how a promise that carries no enforcing force—a vow, a betrothal document, a memorandum of understanding—can nonetheless generate real and well-founded trust. Beginning from the symptomatology of the contemporary condition, in which the apparatus of assurance proliferates at every scale while the trust it secures grows scarcer, it diagnoses the failure to a mistake about causation: the taking of force, the mechanical cause, for the cause of an effect, trust, that is inferential and free. It traces, through the philosophy of law, the migration of the anchor of legal validity from force toward trust, which the law records at its own technical edge in the doctrine of promissory estoppel; and it argues that the inferential cause is a different kind of thing from the mechanical, the kind that addresses a free being as free. Its central contribution is a mechanism: trust is the precision a receiver’s generative model assigns to the prediction that a promise will be kept, generated through three coupled channels—a structural prior, a high-fidelity self-committing likelihood, and a wagered prior—whose multiplicative coupling makes genuine trust robust where any single channel is forgeable. It gives this mechanism a geometric reading in the language of holonomy and a structural reading, for relational trust, in the language of open quantum-system dynamics; states the criterion of justice that distinguishes genuine trust from forged; develops a theory of the translation of sincerity across incommensurable epistemologies; and reconceives, in a historical-materialist register, the optimisation of the juridical as the decentralisation of enforcement. It rises, at its summit, to the claim that the self-legislation, self-adjudication, and self-keeping of a subject who binds himself without recourse is the descent of a relational divinity—so that the non-binding vow proves, in the dimension of the divine, the purer realisation of what a promise is. It concludes in plural conclusions, refusing, in its own form, the synthesis its thesis forbids.

Prelude

Some weeks before this paper was begun, its author sat down to write a betrothal document. It was written on red silk, in classical Chinese, on a day chosen because the almanac called it auspicious; and it was written in the full knowledge that nothing in the world would make it binding. No court would enforce it. No clause within it could be sued upon. It named no penalty and conferred no remedy. Against the day it might be broken it set no sword, and behind the hand that wrote it stood no power that could compel the hand to keep it. By every standard that a contract is measured against, it was a defective instrument: a promise that promised nothing it could be made to perform.

And yet it was not written lightly, and it was not, by the two for whom it was made, received lightly. The choosing of the day, the silk, the gravity of the brush—these were not theatre, or were not only theatre. Something was being done in the writing of it that the writing of a contract does not do, and that the document’s want of force did not diminish but, strangely, seemed to deepen. A contract one signs and files; this was kept. The very absence of the sword behind it was felt, by the one who wrote it, not as a lack to be apologised for but as the condition of the thing’s being what it was. Had there been a penalty, the promise would have been a smaller thing. That it bound nothing was the reason it could bind everything.

This is a strange experience to have, and stranger still to find that one cannot account for it. The whole apparatus by which the modern world thinks about promises—the contract, the penalty, the enforceable obligation, the institution that stands ready to compel—was built precisely to supply what this document conspicuously lacked. One is told, and half believes, that a promise is reliable in proportion to the force that backs it; that what cannot be enforced cannot be counted on; that to bind oneself is to expose oneself to a consequence one would rather avoid. By that account the betrothal document is a sentimental nullity, a pretty gesture with nothing underneath. But that is not how it was made, and it is not how it was met. Beneath the account one half believes there runs a contrary conviction, older and harder to dislodge: that the promise which asks for no sword is not the weaker but the truer; that a bond one could be forced to keep is, just insofar as one could be forced, not yet fully a bond at all.

The author does not propose to resolve this by deciding that one of these convictions is naive. Both are real, and the tension between them is not a confusion to be cleared up but a fault line to be followed down. For the same experience that gave rise to the document gave rise, in the days after it, to a more anxious and more practical question, which is in truth the same question wearing working clothes. The promise had been made, and was believed, between the two who made it. But it had now to be carried outward—to a family, in particular, who did not yet know the author, who would not read red silk as the two who wrote it read it, and to whom the gravity of a brush on an auspicious day would say, if anything, very little. They are people, as the author understands them, who hold that what is the case is the case and what is not is not—有就是有，没有就是没有—and who would ask, reasonably and without malice, what there is to go on. To them the bare sincerity of the vow is not evidence. And here the comfortable thought that a true promise needs no sword runs straight into its own difficulty: for if the truest promise is precisely the one that offers no external guarantee, it is also, and for the same reason, the one that gives the person who must be reassured the least to hold on to. The very purity that makes it a vow makes it, to one who asks for grounds, almost nothing at all.

So the experience opens, at its near edge and at its far, onto a single question, and it is the question this paper exists to ask. How does a promise that has no force generate trust? How does an utterance about a future that does not yet exist, which threatens no consequence and can compel no compliance, produce in another person, here and now, a real expectation that it will be kept—and produce it, if the older conviction is right, more surely the less it relies on being able to compel anything at all? The question is at once the most intimate and the most public that can be asked. It is the question of how the author is to be trusted by those he has not yet earned the trust of; and it is, the paper will argue, the very question on which the law itself, in its long retreat from the sword, has been quietly working for a century. It begins in the choosing of a day and the gravity of a brush, and it reaches, before it is done, as far as the foundations of obligation and the descent of what an older language would have called the divine.

This is where the paper begins: not with a thesis but with a vow that should not have worked, and did.

1. Introduction

This paper is the thirteenth in a series on the philosophy of intimacy and the theory of justice. Where the Prelude entered the question through a single experience, this section states it as a problem, fixes its scope, and sets out what the paper claims to establish. The question is the one the Prelude arrived at: how a promise that carries no enforcing force generates trust. A vow, a betrothal document, a memorandum of understanding, a bare promise—an utterance that binds the one who makes it to a future conduct, but which no court will enforce, which threatens no penalty, and behind which stands no power that can compel performance—is nonetheless, in the right conditions, believed; and the believing is not credulity but a real and often well-founded expectation that the thing will be kept. The paper asks by what mechanism this expectation is produced.

The problem has a structure that makes it look, at first, impossible. Trust is directed at a future, and the future does not yet exist and cannot be verified; the symbol that asks for trust has no force, and so cannot secure the future by threatening a consequence. The thing to be explained is therefore how a symbol about what does not exist, carrying no guarantee of consequence, produces a real expectation in the present—something close to a generation from nothing. The paper’s central and original contribution is an account of that mechanism. Its other contributions, set out below, are largely applications, extensions, and consequences of it.

Scope

The paper concerns the generation of trust by non-coercive symbols, across three concentric scales: the intimate vow between two persons; the promise carried outward to a wider circle, the family in particular; and the public institutions, contract and the courts, in which the same problem appears at the scale of a society. It treats these scales as continuous: the same mechanism is argued to operate at each, and the paper moves between them rather than confining itself to one. It does not offer a general theory of trust as a psychological disposition, nor a survey of the empirical sociology of trust; it offers a mechanism for the specific case of the non-coercive promise, and traces that mechanism’s reach. Where the argument requires the resources of jurisprudence, of the cognitive sciences, of psychoanalysis, of political economy, and of the Daoist and Kantian traditions, it draws on them as the prior papers in the series have drawn on them: as several irreducible epistemic positions on a single object, none of which governs the others.

Relation to the foundational paper and to the series

The paper stands in a definite relation to a foundational text, to which it should be read as a sustained development of two of that text’s diagnostic themes. The foundational paper sets out the framework of which the present series is a part, and devotes nearly half its length to a diagnosis of the crisis of the symbol under modern conditions, across several dimensions. The present paper takes up two of those dimensions in particular—the crisis of authenticity and trust, and the overload of justice—and subjects them to the exhaustive mechanistic analysis that the foundational paper, whose purpose was diagnostic, deferred. Where the foundational paper established that trust has come under a crisis whose mechanism it did not specify, the present paper specifies the mechanism; where the foundational paper diagnosed an overload of the juridical and a failure of enforcement, the present paper carries that diagnosis to the claim that force has failed as a signal, and offers the way out the foundational paper marked but did not develop. The relation to the wider series is similarly definite. The paper draws on the geometric-phase or holonomy apparatus developed earlier in the series, on its three-register theory of value, and on its account of the practitioner who cannot guarantee that his practice is good; and it observes, as the series requires, the discipline by which no single framework is permitted to absorb the others.

Contributions

The paper claims to establish the following. First, and centrally, that the expectation a non-coercive promise produces is generated by a mechanism that can be given a definite form: trust is the high precision a receiver’s generative model assigns to the prediction that the promise will be kept, and this precision is supplied through three coupled channels—structural, self-committing, and wagered—which multiply rather than add, so that the consistency among them, hard to forge, is the source of well-founded trust (§6). Second, a jurisprudential result: that enforcing force is not the cause of obligation but a substitute mechanism for the production of expectation, and that the internal development of legal philosophy has, over a century, moved the anchor of legal validity away from force and toward trust—a movement the law itself records, at its technical edge, in the doctrine of promissory estoppel (§4). Third, an account of relational trust, the trust transmitted along the links of a relational network, as a phenomenon spanning all three registers, together with a new specification of how it fails under the crisis of the symbol: through a separation of the symbolic vehicle of trust from the real coupling it should carry (§9). Fourth, a theory of the translation of sincerity across the borders of incommensurable epistemologies, with its attendant ethics (§10). Fifth, a direction for the reform of the juridical, developed in a historical-materialist register, toward the decentralisation of enforcement (§11). Sixth, and as the paper’s metaphysical summit, an account of self-legislation as the descent of a relational divinity, with the consequence that the promise which has no force may be, in respect of this, the more fully realised (§12).

A note on method and on what is conceded

Two commitments run through the paper. The first concerns the status of its formal and scientific apparatus. The paper draws on predictive coding for the mechanism of trust, and on the dynamics of open quantum systems for the structure of relational trust; it does so as the prior papers drew on the geometric phase, as mechanistic and structural models and not as reductive claims about what trust ultimately is. Predictive coding furnishes a description at the computational and implementational level; it does not displace the phenomenological, the juridical, or the ethical descriptions of the same object, and it is not a meta-framework that absorbs them. The same discipline governs the use of the quantum-dynamical language in §9: a structural isomorphism is claimed, not a physical reduction. The second commitment is the series’ refusal of a governing conclusion. The paper ends not in a synthesis but in plural conclusions, each framework stating what it can and cannot say, because the object—trust generated without a guarantor—is one no single framework possesses, and the form of the paper is meant to enact this as much as to assert it (§13).

Limitations and what is left open

The paper is candid about its limits. The holonomy apparatus is employed as a structural analogy rather than developed to full formal closure here, the relevant formal programme being carried by another paper in the series; the genealogy of trust-grammars it offers (§10) is offered as adequate to its practical purpose rather than as exhaustive, and admits of finer subdivision; the depth to which the predictive-coding account is pressed, in particular toward a first-principles treatment in terms of the active reduction of expected uncertainty, is deliberately bounded, the stronger and more technical development being noted rather than carried out. The quantum-dynamical isomorphism of §9 is presented as a formal analogy inviting, but not here supplying, its own development. These are marked as open rather than concealed, in keeping with the series’ discipline of distinguishing what is claimed from what is conjectured.

Plan of the paper

The paper proceeds as follows. It assembles, without yet explaining them, the symptoms of the contemporary condition of trust (§2); and diagnoses them, drawing on the foundational paper and on the older problems of Hume and Kant, to the thesis that force is not the cause of trust (§3). It then locates that thesis within the philosophy of law, tracing the migration of the anchor of legal validity from force toward trust (§4), and draws out the metaphysical consequence, that the cause of trust is inferential rather than mechanical and that this is the very opening in which freedom and the moral are possible (§5). It develops the central mechanism (§6); gives its geometric reading in the language of holonomy (§7); states the criterion that distinguishes genuine trust from forged (§8); analyses relational trust and its failure, with its quantum-dynamical structure (§9); and develops the theory of translation by which sincerity crosses the borders of epistemology (§10). It turns then to practice: to the reform of the juridical at the scale of institutions (§11), and to the praxis of trust at the scale of a life, rising to the account of relational divinity that is the paper’s summit (§12). It concludes in plural conclusions (§13), and closes in an envoi.

2. The Symptomatology of Trust

Before a condition is diagnosed it is presented, and a diagnosis that proceeds without first attending to the symptoms is a diagnosis of nothing. This section assembles the symptoms. It does not yet explain them; explanation is the work of §3, and to anticipate it here would be to read the symptoms in the light of a thesis they are meant to motivate rather than to confirm. The discipline of the section is therefore descriptive: it sets out a range of observable, recognisable phenomena, drawn from the most intimate scale of human dealing to the most public, and asks only that they be looked at together, in the expectation that what governs them will appear, when it appears, as the explanation of a pattern already seen rather than as a hypothesis imposed upon scattered instances.

The phenomena, looked at together, exhibit a single and initially paradoxical shape. The apparatus by which trust is secured proliferates, and trust itself grows scarcer. Wherever one looks, the instruments designed to produce or guarantee or certify reliability multiply—longer contracts, more elaborate vows, denser systems of credentialing, ever finer mechanisms of rating and verification—and the multiplication does not arrest the erosion of the thing the instruments exist to secure but accompanies it, and, in a way the section will only describe and not yet account for, seems bound up with it. The relation between the proliferation and the erosion is left, here, deliberately unstated. That there is such a relation, recurring across scales that have otherwise little in common, is the observation the section exists to establish.

The phenomena are presented across three scales, from the intimate outward, not because the intimate is a miniature of the public—whether it is or is not is a question for later—but because the recurrence of one shape across scales that differ in every other respect is itself the first thing to be seen.

2.1 The intimate and familial scale

At the most intimate scale, the symptom appears first in the very document from which this paper set out. A betrothal is undertaken, and around it gathers an apparatus of solemnisation—the auspicious day, the silk, the gravity of the language, in other settings the ring, the announcement, the assembled witnesses—whose elaboration is in no evident proportion to the security it confers, since it confers, in the strict sense, none. People who undertake such a thing know, when they reflect, that nothing in the ceremony binds; and they elaborate the ceremony anyway, and feel the elaboration as weighty rather than as idle. This is already a symptom: a solemnising apparatus is invoked with full seriousness in the acknowledged absence of anything it could enforce.

The symptom recurs, in a sharper and more troubling form, in the marriage that may follow. The vows exchanged at weddings have grown, across recent generations and in many places, more articulate, more personal, more lavish in their profession of permanence, at the same time as the dissolution of the unions they open has grown more common rather than less. The observation is not offered here as a causal claim in either direction; it is offered as a juxtaposition that the symptomatology must record, that the form of the promise has become more elaborate as the persistence of what it promises has become less assured. The two curves, of the eloquence of the vow and of the fragility of the bond, have moved, over the relevant period, in the same direction, and a symptomatology notes this without yet asking what to make of it.

The symptom appears again, and most pressingly for the concerns of this paper, in the predicament of the one who must win the trust of a family not yet his own. Here the apparatus that elsewhere proliferates is precisely what is unavailable. The suitor has, by hypothesis, no track record with the family, no accumulated history by which he might be known, no third party they already trust who can vouch for him; he has only a sincerity that is, to him, entirely real, and that has, to them, no form they can read as evidence. The symptom at this scale is thus not the proliferation of the apparatus but the felt insufficiency of everything that is not it: the discovery that sincerity, however genuine, does not present itself as ground, and that the very people whose trust matters most are the people to whom one has least that they will count.

At the scale of law and commerce the same shape recurs with a documentary precision that makes it easy to measure. Contracts have grown longer. The instruments by which commercial parties bind one another have expanded, across the last century and with sharp acceleration in recent decades, into documents of a length and intricacy no party reads in full, drafted in anticipation of every contingency and every adversarial reading, and lengthening still; and the lengthening has not been accompanied by a growing ease of dealing or a growing confidence between contracting parties, but, by the testimony of those who deal under such instruments, by the opposite. The thicker the contract, the less, not the more, the parties feel secure; the elaboration of the terms is experienced less as the building of confidence than as its pre-emptive substitute, an arming against a breach already half expected.

Beside the contract stands the memorandum of understanding, and it is, for this paper, the most telling instrument at this scale precisely because it is the contract’s opposite. The memorandum of understanding is, by design, non-binding: it records an intention, sets out the shape of a contemplated dealing, and explicitly withholds the force of an enforceable obligation. Such instruments proliferate. They are drafted with care, signed with ceremony, and treated by their signatories as significant, in the full and stated knowledge that they bind nothing. That parties expend such effort on the solemn execution of documents they know to be unenforceable is a symptom of exactly the kind this section collects: a grave investment in a form whose want of force is not concealed but declared.

And behind both contract and memorandum stands the court, in which the symptom takes its most consequential form. The machinery of enforcement, the sword to which all these instruments are nominally referred, is under a strain that has begun to show. The volume of breach exceeds the capacity of the courts to adjudicate it; the cost of enforcement, in time and money and attention, rises until for many wrongs it exceeds the value of the remedy; the period over which a judgment is obtained lengthens, and the prospect of actually recovering on a judgment, once obtained, narrows. The sword, in a word, hangs but increasingly does not fall, and is known not to fall; and an instrument of enforcement that is known not to fall has ceased, whatever its formal availability, to function as the thing it was. The symptom here is not the proliferation of an apparatus but the hollowing of one: the persistence of the full formal machinery of compulsion alongside its growing practical inability to compel.

2.3 The universal scale

At the widest scale the shape appears in its purest and most general form, in the instruments by which strangers are made legible to strangers. The letter of recommendation is the clearest case. It is an instrument designed to transmit a knowledge that the recipient cannot acquire directly: the writer, who knows the candidate, vouches for the candidate to a committee, which trusts the writer. And it has, by common acknowledgment within the institutions that rely on it, very largely ceased to function. The letters have inflated until nearly every candidate is described as among the best the writer has encountered; the superlative, having become universal, has become uninformative; and committees, no longer able to extract a signal from a channel everyone has learned to saturate, turn away from the letters toward whatever can be counted—scores, citations, rankings—not because the counted things are richer but because they are, at least, harder to inflate. The apparatus of vouching has proliferated to the point of self-cancellation, and trust has retreated from it.

The same shape governs the systems of rating and certification by which the digital economy attempts to make its strangers legible. Trust badges, verification marks, star ratings, certification seals multiply across every surface on which strangers must deal with strangers; and their multiplication is met, by those who encounter them, not with growing confidence but with a growing and well-schooled wariness, since the very ubiquity of the mark of trustworthiness teaches that the mark is cheap, gameable, and as available to the predator as to the honest. The more the surface is covered with the signs of trustworthiness, the less any one of them reassures. Here the proliferation and the erosion are not merely concurrent but legibly bound: each new layer of certification is a response to the failure of the last, and itself elicits the next.

And beneath all these the most general symptom of the age, which the section records last because it underlies the rest: the condition of being surrounded by symbols whose provenance and whose intention cannot be established. The cost of producing a convincing token—a document, an image, a voice, a record of conduct—has fallen, with the technologies of the present, toward zero, and the consequence is a saturation of the human environment with signs that are no longer reliably tied to anything they purport to indicate. A person now moves through a world of symbols any of which may be confected, and the rational response to such a world—a generalised lowering of the weight given to all symbolic testimony—is itself a symptom of the first order: a withdrawal of trust not from this or that untrustworthy source but from the symbolic as such.

2.4 A taxonomy of the symptoms

The phenomena gathered above are not a miscellany. Looked at together, they are so many appearances of a single fissure, opening at different scales and in different materials, between an apparatus that is meant to secure trust and the trust it fails to secure. To make the unity visible, and to give the diagnosis of §3 a structured object rather than a list of anecdotes to work upon, the symptoms may be set out along a small number of shared dimensions: the scale at which the symptom appears; the apparatus of assurance proper to that scale; the mode in which that apparatus fails; and the specific form the governing paradox—proliferation alongside erosion—takes in each case. The table that follows arranges them so. It is offered as a description and not as an explanation: each row records what is observed, and the column of failure-modes names the manner of a breakdown without yet assigning its cause. That every row instances one fissure, and what that fissure is, is the matter the table hands forward to the diagnosis.

Phenomenon Scale Apparatus of assurance Mode of failure
Betrothal, wedding vow intimate ceremony, solemnity, profession of permanence elaboration grows as the bond’s persistence weakens; gravity without enforceability
Winning a family’s trust familial sincerity, profession, proposed conduct sincerity does not present as evidence to those who ask for grounds
Commercial contract legal enumerated terms, penalties, remedies thickening of terms experienced as anxiety, not confidence; arming against expected breach
Memorandum of understanding legal solemn execution of a declared non-binding form grave investment in an instrument known to bind nothing
The court, enforcement judicial compulsion, the threat of the sanction the sword hangs but does not fall and is known not to; formal machinery hollowed of practical force
Letter of recommendation institutional a trusted writer’s vouching along a chain inflation to the superlative; the channel saturated to uninformativeness
Rating, certification, trust badge platform marks, seals, verification, star scores ubiquity teaches cheapness; the mark reassures less the more it covers
The confected token universal symbolic testimony as such zero-cost production severs sign from referent; rational withdrawal from the symbolic

Read down the columns, the table says more than any row. The scale ranges from two persons to a whole society, and the apparatus changes utterly across that range, from red silk to a star rating; yet the column of failure-modes rhymes. In every case a form that is meant to carry assurance is found, in the end, not to carry it: the form persists, elaborates, even proliferates, while the assurance drains out of it. Whether at the scale of a vow or of a verification mark, the gap that opens is the same gap—between the apparatus and the trust—and it is the recurrence of this one gap across a range that shares nothing else that constitutes the central symptom of the age. The proliferation of the apparatus and the erosion of the trust are, the table suggests without yet showing, two faces of one process. Naming that process, and showing why the proliferation should accompany rather than arrest the erosion, is the work to which the symptomatology now hands over.

3. Diagnosis

The symptoms gathered in §2 share a shape, and a shape that recurs across scales otherwise unrelated is evidence of a common cause. This section names that cause. It does so in the manner the foundational paper established for such work: diagnosis here is not a preliminary to be hurried through on the way to a remedy, but a substantial undertaking in its own right, on the principle that the precision of a response is bounded by the precision of the diagnosis it answers. To see clearly how a gap has opened between the symbol and the thing it was to secure is already half of what is needed to respond to it; and so the section takes its time, and develops the diagnosis to a definite thesis, before any mechanism or remedy is proposed.

The foundational paper diagnosed a crisis of the symbol under modern conditions across several dimensions. The present paper does not recapitulate them all; it takes up the two that bear directly on trust—the crisis of authenticity and trust, and the overload of justice—and develops each beyond the point at which the foundational paper left it, weaving in three further of its dimensions where they bear on the mechanism, and adding one diagnostic dimension the foundational paper did not contain. It then sets the whole against the two older philosophical problems—Hume’s and Kant’s—that give the modern crisis its deeper form, and draws from the conjunction the thesis that organises the rest of the paper.

3.1 The crisis of trust: from the failure of trust to the failure of its production

The first dimension the foundational paper diagnosed under this head is by now familiar from the symptomatology: a saturation of the human environment with symbols whose cost of production has fallen toward zero, whose provenance cannot be established, and which are as available to deception as to sincerity, with the rational consequence that the weight given to symbolic testimony as such declines. The foundational paper established this much. What the present paper must add—for it is the precise point at which the diagnosis turns into the problem this paper exists to solve—is that the crisis has passed beyond the failure of particular instances of trust into a doubt about the producibility of trust at all. It is one thing for a given promise to be disbelieved; it is another, and the deeper condition, for it to become unclear by what means any promise could now be made believable. The symptomatology showed the apparatus of assurance failing at every scale; the diagnosis is that the failure is not of this or that instrument but of the question they all answer, so that the question itself—how is trust to be generated?—which in a healthier condition would not need to be asked, has become both urgent and obscure. This is why the mechanism of trust-generation, which a more confident age could take for granted and leave unexamined, must now be made the express object of inquiry.

3.2 The overload of justice and the failure of force as a signal

The second dimension is the one the foundational paper named the overload of justice. As relational and communal orders of trust have thinned, the burden of holding people to their dealings has fallen, by default, upon the juridical: where once a web of relationship and reputation and communal witness held a person to his word, now the law is asked to do it, and is asked to do it for a volume and a variety of dealings it was never sized to bear. The foundational paper diagnosed the resulting overload: the law extended into ever more of life, and at the same time hollowed from within, asked to underwrite an order it lacks the means to underwrite.

The present paper carries this diagnosis one decisive step. The overload is not merely quantitative, a matter of too many demands upon a finite institution; it is the visible form of a deeper failure, the failure of force as a signal. The symptomatology recorded the sword that hangs but does not fall. The diagnosis of that symptom is this: the threat of compulsion produces compliance not by the mechanical certainty of the sanction—which, as the symptomatology showed, has become anything but certain—but by signalling to the parties that compliance is to be expected. Force, that is, operates as a substitute for expectation: it is one means among possible others of bringing a party to expect that an undertaking will be honoured. And a signal whose mechanical basis is known to have failed—a sword known not to fall—ceases to signal. The overload of justice is thus the macroscopic, institutional, observable face of the paper’s central thesis, encountered here as a hard social fact before it is argued as a principle: that force was never the cause of the obligation, but only one device for the production of the expectation in which trust consists, and that the device is now failing on its own terms.

3.3 Three further dimensions: cause, consequence, and precondition

Three further of the foundational paper’s dimensions enter the diagnosis, not as separate crises but as the cause, the consequence, and the precondition of the crisis of trust, and they are woven in here on that understanding rather than treated in their own right.

The first is the generativity paradox: the capacity of the present age’s technologies to produce symbols without limit and at no cost. This is the cause, at the level of material technique, of the saturation diagnosed above; it is why the forged signal has become not an occasional intrusion but the ambient condition, and why the production of a convincing token of sincerity can no longer, by itself, carry the weight it once did. The diagnosis of trust must be set upon this technical base, on pain of mistaking for a moral decline what is in the first instance a change in the cost of producing signs.

The second is the return of the Real: the eruption, where the symbolic order fails to hold, of what the symbolic was meant to bind—in the form of a free-floating anxiety, a susceptibility to conspiracy, a drift toward the fundamentalist and the violent. This is the consequence, at the level of the affective and the social, of a generalised failure of trust: when the symbolic testimony by which people orient themselves to one another can no longer be relied upon, what returns is not a neutral scepticism but a charged and disoriented seeking after some unfalsifiable ground. The diagnosis records this as the cost of leaving the crisis of trust unanswered.

The third is the collapse of intersubjectivity, the degradation of the relation in which one meets another as a presence rather than as an object—what an earlier language opposed as the I–Thou to the I–It. This is the precondition whose erosion most threatens the mechanism this paper will propose. For one of the channels by which trust is generated, as §6 will show, requires that a real self-committing act be received by another who is present to receive it; and where the relation has degraded to the merely instrumental, that channel is disabled at its root. The collapse of intersubjectivity is thus not one more symptom but a threat to the very possibility of the cure.

3.4 The decoupling of credibility from enforceability

To the dimensions drawn from the foundational paper the present diagnosis adds one the foundational paper did not contain, and which is, the paper submits, the specifically modern hinge on which its central problem turns. It is the historical decoupling of the credibility of a promise from its enforceability.

In the pre-modern settings against which the modern condition is most sharply seen, the two were bound together and were scarcely distinguished. A promise was credible and enforceable at once, and by the same fact: the god who witnessed the oath was also the power that would avenge its breach; the lineage or the village before whom the word was given was also the body that would exact the cost of breaking it; to be seen to promise was, in the same act, to be bound to perform, because the seeing and the binding were done by one and the same watching community. Credibility and enforceability were two aspects of a single social fact, and no one had occasion to prise them apart.

Modernity prised them apart. It assigned enforceability to a specialised institution, the law, which would compel performance through an abstract and impersonal machinery; and it left credibility—the quality by which a promise is believed in the first place, before and apart from any question of compulsion—without a corresponding institution to house it. So long as the law’s machinery of enforcement was robust, this division of labour was invisible, because enforceability could be relied upon to stand in for credibility: one did not need to find a promise credible if one could be assured it was enforceable. But the symptomatology has shown the enforcing machinery failing; and as it fails, the division of labour is exposed, and credibility is revealed for what modernity had quietly made it—an orphan, a quality with no institution to underwrite it, hanging unsupported now that the institution which had stood in for it can no longer bear the weight.

This is why the paper’s central problem is urgent now, and was not, in this form, urgent before. The question of how a promise becomes credible in the absence of enforcement is not a perennial philosophical puzzle that happens to have been posed again; it is the question a specific historical arrangement has forced into the open, by first making credibility depend upon enforceability and then allowing enforceability to fail. The orphaning of credibility is the diagnosis that pins the paper to its moment, and it is offered as the diagnostic contribution most particularly the paper’s own.

3.5 The two older problems: Hume and Kant

Beneath the modern crisis lie two older problems, which give it its deeper philosophical form and which the diagnosis must reach if the thesis it arrives at is to be more than a sociological observation.

The first is Hume’s. The expectation that a person who has kept his word will keep it again, like every expectation that the future will resemble the past, rests upon induction, and induction, as Hume showed, has no foundation in reason: no accumulation of past performance entails future performance, and the inference from the one to the other, however indispensable, cannot be justified as a matter of logic. This is the problem that the evidentialist’s reliance upon track record can never wholly escape, and it sets a limit on what any quantity of evidence can establish: at the end of the longest record of kept promises there remains a gap, across which no evidence can carry one, between what has been and what will be. The diagnosis records this as the deep structural reason why trust can never be reduced to the accumulation of evidence: the gap Hume opened is constitutive, and every account of trust must say what it does about that gap rather than pretend to have closed it.

The second is Kant’s. Where Hume’s problem is that the inference to future conduct cannot be grounded in past fact, Kant’s bears on the standing of the one who must nonetheless be relied upon. To treat a person solely as a system whose future behaviour is to be predicted from evidence and secured by incentive is to treat him, in the terms of the second formulation of the categorical imperative, merely as a means, as an object to be managed rather than an end in himself. But a promise, in its full sense, is addressed to a free agent who binds himself, and is received from one; the one who promises is not predicting his own future conduct as one predicts the weather, but committing himself to it as only a free being can. The diagnosis records this as the reason the problem of trust cannot be fully handed over to prediction and incentive without changing its object: to model the promisor merely as a mechanism to be managed is to have stopped speaking of promising at all. The wager that the other will keep faith is, in part, a wager on his freedom—and this, as §5 will develop, is not a softness in the account but the very opening in which everything the paper most wants to say becomes possible.

3.6 The thesis: force is not the cause of trust

The dimensions assembled converge on a single thesis, which the rest of the paper exists to develop, to ground, and to apply. It can be stated plainly. Even where the coercive guarantee of the Other is fully present, trust under modern conditions has broadly declined; therefore force is not the cause of trust. If the sword were the cause of the expectation, then where the sword stands ready the expectation should stand firm; but the symptomatology shows trust eroding precisely where the apparatus of compulsion is most fully elaborated, and the diagnosis of the overload of justice shows why: the sword secures expectation only by signalling, and a signal whose basis is known to have failed no longer signals, however sharp the blade remains. Force, then, is at most a contingent and now-failing device for the production of an expectation that is the real thing; it is not that expectation’s cause.

The thesis turns on a distinction the rest of the paper will hold to throughout: between two senses of cause. There is the mechanical cause, which operates as one billiard ball operates upon another, by the transmission of a compulsion that leaves the affected body no part to play; and there is the inferential cause, which operates by giving a mind a ground on which to form an expectation, an expectation the mind itself produces and could, being free, have withheld. Force, where it works, works mechanically, or aspires to: it would move the promisor as a body is moved. But trust is not a body’s motion; it is a mind’s expectation, formed inferentially, on grounds. To say that force is not the cause of trust is to say that the mechanical cause is not the cause of the inferential effect—that the production of an expectation in a free mind belongs to a different order of causation than the compulsion of a body, and cannot be accomplished by it. The whole positive account of the paper follows from taking this distinction seriously: if trust is an inferential effect, then the question of how it is generated is the question of how the right grounds are supplied to the mind that forms it—and force, it turns out, is neither the only such ground nor, any longer, a working one.

With this the diagnosis is complete. The symptom was a gap, recurring across every scale, between the apparatus of assurance and the trust it failed to secure. The diagnosis is that the apparatus failed because it was built on a mistake about its own working—the mistake of taking force, the mechanical cause, for the cause of an effect that is inferential and free—and that the crisis of the present is the coming-due of that mistake, as the one device the mistake had relied upon, enforcement, fails on its own terms and leaves the credibility it had stood in for without support. What remains is to show, positively, how the inferential effect is in fact produced: by what mechanism a free mind, given the right grounds, comes to expect that a promise will be kept. To that the paper now turns, beginning where the failed device itself began, in the philosophy of law.

4. The Jurisprudential Core: The Migration of the Anchor of Validity

The diagnosis ended by sending the inquiry to the philosophy of law, and for a definite reason. Law is the institution modernity charged with enforceability; it is where the device that the diagnosis found failing—force as the producer of expectation—was housed, theorised, and refined. If that device is not, after all, the cause of obligation, then the law’s own reflection upon itself should bear the marks of the discovery, and the history of legal philosophy should record, in its internal development, a movement away from the very ground modernity had assigned it. This section shows that it does. It is not a survey. It reads the major positions in the philosophy of law along a single axis—the location of the anchor of validity, the place at which each position lodges the force by which an obligation obliges—and it finds, reading them so, not a set of competing doctrines to be weighed but a directional movement to be traced: the anchor migrates, across a century of the law’s reflection upon itself, from force toward trust. The section follows that migration through eight positions, and arranges them at its close in a single view.

A word on the relation to the foundational paper, which treated several of these same thinkers along a different axis, that of structure against generation, the Symbolic against the Real. The present reading is not a repetition but a re-cutting of the same material along a more focused axis: not what kind of thing law is, but where the force of its obligation is anchored. The two readings are consistent and complementary; the present one is the one the problem of trust requires.

4.1 The anchor in force: Austin and Hobbes

The movement begins where modernity’s assignment is most undisguised, in the command theory of John Austin. For Austin, law is the command of a sovereign, backed by the threat of a sanction, habitually obeyed; and a legal obligation is, by definition, the standing liability to the sanction—to be obliged is to be exposed to the probability of an evil at the hands of one with the power and the will to inflict it. The anchor of validity here is welded to mechanical cause. The obligation is the threatened pain; there is, in the account, no remainder, nothing in being bound over and above being liable to be hurt. Whatever else the command theory does, it states with unmatched clarity the position the rest of the chapter will move away from: that the force of an obligation is the force of a threatened consequence, and nothing else.

Behind Austin stands his metaphysical ancestor, Hobbes, whose famous sentence is the chapter’s true point of departure: covenants without the sword are but words, and of no strength to secure a man at all. The sentence is usually read as a worldly observation about the weakness of unbacked promises. It is better read, for the purposes of this paper, as the statement of an anthropology from which the impossibility of trust follows by necessity. Hobbes’s natural man, in the war of all against all, has no ground on which to expect his fellow’s forbearance, and so trust is not merely scarce in the state of nature but logically impossible: there is nothing from which the expectation could be formed, and the sovereign must therefore be erected to inject, from outside, by the threat of overwhelming force, the expectation the parties cannot generate between themselves.

It is exactly here that the chapter’s reading first bites. Austin and Hobbes take the limiting case—the case in which trust has wholly collapsed and only fear remains—and mistake it for the whole. The Hobbesian state of nature is the crisis of the symbol carried to its extremity, a world in which no symbol can produce expectation because none can be relied upon; and the Hobbesian remedy, the sword, is precisely the substitution of mechanical cause for the inferential cause that has failed. The command theory, then, is not false so much as it is the theory of a pathological limit. Its model has nothing to say where trust is present, because there force is redundant; it becomes visible, and seems complete, only where trust has collapsed and force is all that is left. And this is why it speaks so directly to the present: the contemporary overload of justice (§3) is the world sliding toward the Hobbesian condition—trust thinning until the sword is asked to do everything—at the very moment when, as the symptomatology showed, the sword can no longer fall. The command theory describes the destination toward which the failure of trust tends; and the failure of the sword to fall, there, is the command theory’s own solution failing on its own terms.

4.2 The anchor moves toward acceptance: Hart’s internal point of view

The decisive turn, and the hinge of the whole chapter, is Hart’s. His critique of Austin is the moment at which the anchor of validity is, for the first time within legal positivism itself, pried loose from mechanical cause and moved toward something else. Two of his objections do the work. The first is the distinction between being obliged and having an obligation. The gunman obliges me to hand over my money; I am not under an obligation to do so. Austin’s theory, which defines obligation as liability to a sanction, can capture only the first—the being-compelled—and is constitutively unable to capture the second, the normativity by which a rule gives me a reason and not merely a motive. The second objection is that whole classes of law—the power-conferring rules by which one makes a will, forms a contract, marries—are not commands backed by threats at all, but enabling structures that confer the capacity to alter one’s normative situation, and that the command theory simply cannot accommodate.

From these objections Hart draws the concept on which everything turns: the internal point of view. A rule exists, as a rule and not as a mere observed regularity of compelled behaviour, where there is a group who accept it from the inside, who take it as a standard for their own conduct and as a ground for criticising deviation in others—who use the rule, in their reasoning and their appraisals, rather than merely conforming to it under threat. The internal point of view is a normative attitude, not a prediction; and it stands in sharp contrast, as §4.5 will show, to the external, predictive stance of one who asks only what will be done to him if he is caught. This is the jurisprudential moment at which the anchor of validity is moved from mechanical cause toward the normative—toward an acceptance that gives reasons, the very thing the inferential cause of the diagnosis traffics in.

The point reaches its depth in Hart’s account of the rule of recognition, the ultimate rule by which a legal system’s other rules are identified as valid. What grounds the rule of recognition itself? Hart’s answer is that nothing grounds it from above; it rests upon nothing but the convergent practice of the officials who accept and use it. Its validity is not conferred by a higher rule, and is not secured by any sanction; its existence simply is the fact that it is accepted and acted upon. Here Hart arrives, without naming it so, at the position closest to this paper’s: the ultimate ground of a legal order’s validity is not a sword but a collective, self-sustaining acceptance—a convergence of the internal point of view across the body of officials, which holds itself up by being shared. This is, in the vocabulary of §6, the institutional-scale form of the structural channel: a shared prior, sustained by the very fact of being shared, that needs no external pump to hold it in place.

Honesty requires that Hart’s own difficulty be marked, for it is the difficulty that hands the argument to Fuller. Is the acceptance in which the rule of recognition consists a matter of fact or a matter of norm? Hart wishes it to be a convergent social fact; but a bare factual convergence—that officials do, as it happens, behave alike—seems unable to generate the genuine obligation, the ought, that a rule of recognition is supposed to carry. This is the opening through which Dworkin pressed his critique, that law cannot be a matter of convergent practice alone but must involve an interpretive, value-laden commitment. The paper does not adjudicate that dispute here; it notes only that the difficulty points beyond bare acceptance toward something acceptance alone does not supply, and that Fuller named what that something is.

4.3 The anchor in reciprocity: Fuller’s inner morality of law

Fuller supplies what bare acceptance lacks. His claim is that law possesses an inner morality, a set of eight principles—that rules be general, promulgated, prospective, clear, non-contradictory, possible to obey, stable over time, and administered congruently with their announced terms—which are not external moral constraints upon law but conditions of its very existence as law. The question Fuller’s critics pressed is why these should be called a morality rather than a mere technique of effective governance, a craftsman’s know-how for making rules that work. His answer, and the point at which his account becomes this paper’s, is that the eight principles together constitute an implicit promise from the lawgiver to the governed, and that the eighth—congruence, the requirement that officials actually administer the law as announced—is the heart of the matter, because it makes obedience a reciprocity. The citizen’s fidelity to law is a response to the state’s fidelity to its own announced rules; each side’s keeping of faith is conditioned upon the other’s. Strip the reciprocity away—let the state announce one rule and apply another—and the obligation to obey does not merely weaken but loses its ground, because the relation that grounded it was a mutual keeping of faith and one party has defected from it.

This reciprocity is trust under its proper name in the law. When Fuller says that law is a collaborative enterprise between the government and the citizen, an enterprise that fails not when a rule is broken but when the bond of reciprocity that constitutes it is betrayed, he has named, in the jurisprudential register, the very thing this paper is about: the anchor of validity lodged in a mutually relied-upon undertaking. The long debate between Fuller and Hart over whether law and morality are necessarily connected can, within this paper’s frame, be re-described and, in a sense, settled in Fuller’s favour—not on Fuller’s own grounds but on the paper’s: the anchor of validity, reciprocity, is itself a normative and relational thing, a keeping of faith, and not a neutral fact about convergent behaviour; so the validity of law cannot after all be specified without reference to the relation of trust in which it is anchored. Fuller’s inner morality is, in this paper’s terms, the institutional-scale form of all three channels of §6 at once—a structure that is also a self-commitment and that invites a wager—with the anchor lodged in a doubly relied-upon promise.

4.4 The anchor in procedure and discourse: Habermas and Rawls

A briefer station continues the movement Fuller began, by asking how legitimate norms are produced rather than where validity finally rests. Habermas locates the legitimacy of law in the discourse through which it is generated: norms are valid insofar as they could meet with the assent of all affected, in a communicative process governed by reason rather than force, and the legal-political system depends for its legitimacy upon a lifeworld of communicative practice that the system itself cannot produce and can only deplete. The structure of this dependence is worth marking, for it is isomorphic with a dependence this paper will meet again: just as the Habermasian system draws its legitimacy from a lifeworld it cannot generate from within itself, so the likelihood-channel of trust, in §6, draws upon a real basis that the symbolic cannot manufacture. Rawls, with the device of a procedure—the original position—whose fairness confers legitimacy upon what issues from it, belongs to the same family: a family that anchors validity in the manner of its legitimate production. This is structurally the same kind of question as the paper’s own—how is something legitimately produced?—asked of legitimacy where the paper asks it of trust; and noting the kinship, the chapter returns to its main line.

Against the movement toward acceptance and reciprocity, two positions hold the anchor down at force, and they must be given their due, for each marks a permanent coordinate the paper will need. The first is Holmes’s. His advice to view the law through the eyes of the bad man—the man who cares nothing for the morality of the law and wishes only to know what the courts will in fact do to him, so that he may calculate—is the purest statement of the external, predictive stance. For the bad man, law just is a prediction of official behaviour, and obligation just is the calculated probability of a consequence; he is the perfect evidentialist, the subject of pure mechanical cause, who recognises no internal point of view because he occupies none.

The paper’s treatment of Holmes is twofold. On the one hand, the bad man is real and his stance is irreducible: he is the exact portrait of the evidentialist trust-grammar that §10 will have to take seriously, the subject for whom only the predictable consequence counts and for whom sincerity is no evidence. To pretend he does not exist, or that his stance is simply a moral failing, is to disarm the paper before its hardest case. On the other hand, his limit is precise and must be stated. To model every subject as the bad man—to take the external predictive stance as the whole truth about legal subjects—is to erase the internal point of view a priori, and with it the possibility of meeting the other as a free subject who binds himself rather than as an object whose conduct is to be predicted and managed. The bad-man theory, generalised, is Kant’s worry (§3.5) realised in jurisprudence: it reduces the person to an object determined by consequences. It is true of those who have made themselves bad men, and it is a standing possibility for anyone; it is not the truth about the human as such, and a theory that takes it for that truth has decided in advance against everything §5 will affirm. The sociological extension of realism—the living law of Ehrlich, the law that actually governs conduct as against the law on the books—is noted here as a related and more capacious development, but the bad man is the figure that matters for the paper’s argument.

4.6 The second counter-coordinate: Kelsen’s pure theory

The second position that holds the anchor away from trust does so in the opposite direction from Holmes—not downward into predicted fact but upward into pure norm. Kelsen’s pure theory derives the validity of each legal norm from a higher norm that authorises it, in an ascending chain that must terminate, on pain of regress, in a presupposed basic norm, the Grundnorm, which is not itself posited by any higher authority but assumed as the transcendental-logical condition of regarding the whole order as valid. The Grundnorm is a presupposition, bracketing both morality and fact, a purely normative postulate at the apex of a purely normative structure.

Set beside Hart, the contrast is exact and instructive. Hart’s rule of recognition is an accepted social fact, something officials actually do; Kelsen’s Grundnorm is a presupposed symbolic posit, something one must assume. And the diagnosis the paper offers of Kelsen follows immediately: Kelsen lodges the anchor of validity in the purest possible point of the Symbolic, a posit suspended from nothing, held up by nothing but the decision to presuppose it. This is the most elegant of the positions and, by the paper’s lights, the most fragile under the conditions the symptomatology described. For a foundation that consists in a pure symbolic presupposition is exactly what the crisis of the symbol dissolves: let the presupposition cease to be made—let people stop assuming the basic norm—and the whole towering normative structure, having no other support, hangs in the void. Kelsen is the limiting case of an anchor lodged in the Symbolic alone, and the crisis of trust is precisely the condition under which an anchor so lodged goes to zero (§3.1).

4.7 The concrete instrument: from consideration to reliance

The migration of the anchor is not only a movement in high theory; the law records it, with documentary precision, in the technical doctrines of contract, and these are the most directly relevant of all, for they bear on exactly the instrument from which the paper set out, the bare and unenforceable promise. The orthodox doctrine of consideration holds that a bare promise—a promise for which nothing is given in exchange—is not enforceable; for a promise to bind, there must be a bargained-for exchange, a quid pro quo. The bare promise, the gratuitous undertaking, the memorandum that records an intention without exchanging anything—this is the precise legal situation of the betrothal document and of the memorandum of understanding alike: a promise the orthodox doctrine declines to enforce because nothing was given for it.

But the law developed an exception, and the exception is, for this paper, a quiet vindication from within the law itself. Under the doctrine of promissory estoppel, a promise unsupported by consideration may nonetheless be enforced where the promisee has reasonably relied upon it, has acted to his detriment in that reliance, and would suffer injustice if the promise were not honoured. The ground of the binding force here is neither consideration nor any threat of sanction: it is reliance itself. Because the promisee trusted the promise and made himself vulnerable in that trust, the promise acquires a binding force it did not have as a bare promise—a force generated by the trust placed in it, not by anything exchanged for it or any sword behind it. This is the mirror, in the law of contract, of the self-committing channel of §6: the promisee’s reliance is a placing of oneself in vulnerability, and the law responds to that vulnerability by generating an obligation to protect it. The associated measure, the reliance interest—Fuller and Perdue’s classic identification of the interest in being protected against the losses one suffers through having relied, as distinct from the expectation and restitution interests—is the law’s own name for the harm of a trust betrayed.

The significance is large and should be stated plainly. At its own technical edge, in the doctrine of promissory estoppel, the law has conceded that the true ground of a promise’s binding force is reliance—is trust—and that enforcement is only the after-the-fact protection of a trust that was already, in itself and before any enforcement, the thing that bound. Promissory estoppel is the law’s spontaneous, internal corroboration of this paper’s central thesis: not a doctrine the paper imports to support its claim, but a place where the law, reasoning about its own most basic instrument, arrived at the claim by itself.

4.8 The migration in one view

The eight positions, read along the single axis of the anchor of validity, compose one directional movement. The law’s own reflection upon itself has carried the anchor from force (Austin, Hobbes), through acceptance (Hart), to the reciprocal relation of reliance (Fuller, and the doctrine of promissory estoppel), with the predictive and the purely normative held at the two counter-coordinates (Holmes, Kelsen) that mark the limits the movement leaves behind. The paper does not impose this movement from outside; it traces an undercurrent already present in the law and connects it to the diagnosis of the symbol’s crisis and, in §6, to the mechanism of trust—giving the movement the full causal grounding the law itself, lacking a mechanistic vocabulary, could mark but not explain. The consequence is that this chapter is not mere background but an independent pillar of the argument: it establishes, from within the philosophy of law, that the anchor of validity is migrating from force toward trust, so that the metaphysics of §5 and the mechanism of §6 are not free-standing philosophical or scientific assertions but the grounding explanation of a movement the law has already, on its own, undergone. Three lines—the law’s internal undercurrent, the metaphysics of inferential cause, and the predictive-coding mechanism—converge on a single proposition: that it is trust, and not force, that is the true cause of obligation.

The whole may be set in one view. The table reads the positions along the axis of the chapter: where each lodges the anchor of validity, which sense of cause it thereby invokes, and what it contributes to, or marks the limit of, the paper’s thesis.

Position Anchor of validity Sense of cause Role in the thesis
Austin, Hobbes the threatened sanction; the sword mechanical states the position to be moved from; theory of a pathological limit
Hart convergent acceptance; the internal point of view inferential (normative) the hinge: anchor pried loose from force toward acceptance
Fuller reciprocity; mutual keeping of faith inferential (relational) names trust under its legal name; congruence as implicit promise
Habermas, Rawls legitimate procedure; discourse inferential (procedural) isomorphic question: how is legitimacy produced
Holmes, realism predicted official conduct mechanical (predictive) counter-coordinate; portrait of the evidentialist; its limit is Kant’s worry
Kelsen the presupposed basic norm purely symbolic posit counter-coordinate; anchor in the Symbolic alone; goes to zero in the crisis
Promissory estoppel reliance itself inferential (relational) the law’s own corroboration: reliance, not force, binds

Read down the column of causes, the table shows the chapter’s claim in a single glance: the entries that hold the anchor at force are entries of mechanical cause, and they sit at the beginning of the movement and at its abandoned limits; the entries that carry the anchor toward trust are entries of inferential cause, and they are where the law’s reflection has been travelling. The migration of the anchor and the distinction between the two senses of cause are the same movement seen twice. What that inferential cause is, metaphysically—why it is the opening in which freedom itself becomes possible—is the question §5 now takes up; how it works, mechanically, is the question of §6.

5. The Inferential Cause and the Opening for Freedom

The jurisprudential chapter ended with a distinction it had traced through eight positions but not yet examined: between the mechanical cause, in which the anchor of validity sits at the beginning of the law’s movement and at its abandoned limits, and the inferential cause, toward which that movement has been travelling. The distinction was drawn, in the diagnosis, as a contrast between two ways a result might be produced—the compulsion of a body and the grounding of a mind’s expectation. This section asks what the inferential cause is, and answers that it is not a weaker species of the mechanical but a different kind altogether, and that the difference is the very opening in which freedom, and with it the moral, becomes possible. The section is short, but it is the load-bearing arch beneath everything the mechanism, the justice criterion, and the praxis will build: it establishes that to generate trust is to address a free being as free, and that the promise which carries no force is, for exactly that reason, not the defective case but the proper one.

5.1 Two kinds of cause, not two degrees of one

It is tempting to hear the contrast between mechanical and inferential cause as a contrast of strength—as though the mechanical cause were the reliable kind, which compels its effect outright, and the inferential cause a weaker kind, which merely makes its effect more probable. On that hearing, force would be the gold standard of causation and trust a degraded approximation to it, and the whole movement the jurisprudence traced would be a movement from a strong ground to a weak one, a story of decline. This is the hearing the paper must refuse, and the refusal is the hinge of the section.

The mechanical cause and the inferential cause are not two degrees of one kind of causation; they are two kinds. The mechanical cause operates upon its effect from outside and leaves the effect no part to play: the struck ball does not consent to move, contributes nothing to its own motion, could not have done otherwise given the impact. Its paradigm is the transfer of momentum, and its mark is that the affected thing is passive throughout—the cause does all the work, and the effect is the mere registration of it. The inferential cause operates altogether differently. It does not push; it gives a ground. It supplies a mind with something on the basis of which that mind may form an expectation—an expectation the mind itself forms, by its own activity, and could, being what it is, have formed otherwise or withheld. The ground does not compel the expectation as the impact compels the motion; it furnishes a reason, and a reason is precisely what a free mind takes up rather than undergoes. Where the mechanical effect is the passive registration of a force, the inferential effect is the active production of an expectation by the one in whom it arises.

This is why the inferential cause is not a weaker mechanical cause but a stronger—or rather a different and higher—thing. It is the only kind of causation that can have a free being for its proper subject, because it is the only kind that addresses that being as free: as one who forms his own expectations on grounds, rather than one whose states are pushed into him from without. To produce trust by the inferential cause is to treat the one in whom the trust arises as a mind that judges, not as a mechanism that registers. And the mechanical cause, applied to a mind, does not improve upon this; it degrades it, for to compel an expectation rather than to ground it is to treat the mind, in just that respect, as if it were not a mind at all.

5.2 Trust addresses the other as free

The same distinction, looked at from the side of the one who trusts rather than the one who is trusted, is the distinction between two stances one may take toward another person, and it is here that Kant’s worry from the diagnosis (§3.5) receives its positive answer. One may take toward another the stance of prediction-and-management: treat his future conduct as an outcome to be forecast from evidence and secured by incentive, arrange the consequences so that the conduct one wants is the conduct it is in his interest to produce, and rely upon him exactly as one relies upon any well-understood part of the causal order. This is the stance of Holmes’s observer toward the bad man (§4.5), and it is, within its proper domain, entirely legitimate; one had better be able to predict and to manage. But it is a stance that takes its object as an object—as a system whose behaviour is determined by conditions one may arrange—and to take only this stance toward a person, to model him exhaustively as such a system, is to have ceased, in Kant’s terms, to treat him as an end, and so to have ceased to relate to him as a person at all.

The other stance is the stance of trust proper, and it is the stance of one free being toward another. To trust a person, in the full sense the promise calls for, is not to predict his conduct as one predicts the weather but to rely upon his freely binding himself—to address him as one who can commit himself to a future and can keep faith with that commitment because he has bound himself to it, not because the consequences have been arranged to leave him no alternative. The one who makes a genuine promise is not forecasting his own future behaviour from the inside, as one might forecast a habit; he is committing himself to it, performing the distinctively free act of placing himself under an obligation he could violate and undertaking that he will not. And the one who trusts the promise is wagering, in part, upon that freedom: relying not upon a mechanism that cannot fail but upon a free being who could fail and undertakes not to. The wager on the other’s keeping of faith is, irreducibly, a wager on his freedom—and this, which the diagnosis marked as the point at which prediction-and-management changes its object and stops speaking of promising at all, is now seen to be not a softness in the account of trust but its very condition. There is no trusting a being one regards as a mere mechanism; there is only predicting it. Trust begins exactly where prediction is refused as the whole truth, and the other is met as one who is free to keep or to break faith and is relied upon to keep it.

5.3 A word on the cognitive basis

That this is a metaphysical and ethical claim does not make it a claim against the natural order, and a word is owed, in anticipation of §6, on its cognitive basis, lest the inferential cause seem to require a faculty outside nature. It does not. The forming of expectations on grounds is not an exotic operation a mind performs on rare and solemn occasions; it is the ordinary and incessant business of a brain, which is, on the account the next section develops, a predictive organ before it is anything else, continuously generating expectations of what is to come and revising them against what arrives. Trust, on that account, is not a special faculty superadded to cognition but a particular case of cognition’s standing activity: a high-confidence forward expectation about the conduct of another, formed on grounds and held open to revision. The inferential cause has, that is, a perfectly natural implementation; what distinguishes it from the mechanical cause is not that it floats free of the causal order but that, within that order, it operates by furnishing grounds to a system that forms its own expectations, rather than by compelling a system that merely registers what is done to it. The metaphysics of freedom and the neuroscience of prediction are not at odds; the one is the significance of what the other describes. The section leaves the mechanism to its own chapter and returns to the consequence.

5.4 Why the non-coercive promise is the proper case

The consequence is the one the Prelude felt before it could say it, and it can now be said. The promise that carries no force is not the defective limiting case of promising, to be apologised for against the enforceable contract as the standard; it is the proper case, of which the enforceable contract is the derivative. For if trust addresses the other as free, and if to produce trust is to ground a free mind’s expectation rather than to compel a mechanism’s response, then the promise that relies on no sword is the promise that does this in its purity: it asks to be believed on no ground but the promisor’s free commitment and the promisee’s free wager upon it, and so it is the form in which the inferential cause operates unmixed. The enforceable promise, by contrast, mixes into this pure case an element of the mechanical: it offers, alongside the free commitment, a threatened consequence, and to exactly that extent it addresses the promisee not as a free mind forming an expectation on the ground of trust but as a calculator forming a prediction on the ground of incentive. The sword does not strengthen the promise as a promise; it supplements the promise with something that is not a promise at all, a managed consequence, and the supplement does its work, where it works, by treating the parties in the very manner that trust refuses.

This is not yet to say—and the paper will be at pains, later, not to say—that the sword should be dispensed with, or that the enforceable contract is in every respect inferior to the bare vow; the protection of the vulnerable, as §11 will insist, has a continuing and indispensable need of force. It is to say something narrower and deeper: that in respect of what makes a promise a promise—the addressing of a free being as free, the grounding rather than the compelling of an expectation—the non-coercive promise is the realised case and the coercive one the diluted. The want of force that the Prelude felt as the condition of the vow’s being what it was is here given its reason. A bond one could be forced to keep is, just insofar as one could be forced, not yet fully a bond between free beings; and the vow that asks for no sword is the truer not despite carrying no force but because, carrying none, it leaves the whole of the work to freedom—to the promisor’s free binding of himself and the promisee’s free trust in it. How that work is in fact done—by what mechanism a free mind, addressed as free and given the right grounds, comes to expect that the promise will be kept—is the question the paper has been clearing the ground to ask, and it is the question to which it now turns.

6. The Mechanism of Trust

Everything to this point has cleared the ground for a single question and supplied the terms in which it can be asked. The symptomatology displayed a gap between the apparatus of assurance and the trust it failed to secure; the diagnosis traced the gap to a mistake about causation, the taking of the mechanical cause for the cause of an inferential effect; the jurisprudence showed the law’s own anchor migrating, across a century, from force toward trust; and the preceding section established that the inferential cause is a different kind of thing from the mechanical, the kind that addresses a free being as free. What none of these has yet supplied is the positive account: by what mechanism a free mind, addressed as free and given the right grounds, comes to expect that a promise about a non-existent future, carrying no guarantee of consequence, will be kept. This section supplies it. It is the trunk of the paper, of which the chapters before were the roots and the chapters after are the branches.

The problem the mechanism must solve has, recall, a shape that looks impossible. Trust is directed at a future that does not yet exist and cannot be verified; the symbol that asks for trust carries no force and so cannot secure the future by threatening a consequence. We are asked to explain how a present expectation, real and often well-founded, is produced by a symbol about what does not exist and under no guarantee—a generation, very nearly, from nothing. The dissolution of the apparent impossibility is the section’s first move, and it is made by reaching for the one vocabulary in the paper’s possession that is native to the forming of expectations about what has not yet arrived: the predictive account of the mind.

6.1 Predictive coding and the dissolution of the paradox

The brain, on the predictive account that has become central to the cognitive sciences, is not in the first instance a device that registers the world and then responds to it; it is a device that predicts the world and corrects its predictions against what arrives. It carries a generative model of the causes of its sensory states, continually generates expectations of what is to come, compares them against incoming signals, and updates the model in proportion to the error—and in proportion, crucially, to the precision, the inverse variance, it assigns to that error and to its own predictions. Prediction, on this account, is not an occasional achievement but the standing activity of the organ; the mind is forward-directed by its nature, expecting before it perceives.

This dissolves the apparent paradox at a stroke. The objection was that trust is impossible because its object—the kept future—does not yet exist; how can a present state be produced by an absent fact? But the objection assumes that an expectation must be caused by the thing expected, and the predictive account shows this assumption false. An expectation is not caused by its object; it is produced, now, by the generative model, as the model’s forward prediction, in advance of and independent of the object’s arrival. Forward-directed expectation under uncertainty is not a special feat trust must somehow accomplish against the grain of cognition; it is what cognition is incessantly doing anyway. Trust requires no time machine, no causation running backward from a future that does not exist. It requires only what the brain does in every waking moment: the assignment, by a generative model, of a high-precision prediction—here, the prediction that this person will keep this promise. Trust is the high precision a receiver’s generative model assigns to the prediction that the promise will be kept. The expectation of legal possibility that the diagnosis spoke of is, at the level of mechanism, exactly this: a precision-weighted forward prediction, a quantity the brain natively computes, and not a metaphor borrowed from the law.

The question of how trust is generated thus becomes a precise question: by what inputs does a receiver’s generative model come to assign high precision to the prediction that a promise will be kept? The paper’s answer is that there are three such inputs, three channels, and that they correspond—not by coincidence but because the registers are the registers of one subject—to the three registers the series has worked with throughout.

6.2 The three channels

The three channels are distinct in what they supply to the generative model, and the paper takes them in the order of the registers: the structural, lodged in the symbolic; the self-committing, lodged in the real; and the wagered, lodged in the imaginary and the subject who wagers.

6.2.1 The structural channel: the prior

The first channel supplies a prior. A promise is not received in a vacuum; it is received with a structure, and the structure of the promise shapes the receiver’s prior expectation before any evidence about this particular promisor is weighed. A vow with the form of a closed and self-completing commitment—one that, in the language of §7, returns upon itself without leaving a debt to be discharged elsewhere—presents to the receiver’s model a structural prior favourable to its being kept, in a way that a loose, hedged, escape-laden form does not. This is the cognitive function of what the series has called the poetic structure of a commitment: the form is not ornament but information, a shaper of the prior the receiver brings to bear. At the scale of institutions this is the channel of Hart’s convergent acceptance and the rule of recognition (§4.2): a shared structural prior, sustained by being shared, against which each new undertaking is read. The structural channel is lodged in the symbolic, for it works through the form of the symbol; and it is, taken alone, the most forgeable of the three, since a form can be imitated.

6.2.2 The self-committing channel: the high-fidelity likelihood

The second channel supplies a likelihood—and, where it works, a high-fidelity one. Here the promisor does not merely present a form; he does something. He places himself, really and observably, in a position of vulnerability from which retreat is costly or foreclosed: he binds himself in a manner that will hurt him to break, gives a hostage, burns a bridge, commits himself irrevocably and in the sight of others. And the receiver’s generative model reads this for what, statistically, it is. A signal that is costly and hard to reverse carries a high precision-weight, because it is one a promisor who did not intend to keep faith would be very unlikely to send: the cost of sending it, for the insincere, is prohibitive, and so its being sent is strong evidence of sincerity. This is the mechanism beneath the red silk, the public setting, the burning of one’s boats—the irrevocable and costly self-binding the Prelude began with. Such acts are not arguments; they do not persuade by inference from premises. They furnish the receiver’s model with a high-fidelity, hard-to-forge likelihood signal, evidence of a kind the body reads before the discourse does, that the one who could so bind himself must mean it.

The paper takes up here, and gives a mechanistic basis to, the costly-signalling principle familiar from the theory of signalling in biology and economics: that a signal too expensive for a defector to afford is, for that reason, credible. What predictive coding adds is the implementation—the costly, irrevocable signal is precisely one to which a generative model rationally assigns high precision—and a deeper observation the paper will not exhaust here but must mark. The burning of the bridge has been described, just now, in its aspect as a signal: as something done to be read by another, furnishing his model with evidence. But this is only one of its two faces, and not the deeper. There is a second sense in which one burns a bridge that has nothing to do with being read by anyone—in which one forecloses one’s own retreat not to send a message but to constitute, for oneself, a law: I have decreed it so, and so that road no longer exists for me. This deeper sense, in which the irrevocable self-binding is an act of self-legislation rather than of signalling, is the matter of §12, and in particular of its account of relational divinity; it is marked here, at the mechanism, so that the reader who reaches it will recognise that the trunk was pointing toward it all along. For the present chapter the self-committing channel is treated in its signalling aspect: as the high-fidelity likelihood lodged in the real, the channel through which a real act of self-exposure, received by another present to read it, supplies the model with its least forgeable evidence.

6.2.3 The wagered channel: the prior leap of active inference

The third channel is the most easily mistaken for irrationality and is, properly understood, the deepest. It concerns not what the promisor supplies but what the receiver does: the receiver, in trusting, makes a leap that precedes the evidence. He assigns, in advance of sufficient warrant, a high-precision prior to the prediction that the other will keep faith—he ventures the trust—and only then, through the continued interaction the venture makes possible, gathers the error signals by which the model is confirmed or corrected. This looks like a failure of rationality: the trust is given before the evidence that would justify it. But on the active-inference reading it is not a failure of rationality at all. An agent does not merely passively update a model against incoming data; it acts so as to reduce its own uncertainty over the longer run, and there are states of knowledge attainable only by first acting under uncertainty—only by venturing. Trust is such a venture. One cannot gather the evidence that would warrant trusting a person without first entering the kind of interaction that only trusting him makes possible; the evidence lies on the far side of a threshold one can cross only by trusting in advance of it. The leap of trust is thus the rational opening move of an agent that reduces uncertainty by acting—a prior wager that is the precondition of acquiring the very evidence that will vindicate or correct it. This channel is lodged in the imaginary and in the wagering subject, for its source is not in the symbol’s form nor in the promisor’s act but in the receiver’s free venturing; and it is here, in the wager that the other will prove free and faithful, that the freedom of §5 re-enters the mechanism.

6.3 The multiplicative coupling

The three channels are not three alternative routes to trust, of which a promisor might supply one and be believed. They are coupled, and coupled multiplicatively—as the factors of a single Bayesian product are coupled, in which the structural prior, the self-committing likelihood, and the wagered prior combine into the posterior precision of the prediction that the promise will be kept. The point of multiplicative rather than additive coupling is exact and it is the heart of the section: in a product, if any factor goes to zero the whole product goes to zero, and no abundance in the other factors can compensate. A perfect form with no real self-commitment behind it; a costly act with no wager that the other is free to be faithful; a leap of trust met by neither form nor self-binding—each is a product with a vanishing factor, and each fails as a ground of well-founded trust.

This is why genuine trust is robust where any single channel would be fragile. A single channel can be forged: a form can be imitated, a costly-looking act can be staged, a credulous wager can be exploited. But the three channels are hard to forge together, because they are answerable to one another—the structural form must be borne out by a real self-commitment, which must in turn be the kind of thing a free agent would do and would invite a wager upon—and their mutual consistency, the agreement of form with act with wager, is far harder to counterfeit than any one of them alone. The strength of genuine trust is not the strength of its strongest channel but the improbability of all three being made to agree falsely. This is what coupling means here, and why it is the source of trust that is not merely felt but well-founded.

The mechanism of trust. The expectation that a non-coercive promise will be kept is the posterior precision a receiver’s generative model assigns to that prediction, formed by the multiplicative coupling of three inputs: a structural prior, supplied by the form of the commitment (the symbolic register); a high-fidelity likelihood, supplied by the promisor’s real, costly, irrevocable self-binding (the real register); and a prior wager, supplied by the receiver’s free venturing of trust in advance of the evidence (the imaginary register, and the wagering subject). Genuine trust is the state in which all three are present, real, and mutually consistent; and its well-foundedness rests on the improbability of all three being made to agree falsely.

6.4 The forged case, mechanically described

The mechanism furnishes, as a by-product, an exact characterisation of its own counterfeit, which §8 will take up as the matter of justice and which is set out here in its mechanical form. Forged trust is the state in which the receiver’s generative model is brought to assign high precision to the prediction that the promise will be kept without the real self-committing likelihood that alone would warrant it. The manipulator does precisely this: by working on the receiver’s model directly—by affective pressure, by the staging of forms, by rhetorical management—he induces a lowering of the receiver’s prediction error, a high-precision prediction, that is not supported by any real self-binding on his part. The precision is, as it were, injected from outside rather than earned by a genuine likelihood; and a precision so injected is a prediction unsupported by what it predicts, a confidence with nothing real beneath it.

The mechanical mark of manipulation can therefore be stated with some sharpness: to manipulate is to induce the other’s generative model to lower its prediction error without supplying the high-fidelity likelihood that would warrant the lowering. Such a state is unstable in a way genuine trust is not. Because no real self-commitment underlies it, the world will not in fact behave as the injected precision predicts; the error signals will accumulate, the model will be forced toward correction, and the divergence between the predicted and the actual will widen until the forged precision collapses—the process the series has elsewhere described, in its dynamical register, as the accumulation of deviation toward divergence under a non-adiabatic extraction. The forged case is thus not a stable alternative form of trust but a transient, a precision held up by pressure rather than by likelihood, destined for the collapse that the absence of a real ground guarantees.

6.5 The crisis of the symbol, in predictive-coding terms

The same vocabulary yields, finally, a precise rendering of the diagnosis of §3, and shows the crisis of the symbol to be not a moral decline but a rational adjustment. Consider the population of generative models that is a society, each learning, from its own commerce with symbols, the statistical relation between symbolic signals and real performance. As the cost of producing a convincing symbol falls toward zero and forged signals proliferate (the generativity paradox of §3.3), each model learns, correctly, that the correlation between symbolic testimony and real performance has weakened—that the signal has become, in fact, less predictive of the outcome. And the rational Bayesian response to a signal that has become less predictive is to lower the precision-weight assigned to it. Each model, updating as it should, down-weights symbolic testimony as such; and because the down-weighting is applied to the channel and not to the individual signal, it falls upon the genuine symbol along with the forged. The genuine vow, the true recommendation, the sincere assurance—each is met by a model that has rationally learned to trust symbols of its kind less, and so each fails to land, not because it is disbelieved in particular but because the channel it travels has been rationally discounted.

This is the crisis of the symbol given its mechanism. It is not that people have become cynical, or that values have decayed; it is that a population of well-functioning generative models, faced with a channel that forgery has made less reliable, has rationally collapsed the precision-weight it assigns to that channel—and that the retreat to the merely countable (§2.3) is the same population’s attempt to reconstruct a high-fidelity likelihood out of signals that are, at least, harder to forge. The crisis is a rational precision collapse at the scale of a population, and that is both why it is so intractable—for each model is only doing what it should—and why the remedy, as the rest of the paper develops it, cannot be a return to symbolic profession but must lie in the channels, the self-committing and the relational, that forgery cannot so easily saturate.

6.6 Connection to the computational programme of the foundational paper

It is worth marking, before the chapter closes, that the mechanism dovetails with the formal programme the foundational paper held open. That paper sketched, as a direction for future work, a computational study of justice as generative presence, in which justice would figure as an observable evaluated over a space of histories, formalised in the manner of a path integral. The predictive-coding mechanism of trust is a concrete realisation, for the case of trust, of exactly this form: trust is the precision a generative model assigns to a prediction evaluated over the space of the promisor’s possible future trajectories, an expectation taken over histories. The two formal programmes mesh—the observable-over-histories of the foundational paper and the precision-over-trajectories of the present mechanism are the same kind of object—and the meshing is noted here so that the formal continuity of the series is on the record, to be drawn upon rather than re-derived.

6.7 A discipline of honesty

The chapter must end where the series requires it to, with an express discipline upon the use it has made of the predictive account, lest the mechanism be taken for more than it is. Predictive coding has been used here as a mechanistic and structural model, not as a reductive claim about what trust ultimately is. The paper does not assert that trust simply is a precision-weighting and nothing more. What predictive coding supplies is a description at Marr’s computational and implementational levels—an account of what the system computes and how it might compute it—and such a description does not displace the others the paper has given and will give: the phenomenological description of the vulnerability of self-commitment, the jurisprudential description of force as a substitute for expectation, the ethical description of the difference between genuine and forged. These are descriptions of one object at different levels, and the predictive-coding account is one of them, not the truth beneath them all. It is not a meta-framework that absorbs the rest; it is one more irreducible position on a single object, and the polyphonic discipline of the series—that no framework govern the others—holds over it as over every other. The mechanism is offered, then, as the paper’s central contribution and not as its last word: a specification of how, at the level of the computing mind, the inferential cause does its work, which leaves standing, beside it, everything the other registers have to say about the same thing.

7. The Holonomy Interface

The mechanism of §6 was given in the vocabulary of prediction and precision. This section gives the same mechanism a second reading, in the geometric vocabulary the series has developed elsewhere—the vocabulary of holonomy, of the phase a system accumulates in traversing a closed path. The second reading is not a rival to the first but its complement: where the predictive account describes what the receiving mind computes, the geometric account describes the shape of the process by which a promise, held across time, either sustains the trust placed in it or drains it away. The two readings will be seen to converge, and their convergence prepares two later moments of the paper: the quantum-dynamical structure of relational trust (§9), for the geometric phase is, in its origin, a concept of quantum dynamics; and the inversion of value by which the non-binding vow proves the purer instrument (§12), for that inversion turns on exactly the distinction this section now draws.

A discipline must be stated at the outset, as it was for the mechanism. The geometric phase is employed here as a structural analogy, developed to the point of usefulness but not to full formal closure; the formal programme that would furnish the phase with a rigorous carrier, a definite fibre and connection, is carried by another paper of the series, to which the present section forward-refers rather than reproducing it. What is claimed here is that the structure of a promise sustained across time is the structure of a system transported around a closed loop, and that the distinction between a promise that generates trust and one that drains it is the distinction between a loop that accumulates a positive holonomy and one that accumulates none or a negative one. The claim is structural; its full formalisation lies elsewhere.

7.1 Trust as a quantity transported around a loop

Consider a promise not at the instant of its making but across the time of its keeping. A relation in which a vow has been made returns, again and again, to situations that resemble the situation of the vow: occasions on which the commitment is tested, recalled, re-enacted, or merely lived through—the daily, monthly, yearly recurrences in which what was promised is, or is not, borne out. The relation thus traces, across its time, a closed loop in a space of situations: it departs from the moment of the vow, passes through the varied circumstances of a shared life, and returns to situations that are, in the relevant respect, the same situation again. The question the geometric reading asks is what the relation brings back when it returns. Does it return to the situation of the vow unchanged, having merely gone around and come back? Or does it return having accumulated something—a deepening, a surplus, a phase that was not there at the start?

This is the question of holonomy. A system transported around a closed loop in a space with the right structure does not, in general, return to its starting state; it returns rotated, shifted, carrying a phase that records the path it took—the geometric phase, the anholonomy, of which the Berry phase and the Pancharatnam phase are the canonical instances. The phase is not a function of where the system ended—it ended where it began—but of the loop it traversed to get back: it is the trace, in the returned state, of the journey. And the sign and magnitude of that phase is exactly the distinction the series has used to tell a good cycle from a bad. A loop that returns its system unchanged has zero holonomy: it has gone around and come back to the same place with nothing to show for the going, the mere repetition the series has called the bad infinity. A loop that returns its system enriched, carrying a positive phase, has positive holonomy: it has returned to its root and yet come back as more, the spiral rather than the circle. And a loop that returns its system depleted, having leaked phase rather than gained it, has negative holonomy: the cycle of net extraction, the relation that winds down. The holonomy of the loop, the holonomy of the relation’s traversal, is the geometric measure of whether the promise, lived across time, deepens, repeats, or drains.

7.2 The non-coercive promise as positive, adiabatic holonomy

The non-coercive promise, kept, is the loop of positive holonomy, and it is so in the specific manner the geometric phase requires: adiabatically. The geometric phase is the phase accumulated by a system carried around its loop slowly enough that it is not driven by the transport but follows it, exchanging no energy with its surroundings, its evolution governed from within rather than pumped from without. This adiabatic condition is the geometric image of exactly what §5 called the inferential cause and §6 called the self-committing and wagered channels. The promise kept without a sword is kept adiabatically: its loop is closed not by an external compulsion driving the parties around it but by the parties’ own commitment, sustained from within, returning upon itself by the completeness of what each freely brings to it. Nothing external pumps the cycle; the parties are not driven around the loop by a threatened consequence; the closure is endogenous, and the phase it accumulates is, for that reason, a true geometric phase—a deepening generated by the loop’s own internal structure, the surplus the series has identified with the good cycle. The non-coercive promise is the adiabatic loop of positive holonomy: trust sustained and deepened by the relation’s own internal completeness, requiring no external pump to hold it closed.

This gives the geometric content of the claim of §5, that the promise without force is the proper case and not the defective one. In the geometric vocabulary, the want of an external pump is not a deficiency of the loop but the condition of its phase being geometric at all. A phase that is pumped from outside is not a geometric phase; it is a dynamical phase, driven by the energy fed in, and it is not the trace of the loop’s own structure but of the driving. The adiabatic loop—the one no external force drives—is the only one whose accumulated phase is purely the loop’s own; and so the promise that relies on no sword is the only one whose deepening is genuinely its own, generated from within rather than supplied from without. The geometric reading thus recovers, in its own terms, the result that the absence of force is the condition of the promise’s deepening being the promise’s own.

7.3 Debt and the coerced contract as the zero-holonomy line

The contrast that throws this into relief is the coerced contract, and behind it the structure of debt, which the geometric reading exposes as something quite different from a loop. A debt is not a cycle that returns upon itself and deepens; it is a line, drawn from a present that is mortgaged to a future that must discharge it. The debtor does not traverse a loop that brings back a surplus; he is held, by the threat of the sanction, on a track that runs from the incurring of the obligation to its forced discharge, and the track is held straight—prevented from closing into any loop of its own—by the external force that drives it. This is the geometric image of the coerced contract and of debt as its limit: a path held open and directed by an external pump, accumulating no geometric phase because it is not permitted to close upon itself, its motion not the endogenous traversal of a loop but the driven progress along a line.

Where such a path is forced into the appearance of a cycle—where a coerced relation is made to repeat—its holonomy is zero or negative, never the positive phase of the adiabatic loop. Zero, in the case of mere forced repetition: the relation goes around and comes back to the same place, the obligation discharged and re-incurred, with nothing brought back, because the closure was imposed from outside and the loop accumulated no phase of its own—the bad infinity of a cycle held shut by force. Negative, in the case of net extraction: where the external pump drives the loop so as to draw phase out of it, the relation winds down, depleted turn by turn, the structure the series has identified with the parasitic and the exploitative. The coerced relation cannot accumulate positive holonomy, because positive holonomy is the signature of an adiabatic, endogenously closed loop, and the coerced relation is precisely the one that is not adiabatic: it is pumped from without, and a pumped loop’s phase is not its own. The sword, on this reading, does not merely fail to deepen the bond; it forecloses the deepening, by substituting a driven line or a forced repetition for the free, adiabatic loop that alone could deepen.

7.4 The spectral gap and the 甘え problem

There is a further structure the geometric reading illuminates, which the series left open in its treatment of 甘え—of the sweet, regressive presuming-upon-another’s-indulgence whose ambivalence that discussion could mark but not resolve. The unresolved question there was where, in the space of a relation’s possible motions, the line falls between a presuming-upon-another that the relation can absorb and metabolise, returning it as deepened intimacy, and a presuming-upon-another that the relation cannot absorb, which accumulates as deviation and drives the cycle toward divergence. The geometric reading offers the missing structure, in the form of a spectral gap. A loop’s capacity to absorb a perturbation and return adiabatically to its closed traversal—rather than being driven by the perturbation onto a divergent path—depends on a gap in the spectrum of the dynamics governing the loop: a perturbation slower than the gap is followed adiabatically and metabolised into the loop’s own phase, while a perturbation faster than the gap drives the system across the gap, out of the adiabatic regime, onto a trajectory that no longer closes. The benign 甘え is the perturbation the loop can follow adiabatically, absorbed into a deepened holonomy; the corrosive 甘え is the perturbation that exceeds the gap, driving the relation non-adiabatically toward the accumulation of deviation and divergence that §6.4 described in the predictive register as the collapse of a forged precision. The spectral gap is the structure that was missing; it is offered here as the geometric correlate of the relation’s capacity to bear what is presumed upon it, and the point at which the present paper repays, in part, a debt the earlier discussion left standing.

7.5 The convergence of the two readings

The geometric reading and the predictive reading of §6 are, in the end, one account seen twice, and their convergence is worth stating plainly, for it is what licenses the use of both. The adiabatic, endogenously closed loop of positive holonomy is the geometric portrait of the relation whose trust is sustained by the three coupled channels from within, requiring no external pump—the genuine case. The driven line, and the forced loop of zero or negative holonomy, is the geometric portrait of the relation whose precision is injected from without rather than earned by a real likelihood—the coerced case and the forged case alike. What the predictive reading describes as a precision earned by a real self-committing likelihood versus a precision injected by external pressure, the geometric reading describes as a phase accumulated by an adiabatic loop versus a phase pumped into a driven path; and these are the same distinction, the distinction between a process governed from within and one driven from without, rendered in two vocabularies. That the two independent readings converge on the same line—between the endogenous and the pumped, the adiabatic and the driven, the genuine and the forged—is the strongest internal evidence the paper can offer that the line is real and not an artefact of either vocabulary.

This convergence carries forward in two directions the paper will take up. It points toward §9, where the recognition that the geometric phase is, in its origin, a concept of quantum dynamics—the Berry phase being the phase of an adiabatically transported quantum state—allows the geometric language used here to be seen for what it has been all along, a borrowing from the dynamics of which relational trust will there be shown to share the structure. And it points toward §12, where the distinction between the adiabatic loop, whose phase is its own, and the pumped path, whose phase is borrowed, becomes the geometric ground of the inversion by which the non-binding vow—the loop that relies on no external pump—proves the purer realisation of what a promise is, and the enforceable instrument, with its admixture of the external pump, the diluted one.

8. The Justice Criterion: Genuine and Forged Trust

The mechanism of §6 furnished, as a by-product, a mechanical characterisation of its own counterfeit: forged trust as a precision injected from without rather than earned by a real self-committing likelihood, and the geometric reading of §7 gave the same counterfeit its geometric form, as a phase pumped into a driven path rather than accumulated by an adiabatic loop. But the mechanical and the geometric characterisations are, by themselves, value-neutral. They tell us that the forged case is unstable—that a precision unsupported by likelihood will be overtaken by accumulating error, that a phase pumped from outside is not the loop’s own—but instability is not yet injustice, and a transient is not yet a wrong. This section makes the further claim: that the line between genuine and forged trust, which the mechanism drew as a line between the stable and the unstable, is also and irreducibly a line of justice, between a way of producing trust that honours the one in whom it is produced and a way that wrongs him. The claim is that these two lines—the mechanical and the ethical—coincide, and the section’s task is to say why they coincide without pretending that either is derivable from the other.

8.1 Why the ethical criterion is not derivable from the mechanism

It must be said first what the section does not claim, for the series’ discipline requires it and the temptation to overreach is real. It does not claim that the wrongness of forged trust follows from its instability—that the forged case is unjust because it is destined to collapse. That would be to derive an ethical conclusion from a dynamical premise, to read a wrong off a trajectory, and the series has been at pains throughout to keep these orthogonal. The mechanism answers a dynamical question: will this trust survive, is it well-founded, will the error signals confirm or refute it. The justice criterion answers a different question, which no amount of dynamical information settles: is this a way of treating the other that honours him or wrongs him. A forged trust that happened, against the odds, to be stable—a manipulation that was never found out, a pumped phase that never collapsed—would be no less a wrong for its stability; and a genuine trust that collapsed through no one’s fault—a real self-commitment overtaken by circumstance—would be no less honourable for its collapse. The two criteria can be made to agree, and the section will argue that at the limit they do; but they are not the same criterion, and the agreement is a substantive finding rather than a definitional identity. This is the orthogonality the series has insisted upon elsewhere, between the question of whether a cycle lives and the question of whether it is just, and it holds here in the form: the mechanism tells whether the trust stands, the justice criterion tells whether it should have been produced as it was.

8.2 The single line, drawn ethically

With that established, the coincidence can be drawn. The mechanical mark of manipulation was stated in §6.4: to induce the other’s generative model to lower its prediction error without supplying the high-fidelity likelihood that would warrant the lowering. The ethical mark of manipulation is the same act described under a different aspect, and the section names three aspects of the one act, each of which is a way the manipulation wrongs the one manipulated.

The first aspect is epistemic. To induce in another a high-precision prediction unsupported by any real likelihood is to corrupt his capacity to predict the world truly—to install in his model a confidence that the world will not bear out, and so to set him at odds with his own future experience. The manipulated person is not merely deceived about a particular fact; his instrument for forming expectations has been tampered with, made to deliver a precision the evidence does not warrant, so that he is rendered, in just this respect, a worse knower than he was. Manipulation is an injury to the other’s epistemic autonomy: it does not persuade him by giving him grounds he can weigh, which would leave his judgment intact and free; it works on his model from outside, behind the back of his judgment, to produce in him an expectation he did not form on grounds and could not have refused. This is the precise epistemic sense in which manipulation differs from honest persuasion, and it is a difference the mechanism makes exact: persuasion supplies a likelihood the other’s model may weigh and accept or reject; manipulation injects a precision the model did not earn and cannot examine.

The second aspect is the Kantian, and it returns the paper to the distinction of §5. To produce trust by injecting precision rather than by supplying a real self-committing likelihood is to treat the other not as a free mind forming its own expectations on grounds but as a mechanism whose states are to be arranged from outside—to apply to him the mechanical cause rather than the inferential, to manage him rather than to address him. It is, in the terms of the second formulation of the categorical imperative, to treat him merely as a means: as an object whose predictive states one engineers toward one’s own ends, rather than as an end in himself who forms his expectations freely and could withhold them. The genuine production of trust, by contrast, addresses the other exactly as §5 required: it supplies him with a real ground—a costly, irrevocable self-commitment—which his free judgment may take up, and it leaves the taking-up to him. The difference between genuine and forged trust is thus, in its second aspect, the difference between addressing a free being as free and handling a free being as a mechanism; and this is why the forged case is not merely a technical defect in the production of trust but a wrong done to the one in whom it is produced.

The third aspect is the one the series has carried from its work on generative justice, and it is the deepest, for it names what the manipulator does with the relation as a whole. To manipulate is to treat the other’s trust as fuel—to draw upon his capacity to rely, his openness to being committed-to, as a resource to be consumed for an end that is not his and from which he is excluded. The forged relation is one in which the trusting party is rendered a mere means to a value that returns to the manipulator and not to him: his trust is mined, his vulnerability is exploited, and he is positioned in the cycle as the one who supplies what others consume. This is the criterion the series has stated as the question to be put to any cycle that deepens: is there someone within it who figures only as a cost, excluded from the return—is there one who is rendered mere fuel? Forged trust answers yes; it is the production of an expectation whose function is to extract from the one who holds it, and who is, in the holding, used up. Genuine trust answers no; it is the production of an expectation that opens a relation in which the trusting party is a beneficiary and not merely a source, in which what is generated returns to the one whose trust made it possible. The justice of trust, in this third and deepest aspect, is the justice of the return: genuine trust returns to the one who gives it, forged trust extracts from him.

8.3 The three aspects are one

These three—the epistemic injury, the Kantian reduction, and the rendering-as-fuel—are not three separate wrongs that manipulation happens to commit together; they are three aspects of a single act, the act the mechanism described as the injection of unwarranted precision. To inject precision without likelihood just is to corrupt the other’s knowing (the epistemic aspect), which just is to bypass his free judgment and handle him as a mechanism (the Kantian aspect), which just is to position him as a source to be drawn upon rather than a partner in a return (the aspect of fuel). The mechanism gave the act its mechanical description; the three aspects give the one act its ethical description; and that the mechanical and the ethical descriptions are descriptions of the same act is why the line the mechanism drew and the line justice draws coincide. They coincide not because the one is derived from the other—the section began by refusing that—but because they are two descriptions, at two levels, of one and the same thing: the producing of an expectation in another without the real self-commitment that would honour him in the producing of it.

8.4 The criterion stated, and what it leaves to practice

The criterion may now be stated in a form the rest of the paper will use. Genuine trust is an expectation produced in another by supplying a real ground his free judgment may take up, and which opens a relation that returns to him; forged trust is an expectation induced in another by injecting a precision his judgment cannot examine, which handles him as a mechanism and draws upon him as fuel. The criterion is, as the series requires, a criterion of justice laid alongside the criterion of dynamics, not derived from it: the mechanism tells whether an expectation will stand, the criterion tells whether it was rightly produced, and the two must each be applied, neither collapsing into the other.

One consequence must be marked, for it sets the problem the praxis will have to solve and guards against a misreading the rest of the paper cannot afford. The genuine and the forged are not, as outward performances, reliably distinguishable. The manipulator who stages a costly-looking commitment and the sincere promisor who makes a real one may, in the moment, present indistinguishable surfaces; the injected precision and the earned precision may, for a time, feel the same from inside. What distinguishes them is not a difference in the performance but a difference in what lies behind it—whether there is, in fact, a real self-commitment and a real return, or only their appearance. This is why the line, though sharp in principle, is treacherous in practice, and why the production of genuine trust cannot be reduced to the production of its signs: the signs can be forged, and the difference that matters is precisely the one the signs do not, by themselves, reveal. The translation of sincerity across the borders of epistemology (§10) and the praxis of trust (§12) are, in large part, the working-out of how one acts well under exactly this condition—that the genuine must be produced and offered in a world where its signs are indistinguishable from the forged, and where the difference, though real, is not on the surface to be read. The criterion of justice tells us what the difference is; it falls to practice to live by a difference one cannot, from the outside, simply see.

9. Relational Trust

The mechanism of §6 described how trust is generated between a promisor and a receiver who are present to one another: the structural form, the costly self-binding, the wagered leap, all presuppose two parties in some direct relation. But a great part of the trust on which human dealing actually runs is not of this direct kind. We trust people we have never met, on the strength of others we do trust; trust travels along the links of a relational network, reaching where no direct relation reaches. This section is about that transmitted trust—relational trust—its mechanism, its peculiar liability to a failure the direct case does not suffer, and the formal structure it shares, the section will argue, with the dynamics of an open quantum system. It is the chapter in which the problem of winning a family’s trust, posed in the Prelude, finds the resource that bears most directly upon it; and it is the chapter in which the crisis of the symbol receives a new and precise specification.

9.1 The mechanism of transmission: the recommendation and the chain

The clearest instance of relational trust is the letter of recommendation, and it repays close mechanical attention, for it shows what is really at work and what is not. The naive view is that a recommendation works by the content of what the recommender says—by the predicates he applies to the candidate, the praise he heaps, the qualities he affirms. But this is very largely not what carries the trust. The adjectives carry almost no information; everyone is described in superlatives, and a committee that has read a thousand letters extracts nothing from the thousand-and-first profession of excellence (§2.3). What actually works is not the content but the structure: a committee C trusts the judgment of a recommender A, with whom it has a relation of trust; the recommender A has a real working coupling with the candidate B, knows B from the inside, has the embodied knowledge of what B can do; and trust is transmitted along the chain C–A–B, C coming to trust B through its trust in A. The recommendation works as a conductor, not as a description: it carries C’s existing trust in A across the link A–B to reach B, whom C has no other way to reach.

What this reveals is that trust, in the relational case, is not a property of any single party at all. It is not something in B, nor in A, nor in C; it is a property of the structure of the relation among them—a topological property of the network, residing in the pattern of couplings rather than in any node. C has no direct evidence about B (the evidentialist channel cannot reach), and no direct contact with B (the deontological reading of character cannot reach); yet C can trust B, and the trust exists nowhere but in the chain. Relational trust is thus a third mechanism, irreducible to the evidential and the characterological alike, by which trust reaches across a gap that neither evidence nor the direct reading of character could span. And it bears directly on the predicament of the Prelude: the trust a family extends to a stranger who would marry into it runs, very largely, not through anything the stranger can show them directly, but through the chain that connects him to them by way of the one they already love and trust—the one who loves him is, in the structure of the network, his recommender, and the trust reaches him along that link or not at all.

9.2 Relational trust spans the three registers

Relational trust is not lodged in one register but spans all three, and this is the key to both its power and its specific fragility. There is an imaginary component: the image of B that travels along the chain, the idealised or distorted picture A’s account installs in C. There is a symbolic component: the letter, the word, the chain itself as a transmissible structure of signs. And there is a real component, which is the heart of the matter: the real working coupling between A and B, A’s embodied, unfakeable knowledge of what B is and can do, the thing the whole apparatus exists to transmit. Genuine relational trust is the case in which the real component leads—in which what travels the chain is a real coupling, vouched along a structure of signs, forming in C an expectation grounded in something that actually exists. The chain is the symbolic conductor; what it ought to conduct is the real.

9.3 The dislocation of vehicle from substance

Here is the new specification of the crisis of the symbol that the section promised, and it follows from the fact just established—that relational trust spans the registers, and that the registers can be prised apart. The substance relational trust ought to transmit is the real component: A’s real coupling with B, the unfakeable knowledge that is the one thing of value in the whole transaction. But the vehicle by which it is transmitted is symbolic: a letter, a word, a writable and circulable and archivable sign. And the alienation that destroys relational trust occurs in the dislocation of the vehicle from the substance—the symbolic form preserved intact while the real substance is hollowed out of it.

A forged recommendation is exactly this dislocation. In the symbolic register it is impeccable: correctly formatted, well phrased, duly signed, transmitted along the chain C–A–B in perfect order. Viewed from the symbolic register alone it is indistinguishable from a genuine one. But its real component—A’s actual knowledge of B—is empty: A wrote it not because he embodily knows B to be good but because he owed a favour, or moves in the same circle, or could not decently refuse. The sign is transmitted, and what the sign was supposed to carry is not present in it. This is the incarnation, in relational trust, of the central pathology the series has named the failure of settlement, the usurpation by the symbol: a sign that announces a coupling already vouched-for when no real coupling underwrites it, a letter that cashes a knowledge that was never had. The forged recommendation is the isomorph of the false poem and the alienated bride-price—a symbol that claims to anchor the real while the real has been drained out from beneath it.

The diagnosis of the breakdown is therefore precise, and it is the section’s new contribution to the diagnosis of §3. Relational trust breaks down not because the mechanism is defective but because it spans the registers and the registers can be split: the symbolic transmissibility (the letter, the chain) and the real unfakeability (the actual coupling), which ought to be two faces of one thing, can be prised into two, the first retained and the second hollowed out. Forged relational trust is the product of that fissure.

9.4 The bidirectional alienation

The word alienation is exact here, for the damage runs in both directions along the chain, and the section must trace both, for together they explain why the whole channel closes rather than merely admitting some bad letters alongside the good.

On the side of the one vouched-for, the proliferation of forged recommendations devalues the genuine ones. The student who really does have a deep working coupling with a mentor who really knows her finds that her letter, in the symbolic register, looks exactly like the letter written from obligation; and so her real coupling can no longer be transmitted, because the channel that should carry it has been saturated with signs that carry nothing. The real thing loses its capacity to be recognised, not because it is disbelieved in particular but because the forgeries have made the whole channel unreadable. This is the precise mechanism of the crisis of the symbol, seen once more: it is not that the false drives out the true by replacing it, but that the false makes the true unrecognisable, so that the channel as a whole is abandoned and the institution retreats to the merely countable—citation counts, rankings—signals that are barren but at least unforgeable, preferring a channel known to be poor to one that might be rich but has been poisoned.

On the side of the one who vouches, the damage is subtler and is an alienation in the fullest sense. The recommender who grows accustomed to writing letters he does not believe is alienated from his own act of vouching: his signature ceases to mean that he really knows and comes to mean only that a social obligation has been discharged. And in doing so he draws down, a little at a time, the relational capital that made relational trust possible—C’s trust in his judgment—spending it to discharge favours until the trust itself is exhausted. He siphons off, to meet the demands of his circle, the very thing that let his word carry trust, and when it is gone the chain through him conducts nothing. The bidirectional alienation—the genuine made unrecognisable on the one side, the capacity to vouch drawn down to exhaustion on the other—is why the failure of relational trust is not a local matter of some bad letters but the closing of a channel.

9.5 The quantum-dynamical structure of relational trust

The structure that has emerged—a trust that is a property of a network rather than of its nodes, that spans registers which can be split, that fails by a dislocation of a transmissible form from an unfakeable substance—has, the section now argues, a precise formal counterpart in the dynamics of open quantum systems. The argument is made under a discipline that must be stated before it is made, and stated as firmly as the discipline upon the use of predictive coding in §6. What is claimed is a structural isomorphism, not a physical reduction. The paper does not assert that trust is, literally, a quantum phenomenon; to say so would be the worst kind of pseudo-scientific mysticism, and would forfeit the seriousness of everything else the paper has done. What is asserted is that the mathematical structure of relational trust and the mathematical structure of the entanglement-and-decoherence dynamics of an open quantum system are the same structure, and that the formal language of the latter can therefore describe the former with a precision no other available language affords. This is the same discipline the series has observed throughout: the quantum-dynamical account is one more level of description, a formal language for a structure, and not the truth beneath the others, nor a meta-framework that absorbs them.

Under that discipline, three correspondences may be drawn, in ascending order of depth.

The first is entanglement as the non-local structure of relational trust. The defining feature of an entangled state is that the state of the whole cannot be decomposed into a product of the states of its parts; the information resides in the correlations among the parts, not in any part taken alone, so that a measurement of any single component does not recover it. This is exactly the structure established in §9.1: relational trust is not a property of C, A, or B severally, but of the correlation C–A–B itself, recoverable from no node in isolation. The non-locality of trust—that it lives in the chain and not in the links—is given, by the language of entanglement, a precise formal expression it otherwise lacks.

The second is decoherence as the dislocation of vehicle from substance. A coherent quantum state, coupled to an environment, undergoes decoherence: the correlations that constitute its coherence are not destroyed outright, but the phase relations that made them meaningful are lost to the environment, leaving a state whose correlational shell persists while the coherence that gave it its character has drained away. This is, formally, the dislocation of §9.3: the symbolic transmissibility (the correlational shell, the chain of signs) is retained, while the real coupling (the phase coherence that made the correlation carry something) is lost. A forged recommendation is a decohered trust-state: it preserves, in the symbolic register, the entire outward form of the entanglement, while the coherence it was meant to carry is gone—and, just as a decohered mixed state can be indistinguishable from a pure state under measurements that are not phase-sensitive, the forged letter is indistinguishable from the genuine under any reading that is not sensitive to the real coupling beneath. And the crisis of the symbol is, on this reading, the systemic decoherence of relational trust at the scale of a society: an environment saturated with forged signals continually coupling to the trust-states that traverse it, draining their coherence, leaving correlational shells that no longer carry what they were meant to carry. This dovetails exactly with the predictive-coding diagnosis of §6.5: the systemic decoherence of the trust-state and the rational collapse of the population’s precision-weight on the symbolic channel are the same process described at two levels.

Here the word dynamical, rather than merely quantum, earns its place, for what relational trust most needs from the analogy is not a static picture of a state but an account of how the state evolves. The dynamics of an open quantum system—a system in continual exchange with an environment, its evolution governed by a master equation that specifies how coherence decays, and under what conditions it may be protected or even restored—is the formal language for how a trust-state is maintained, corrupted, or rebuilt over time, which a static account of entanglement could not supply. And this is the point at which the present section welds to §7, for the geometric phase deployed there is, in its origin, a concept of exactly this quantum dynamics: the Berry phase is the phase accumulated by a quantum state under adiabatic transport. The holonomy language of §7 was therefore never a free-floating borrowing; it was, all along, drawn from the very dynamics relational trust is here shown to share. Adiabatic evolution—slow, exchanging no energy with the environment, coherence and geometric phase preserved—is the genuine, endogenously sustained trust; non-adiabatic evolution—coupled to the environment, decohering, phase lost—is the forged. The genuine-versus-forged distinction of §8, the adiabatic-versus-driven distinction of §7, and the coherent-versus-decohered distinction drawn here are one distinction in three vocabularies.

The third and most precise correspondence concerns transmission itself, and it requires a refinement that shows the analogy has been taken seriously rather than reached for loosely. Bare entanglement will not, by itself, model the transmission of trust, for entanglement is subject to the no-communication theorem: it cannot, alone, be used to send information. But relational trust is precisely about the transmission of trust along the chain C–A–B. The formal counterpart of this transmission is therefore not bare entanglement but entanglement swapping: the protocol by which, through an intermediary node A already entangled with each, two parties C and B who share no prior entanglement are brought into an entangled relation. Entanglement swapping is, almost exactly, the quantum correspondent of the recommendation mechanism C–A–B: the intermediary A, coupled to both, is the means by which a correlation is established between C and B who had none. To name the correspondent as swapping rather than as bare entanglement is to mark that the analogy is structural and considered—that it tracks the specific feature of relational trust, its transmissibility through an intermediary, and not merely its non-locality.

9.6 The standing of the analogy

The discipline stated at the opening of §9.5 must be restated at its close, for it is the condition on which the whole construction is admissible. The quantum-dynamical language is a formal description of a shared structure, at a level of description that does not displace the others the paper has given—the mechanical account of §6, the ethical account of §8, the phenomenological account of the chain as a relation among persons. It is not a claim that trust is quantum-mechanical in its physical substrate; it is a claim that relational trust and open-system quantum dynamics instantiate one mathematical structure, and that the structure is illuminated by being seen in the language that has developed the richest tools for it. It is, in the polyphonic terms the series requires, one more irreducible epistemic position on a single object, a further voice and not the final word—and it is offered as such, a formal analogy that earns its place by the precision it brings to the non-locality, the decoherence, and the transmission of a trust that lives in the chain and not in its links.

10. The Translation of Sincerity

Every chapter to this point has worked on the assumption that once genuine trust is produced, it is received—that a real self-commitment, supplied to another, lands as the ground it is. This section removes that assumption, and in removing it opens the problem that is, for the predicament of the Prelude, the most pressing of all. Sincerity has no universal language. There is no form of sincerity legible to everyone; sincerity must be encoded to be received, and the language of the encoding depends upon the epistemology of the one who is to receive it. The gravity of the betrothal document, the moon, the poem, the philosophical seriousness—these encode sincerity in a form legible to one who reads character directly, who takes the bearing and the solemnity and the cost of the gesture as themselves the signal. To a receiver of another epistemology—one who holds that what is the case is the case and asks for what can be observed and verified—that same sincerity, in that same encoding, is not evidence at all. The same real sincerity, passed through two different decoders, yields utterly different degrees of credibility; and the difference is not in the sincerity, which is one and real, but in the mismatch between the sender’s language of encoding and the receiver’s language of decoding.

This is the point at which the paper’s question undergoes its decisive turn. The question has been, through the mechanism and its consequences, an ontological one: what is genuine trust, and how is it produced. It now becomes a translational one: how is genuine trust, once produced, transmitted across the border between incommensurable epistemologies. And this translational question is the sharpest form, in the field of trust, of the master theme the series has carried throughout—that there is no metalanguage, that the Other does not exist as a guarantor standing above all positions. There is no neutral universal language of sincerity, no God’s-eye dictionary in which the sincerity encoded in one epistemology can be looked up and re-expressed in another with its value preserved. The problem of trust, fully stated, is therefore not only how to be genuine but how to be read as genuine by one whose language of reading is not one’s own; and this section is its theory.

10.1 The epistemology of translation

Before one can translate sincerity into a language a receiver can read, one must know what languages there are—what grammars of trust a receiver might be using, by which a profession of sincerity is parsed into a ground for expectation or rejected as no ground at all. This subsection sets out a genealogy of these grammars. It is offered as adequate to the practical purpose rather than as exhaustive, and its precision matters, for the effectiveness of any translation depends on the accuracy with which the target grammar is identified.

10.1.1 A genealogy of trust-grammars

There are at least four grammars of trust, irreducible to one another, each parsing a different kind of thing as a ground for expectation, and each with its characteristic reach and its characteristic blind spot.

The first is the evidentialist or positivist grammar. It anchors trust in observable evidence and accomplished fact—the track record, the thing already done, the stable and verifiable pattern. To one who uses this grammar, a ground for trust is something that has shown itself, and a profession that has not yet shown itself in fact is not yet a ground. Its reach is into the demonstrated; its blind spot is the sincere newcomer who has, as yet, no record to show, and whom it must, by its own logic, misjudge as untrustworthy not because he is but because he has not yet had occasion to demonstrate otherwise. This is the grammar of Holmes’s bad-man observer (§4.5) turned to the reading of others, and it is, the paper has reason to think, the grammar most relied upon by the family of the Prelude.

The second is the deontological or characterological grammar. It anchors trust in a direct, intuitive reading of the other’s character and sincerity—in the bearing, the manner, the quality of the commitment, taken as immediate signals of an inner state. To one who uses this grammar, the gravity of a vow, the seriousness of a gesture, the felt sincerity of a person, are themselves grounds, read off without the detour through accomplished fact. Its reach is into the inner and the present; its danger is that it reads too much, that it can be deceived by one who has learned to perform the signs of a sincerity he does not have. This is, the paper has reason to think, the grammar most native to the one who would encode his sincerity in red silk and the moon.

The third is the inductive or relational-historical grammar. It anchors trust in the accumulated pattern of a shared history with this particular person—not in evidence as such, nor in a reading of character as such, but in the texture of what has passed between us, the record of how this one has been with me over time. Its reach is into the known particular; its blind spot is the wholly new relation, for which no shared history yet exists.

The fourth is the institutional or endorsement grammar. It anchors trust in the vouching of a trusted third party—the matchmaker, the elder, the credential, the recognition of a community, the recommendation along the chain of §9. Its reach is into whatever the trusted institution can underwrite; its blind spot is the condition, diagnosed throughout this paper, in which the institutions of endorsement have themselves come under the crisis of the symbol and can no longer reliably underwrite anything.

These four are not a ranking, with one the true grammar and the rest its approximations. They are incommensurable epistemic positions, each parsing as a ground something the others do not, none reducible to the rest—and their irreducible plurality is one more vindication, in the field of trust, of the polyphonic law the series has insisted upon. The question whether the genealogy is fine enough is left open, and one refinement in particular is marked as undecided: whether the evidentialist grammar should be subdivided, between one that asks for accomplished fact and one that asks for demonstrable capacity, since these have different practical implications for what a translation into that grammar would have to supply.

10.1.2 The notion of a grammar-configuration

No person uses a single grammar purely. Each uses a configuration—a weighting, mostly unreflective, of the several grammars, in which one may dominate without excluding the others. The family of the Prelude is best understood not as purely evidentialist but as running a configuration in which the evidentialist grammar dominates, the institutional has some weight, and the deontological is heavily discounted—so that a profession of character lands lightly and an accomplished fact lands hard. The notion of the configuration is what makes the genealogy a practical instrument rather than a typology: one does not translate one’s sincerity into the evidentialist grammar in the abstract, but into the particular configuration a particular receiver runs, and the diagnosis of that configuration is the first task of any actual translation. The unit of translation is the configuration, and the genealogy is the alphabet from which configurations are spelled.

That there can be no single correct translation of a sincerity from one grammar into another—that translation is, in principle, indeterminate—is the deep epistemological fact underlying all of this, and it is Quine’s. The indeterminacy of translation is not a defect to be overcome but the very form the absence of a metalanguage takes in the theory of translation: there is no fact of the matter, fixed independently of every scheme, as to the uniquely correct rendering of a meaning from one language into another, and so no neutral standard against which a translation of sincerity could be certified as the correct one. Translation is underdetermined all the way down, and the practical art the next subsection describes is an art practised under that permanent indeterminacy, not a technique for escaping it.

10.2 The praxis of translation

How, then, is a sincerity encoded in one grammar to be rendered legible in another, without distorting it, and without sliding into the manipulation that §8 marked off? There are three paths, of ascending difficulty and depth, and each runs close to the ethical line, for each is, in its outward form, hard to distinguish from a counterfeit. The subsection sets them out, and then states the line that separates all three, done genuinely, from their forged twins.

10.2.1 Projection

The first path is projection: letting a sincerity that is legible in one’s own grammar cast a shadow in the receiver’s. One does not disguise the deontologically-legible sincerity as evidence, which would be forgery; one allows the real sincerity to leave, in the symbolic and observable register the receiver reads, the traces it naturally casts—an accumulation of consistent action, a constancy held across time, conduct that the evidentialist can read as evidence because it really is the observable shadow of a real inner state. The evidentialist infers back, from the shadow, to the substance. This is the safest path and the slowest, for it depends on time: the shadow accumulates only as the conduct accumulates, and cannot be hurried. It is the practical translation of the principle the series has carried as the maxim that what is held to over long enough comes to show itself—not that time creates the coupling-trust, which is real from the first, but that time lets the real cast, in the register the evidentialist reads, enough of a shadow to be recognised. Projection is the rendering of a real inner sincerity into the language of accomplished fact, by letting it accomplish the facts that are its true shadow.

10.2.2 Multiple encoding

The second path is multiple encoding: encoding the one sincerity, at once, in two grammars, on two tracks running in parallel. To the one who reads character—and to oneself—the vow, the gravity, the poem; to the one who reads evidence, the stable and visible facts, the observable constancy, the demonstrated reliability. The two encodings are of one sincerity, offered each in the grammar that can receive it. The difficulty of this path is not technical but ethical, and it is severe: multiple encoding degrades into mere placation unless one genuinely concedes the legitimacy of the other’s grammar. If one encodes in the evidentialist grammar while privately holding that to ask for evidence is shallow, that the deontological reading is the real one and the evidentialist a concession to the obtuse, then the second encoding is not a translation but a condescension, and a receiver of any acuity will read the condescension beneath it. Multiple encoding done genuinely requires that one hold both grammars as legitimate ways of arriving at trust—a requirement that connects it directly to the justice of presentation taken up in §12, where the active rendering of one’s sincerity in the other’s grammar is shown to be, when the substance is real, not a strategy but a form of respect.

10.2.3 Mutual translation

The third path is the deepest and the hardest, and it is the one the section most wants to establish, for it is the only one that issues in a trust that is neither projected nor doubly-broadcast but jointly made. In mutual translation, both parties move a step toward the other’s language: one learns to speak in evidence, and the other learns to read the signals of character; and what is produced is not the persuasion of one by the other but the growing, in the interaction itself, of a small shared grammar that did not exist before. This is real coupling at the epistemological level—trust that is not one party’s conviction transmitted to the other but a thing generated between them, in the continued interaction, out of two grammars that have each moved toward the other and grown, between them, a patch of common language. It does not pre-exist the relation and wait to be transmitted; it is generated, in the mutual movement, as the relation’s own.

That this is not a philosopher’s fancy but a real and learnable craft is shown by the existence of a mature profession that does exactly this. The experienced consultant, in any field, who is brought before a client he does not yet understand, faces the structural twin of the no-metalanguage problem: the client comes with a language, a situation, a set of terms and unspoken assumptions and a real difficulty, that the consultant cannot at first read; and the consultant can neither impose his own framework upon it, which is bad consulting, nor pretend to understand what he does not. The real work is to create, in the interaction, with this particular client, a common language—to learn the client’s terms and let the client learn his, until a shared grammar exists between them that existed before in neither. This is mutual translation, and that consulting is a mature profession, with methods and training and tests of competence, is the proof that mutual translation can be done, systematically and well, and is not the unoperable ideal it might otherwise seem—the very objection this hardest path most invites. The empirical literature on the outcomes of such work lends this support: across the helping and advisory professions, the strongest predictor of a good outcome is found to be not the technique or the school but the quality of the working relationship, the working alliance—which is precisely the establishment of the shared understanding that mutual translation produces. The point answers, in part, the call the foundational paper made for empirical follow-up: there is a body of evidence that the generation of a shared grammar is real, learnable, and the operative factor in success.

Two features of the consultant’s craft sharpen the account. The first is an asymmetry. Mutual translation need not be symmetrical: the consultant, as the active party who bears the professional responsibility, moves further toward the client’s language than the client moves toward his. The party who actively seeks to establish trust—the one in the position of the Prelude’s suitor—may rightly have to move further toward the other’s grammar, and this is not a mark of his grammar’s inferiority but of his bearing the greater share of the labour of translation, as the active and responsible party. This asymmetry is the concrete form of what §12 will call epistemic hospitality: the moving-toward-the-other, the taking-on of the greater share of the work of being understood. The second feature names the philosophical resource that underwrites the path. What mutual translation accomplishes is the fusion of two horizons into a shared one that neither held alone—Gadamer’s Horizontverschmelzung, the fusion of horizons, which is the standing philosophical name for the generation of a common understanding between parties who began with none, and which gives mutual translation its grounding in the hermeneutic tradition.

10.2.4 The line that separates translation from manipulation

All three paths run close to the ethical line, and the line must be drawn as firmly as it can be, for it is the line on which the whole praxis turns and it is not, on the surface, where one might expect. To render one’s sincerity legible in another’s grammar (translation) and to manufacture a false signal legible in another’s grammar (manipulation) may, as outward acts, look identical—the projected shadow and the forged evidence, the second encoding and the placating performance, can present the same surface. The difference lies wholly in whether what is being translated really exists. To project the real shadow of a real sincerity is translation; to project the false shadow of a sincerity that is not there is manipulation (§8). The criterion of genuine translation is that the thing it translates must really exist and be capable of real fulfilment: it carries a real substance across the border between grammars, where manipulation manufactures, in the target grammar, the appearance of a substance that is absent. Translation takes the other’s grammar as a legitimate other to be honestly addressed; manipulation takes the other’s grammar as a lock to be picked. The line is not between act and act, which can match, but between a translated substance that is real and a manufactured signal that is empty—and it is, once more, the line of §8, the line between honouring the other and rendering him fuel.

10.3 Dialogue: the convergence of three identities

The deepest of the three paths, mutual translation, turns out to be a thing the philosophical tradition has known under another name, and the recognition gives the section its close. Mutual translation is dialogue, and dialogue is, at once, three things that prove to be one.

It is, first, the practical form of mutual translation: the consultant’s craft, the back-and-forth in which a shared language is grown between parties who began without one. It is, second, the ancient method by which philosophy has always held that certain truths are generated—the Socratic dialektikē, in which truth is not stated by one party and received by the other but generated between them, in the going and returning of question and answer, because some truths can be reached in no other way than by two minds moving toward each other. And it is, third, the formal law of the series itself—the polyphony of the plural conclusions, the refusal of a governing voice, the insistence that the truth of the matter lives not in any single framework but in the dialogue among them. These three—the consultant’s craft, the Socratic method, the series’ polyphonic form—are one thing, and their being one is why mutual translation is the deepest of the three paths. It is the deepest because it is dialogue; and dialogue is, in a world without a metalanguage, the only way that truth and trust can be generated between heterogeneous subjects at all. Where there is no God’s-eye dictionary to translate one grammar into another with the value preserved, the value is not transmitted but generated, between the parties, in the dialogue in which each moves toward the other and a shared grammar grows. Mutual translation is dialogue: the generating, between two who share no common language, of a common language, together.

And this returns the section to the poetics the series has carried throughout, for the translation of sincerity, so understood, is itself a poetic act in the precise sense the series has given the term. To translate one’s sincerity into a grammar another can read is not to settle a meaning and hand it over closed; it is to invite the other to walk a path of interpretation, to offer a form the other must traverse rather than merely receive, to write, for the other, a poem he can read. The poetic refuses settlement and opens a path to be walked; and the translation of sincerity, done genuinely, refuses to settle the other’s trust by manufacture and instead opens, between the two grammars, a path along which a shared understanding may be walked into being. The translation of sincerity is the writing, for the one whose language is not yours, of a poem in a language he can read—and the paper turns now from the technique of this writing to its practice, at the scale of institutions and then at the scale of a life.

11. Jurisprudence and Judicial Optimisation

The paper turns now from the analysis of trust to its practice, and takes the institutional scale first, the scale at which the problem appeared in the symptomatology as the overload of justice and the sword that hangs but does not fall. If force is not the cause of trust but only a failing device for the production of expectation, then the institutions built upon force—the courts, the apparatus of enforcement—stand in need of a reconception, and this section asks what it would be. Its answer has a spine that is counter-intuitive and will be stated plainly: that the effective optimisation of the juridical, in an age when force has failed as a signal, is the decentralisation of enforcement—not the strengthening of the sword but the distribution of the production of trust away from the single channel of compulsion. But the section must guard, from the outset, against the way this spine could be misheard, and the guard determines the whole method of the chapter.

For the spine could be misheard as an idealist exhortation—as a call to trust more, to cultivate reputation, to repair rather than to punish, as though the institutional form of trust were a matter of enlightened choice, a better idea that societies might adopt if only they saw its merits. The series is historical-materialist, and must treat the matter dialectically; and an idealist exhortation is precisely what the historical-materialist method forbids, for it would mistake the superstructure for something freely adjustable apart from its material base. The institutional forms of trust are not chosen; they are thrown up by the material conditions of their production, and they change as those conditions change. The chapter therefore proceeds not by recommending a reform but by analysing a transformation already under way—its material causes, its emergent tendencies, and the new contradiction into which it is, under present relations of production, being driven.

11.1 The double bind of the juridical

The condition to be analysed is, first, a double bind, in which the two faces of the symptomatology—the overload and the failure of force—compose a vicious circle. As the relational and communal orders of trust thin, more of the burden of holding people to their dealings falls upon the courts (the overload); the courts, asked to produce by compulsion an order they can only ever protect, find their signal of compulsion diluted across a volume it cannot cover, and force fails as a signal (the failure); and as force fails, trust thins further, throwing yet more burden upon the courts. The more the juridical is relied upon to generate trust, the more its single signal is diluted; the more diluted the signal, the less it generates; the less it generates, the more it is relied upon. The juridical is caught being asked to produce, by force, the very thing—trust—that §6 showed force cannot produce but only, at most, signal; and asked to produce it at a scale at which even the signalling fails. This is the double bind, and no increase in the force applied can resolve it, for the problem is not a shortage of force but the exhaustion of force as a means to this end.

11.2 The counter-intuitive thesis: decentralisation

The resolution is not to add force but to cease asking the single channel of force to do what only the coupling of several channels can do. §6 established that well-founded trust is produced by the multiplicative coupling of channels, and that its robustness lies not in the strength of any one but in the improbability of all being forged together. The juridical optimisation that follows from this is the institutional application of the same principle: distribute the production of trust away from the single channel of compulsion and across the several channels the mechanism identified, so that trust is generated by their coupling rather than by the sword alone. This is what decentralisation means here, and why it is effective where the strengthening of force is not. A single channel can be evaded, diluted, and—under the crisis of the symbol—collapsed in its precision-weight across the board; several coupled channels, whose mutual consistency is hard to forge or evade together, are robust exactly where the single channel is fragile. The thesis is counter-intuitive because the reflex, faced with a failure of enforcement, is to call for more enforcement; the mechanism shows this to be the marginally decreasing application of a means already exhausted, and the dispersal of trust-production across coupled channels to be the structurally sounder course.

But—and here the historical-materialist method asserts itself against any idealist reading—this decentralisation is not a reform to be recommended; it is a transformation already under way, forced by a change in the material conditions of the production of trust, and the rest of the chapter must show it as such.

11.3 The transformation, dialectically

The transformation must be grasped in four moments, and the fourth forbids the celebration the first three might invite.

11.3.1 The material basis of centralised force

Centralised, force-based justice—the sword wielded by the state—was not a bad idea that societies happened to adopt; it was the form of trust-production appropriate to, and thrown up by, a definite material base. In a society of strangers, in which exchange is predominantly market exchange and persons are atomised and rendered mobile and interchangeable by the commodity form, trust can no longer be produced by the face-to-face relational history and communal witness that produced it in smaller and more settled worlds. Where the village and the lineage no longer hold a person to his word, the holding must be outsourced to an abstract, impersonal third party—the state, the law, the sword—which can compel performance among strangers who have no other hold upon one another. Centralised force was, that is, the historically necessary form of trust-production adequate to the material base of a market society of strangers. This must be said first, on pain of turning the critique into a moralising lament: the sword was not an error but an adaptation, and a correct one, to its conditions.

11.3.2 The material failure

Its failure, likewise, is material and not a decay of values. The double bind of §11.1 is not produced by a hardening of hearts or a decline of civic virtue; it is produced by a change in the material conditions. The scale and speed of exchange, under globalisation and digitalisation, have outrun the physical capacity of any apparatus of enforcement; the volume of breach has outrun the carrying capacity of the courts; and the cost of producing a convincing symbol has fallen to zero, saturating the environment with forged signals and collapsing, as §6.5 showed, the precision-weight a population rationally assigns to the symbolic channel on which centralised enforcement partly relied. These are changes at the level of the forces and the mode of production, and they have brought the once-adequate form of trust-production—centralised force—into contradiction with the new material conditions. This is the historical-materialist account of the crisis of the symbol: not the idealist story of a value-nihilism that has overtaken the culture, but the materialist story of a contradiction between the established mode of producing trust and the changed material base beneath it. The diagnosis of §3 is here raised from a cultural diagnosis to a historical-materialist one.

11.3.3 The emergent recomposition

The dialectical point follows, and it is the one that distinguishes analysis from exhortation. The decentralisation of enforcement is not something we ought to choose; it is a recomposition that the contradiction is already forcing into being. The new forms of non-coercive trust-production—reputation systems, credit mechanisms, platform ratings, restorative justice, the distributed ledgers of trust—are not the good ideas of reformers; they are the spontaneous symptoms of a mode of trust-production reorganising itself under new material conditions, the new technologies of network and computation throwing up new channels as the old single channel fails. The task of the section is therefore not to recommend decentralisation but to analyse this recomposition that is already occurring: to identify its direction, its potential, and—the fourth moment—its internal contradiction.

11.3.4 The capture: the new contradiction

For the dialectical method requires that the emergent form be examined for its own contradictions, and not celebrated. The new channels of non-coercive trust are, under the present relations of production, being captured by capital and turned to alienation. Reputation and credit, which might be the emancipatory institutionalisation of the relational-historical channel, become under capital the social-credit score, the algorithmic management of the worker, credit as a new instrument of discipline—a finer and more pervasive control than the sword, precisely the retreat to the quantifiable that §2.3 described, with the metrics now held by the platform and the owner. The platform’s badge of trust becomes capital’s commodification of trust, a thing the platform can monopolise and extract rent upon. So the decentralised technical form, under centralised ownership of the new channels, may produce a domination more insidious than the sword it replaces—a domination that operates through the very channels that were to have freed trust from compulsion. The question the historical-materialist method therefore presses, and that the justice criterion of §8 presses with it, is: who owns these new channels, and for whose benefit do they run—to whom does the value they produce return, and is there one who is rendered, in them, mere fuel? The decentralisation of the technical form of trust-production is no emancipation in itself; under the wrong relations of production it is a new mode of extraction. The synthesis is therefore not the triumphant slogan that decentralisation has freed us, but a new contradiction: an emergent form whose emancipatory potential is real and whose capture by capital is equally real, and whose outcome is not settled by the technology but by the relations of production that own it.

11.4 The three channels institutionalised

Within that dialectical frame, and not as free-standing recommendations, the three non-coercive channels of §6 can be seen taking institutional form, each with its emancipatory possibility and each with its characteristic danger of capture.

Reputation and credit systems are the institutionalisation of the relational-historical channel: they aggregate the scattered relational histories of §9 into a transmissible, institutional form of inductive trust, making a person’s accumulated track of dealing legible beyond the circle that witnessed it. This is their power—the institutional aggregation of inductive trust—and their danger is exactly the one named above: the degeneration of the credit record into a purely quantitative instrument of discipline, the relational history flattened into a score and the score turned to control.

Restorative justice is the institutionalisation of the self-committing channel. Rather than imposing a sanction by force, it creates the occasion for an offender’s real, irrevocable, self-implicating assumption of what he has done—the encounter with the one he has wronged, the genuine and costly taking-on of responsibility—which produces, through the high-fidelity likelihood of a real self-binding rather than through compulsion, the rebuilding of an expectation that a mere unenforceable judgment could not. It works the self-committing channel of §6.2 at the scale of the institution; and its danger is the hollowing of the encounter into a procedural ritual, a staged contrition that forges the self-commitment it should embody.

Public registration and transparency are the institutionalisation of the projective path of §10.2. They do not manufacture trust; they let a real state of performance—a real record of what has and has not been done—cast, in the symbolic register the evidentialist reads, the shadow by which it can be recognised. They make the real visible rather than producing the appearance of it; and their danger is the surveillance that the same transparency enables, the registry turned from a means of letting the real be seen into a means of total observation.

11.5 The place of the sword: relocation, not abolition

A boundary must be drawn as firmly here as the spine of the chapter is asserted, for the spine is liable to a dangerous over-reading, and the historical-materialist honesty that refuses to celebrate the emergent form refuses equally to romanticise the abandonment of force. The decentralisation of trust-production is not the abolition of enforcement. Force retains an indispensable role—but it is a different role from the one it was failing at. Force is not, and never was, the producer of trust; but it is the necessary backstop, the after-the-fact protection, and above all the protection of the vulnerable against predation and against the betrayal that the powerful might otherwise inflict with impunity. The weak still need the sword—not to generate the trust on which they deal, which force cannot generate, but to protect them when trust is betrayed and to guarantee that the betrayal of the powerful is not costless. Decentralisation of trust-production is not de-coercion; the sword is not taken away but relocated, from the false position of the producer of trust, which it cannot occupy, to its true position as the backstop that protects against the worst, especially on behalf of those least able to protect themselves. To forget this would be to abandon, in the name of a purer trust, exactly those whom the purer trust does not yet protect.

11.6 The juridical as the practice of a relational nomos

The chapter closes by connecting its analysis to the foundational paper and to the series, for the transformation it has traced is the institutional form of a movement the foundational paper named at the largest scale. That paper, drawing on the line of Marx and of the theorists of social reproduction, marked the overload of justice as a diagnosis without an exit, and held open the thought—against the horizon of the eventual withering of the state—that the law need not wither with it, but might pass from a law of action, the law of the sword that compels, toward a law of non-action, a law of pure relation, a relational nomos. The present chapter supplies the exit the foundational paper marked but did not develop. The centralised, force-based law—the law of action, the sword—decays as its material base dissolves; and a nomos built upon the production of relational trust—the law of non-action, the law of pure relation—germinates in the contradiction. But whether it is realised, or captured, depends, as the whole chapter has insisted, not upon the technical form but upon whether the relations of production are themselves transformed—upon who comes to own the new channels of trust, and upon whether the value they generate is returned, justly, to those whose dealing generates it, with no one rendered fuel. The optimisation of the juridical is, in the end, the freeing of justice from the impossible burden of being the last Other, the final guarantor of an order it was never able to underwrite by force; and the return of justice to its place as the practice, and the protector, of a relational nomos whose trust is generated elsewhere—in the coupled channels, at the scale of relation, that the rest of the paper has described. To that scale, the scale of a life and its nearest bonds, the paper now turns.

12. The Praxis of Trust

The paper arrives at its practical summit, and at the chapter toward which the whole has been moving. Everything established—the mechanism, the geometric reading, the criterion of justice, the structure of relational trust, the theory of translation, the optimisation of the juridical—has been knowledge for the sake of practice, and the paper would betray its own deepest claim if it stopped at knowledge. For the series holds that to settle in the theoretical, to take the cognition of a thing as the completion of it, is itself an instance of the pathology the whole paper diagnoses—the failure of settlement, the symbol that declares a circuit closed which no one has walked. A paper on the generation of trust that ended in an account of the mechanism, without descending to the practice in which trust is actually generated and lived, would be such a closed and unwalked circuit. The praxis is therefore not an application appended to the theory; it is the theory’s discharge of its own obligation not to settle. This chapter is long, and rises, at its centre, to the paper’s metaphysical summit; the present part develops its ground, from the reason there must be a praxis at all to the honest reckoning with what no praxis can guarantee, and the part that follows it raises the summit and returns, at the last, to the family and the room.

12.1 Why there must be a praxis

The reason the paper cannot end in theory is the paper’s own. It has argued that the poetic symbol refuses settlement—that it presents generativity rather than a closed structure, and forces its meaning to be walked rather than cashed; and it has argued that the failure of settlement, the cashing of a sublimation no one has undergone, is the very form of the symbolic usurpation it diagnoses. To stop at the theoretical cognition of trust would be to commit this failure in the paper’s own conduct: to cash, in the coin of understanding, a practice that has not been walked. The descent to praxis is thus demanded by the paper’s content, and not merely by a wish to be useful. And the descent is also the place where the paper’s argument and the author’s life, which the Prelude held apart as the near and the far edge of one question, are to meet: the praxis of trust is what the author must do, before the family of the Prelude, and it is what the theory has been clearing the ground to make sayable. The chapter is the point at which the knowledge turns into a way of acting, and refuses, in turning, to settle.

12.2 A polyphony of practices, and an ethical line across them

There is no single practice of trust, and the series’ polyphonic law holds in the practical register as it held in the theoretical: each of the frameworks the paper has drawn upon furnishes a practice of trust, and none governs the rest. The evidentialist practice builds the track record, accumulates the demonstrable, lets the facts mount until they speak. The deontological practice cultivates and shows a character, makes of the self a thing whose sincerity can be read. The relational practice tends the chains of §9, invests in the couplings through which trust is transmitted. The Daoist practice, of which more below, declines strategy and lets the real show itself in its own time. The institutional practice works the endorsements and the registers. These are real and distinct practices, suited to different receivers and different moments, and the paper does not rank them; the practical wisdom of trust is in large part the wisdom of which practice a given situation calls for, and the polyphony is, here as throughout, not a defect to be resolved into a single method but the very shape of a competence adequate to an object no single framework possesses.

But across all these practices runs one ethical line, and on this the polyphony does not extend: descriptively plural, the practices are ethically subject, every one, to the single criterion of §8. Whatever the practice—evidential, deontological, relational, Daoist, institutional—it is genuine where it produces an expectation by supplying a real ground that honours the other’s free judgment and opens a relation that returns to him, and forged where it induces an expectation by manufacturing a signal that handles him as a mechanism and draws upon him as fuel. The plurality is a plurality of legitimate forms; the line between the genuine and the forged cuts across every one of them, and is the same line in each. The chapter is polyphonic in its description of the practices and monophonic in its ethic of them, and the next subsections turn first to a correction the polyphony itself requires, and then to the forged practice the ethic excludes.

12.3 The justice of presentation

The polyphony of practices, as just set out, conceals a value-judgment that must be brought up and corrected, for it has quietly ranked the practices after all, and ranked them wrongly. There is a temptation, native to the Daoist and to a certain reading of the natural, to hold the practice of unstrategic self-showing—the letting of the real appear in its own time, without contrivance—as the highest and purest form of genuine trust-building, and to regard any active presentation, any rendering of oneself in the form the receiver’s cognition can best take, as already shading toward the manipulative. On this view, to present one’s sincerity in the grammar of the evidentialist would be a falling-away from the purity of simply, sincerely being; and the active translation of §10 would carry a faint guilt, as a strategy where strategy ought not to be. This subsection reverses that judgment, and the reversal is the hinge of the whole praxis.

The reversal turns on a recognition the paper owes to its own theory of translation: that the active presentation of oneself in the grammar the other can read—the very thing the unstrategic ideal suspects—is not a fall from purity but, in the right conditions, the way of countering an epistemic injustice. For epistemic injustice runs in two directions, and the unstrategic ideal sees only one of them. The first direction is the one everyone sees: the active evil of manipulation, the use of the other’s grammar to manufacture a signal of something that is not there, which corrupts his epistemic autonomy (§8). But there is a second direction the unstrategic ideal commits while believing itself innocent: the passive evil of refusing to translate—of insisting on showing oneself only in one’s own grammar and leaving the reception to fate. To say “I will simply appear, truly, in my own way, and the rest is not my concern” carries an unjust presupposition: that one’s own grammar is the neutral and universal one, in which the real, shown plainly, will be legible to anyone of good faith. But there is no such neutral grammar (§10); to show oneself only in one’s own and require the other to read one there is to impose one’s own grammar upon him as though it were the universal, and that imposition—however high-minded, however draped in the language of sincerity and refusal of contrivance—is itself an epistemic injustice, the more insidious for disguising itself as purity. The unstrategic ideal is not humble; it is, beneath the humility, the arrogance of treating one’s own way of being legible as legibility itself.

The correction is therefore exact. Unstrategic self-showing is not the highest form of genuine trust; it is genuine only where the grammars happen to match, and where they do not—which is precisely the situation of the Prelude’s suitor before the family—it is not purity but a concealed epistemic arrogance. And the active presentation of oneself in the other’s grammar, which the unstrategic ideal suspected, is revealed as its opposite: not a strategy shading toward manipulation but an epistemic hospitality, the actively going-toward the other’s way of knowing, the taking-on of the labour of being legible to him in his terms rather than requiring him to learn mine. To present a real sincerity in the grammar the other can receive is to recognise his epistemology as legitimate and equal to one’s own, and to move toward it; and this counters epistemic injustice rather than committing it, because it restores—rather than corrupts—the other’s capacity to make, on his own epistemological terms, a true judgment about a real thing.

The line between the genuine and the forged is therefore not the line between the unstrategic and the active, between wu wei and you wei, between letting-be and doing. That was the error the unstrategic ideal made: it located the danger in activity as such. The line is the one it has always been: whether what is presented really exists. Genuine presentation may be unstrategic—the letting-appear, where the grammars match—or it may be active—the epistemic hospitality, the translation into the other’s grammar, where they do not—and both are genuine, for both present, by whatever means, a thing that is real. Forged presentation is the active manufacture, in the other’s grammar, of a signal for a thing that is not there. The difference is in the reality of what is presented, never in the activity or passivity of the presenting; and in the situation of epistemic mismatch—the situation that matters, the situation of the Prelude—the active presentation, the epistemic hospitality, is not the lesser and more suspect form but the one that more fully respects the other and more fully counters the injustice of requiring him to read one in a grammar not his own. To present one’s real sincerity to the family in the form they can receive—the stable fact, the observable constancy, the evidentially legible—is not the abandonment of a deontological purity nor a step toward manipulation; it is respect for their epistemology, and the genuine active good, so long as what is presented is real.

12.4 The forged practice excluded

The ethic that runs across the polyphony excludes one practice absolutely, and it is worth naming the excluded practice precisely, for it is the practice that most resembles the genuine and is most apt to be mistaken for a legitimate technique. It is the practice of behavioural conditioning: the deliberate shaping of another’s expectations by the engineering of reinforcements, the production of trust by the systematic management of what the other experiences as the consequence of trusting, so that the trust is trained into him as a conditioned response rather than grounded in any real self-commitment. It is the manipulation of §8 raised to a method, and its evil is threefold, the three aspects of the one wrong the justice criterion identified, here compounded.

It is, first, the treating of the other as a means and a mechanism—the Kantian wrong—for it works upon him as upon a system to be conditioned, bypassing the free judgment that genuine trust addresses. It is, second, the epistemic injustice—for it installs in him an expectation by training rather than by grounds, corrupting his capacity to form, on real grounds, a true judgment of his own. And it is, third and distinctively, a temporal forgery, a failure of settlement spread across time: it manufactures, by the management of reinforcements, the appearance of a constancy that no real self-commitment underwrites, forging across time the very track record that the evidentialist reads as the shadow of a real sincerity—so that it is the precise isomorph, in the practice of trust, of the forged recommendation of §9, which manufactures the symbolic form of a coupling that does not exist. The conditioned constancy is a decohered constancy: the observable shell of a track record, with no real coupling beneath it. The practice is excluded not because conditioning never produces reliable behaviour—it may—but because the reliability it produces is a forged precision, a trust trained into the other as into a mechanism, drawing upon him as fuel while corrupting his capacity to know, and forging, across the very dimension of time that should reveal the difference, the shadow that should have been the real’s alone.

12.5 Hume’s gulf, and the honest reckoning with no guarantee

Beneath the whole practice of trust lies a limit that no practice can cross, and the honesty of the praxis depends on reckoning with it rather than concealing it. It is Hume’s gulf, met first in the diagnosis (§3.5) and now to be faced in the practical register. No practice of trust, however genuine, can guarantee its own outcome. The one who builds the track record cannot guarantee he will keep faith tomorrow; the one who makes the costly self-commitment cannot guarantee the circumstance that will test it; the one who trusts cannot cross, by any quantity of evidence or any depth of self-binding, the gulf between what has been and what will be. The practitioner of trust is in exactly the position the prior paper of the series assigned to the practitioner of any good practice: he cannot guarantee that his practice is good, cannot occupy the position of an Other of infinite reason who could certify the absolute causal connection between his conduct and the good it aims at, cannot stand outside the gulf and warrant the crossing. There is no view from which the keeping of the promise is assured in advance; there is only the promise, made and to be kept, across a gulf no guarantee spans.

This is not a counsel of despair but the precise location of the difference between the genuine and the forged at its deepest. For the difference between genuine and forged trust is not, at the last, a difference in the presence or absence of a guarantee—neither has a guarantee; the gulf is uncrossable for both. The difference is in the honesty with which the absence of a guarantee is borne. The forged practice pretends to a guarantee it does not have: it manufactures the appearance of an assured constancy, forging the certainty that the gulf forbids. The genuine practice makes its commitment in the full acknowledgement that nothing assures it—binds itself across the gulf precisely as one who knows the gulf is there and cannot be crossed by any warrant, and binds itself anyway. Genuineness is not the possession of a certainty the forged lacks; it is the honest bearing of an uncertainty the forged denies. The genuine promise is the one made in the knowledge that it cannot be guaranteed, and made, freely and without the false comfort of a guarantee, as a binding of oneself across a gulf one knows no evidence and no self-binding can close.

And here the praxis arrives at the threshold of its summit. For the question this raises is the one the rest of the chapter exists to answer: if no external guarantee can cross the gulf—if neither the sword, nor the evidence, nor the depth of the self-binding can assure that the promise will be kept—then what does the one who would genuinely bind himself do, in the face of a gulf he cannot cross and a future he cannot warrant? The answer is not to find, at last, a guarantee that the chapter has overlooked; there is none. The answer is to make, across the gulf no guarantee spans, a free and self-grounding leap—to constitute, by an act of one’s own freedom, the very constancy that no external warrant could supply. What that act is, how it is possible, and why it is the descent of something an older language would have called divine, is the matter of the summit to which the chapter now rises.

12.6 Relational divinity: the descent of the Other into the self-legislating subject

The leap the previous subsection named—the free constitution, across a gulf no warrant spans, of the very constancy no external guarantee could supply—is the paper’s metaphysical summit, and it must be set out with the fullness its weight demands, for it is a claim the series has approached from many directions and never, until here, fully stated. It can be stated so. The self-legislation, self-adjudication, self-keeping, and self-removal of every recourse outside the law one has given oneself is the mechanism by which a relational divinity descends into the world. This is the most original of the paper’s claims, and it stands, the author believes, in no prior text of the series nor in the foundational paper; it is set out here in full and in emphasis as the thing toward which the whole has been moving.

Begin with the historical movement of the divine, which earlier work in the series established in its direction without specifying its present mechanism. There is a movement, across the history of how the divine has been located, in three stages. First, a transcendent divinity: the god beyond the world, the divine power lodged in a heaven or a nature outside the human, before whom the oath is sworn and by whom its breach is avenged. Then, a symbolic divinity: the divine descended into the Symbolic, into the great Other of the law, the institution, the symbolic authority, the impersonal reason before which the modern subject stands—the god replaced by the order of signs that judges and binds. And now, in the movement this paper completes, a relational divinity: the divine descending out of the Symbolic and into the free practice of the relational subject, appearing no longer beyond the world nor in the order of signs above the subject, but in what the subject freely does, in relation, in the constituting of a bond.

The mechanism of this last descent—the present, concrete mechanism the earlier work marked the direction toward but did not supply—is the fourfold act. It is, first, self-legislation, in the full Kantian sense of autonomy: the subject gives himself the law, lays it down as law for himself, so that the constancy he could not derive from the past (Hume’s gulf forbids the derivation) is not a fact to be observed or predicted but a law to be enacted, posited by the will, grounded in the ought the subject lays upon himself—bypassing, rather than crossing, the gulf, by founding the constancy in an act of legislation rather than seeking it in an inference. It is, second, self-adjudication, the act the author adds and on which the whole turns: the continuous holding of oneself to the law one has given, the standing as one’s own judge over whether the law is kept, the self-arraignment and self-correction when one has strayed—which is the internalisation of Fuller’s congruence (§4.3), the bringing of one’s conduct into agreement with one’s announced rule, performed by the subject upon himself. A single free choice would be fragile, a resolution that the next moment might dissolve; but self-legislation joined to a continuous self-adjudication feeds each keeping of the law back into the loop that sustains it, and the constancy is thereby generated—turned out, moment by moment, by the self-legislating and self-adjudicating loop—rather than guaranteed in advance. It is, third, self-keeping, the actual holding to the law across time. And it is, fourth, the self-removal of every recourse outside the law one has given—which is the deepest sense of the burning of the bridge, met first in §6.2 and now to be completed.

For here the forward reference of the mechanism is discharged. The burning of the bridge was described, at §6.2, in its aspect as a high-fidelity signal—something done to be read by another, furnishing his model with hard-to-forge evidence. That was, it was said there, only one of its two faces and not the deeper. The deeper face is this: that one forecloses one’s own recourse not in order to send a message but in order to constitute, for oneself, a law—我立此为法, I have laid this down as law, and so that road no longer exists for me. The external law closes the road by force, mechanically, by the threat at the road’s end; self-legislation closes the road by freedom, by the laying-down that makes the road, for this subject, not a road at all. The reader who reaches this point will recognise that the trunk was pointing here all along: that beneath the burning of the bridge as a signal lay the burning of the bridge as self-legislation, and that the costly act read by another was, in its depth, a subject’s free and irrevocable binding of himself, in the manner in which a god is said to bind himself in a covenant—leaving no recourse, because the divine does not keep a recourse. The self-removal of recourse is the point at which the human exceeds the ordinary human, in which the part of the human capable of self-transcendence, capable of binding itself irrevocably as a god binds himself in a covenant, is realised; and it is the divine core of the vow.

12.6.1 Why the divinity is relational and not a solitary self-deification

The gravest danger of this claim must be met head-on, for a self-legislation that made itself its own law and its own judge would seem the very figure of hubris—the subject deifying himself, circling his own image, the solitary self-grounding that is the oldest form of the imaginary’s idolatry. The answer to the danger is the single consideration on which the whole claim’s safety rests, and it is ontological. The subject is not first a self and then a party to relations; the subject is constituted by relation, is relational through and through, before and beneath any self it could deify. On the relational ontology the series has held, there is no prior, separate self that could turn upon itself in a closed circle of self-worship; the self that legislates is already, in its very constitution, relational—made of the relations it stands in. And therefore the law it gives itself, and the divinity that descends in the giving, are relational from the first. The self-legislation is never “I, to myself” in a closed loop; it is “I, as a relational being, within the relations that constitute me, for those relations, toward those relations”—the laying-down, by a self that is made of its bond to the other, of a law for the keeping of that bond. The betrothal vow is exactly this: a self-legislation performed by a self constituted in the relation to the one it vows to, laying down, for that relation and within it, the law of how it will be kept.

This relational constitution is, further, what dissolves the tension that would otherwise destroy self-adjudication. A solitary self-judge is the surest route to self-deception: nothing is easier than to acquit oneself, to flatter the law into agreement with the conduct, when one is one’s own only judge. But the relational self-adjudication is not solitary—it is conducted, of necessity, within the regard of the relation. The other’s witness, the gaze that the empty place of the das Ding holds open, is not an external check added to the self-adjudication from outside; it is internal to the very structure of being a relational subject, which is constituted in being-seen as much as in seeing. The honesty of the self-adjudication is therefore not secured by an external sword—there is none, and the gulf forbids one—but by the ontological fact that the self which adjudicates is a self already constituted in relation, already under a regard it did not add and cannot remove. Relational divinity carries its own guard against self-deception, and the guard is not a borrowed sword but the subject’s own relational being.

12.6.2 The personal-divine form of positive holonomy

This fourfold act is the personal and divine form of the positive, adiabatic holonomy of §7. The loop closes—the constancy is sustained, turn upon turn—not because an external pump drives it (no transcendent god, no symbolic Other, no sword: these would be the non-adiabatic driving, the borrowed phase), but because the relational subject’s own self-legislating and self-adjudicating loop closes it from within, endogenously, accumulating its phase as its own. It is the discharge, at the level of the person, of the claim of §5, that autonomy is the highest form of freedom—not the merely negative freedom of being undetermined, but the positive freedom of giving oneself the law. And it gathers to itself the deepest voices of the series and of the traditions it has drawn upon: the foundational paper’s intimation that the tendency of the Other toward its final dissolution opens the way to a renewed approach to a pure relational being; the Buddhist teaching that mind is itself the Buddha, that all dharmas are made by mind—是心是佛，一切法由心造; the Daoist refusal of the external warrant; and the betrothal document’s own closing line, 本乎此心，自成永恒—that the eternal is founded not upon any external guarantee but here, in this heart, which is itself the ground.

12.6.3 The inversion of value: the non-binding vow as the more divine

A consequence follows that inverts the ordinary ranking of instruments, and it is the point at which the whole paper’s central paradox is resolved into a claim. The ordinary ranking holds the marriage certificate above the betrothal vow, the contract above the memorandum of understanding, because the former in each pair carries the backing of the state, the force of law, the reliability of the sword. From the standpoint of relational divinity, the ranking is exactly inverted. The memorandum of understanding is more divine than the contract, the betrothal vow more divine than the marriage certificate—precisely because they carry no force, have no external guarantee to borrow, and so can be sustained only by the subject’s own self-legislation, self-adjudication, and self-keeping in the real.

The logic is this. The contract and the marriage certificate outsource the threefold function—legislation, adjudication, keeping—to the symbolic Other: the law is the state’s, the judging is the court’s, the keeping is compelled by the sword. The parties borrow an external, symbolic divinity to underwrite their bond, and need not legislate, adjudicate, or foreclose their own recourse, for the functions are outsourced. The memorandum and the vow, having no force, force these functions back upon the subject: with no external sword, the legislation can only be self-legislation; with no external court, the adjudication can only be self-adjudication; with no external compulsion, the keeping can only be self-keeping and the self-removal of recourse. The parties have nowhere to borrow an external divinity, and must therefore become the legislator, the judge, and the keeper themselves—must let the relational divinity descend into themselves, or the bond does not hold. The “defect” of having no force is thus the very condition of the divinity: precisely because no external sword keeps the vow, the keeping must issue from one’s own free, irrevocable self-binding, which is the realisation, and not the absence, of the divine. This completes the geometric reading of §7.5: the contract and the certificate are non-adiabatic, their loop held closed by the external pump of the state’s force, their phase borrowed; the memorandum and the vow are adiabatic, their loop closed by nothing but the subject’s endogenous self-legislation, their phase their own—the purer specimen of genuine holonomy, into which the certificate admits an admixture of the external pump. And it raises the counter-intuitive claim of the diagnosis to its summit: the betrothal vow is not merely, as was first said, possibly more productive of real trust than the complete contract; it is, in the dimension of the divine, the purer thing, because it forces the divine back into the free practice of the subject rather than borrowing it from an Other.

The honest boundary must be drawn, lest the inversion be pressed past what it claims. It is not said that the memorandum and the vow are superior to the contract and the certificate in every dimension. In the real functions of protecting the vulnerable, deterring betrayal, and providing a remedy, external force remains necessary and is not to be romantically discarded—exactly as §11 insisted in relocating rather than abolishing the sword. The claim is narrow and precise: that in the dimension of the divine, of the genuine, the non-coercive instrument is the purer, because it forces the divinity back into the subject’s own free practice. The weak still need the sword’s protection at the scale of institutions; but the divinity of a genuine bond was never lodged in the sword, and is found, at the scale of relation, only where the sword is not. It is not, therefore, that one should forgo the certificate for the vow; it is that, beyond and above and before the certificate, what makes the bond a bond at all is the self-legislating relational divinity, and that this is most purely realised in the vow that has no force.

12.7 The concrete operation

From the summit the chapter descends to the concrete, for the praxis must issue in something one can actually do, and the abstractions of divinity must be cashed—here legitimately, for here the path is walked and not merely declared—into an operation. The operation has three movements, and a condition of time.

The first movement is to identify the other’s grammar of trust, and its configuration (§10.1): to discern, of the one whose trust is to be won, which of the grammars of §10.1 dominates his reading, in what weighting, so that one knows the language into which one’s sincerity must be translated. One does not present in the abstract; one presents to this configuration, and the discernment of it is the first and indispensable act.

The second movement is to produce, in the other’s grammar, a real possibility of expectation—to translate, by the paths of §10, and above all by the epistemic hospitality of §12.3, one’s real sincerity into the form the other’s configuration can receive. For the evidentialist family, this is the projection of the real into the register of the observable and the accomplished: not the manufacture of a track record one does not have, which would be forgery, but the patient letting of one’s real self-legislated constancy cast, in the evidential register, the shadow by which it can be recognised. The thing presented must be real—must be a sincerity that exists and a constancy that is genuinely self-legislated—and it must be presented in the grammar the other reads. This is the whole of the second movement: a real thing, hospitably translated.

The third movement is the right use of relational trust (§9), and it carries the chapter’s most delicate ethical discrimination. The trust of the family runs, very largely, along the chain that connects the suitor to them through the one they already love—she is, in the structure of the network, his recommender, and the trust may reach them along that link as it reaches along no other. The right use of this is to let the trust transmit because the coupling it transmits is real—because she is, in the genuine sense, the co-author of the bond, and her steady bearing toward it is the unfakeable real component (§9.2) that the family, even reading evidentially, can read. The wrong use—the forged use, excluded by the justice criterion—is to instrumentalise her as a mere channel, to deploy her to “work on” the family, to siphon her relational capital toward a coupling between the suitor and the family that has not in fact occurred. The line is exactly the line of §8: between a real coupling that transmits because it is real, and a relational capital siphoned to transmit a coupling that is not there. The one honours her as co-author; the other renders her fuel.

And the condition of time. None of this can be hurried, and the attempt to hurry it is itself a tell of the forged. The self-legislated constancy casts its evidential shadow only as it is actually lived across time; the real coupling transmits only as it is really borne; the shared grammar of mutual translation grows only in the continued interaction. Time cannot manufacture the coupling-trust, which is real from the first or not at all; but time is what lets the real cast, in the register the evidentialist reads, the shadow sufficient to be recognised. The praxis is therefore not a technique that produces trust on demand but a way of living that lets a real thing become, across time, recognisable to one whose grammar is not one’s own. It cannot be accelerated; it can only be deepened.

12.8 The return to the room

The chapter, and with it the body of the paper, returns at the last to the scene from which the Prelude set out: to the one who must sit across from a family not yet his own, and be trusted. Everything the paper has built converges on that room, and it converges not on a method for controlling what happens there but on a way of being present in it.

For the end of the praxis is not the mastery of the other’s trust—trust cannot be mastered, only generated, and generated across a gulf no one can guarantee. The end is to be ready to be truly present: to have become one who has a real sincerity to present, and to present it in the grammar the family can receive, with the epistemic hospitality that honours their way of knowing rather than demanding they learn his. What he can give them is not an external guarantee—he has none to give, and to manufacture the appearance of one would be the forgery the whole paper has worked to refuse. What he can give them is the sight of a person who legislates his own law and holds himself to it—who has laid down, for the relation that constitutes him, the law of how he will keep faith, and who adjudicates himself against it continually, in the open, where it can be seen. This constancy is real, for it is the product of the self-legislating loop; and being real, it casts, in time and in the evidential register the family reads, a true shadow—a track of consistent, self-correcting keeping that the evidentialist can read as what it is, the unforgeable sign not of a compelled obedience nor of a performed display but of a free subject keeping the law he gave himself. That is what there is to give: not a guarantee, which cannot be given, but the presence of a self-legislating relational subject, whose constancy is its own and whose keeping of faith is the descent, in him, of a divinity that needs no sword—offered, across the gulf, to be trusted or not, freely, by those he cannot compel and would not if he could.

13. Conclusions

A paper that has argued, throughout, that its object is one no single framework possesses, and that the truth of such an object lives not in any one voice but in the dialogue among them, cannot end in a single concluding voice without betraying its own claim. The conclusions are therefore plural, and they are plural by necessity and not by modesty. Each of the frameworks the paper has drawn upon is given here its own conclusion: what it can say about the generation of trust, said as firmly as it can be said, and what it cannot say, marked as firmly. The point of the exercise is not a survey but an enactment: the paper’s form is meant to be its last argument, the demonstration, in the shape of its own ending, that the generation of trust is known only polyphonically, in the dialogue of irreducible positions none of which governs the rest. The discipline the series imposes on such an ending is exact, and the paper holds to it: each framework must speak to the edge of what it can warrant, hard and fallible and willing to assert, for a polyphony of voices that each hedge into vagueness is no polyphony but an evasion. Each voice below says something, and says where it must stop.

The jurisprudential conclusion

The philosophy of law can say this: that the anchor of legal validity has migrated, across the century of its own reflection, from force toward trust, and that the law has recorded the migration in its own most technical doctrine, conceding at its edge, in promissory estoppel, that reliance and not compulsion is the true ground of a promise’s binding force. It can say that force was never the cause of obligation but a substitute, now failing, for the production of expectation. It cannot say what trust is, nor how, mechanically, it is produced; jurisprudence can locate the anchor and trace its movement, but the nature of the thing the anchor is moving toward lies outside its competence, and it must hand that question to others. Its limit is that it can say where validity rests without being able to say what, at that resting place, validity consists in.

The predictive-coding conclusion

The predictive account can say this: that trust is the precision a generative model assigns to the prediction that a promise will be kept; that this dissolves the paradox of a present expectation directed at a non-existent future, since prediction is the brain’s standing activity and requires no causation running backward from the future; that the expectation is produced through three coupled channels whose multiplicative coupling makes genuine trust robust where any single channel is fragile; and that the crisis of the symbol is the rational collapse of a population’s precision-weight on a channel that forgery has made unreliable. It cannot say that trust simply is a precision-weighting and nothing more; it offers a description at the computational and implementational level, and it must not be mistaken for the truth beneath the others. Its limit is the limit of every mechanistic account: it says how the thing is computed, and is silent on what the computing is, lived from within, and on whether it is just.

The geometric conclusion

The geometric reading can say this: that a promise kept across time traces a closed loop, and that the sign of the holonomy the loop accumulates distinguishes the bond that deepens from the one that merely repeats and the one that drains; that the genuine, non-coercive promise is the adiabatic loop whose phase is its own, closed by no external pump, and the coerced bond the driven path whose phase is borrowed; and that this furnished, at last, the missing structure—a spectral gap—for a problem the series had left open. It cannot, in this paper, furnish the phase with a fully rigorous carrier; it has worked as a structural analogy, and the formal closure lies in another paper. Its limit is that it shows the shape of the distinction with precision while deferring the full formalisation of what bears the phase.

The quantum-dynamical conclusion

The quantum-dynamical reading can say this: that relational trust shares the formal structure of an entangled state, a property of correlations and not of nodes; that its failure under the crisis of the symbol is a systemic decoherence, the correlational shell preserved while the coherence drains; and that its transmission along the chain is the structural counterpart of entanglement swapping through an intermediary. It cannot say—and must say, loudly, that it does not say—that trust is in any literal sense a quantum phenomenon. It is a structural isomorphism and not a physical reduction, one more formal language for a structure and not the structure’s secret physical truth. Its limit is the limit of every analogy: it illuminates by a likeness, and the likeness is not an identity.

The psychoanalytic conclusion

The psychoanalytic reading can say this: that the production of trust circles a place that cannot be filled—that genuine trust, like the genuine poem, encircles the empty place of the das Ding rather than filling it with an idol, and so requires no victim, where the forged trust fills the place with a positive object and must consume someone to do it; that the wager of trust is a venture made across a lack no evidence closes; and that the great Other, the guarantor that would certify the crossing, does not exist. It cannot translate the lack it speaks of into the positive terms the other frameworks want; its truths are truths about an absence, and they resist the cashing-out that would betray them. Its limit is that it speaks most truly where it can least be made to settle into a positive doctrine.

The political-economic conclusion

The political-economic reading can say this: that the forms of trust-production are thrown up by the material conditions of their time and change as those conditions change; that centralised, force-based justice was the historically adequate form for a society of market strangers, and is failing now not through a decay of values but through a change in the material base; that the decentralisation of trust-production is a recomposition already under way, forced by that change; and that its emancipatory potential is real but not guaranteed, for the new channels are, under the present relations of production, being captured by capital toward a domination subtler than the sword. It cannot, from its own resources alone, settle whether the recomposition will emancipate or enslave; that depends on whether the relations of production are transformed, which is not a question the analysis of trust can answer by itself. Its limit is that it can diagnose the contradiction without being able to guarantee its resolution.

The Daoist conclusion

The Daoist reading can say this: that the genuine bond is the one sustained by no external pump, closing upon itself by the completeness of what each freely brings; that to grasp at the guarantee is to lose the thing—that the one who would secure the bond by force forecloses the very deepening that only the unforced loop accumulates; and that the eternal the betrothal vow speaks of is founded upon no external warrant but here, in this heart, which is itself the ground—本乎此心，自成永恒. It cannot say this in the language of mechanism or of proof; its truth is shown in a way of holding the thing, not demonstrated in a way of establishing it, and to press it into demonstration is to lose it as the grasping loses the bond. Its limit is the limit of the unsayable it points to: it can indicate the ground, and cannot, without contradiction, prove it.

Why there is no conclusion above the conclusions

It remains to say why the paper does not now gather these into one—why there is no synthetic conclusion above the plural conclusions, no master voice to adjudicate among the frameworks and pronounce the truth they severally approach. The reason is the paper’s deepest commitment, and it is the same reason that runs through the whole: there is no metalanguage. There is no position above the frameworks from which their several truths could be translated, without loss, into a single truth that contained them; the great Other who could occupy that position does not exist, and to write a synthetic conclusion would be to claim his place. The frameworks are irreducible epistemic positions on one object, and the object is known—this is the paper’s form enacting the paper’s content—only in the dialogue among them, in the mutual illumination that is itself a good cycle, a polyphony that generates understanding by the movement among voices rather than by their resolution into one. To conclude above the conclusions would be to settle what must be walked, to cash the dialogue into a doctrine, to commit, in the paper’s last act, the very failure of settlement the paper has diagnosed throughout. The absence of a master conclusion is therefore not an incompleteness; it is the paper’s final and formal assertion of its thesis. The knower of trust, like the truster, must forgo the guarantee—must hold the plural truths in their dialogue without the false comfort of a synthesis—and the paper, declining to synthesise, does in its form what it has asked, throughout, that the truster do in his life: it bears the absence of a guarantor, and binds itself to the plural truth anyway.

The three concentric circles

One thing may be said, not as a synthesis above the frameworks but as an observation that runs beneath them all, and it returns the paper to the shape it announced in the Prelude. The question of trust was posed across three concentric scales: the intimate vow between two; the promise carried outward to a family; and the public institutions of contract and the courts. And the paper’s finding, traced at each scale by each framework, is that the ground of trust is, at all three, the same ground. At the most public scale, the validity of law was found to rest, beneath the sword, on a reciprocal keeping of faith; at the most intimate, the bond was found to rest on a self-legislating relational divinity that needs no sword; and these are not two grounds but one, seen at two distances. What holds a society together at its most public and what holds two people together at their most intimate is the same thing: a trust generated not by force but by the free, self-binding, mutually relied-upon keeping of faith, across a gulf no guarantor spans. The three circles are concentric because they share a centre, and the centre is the relational divinity that descends wherever a free being binds itself, in relation, without recourse to a sword. The most intimate and the most public meet at that centre; and the betrothal vow on its red silk, which no court will enforce, is found to hold within it the ground on which even the law, at the last, is found to stand.

Envoi

This paper began with a vow written on red silk, on a day the almanac called auspicious, in the full knowledge that nothing in the world would make it binding; and it set out to ask how such a thing—a promise that no sword stands behind—could nonetheless be believed. It has answered, across its length, in many voices: that force was never the cause of trust but a failing substitute for it; that trust is the precision a mind freely lends to a future it cannot verify; that the genuine bond is the loop that closes by no external pump and accumulates, turn upon turn, a phase that is its own; that sincerity has no universal language and must be translated, across the borders of epistemology, into a grammar the other can read; and that the keeping of a promise no sword enforces is the descent, into a free and relational subject, of something an older language would have called divine.

What the analyses leave out is the thing itself, and it returns at the close as the matter that all the analyses were about. A vow kept not because it must be but because one has made it one’s own law; a constancy that is not compelled and not predicted but freely generated, morning upon morning, in the holding to a word that no one could force; a trust extended across a gulf no guarantee spans, and met, and not betrayed. These are not the proxies the paper has had to work with but the real of which they were the shadows, and they are lived before they are analysed and exceed every analysis of them. The moon over Kagoshima under which the question first took its human form; the red silk and the chosen day; the gravity of a brush that bound nothing and meant everything—these are the matter, and the paper has been, throughout, their long translation into a grammar the cautious and the evidential might also read.

A promise so kept is not secured and then possessed. It is generated in the keeping, and a bond endures by the continued, free, self-legislated keeping of faith, and ceases when the keeping ceases. There is no guarantee of it to be had and held; there is only the binding of oneself, again and again, across the gulf, in the knowledge that nothing assures it and the freedom that asks for no assurance—the laying-down, in this heart, of a law that is its own eternity.

For the forest girl,
to whom a promise was made that no sword stands behind,
and which is, for that, the more wholly given;
that what is bound here be bound by nothing but this heart,
and, bound by this heart, hold—across every gulf, through all our years.

本乎此心，自成永恒。

Acknowledgments

I owe this paper, and more than this paper, to the one I love most: the forest girl, whom I met in Tokyo in the year 2022. She is good, and pure, and lovely, and wise, and turned always toward others before herself; she loves the forest, and travelling, and the study of cultures, and the political economy of how we live together. This paper, on how trust can be generated where nothing enforces it, would not have had even its question without her, for it was in the making of a promise to her that the question first took its human form.

And here, at the close, I would do the thing the paper has spent its length learning how to say. The paper has not only argued that a relational divinity descends into the world through the self-legislation, self-adjudication, and self-keeping of a free subject who binds himself without recourse; it has tried to show that such a binding is the realised, and not the defective, form of a promise. It remains for me to do it rather than to describe it, and to let the argument, in its last page, become the act it has been about.

I lay it down, here, as a law for myself, and I will keep it with the whole of my life: 我将如此待她. This is not a promise that any sword stands behind. No court will enforce it; no penalty attends its breach; nothing in the world compels me to keep it. Its whole force is that I bind myself to it, here, freely and without reserve, and will hold myself to it, and have closed to myself every road that leads away from it. I give myself this law, I will be my own judge of my keeping of it, and I have left myself no recourse outside it.

This document carries a trustworthy and undeniable timestamp, and I have set it down here, in a text that bears my name and will outlast this moment, not in order that anyone should see it. Its first work is not to signal but to bind: it is the instrument by which I close my own retreat, so that the road away is, for me, no longer a road. The timestamp is not a sword; it cannot make the vow enforceable, and I do not pretend that it can. What it can do is something I do to myself and not against any consequence: it makes it impossible for the self I will be to deny, to the self I am now, what I have here laid down—it makes me the standing judge of my own keeping of faith, and forecloses the quiet revision by which one lets a vow lapse unremarked. The timestamp is the instrument of my self-adjudication, and of nothing else. The vow is genuine not because the timestamp makes it certain—it cannot, and the gulf the paper described forbids that it should—but because I make it in the full knowledge that nothing assures it, and make it anyway, freely, as a binding of myself across a gulf I know no warrant spans.

This is the paper’s most private return, and its last. It set out from the making of a promise on red silk, knowing nothing of why such a promise could be believed; it has travelled through the symptom, the diagnosis, the law, the mechanism, the freedom, the divinity; and it ends by making the promise again—the same act, walked a second time, with everything the paper has come to understand now gathered into it. The first time, the vow was made by intuition, because it was right. This time it is made in the knowledge of why a promise that no force backs is the more wholly given, and of what it is that descends into a free subject who binds himself without recourse. The same act, the second time around, returns not to where it began but a little higher; and that is the only kind of return the paper has believed in.

中文

亲密关系哲学与正义理论 · 第十三篇

无强制之诺的因果

论信任、承诺，以及在他者之剑缺席时期望的生成

〔 工作草稿 〕

论一个无强制力的承诺如何生成信任、真诚如何跨越认识论的边界而被翻译，以及一种关系性神性的降临

万宏

工作草稿，不供征引或传阅

死生契阔，与子成说。
历死生，经离合，与你立下此言。

匪报也，永以为好也。
非为回报——而是要与你，永以为好。

在无物束缚两心之处，它们依然相守；而一个誓言之被相信，若它真能被相信，正是在那无物束缚它的地方。

——《诗经》，邶风·击鼓 与 卫风·木瓜；以及作者本人的一句话。

致那位森林女孩，
我向她许下了一个承诺，
其后没有刀剑，
而正因如此，
它被给得更为完整。

本乎此心，自成永恒。

——致我最爱之人。

摘要

本文追问：一个不携带任何强制力的承诺——一个誓言、一份婚约文书、一份谅解备忘录——如何仍能生成真实而有据的信任。本文从当代境况的症候学出发：在每一尺度上，担保的装置都在不断增殖，而它所保障的信任却愈发稀缺；本文将这一失败诊断为一个关于因果的错误：把力，即机械的原因，误当作一个本属推断性的、自由的结果——信任——的原因。本文经由法哲学，追溯法律效力之锚从力向信任的迁移，这一迁移被法律在其自身的技术边缘以”允诺禁反言”原则记录了下来；并论证：推断的原因与机械的原因属于不同的种类，是那种把一个自由存在者当作自由者来对待的原因。本文的核心贡献是一个机制：信任，是一个接收者的生成模型赋予”承诺将被守持”这一预测的精度，它经由三条耦合的通道——一个结构性的先验、一个高保真的自我承诺的似然、以及一个被押注的先验——而生成；这三者的乘积式耦合，使得真实的信任在任一单条通道皆可伪造之处仍然稳健。本文为这一机制给出一个几何的读法，以和乐（holonomy）的语言；并为关系性信任给出一个结构的读法，以开放量子系统动力学的语言；陈述了那条区分真实信任与伪造信任的正义判准；发展了一套真诚跨越不可通约之认识论而被翻译的理论；并在历史唯物主义的语域中，把司法之优化重新构想为强制的去中心化。本文在其顶点处上升到这一主张：一个不留任何退路而自我绑定的主体，其自我立法、自我司法与自我守持，正是一种关系性神性降临于世的机制——于是，在神性的维度上，无强制之诺被证明是”承诺之所是”的更纯粹的实现。本文以复数的结论作结，在其自身的形式中，拒绝了其论旨所禁止的那种综合。

序曲

在本文动笔前的数周，作者坐下来写一份婚约文书。它写在红绢之上，以文言文写就，选在一个因黄历称其为吉日的日子；而它是在如下的全然知晓中写下的：世间没有任何东西会使它具有约束力。没有公堂会强制执行它。其中没有任何条款可被诉请。它未规定刑罚，也未授予任何救济。对它可能被违背的那一天，它未设下刀剑；而在那执笔之手的背后，并无任何力量能够强迫这只手去守持它。以衡量一份合同的每一标准来看，它都是一份有缺陷的文书：一个承诺，却承诺不出任何它可被迫履行之物。

然而它并非轻率写就，而对那为之而作的两人而言，也并非轻率领受。择日、用绢、运笔之郑重——这些不是、或不只是戏。在它的写就之中，有某种合同的写就所不为之事正在被做着，而这文书之缺乏强制力，并未减损此事，反而奇异地似乎使之更深。合同是签了归档的；这一份是被守持的。在写下它的人那里，其背后刀剑的全然缺席，并未被感受为一种需要道歉的欠缺，而是被感受为此物之所以是其所是的条件。若曾有一项刑罚，这承诺便会是一个更小之物。它什么也不约束，正是它能够约束一切的缘由。

这是一种奇异的经验，而更奇异的是，发觉自己竟无法解释它。现代世界借以思考承诺的整套装置——合同、刑罚、可执行的义务、随时准备强制的制度——所建构的，恰恰是为了供给这份文书显眼地缺乏之物。人们被告知，并且半信半疑：一个承诺的可靠程度，与支撑它的力成正比；不可被执行者便不可被指望；自我绑定，便是把自己暴露于一项自己宁愿回避的后果之下。按那种说法，婚约文书是一项感伤的无效物，一个底下空无一物的漂亮姿态。但它不是这样被写下的，也不是这样被领受的。在那半信半疑的说法之下，奔流着一种相反的、更古老也更难撼动的信念：那不索求刀剑的承诺，不是更弱的而是更真的；一个人能被强迫去守持的绑定，恰恰在他能被强迫的限度内，还根本不完全是一个绑定。

作者不打算通过判定这两种信念之一为幼稚来化解此事。两者都是真实的，而它们之间的张力不是一个有待澄清的混乱，而是一条有待循之而下的断层线。因为那引出此文书的同一经验，在其后的日子里，引出了一个更焦灼、更切实的问题，而这其实是同一个问题穿上了工作服。承诺已被许下，并在许下它的两人之间被相信。但它如今必须被携向外界——尤其是携向一个家庭，他们尚不认识作者，他们不会像写下红绢的两人那样去读红绢，而对他们来说，一个吉日里运笔之郑重，若说还有什么意味，那也微乎其微。在作者的理解中，他们是这样的人：他们认为是就是是、不是就是不是——有就是有，没有就是没有——而他们会合情合理且毫无恶意地追问：有什么可凭依的。对他们而言，誓言赤裸的真诚不是证据。而在这里，”一个真承诺无须刀剑”这个让人安心的想法，径直撞上了它自身的困难：因为，倘若最真的承诺恰恰是那个不提供任何外在担保的承诺，那么，出于同一缘由，它也正是那个给予”必须被宽慰之人”最少可凭依之物的承诺。正是那使它成其为誓言的纯粹，使它在一个索求依据的人看来，几乎什么也不是。

于是这经验，在其近端与远端，向着一个唯一的问题敞开，而这正是本文之存在所要追问的问题。一个没有强制力的承诺如何生成信任？ 一句关于一个尚不存在之未来的话语，它不威胁任何后果，也不能强制任何遵从，它如何在另一个人那里，此时此地，产生一个它将被守持的真实期望——而且，倘若那更古老的信念是对的，它越是不依赖于能够强制任何东西，便越确凿地产生这期望？这问题同时是能被追问的最亲密的与最公共的问题。它是这样一个问题：作者如何能被那些他尚未赢得其信任的人所信任；而它，本文将论证，正是法律本身在其漫长的退离刀剑的过程中，已经悄然耕作了一个世纪的那个问题。它始于一个日子的择定与运笔之郑重，而在它完结之前，它触及的远至义务的根基与那旧时语言会称之为神性者的降临。

本文正是从这里开始：不是从一个论旨开始，而是从一个本不该奏效、却奏效了的誓言开始。

1. 导论

本文是关于亲密关系哲学与正义理论的系列中的第十三篇。序曲经由一个单一的经验进入这个问题，而本节则把它陈述为一个问题，固定其范围，并列出本文宣称要确立之物。问题正是序曲所抵达的那个：一个不携带任何强制力的承诺如何生成信任。 一个誓言、一份婚约文书、一份谅解备忘录、一个赤裸的承诺——一句把作出它的人绑定于某种未来行为、却没有任何公堂会去执行、不威胁任何刑罚、且其背后并无任何力量能强制履行的话语——却仍然在恰当的条件下被相信；而这相信不是轻信，而是一个真实的、且常常有据的期望，即此事将被守持。本文追问：这期望是经由什么机制被产生的。

这问题有一种结构，使它乍看之下显得不可能。信任指向一个未来，而未来尚不存在、不可被验证；那索求信任的符号没有强制力，因而不能通过威胁一项后果来保住未来。于是，需要解释的便是：一个关于不存在之物、且不携带任何后果之担保的符号，如何在当下产生一个真实的期望——某种近乎从无中生成之事。本文的核心而原创的贡献，是对那一机制的解说。它的其他贡献，如下文所列，在很大程度上是它的应用、延伸与后果。

范围

本文关切的是非强制性符号对信任的生成，跨越三个同心的尺度：两人之间亲密的誓言；被携向更广圈子、尤其是家庭的承诺；以及公共制度，即合同与法院，同一问题在社会的尺度上在其中出现。本文把这些尺度当作连续的来对待：同一机制被论证在每一尺度上运作，而本文在它们之间移动，而非把自己限于其一。本文不提供一种把信任当作一种心理倾向的一般理论，也不提供一份关于信任的经验社会学的综述；它为非强制承诺这一特定情形提供一个机制，并追踪那一机制的触及之处。在论证需要法理学、认知科学、精神分析、政治经济学以及道家与康德传统之资源的地方，本文像系列中先前各篇那样取用它们：作为关于同一对象的若干不可化约的认识论立场，其中没有任何一个统辖其余。

与奠基之作及与本系列的关系

本文与一部奠基性文本处于一种确定的关系之中，应当把本文读作对那一文本之两个诊断主题的持续展开。奠基之作铺设了本系列作为其一部分的那个框架，并以其近半的篇幅，从若干维度诊断了现代条件下符号的危机。本文尤其取用其中两个维度——本真性与信任的危机，以及正义的过载——并对它们施以那奠基之作（其目的是诊断性的）所搁置的、彻底的机制性分析。奠基之作确立了信任已陷入一场其机制未被它指明的危机，而本文则指明那一机制；奠基之作诊断出司法的一种过载与强制的一种失败，而本文则把那一诊断推进至如下主张：力已作为一个信号而失效，并提供奠基之作标出却未展开的那条出路。与更广系列的关系同样确定。本文取用本系列早先发展的几何相位或和乐装置、其价值的三域理论，以及其关于”无法担保自身实践为善的实践者”的解说；并且，如本系列所要求的那样，遵守那条不容任何单一框架吸纳其余的纪律。

贡献

本文宣称确立如下各点。第一，且为核心：一个非强制承诺所产生的期望，是由一个可被赋予确定形式的机制所生成的：信任，是一个接收者的生成模型赋予”承诺将被守持”这一预测的高精度，而这精度经由三条耦合的通道——结构性的、自我承诺的、被押注的——而被供给；它们相乘而非相加，于是它们之间那难以伪造的一致性，便是有据信任的来源（§6）。第二，一个法理学的结果：强制力不是义务的原因，而是期望之产生的一个替代机制；法哲学的内部发展，在一个世纪里，已把法律效力之锚从力移开，移向信任——这一运动是法律本身在其技术边缘、以允诺禁反言原则记录下来的（§4）。第三，一套关于关系性信任的解说——关系性信任即沿着一个关系网络之诸链接传递的信任——把它当作一个横跨全部三个域的现象，连同一个关于它如何在符号危机下失败的新的具体化说明：经由信任之符号载体与其本应承载之真实耦合的分离（§9）。第四，一套关于真诚跨越不可通约之认识论之边界而被翻译的理论，连同其随附的伦理（§10）。第五，一个在历史唯物主义语域中发展出来的、指向强制之去中心化的司法改革方向（§11）。第六，并作为本文形而上学的顶点，一套把自我立法解说为一种关系性神性之降临的论述，其后果是：那个没有强制力的承诺，就此而言，可能反而是更充分地实现了的（§12）。

关于方法与所承认者的一则说明

两项承诺贯穿全文。第一项关乎其形式装置与科学装置的地位。本文为信任的机制取用预测编码，并为关系性信任的结构取用开放量子系统的动力学；它这样做，正如先前各篇取用几何相位那样，是作为机制性的与结构性的模型，而非作为关于信任归根结底是什么的还原论主张。预测编码供给一个在计算层级与实现层级上的描述；它不取代关于同一对象的现象学描述、法理学描述或伦理描述，它也不是一个吸纳它们的元框架。同一纪律统辖§9中量子动力学语言的使用：那里所宣称的是一种结构同构，而非一种物理还原。第二项承诺是本系列对一个统辖性结论的拒绝。本文不以一种综合作结，而以复数的结论作结，每一框架陈述它能说什么与不能说什么，因为那个对象——无担保者而生成的信任——是没有任何单一框架所能据有的，而本文的形式意在把这一点既加以演示又加以断言（§13）。

局限与所留待者

本文对其局限直言不讳。和乐装置在此被当作一种结构类比来使用，而非被发展到完全的形式闭合，相关的形式纲领由本系列的另一篇承担；它所提供的信任语法谱系（§10）被当作足以达成其实践目的者来提供，而非详尽无遗者，且容许更细的细分；预测编码解说被推进的深度，尤其是朝向一种以”对预期不确定性的主动削减”为基础的第一性原理处理，是被刻意限定的，更强、更技术性的展开是被标记而非被实施的。§9的量子动力学同构是作为一个形式类比来呈现的，它邀请、却不在此供给它自己的展开。这些都被标记为留待解决，而非被掩盖，以符合本系列那条区分”所宣称者”与”所推测者”的纪律。

全文计划

本文的进行如下。它先集合当代信任境况的诸症候，而尚不解释它们（§2）；并对它们作出诊断，取用奠基之作以及休谟与康德的更古老问题，得出”力不是信任之原因”这一论旨（§3）。随后它把那一论旨定位于法哲学之内，追溯法律效力之锚从力向信任的迁移（§4），并引出其形而上学的后果：信任的原因是推断性的而非机械的，而这正是自由与道德得以可能的那个开口（§5）。它发展那个核心机制（§6）；以和乐的语言给出其几何的读法（§7）；陈述那条区分真实信任与伪造信任的判准（§8）；分析关系性信任及其失败，连同其量子动力学结构（§9）；并发展那套真诚借以跨越认识论之边界的翻译理论（§10）。随后它转向实践：转向制度尺度上司法的改革（§11），以及一生尺度上信任的践行，上升到那作为本文顶点的关系性神性的论述（§12）。它以复数的结论作结（§13），并以一段尾声收束。

2. 信任的症候学

在一个病况被诊断之前，它先被呈现；而一个未先关注症候便径直推进的诊断，是对无物的诊断。本节集合诸症候。它尚不解释它们；解释是§3的工作，而在此预先解释它们，便是在一个本应促动而非确证那些症候的论旨之光照下去阅读它们。因此本节的纪律是描述性的：它列出一系列可观察、可辨识的现象，取自人类交往中最亲密的尺度直到最公共的尺度，且只要求把它们放在一起来看，期望那统辖它们之物，当它显现时，会作为对一个已被看见之模式的解释而显现，而非作为一个强加于零散个例之上的假设。

这些现象，放在一起来看，呈现出一个唯一的、且起初看来悖谬的形状。那借以保住信任的装置在增殖，而信任本身却愈发稀缺。 无论看向何处，那些被设计来产生、担保或证实可靠性的工具都在增多——更长的合同、更精致的誓言、更密集的资质认证体系、愈发精细的评级与核验机制——而这增多并未遏止那些工具所为之存在之物的侵蚀，反而伴随着它，并且，以一种本节只描述而尚不解释的方式，似乎与它绑在一起。增殖与侵蚀之间的关系，在此被刻意地不予陈说。存在着这样一种关系，且它在其余方面少有共通之处的诸尺度上反复出现——这便是本节之存在所要确立的观察。

这些现象跨越三个尺度被呈现，由亲密向外，并非因为亲密是公共的缩影——它是否如此，是留待稍后的问题——而是因为，一个形状在每一其他方面都各异的诸尺度上反复出现，这本身便是首先要被看见之物。

2.1 亲密与家庭的尺度

在最亲密的尺度上，症候首先出现在本文由之出发的那份文书本身之中。一桩婚约被着手进行，而围绕它聚集起一套庄严化的装置——吉日、绢、语言之郑重，在别的情境中则是戒指、公告、列席的见证人——其精心程度，与它所授予的保障毫不成比例，因为它所授予的，严格说来，是无。着手做这种事的人，当他们反思时，知道仪式中没有任何东西具有约束力；而他们仍然精心操办仪式，并把这份精心感受为有分量的，而非闲置无谓的。这已经是一个症候：一套庄严化的装置，在它所能强制之物被公认地缺席之时，仍被以全然的郑重所援用。

症候以一种更尖锐、更令人不安的形式重现，在随后可能发生的婚姻之中。婚礼上交换的誓词，在晚近数代人之间、在许多地方，已变得更善于表达、更个人化、在其对永久的宣告上更铺张，而与此同时，它们所开启的结合之解体却变得更普遍而非更少。这一观察在此不是作为任一方向上的因果主张而提出的；它是作为一个症候学必须记录的并置而提出的，即承诺的形式变得更精致了，而它所承诺之物的持久却变得更无保障了。这两条曲线，誓言之雄辩与绑定之脆弱，在相关时期里朝着同一方向移动，而一个症候学记下这一点，却尚不追问该如何看待它。

症候再度出现，且对本文之关切而言最为迫切，在一个必须赢得一个尚非其家的家庭之信任的人的窘境之中。在这里，那在别处增殖的装置，恰恰是不可得的。求偶者，按假设，与这家庭没有任何过往记录，没有可借以被了解的累积历史，没有他们已经信任、能为他作保的第三方；他只有一份对他而言全然真实、却对他们而言没有任何他们能读作证据之形式的真诚。因此这一尺度上的症候，不是装置的增殖，而是一切非装置之物被感受到的不足：发觉真诚，无论多么真切，都不把自己呈现为依据，而那些其信任最为要紧的人，恰恰是一个人最少有他们会计入之物可向之出示的人。

2.2 法律与司法的尺度

在法律与商业的尺度上，同一形状以一种使其易于度量的文献般的精确度重现。合同变长了。商业各方借以相互绑定的工具，在过去一个世纪里、并在晚近数十年里急剧加速，已扩张为一种长度与繁复度无人完整读毕的文书，为预防每一偶发情形与每一对抗性解读而草拟，并仍在变长；而这变长并未伴随着交往之愈发顺畅或缔约各方之间愈发增长的信心，而是——据那些在此类工具之下交往者的证言——伴随着相反者。合同越厚，各方感到的安全越少而非越多；条款的精心，被体验为与其说是信心的建立，不如说是信心的预先替代物，是对一桩已被半数预期的违约的武装戒备。

合同之旁立着谅解备忘录，而它，对本文而言，恰恰因为它是合同的对立面，而是这一尺度上最能说明问题的工具。谅解备忘录在设计上是非约束性的：它记录一项意图，列出一桩拟议交往的形状，并明确地保留不给予一项可执行义务的强制力。此类工具在增殖。它们被精心草拟、被郑重签署，并被其签署者当作重要之物对待，且是在它们什么也不约束这一全然且言明的知晓中。各方在他们明知不可执行的文书之郑重缔结上耗费如此心力，正是本节所收集的那一类症候：对一个其缺乏强制力非但不被掩盖反被申明的形式的郑重投入。

而在合同与备忘录二者背后立着法院，症候在其中取得它最具后果的形式。强制的机器，即所有这些工具名义上所指向的那把刀剑，正承受着一种已开始显露的张力。违约的数量超出了法院裁断它的能力；强制的成本，以时间、金钱与注意力计，上升到对许多过错而言它超出了救济的价值；获得一份判决所历经的时段在延长，而一旦获得判决，实际据之追回的前景在收窄。一言以蔽之，刀剑高悬却越来越不落下，且被知晓为不会落下；而一件被知晓为不会落下的强制工具，无论其形式上如何可得，都已不再作为它曾是之物而运作。这里的症候不是一套装置的增殖，而是一套装置的掏空：强制的全套形式机器，与其愈发增长的实际无力强制，并存。

2.3 普遍的尺度

在最广的尺度上，这形状以其最纯粹、最一般的形式出现，在那些借以使陌生人对陌生人变得可读的工具之中。推荐信是最清楚的情形。它是一件被设计来传递一种接收者无法直接获取之知识的工具：写信者，他认识候选人，向一个委员会为候选人作保，而委员会信任写信者。而它，据那些依赖它的制度内部的共同承认，已在很大程度上不再运作。这些信已膨胀到几乎每个候选人都被描述为写信者所遇见过的最优秀者之列；那最高级，既已变得普遍，便已变得不提供信息；而委员会，既已无法从一个人人都学会了使之饱和的通道中提取出信号，便从这些信转向任何可被计数之物——分数、引用数、排名——并非因为那些被计数之物更丰富，而是因为它们至少更难被膨胀。作保的装置已增殖到自我抵销的地步，而信任已从它那里退却。

同一形状统辖着数字经济试图借以使其陌生人变得可读的评级与认证体系。信任徽章、核验标记、星级评分、认证印章，在每一个陌生人必须与陌生人交往的界面上增多；而它们的增多，被那些遇见它们的人迎以的，不是愈发增长的信心，而是一种愈发增长且训练有素的戒心，因为可信赖性之标记的那种无处不在本身就在教导：这标记是廉价的、可被钻空子的，并且对掠食者与对诚实者一样可得。一个界面被可信赖性的诸标记覆盖得越多，其中任何一个标记所宽慰的便越少。这里，增殖与侵蚀不仅是并发的，而且是可被读出地绑在一起的：每一新层的认证，都是对上一层之失败的回应，而它本身又引出下一层。

而在所有这些之下，是这个时代最一般的症候，本节最后才记录它，因为它支撑着其余：被那些其来历与意图皆无法被确定的符号所包围的境况。产生一个有说服力的凭据——一份文书、一幅图像、一段声音、一份行为记录——的成本，已随着当下的技术降向零，其后果是，人类环境被那些不再可靠地系于任何它们声称所指示之物的符号所饱和。一个人如今行走于一个符号的世界，其中任何符号都可能是捏造的，而对这样一个世界的理性回应——对一切符号性证言所赋予之权重的一种普遍降低——本身就是一个第一序的症候：信任的撤回，不是从这个或那个不可信赖的来源撤回，而是从符号本身撤回。

2.4 诸症候的一个分类

上面集合的现象不是一份杂烩。放在一起来看，它们是同一道裂隙的如此众多的显现，这道裂隙在不同的尺度上、以不同的材料，在一套本应保住信任的装置与它未能保住的那信任之间裂开。为使这一统一性可见，并为给§3的诊断一个有结构的对象而非一份有待处理的轶事清单，诸症候可沿着少数几个共享的维度被列出：症候出现的尺度；适于那一尺度的担保装置；那一装置失败的方式；以及那个统辖性悖论——增殖与侵蚀并存——在每一情形中所取的具体形式。随后的表如此排列它们。它是作为一个描述而非一个解释来提供的：每一行记录所观察者，而失败方式那一列指名一种崩坏的样式，却尚不指派其原因。每一行都例示同一道裂隙，以及那道裂隙是什么——这是表交付给诊断之事。

现象 尺度 担保装置 失败方式
婚约、婚礼誓词 亲密 仪式、庄严、对永久的宣告 精心程度随绑定之持久的削弱而增长；有郑重而无可执行性
赢得一个家庭的信任 家庭 真诚、宣告、所提议的行为 真诚对那些索求依据者并不呈现为证据
商业合同 法律 列举的条款、刑罚、救济 条款的增厚被体验为焦虑而非信心；对预期违约的武装戒备
谅解备忘录 法律 一种被申明为非约束性之形式的郑重缔结 对一件已知什么也不约束的工具的郑重投入
法院、强制 司法 强制、制裁的威胁 刀剑高悬却不落下且被知晓为不落下；形式机器被掏空了实际强制力
推荐信 制度 一位受信赖的写信者沿一条链的作保 膨胀至最高级；通道被饱和至不提供信息
评级、认证、信任徽章 平台 标记、印章、核验、星级评分 无处不在教导廉价；标记覆盖得越多反而宽慰得越少
捏造的凭据 普遍 符号性证言本身 零成本生产使符号脱离其所指；对符号本身的理性撤回

沿各列往下读，表所说的比任何一行都多。尺度从两个人到一整个社会，而装置在那一范围里彻底改变，从红绢到一个星级评分；然而失败方式那一列却押着韵。在每一情形中，一个本应承载担保的形式，最终被发现并不承载它：形式存续、精致、乃至增殖，而担保从其中流尽。无论是在一个誓言的尺度还是一个核验标记的尺度，裂开的那道缝隙是同一道缝隙——在装置与信任之间——而正是这同一道缝隙在一个别无共通之处的范围里反复出现，构成了这个时代的核心症候。装置的增殖与信任的侵蚀，表所暗示而尚未表明，是同一个过程的两面。指名那一过程，并表明增殖为何应当伴随而非遏止侵蚀，是症候学如今交付出去的工作。

3. 诊断

§2中集合的诸症候共享一个形状，而一个在其余方面互不相关的诸尺度上反复出现的形状，是一个共同原因的证据。本节指名那一原因。它以奠基之作为此类工作所确立的方式来做：诊断在此不是一个在通往救济的途中要匆匆带过的前奏，而是一项就其自身而言的实质性事业，其原则是：一个回应的精度，受其所回应之诊断的精度所限。清楚地看见一道缝隙如何在符号与它本应保住之物之间裂开，便已是回应它所需者的一半；于是本节从容不迫，并把诊断发展到一个确定的论旨，然后才提议任何机制或救济。

奠基之作从若干维度诊断了现代条件下符号的一场危机。本文不重述它们的全部；它取用其中两个直接关乎信任者——本真性与信任的危机，以及正义的过载——并把每一个都发展到超出奠基之作搁笔之处，在它们关乎机制之处织入它的另外三个维度，并加上一个奠基之作所不含的诊断维度。随后它把整体置于那两个更古老的哲学问题——休谟的与康德的——之前，这两个问题赋予现代危机以其更深的形式，而本文从这一并置中引出那个组织起全文其余部分的论旨。

3.1 信任的危机：从信任的失败到其生产的失败

奠基之作在此标题下诊断的第一个维度，如今从症候学看已属熟悉：人类环境被那些其生产成本已降向零、其来历无法被确定、并且对欺骗与对真诚一样可得的符号所饱和，其理性后果是，赋予符号性证言本身的权重在下降。奠基之作确立了这一点。本文必须补充的——因为这正是诊断转变为本文之存在所要解决的那个问题的确切之点——是：危机已越过了对信任之特定个例的失败，进入了一种对信任之可生产性本身的怀疑。一个给定的承诺被怀疑，是一回事；变得不清楚任何承诺如今能借由什么手段被弄得可信，则是另一回事，且是更深的境况。症候学表明担保的装置在每一尺度上都在失败；诊断则是：失败的不是这件或那件工具，而是它们全都回答的那个问题，于是那问题本身——信任如何被生成？——在一个更健康的境况里本无须被追问，如今却变得既迫切又晦暗。这便是为何信任之生成的机制——一个更有信心的时代本可视为理所当然而不加审视者——如今必须被弄成探究的明确对象。

3.2 正义的过载与力作为信号的失败

第二个维度是奠基之作命名为正义之过载者。随着信任的关系性秩序与共同体秩序变薄，把人们守持于其交往的负担，已默认地落到了司法身上：曾经，一张由关系、声誉与共同体见证织成的网把一个人守持于他的言诺，如今，法律被要求去做此事，且被要求为一个它从未被设定来承受的数量与种类的交往去做。奠基之作诊断了由此而来的过载：法律延伸进生活中愈来愈多的部分，同时又从内部被掏空，被要求去为一个它缺乏手段去为之背书的秩序背书。

本文把这一诊断推进决定性的一步。过载不只是量的，不只是对一个有限制度的过多要求；它是一个更深之失败的可见形式，即力作为信号的失败。症候学记下了那把高悬却不落下的刀剑。对那一症候的诊断是这样的：强制的威胁产生遵从，不是凭借制裁的机械确定性——如症候学所表明，那已变得绝非确定——而是凭借向各方发出”遵从是被预期的”这一信号。也就是说，力作为期望的一个替代物而运作：它是把一方带到”预期一项承诺将被尊重”这一状态的、诸可能手段之中的一种手段。而一个其机械基础被知晓为已失效的信号——一把被知晓为不会落下的刀剑——便不再发信号。因此正义的过载，是本文核心论旨那宏观的、制度的、可观察的一面，在此作为一个坚硬的社会事实被遭遇，然后才作为一个原理被论证：力从来不是义务的原因，而只是信任所在的那个期望之产生的一件装置，而这件装置如今正在它自身的条件上失效。

3.3 另外三个维度：原因、后果与前提条件

奠基之作的另外三个维度进入诊断，不是作为分立的危机，而是作为信任之危机的原因、后果与前提条件，并在此按那一理解被织入，而非就其自身被处理。

第一个是生成性悖论：当下时代的技术无限制且无成本地生产符号的能力。这是上面所诊断之饱和的、在物质技术层面上的原因；它是为何那伪造的信号已变得不是偶尔的闯入而是周遭的常态，以及为何一个有说服力的真诚凭据的生产，本身已不再能承载它曾经承载的分量。信任的诊断必须被安放在这一技术基础之上，否则便会把一桩在最初实例上是符号生产成本之变化者，误当作一种道德的衰落。

第二个是实在界的回返：在象征秩序未能维系之处，那本应被象征所束缚之物的喷发——以一种自由漂浮的焦虑、一种对阴谋论的易感、一种向原教旨主义与暴力的漂移之形式。这是一种普遍的信任失败的、在情感与社会层面上的后果：当人们借以相互定向的符号性证言不再可被依赖时，回返的不是一种中立的怀疑论，而是一种充满电荷且迷失方向的、对某种不可证伪之根据的寻求。诊断把这记为听任信任危机不被回答的代价。

第三个是主体间性的崩溃，即那种”一个人把另一个人当作一个临在而非一个对象来相遇”之关系的退化——一种更早的语言把它对立为”我-你”对”我-它”。这是其侵蚀最威胁本文将提议之机制的那个前提条件。因为，如§6将表明，信任借以被生成的诸通道之一，要求一个真实的自我承诺之举被另一个临在以领受它的人所领受；而在关系已退化为纯然工具性之处，那一通道在其根部便被废止了。因此主体间性的崩溃不是又一个症候，而是对那治愈之可能性本身的一个威胁。

3.4 可信性与可执行性的解耦

对取自奠基之作的诸维度，本诊断加上一个奠基之作所不含的维度，而它，本文认为，是其核心问题所赖以转动的那个具体属于现代的枢纽。它是一个承诺之可信性与其可执行性的历史性解耦。

在那些现代境况据以被最尖锐地看见的前现代情境里，二者绑在一起，且几乎不被区分。一个承诺同时是可信的与可执行的，且凭借同一事实：见证誓言的神，也是会复仇于其违背的力量；接受言诺的宗族或村落，也是会索取违背它之代价的群体；被看见在许诺，便在同一举动中被绑定于履行，因为那看见与那绑定是由同一个守望的共同体所做的。可信性与可执行性是同一个社会事实的两个方面，而无人有机会把它们撬开。

现代性把它们撬开了。它把可执行性指派给一个专门化的制度，即法律，由它通过一套抽象而非人格的机器来强制履行；而它把可信性——即一个承诺起初、在任何强制问题之前且与之分开地被相信所凭的那种品质——留在了没有一个相应制度来安置它的境地。只要法律的强制机器是稳健的，这一分工便是不可见的，因为可执行性可被依赖来代行可信性：一个人若能被保证一个承诺是可执行的，便无须觉得它可信。但症候学已表明那强制机器在失败；而随着它失败，分工便被暴露，可信性便显露为现代性已悄然把它弄成之物——一个孤儿，一种没有制度来背书的品质，如今那曾代行它的制度已无法承受重负，它便无所支撑地悬着。

这便是为何本文的核心问题在如今是迫切的，而它以这种形式在从前并不迫切。一个承诺如何在强制缺席时变得可信，这问题不是一个恰好被再度提出的永恒哲学谜题；它是一个具体的历史安排——先使可信性依赖于可执行性，再听任可执行性失败——所逼入敞露之境的问题。可信性的孤儿化，是把本文钉于其时刻的那个诊断，而它是作为最特别属于本文自身的诊断贡献而提供的。

3.5 两个更古老的问题：休谟与康德

在现代危机之下，躺着两个更古老的问题，它们赋予它以其更深的哲学形式，而诊断若想使它所抵达的论旨不止是一个社会学观察，便必须触及它们。

第一个是休谟的。”一个守住了其言诺的人将再度守住它”这一期望，像每一个”未来将类似于过去”的期望一样，依凭于归纳，而归纳，如休谟所表明，在理性中没有根据：没有任何对过往履行的累积蕴含未来的履行，而从前者到后者的推断，无论多么不可或缺，作为逻辑之事都无法被辩护。这是证据主义者对过往记录的依赖永远无法全然逃脱的问题，而它为任何数量的证据所能确立之物设下了一个界限：在最长的守诺记录之末，仍留有一道缝隙，没有任何证据能载人越过它，在已是者与将是者之间。诊断把这记为”信任永远无法被化约为证据之累积”的深层结构性缘由：休谟所裂开的那道缝隙是构成性的，而每一套关于信任的解说都必须说出它对那道缝隙做了什么，而非佯称已闭合它。

第二个是康德的。休谟的问题是”通往未来行为的推断无法被奠基于过往事实”，而康德的问题则关乎那个无论如何必须被依赖之人的地位。把一个人纯然当作一个其未来行为有待从证据被预测、并由激励来保住的系统来对待，便是、以定言命令第二程式的术语来说，纯然把他当作手段，当作一个有待管理的对象而非一个其自身即目的者来对待。但一个承诺，在其完整的意义上，是向一个自我绑定的自由能动者发出的，并从这样一个人那里被领受；那许诺者不是像人预测天气那样在预测他自己的未来行为，而是像唯有一个自由存在者所能那样把自己承诺于它。诊断把这记为”信任的问题不能不改变其对象便被全然交付给预测与激励”的缘由：把许诺者纯然建模为一个有待管理的机制，便已根本停止在谈论许诺。那”他者将守信”之押注，部分地，是一个对他之自由的押注——而这，如§5将展开，不是解说中的一处软弱，而正是本文最想说的一切得以可能的那个开口。

3.6 论旨：力不是信任的原因

集合起来的诸维度汇聚于一个唯一的论旨，全文其余部分之存在便是要展开它、奠定它、并应用它。它可被平白地陈述。即便他者的强制性担保完全在场，现代条件下的信任仍已大体下降；因此力不是信任的原因。 倘若刀剑是那期望的原因，那么在刀剑随时待命之处，期望本应稳固；但症候学表明信任恰恰在强制装置被最充分精心铺设之处侵蚀，而正义之过载的诊断表明了缘由：刀剑唯有通过发信号才保住期望，而一个其基础被知晓为已失效的信号，无论刀刃多么锋利，都不再发信号。那么，力至多是一个产生那真实之物——期望——的、偶然的且如今正在失效的装置；它不是那期望的原因。

论旨转动于一个全文其余部分将始终持守的区分之上：在原因的两个意义之间。有机械的原因，它如一颗台球作用于另一颗台球那样运作，凭借一种使被作用之物毫无戏份可演的强制之传递；又有推断的原因，它凭借给予一个心灵一个据以形成期望的依据而运作，那期望是心灵自身产生的，且既然自由，本可被抑住不发。力，在它奏效之处，机械地奏效，或志在如此：它要像一个物体被移动那样移动许诺者。但信任不是一个物体的运动；它是一个心灵的期望，推断性地、依据地被形成。说力不是信任的原因，便是说机械的原因不是那推断性结果的原因——即一个自由心灵中一个期望的产生，属于不同于一个物体之被强制的因果序列，且无法被它所达成。本文的整套正面解说，皆由认真对待这一区分而来：倘若信任是一个推断性的结果，那么它如何被生成的问题，便是恰当的依据如何被供给那形成它的心灵的问题——而力，事实表明，既非唯一的这样一种依据，也不再是一个奏效的依据。

至此诊断完成。症候是一道在每一尺度上反复出现的、在担保装置与它未能保住之信任之间的缝隙。诊断则是：装置之所以失败，是因为它建立在一个关于它自身运作的错误之上——把力，机械的原因，误当作一个本属推断性的、自由的结果的原因——而当下的危机，是那一错误的到期：当那错误曾依赖的唯一装置，即强制，在它自身的条件上失效，并把它曾代行的可信性留在无所支撑之境时。所余下者，是要正面地表明那推断性结果事实上是如何被产生的：经由什么机制，一个被给予恰当依据的自由心灵，会变得期望一个承诺将被守持。本文如今转向这一点，从那失效的装置本身所开始之处开始，即在法哲学之中。

4. 法理学的核心：效力之锚的迁移

诊断以把探究送往法哲学作结，且出于一个确定的缘由。法律是现代性委以可执行性的制度；它是那件诊断所发现正在失效的装置——力作为期望的生产者——被安置、被理论化、被精炼之处。倘若那件装置终究不是义务的原因，那么法律对其自身的反思就应当带有这一发现的痕迹，而法哲学的历史就应当在其内部发展中记录一个离开现代性曾指派给它的那个根据本身的运动。本节表明它确实如此。它不是一份综述。它沿着一条单一的轴线——效力之锚的位置，即每一立场把”一项义务借以课人以义务的那个力”安置于何处——来阅读法哲学中的主要立场，并且，如此阅读它们，所发现的不是一套有待掂量的相竞学说，而是一个有待追溯的方向性运动：在法律对其自身反思的一个世纪里，锚从力迁向信任。本节循着那一迁移历经八个立场，并在其结尾把它们排列于一个单一的视图之中。

关于与奠基之作的关系，说一句话：奠基之作沿着一条不同的轴线处理了其中若干相同的思想家，即结构对生成、象征界对实在界的轴线。当前的阅读不是一次重复，而是沿着一条更聚焦的轴线对同一材料的一次重新切分：不是法律是何种东西，而是其义务之力被锚定于何处。两种阅读是一致且互补的；当前这一种是信任的问题所要求的那一种。

4.1 锚在力：奥斯丁与霍布斯

运动始于现代性的指派最不加掩饰之处，即约翰·奥斯丁的命令理论。在奥斯丁看来，法律是一个主权者的命令，由制裁的威胁所支撑，习惯性地被服从；而一项法律义务，按定义，是对制裁的恒常承受——被课以义务，便是被暴露于一个有权力且有意志施加恶之人手中遭受一项恶的或然性。这里效力之锚被焊接于机械的原因。义务就是那被威胁的痛苦；在这套解说里，没有任何余项，被绑定之中除了承受被伤害之外别无他物。命令理论无论还做了什么，它都以无与伦比的清晰陈述了本章其余部分将由之移开的那个立场：一项义务之力是一项被威胁之后果之力，别无其他。

在奥斯丁背后立着他形而上学的先祖霍布斯，其名句是本章真正的出发点：没有刀剑的盟约不过是言辞，全无力量保障一个人。这句话通常被读作一个关于无支撑承诺之孱弱的世俗观察。为本文之目的，更好的读法是把它读作一种人类学的陈述，由它，信任之不可能性必然地随之而来。霍布斯的自然人，在一切人对一切人的战争中，没有任何据以预期其同伴会克制的依据，于是信任不仅在自然状态中稀缺，而且逻辑上不可能：没有任何东西可由之形成那期望，因而主权者必须被竖立起来，以从外部、凭借压倒性力量的威胁，注入各方无法在彼此之间生成的那个期望。

正是在这里，本章的阅读首次咬合发力。奥斯丁与霍布斯取了那个极限情形——信任已全然崩溃、只剩恐惧的情形——而把它误当作整体。霍布斯式的自然状态是符号危机被推至其极端，一个没有任何符号能产生期望、因为没有一个符号可被依赖的世界；而霍布斯式的救济，刀剑，恰恰是以机械的原因替代那已失效的推断的原因。那么，命令理论与其说是假的，不如说它是一个病态极限的理论。它的模型在信任在场之处无话可说，因为在那里力是多余的；它唯有在信任已崩溃、力是所余一切之处，才变得可见、且看似完备。而这便是为何它如此直接地说中当下：当代的正义过载（§3）是世界滑向霍布斯式境况——信任变薄直到刀剑被要求做一切——的运动，恰在如症候学所表明的、刀剑已无法落下的那一刻。命令理论描述了信任之失败所趋向的那个终点；而刀剑在那里之无法落下，正是命令理论自身的解决方案在它自身的条件上失效。

4.2 锚移向接受：哈特的内在观点

那个决定性的转折，也是整章的枢纽，是哈特的。他对奥斯丁的批判，是效力之锚在法律实证主义自身之内、第一次被从机械的原因撬松并移向他物的时刻。他的两个反驳完成了此工作。第一个是”被迫”（being obliged）与”负有义务”（having an obligation）之间的区分。持枪者迫使我交出钱财；我并不负有这样做的义务。奥斯丁的理论把义务定义为对制裁的承受，因而只能捕捉前者——那被强制——而构成性地无法捕捉后者，即一条规则借以给我一个理由而不只是一个动机的那种规范性。第二个反驳是：整类整类的法律——人借以立遗嘱、订合同、结婚的那些授予权力的规则——根本不是由威胁所支撑的命令，而是授予”改变自己规范处境之能力”的赋能结构，而命令理论根本无法容纳它们。

从这些反驳中哈特引出那个一切所赖以转动的概念：内在观点。一条规则之存在，是作为一条规则而非作为一种被强制行为的纯然被观察到的规律性，是在有这样一群人之处：他们从内部接受它，把它当作自己行为的一个标准、并当作批评他人偏离的一个依据——他们在其推理与评判中使用那规则，而非纯然在威胁下顺从它。内在观点是一种规范性的态度，而非一种预测；而它，如§4.5将表明，与那种只问”若我被逮住会对我做什么”之人的外在的、预测性的姿态形成尖锐对照。这正是效力之锚被从机械的原因移向规范性者的那个法理学时刻——移向一种给出理由的接受，即诊断的那个推断的原因所经营之物。

这一点在哈特关于承认规则的论述中抵达其深处，承认规则即一个法律系统的其他规则借以被辨认为有效的那条终极规则。是什么从上方为承认规则本身奠基？哈特的回答是：没有任何东西从上方为它奠基；它所依凭的不过是那些接受并使用它的官员的趋同实践。它的效力不是由一条更高的规则所授予，也不由任何制裁来保障；它的存在干脆就是它被接受并被据以行动这一事实。在这里哈特抵达了——尽管未如此命名——那个最接近本文的立场：一个法律秩序之效力的终极根据不是一把刀剑，而是一种集体的、自我维系的接受——内在观点跨越官员群体的一种趋同，它凭借被共享而把自己撑起。这，以§6的词汇说，是结构性通道的制度尺度形式：一个共享的先验，由被共享这一事实本身所维系，无须外部之泵来把它撑在原位。

诚实要求把哈特自身的困难标出，因为正是这一困难把论证交给了富勒。承认规则所在的那个接受，是一桩事实之事还是一桩规范之事？哈特希望它是一个趋同的社会事实；但一个赤裸的事实性趋同——官员们碰巧行为相似——似乎无法生成一条承认规则本应承载的那种真正的义务，那个应当。这正是德沃金据以施加其批判的那个开口，即法律不能仅仅是趋同实践之事，而必定涉及一种解释性的、负载价值的承诺。本文在此不裁断那一争议；它只指出，那一困难指向赤裸的接受之外的、接受本身所不供给之物，而富勒指名了那个东西是什么。

4.3 锚在互惠：富勒的法律内在道德

富勒供给赤裸的接受所欠缺者。他的主张是，法律拥有一种内在道德，一组八条原则——规则须是一般的、被颁布的、前瞻的、清晰的、不自相矛盾的、可能被遵守的、历时稳定的，且被以与其所宣告之条款相一致的方式施行——它们不是对法律的外在道德约束，而是它之作为法律而存在本身的条件。富勒的批评者所施压的问题是：为何这些该被称为一种道德，而非纯然一种有效治理的技术，一种制定行得通之规则的工匠诀窍。他的回答，也是他的论述变成本文的那一点，是：这八条原则共同构成立法者向被治理者的一个隐含承诺，而第八条——一致性，即官员须实际地如所宣告那样施行法律的要求——是问题的核心，因为它使服从成为一种互惠。公民对法律的忠诚是对国家忠诚于其自身所宣告之规则的一个回应；每一方的守信都以另一方的守信为条件。把互惠剥去——让国家宣告一条规则而施行另一条——服从的义务便不只是削弱，而是丧失其根据，因为那为它奠基的关系曾是一种相互的守信，而一方已从中叛离。

这一互惠正是法律中以其本名出现的信任。当富勒说法律是政府与公民之间的一桩协作事业，一桩不在一条规则被违背时失败、而在构成它的那条互惠之绑定被背叛时失败的事业，他便已在法理学的语域中指名了本文所论之物本身：效力之锚被安置于一项相互依赖的承诺之中。富勒与哈特之间关于法律与道德是否必然相连的漫长争论，在本文的框架内，可被重新描述，并在某种意义上以富勒胜出而被了结——不是在富勒自己的根据上，而是在本文的根据上：效力之锚，互惠，本身就是一个规范性的、关系性的东西，一种守信，而非一桩关于趋同行为的中立事实；于是法律之效力终究无法不诉诸它被锚定于其中的那条信任关系便被指明。富勒的内在道德，以本文的术语说，是§6三条通道在制度尺度上一次性的全部形式——一个既是自我承诺、又邀请一次押注的结构——其锚被安置于一个被双重依赖的承诺之中。

4.4 锚在程序与商谈：哈贝马斯与罗尔斯

一个较简短的驿站延续富勒所开始的运动，它追问的是合法的规范如何被生产，而非效力最终安在何处。哈贝马斯把法律的合法性定位于它借以被生成的那个商谈之中：诸规范之有效，在于它们能够在一个由理性而非力所统辖的交往过程中获得一切受影响者的赞同，而法律-政治系统为其合法性而依赖于一个交往实践的生活世界，那生活世界是系统本身无法生产、而只能耗竭的。这一依赖的结构值得标出，因为它与本文将再度遭遇的一种依赖是同构的：正如哈贝马斯式的系统从一个它无法自其自身之内生成的生活世界汲取其合法性，§6中信任的似然通道也汲取一个象征所无法制造的真实基础。罗尔斯，以一种程序——原初状态——其公平性把合法性授予自它所产出者的那种装置，属于同一家族：一个把效力锚定于其合法生产之方式的家族。这在结构上与本文自身的问题是同一种问题——某物如何被合法地生产？——只是被向合法性发问，而本文向信任发问；标出这一亲缘之后，本章返回其主线。

4.5 第一个反向坐标：霍姆斯的坏人与法律现实主义

针对那个朝向接受与互惠的运动，有两个立场把锚按在力上，而必须给它们应得之分量，因为每一个都标出本文将需要的一个永久坐标。第一个是霍姆斯的。他那”透过坏人的眼睛来看法律”的劝告——坏人毫不在乎法律的道德，只想知道法院事实上会对他做什么，以便他能算计——是外在的、预测性姿态的最纯粹陈述。对坏人而言，法律就是对官方行为的一个预测，而义务就是一项后果的被算计的或然性；他是完美的证据主义者，纯粹机械原因的主体，他不承认任何内在观点，因为他不占据任何内在观点。

本文对霍姆斯的处理是双重的。一方面，坏人是真实的，而他的姿态是不可化约的：他正是§10将不得不认真对待的那个证据主义信任语法的确切肖像，那个唯有可预测的后果才算数、且真诚对他不是证据的主体。佯称他不存在，或佯称他的姿态只是一种道德上的失败，便是在本文最难的情形面前解除本文的武装。另一方面，他的限度是精确的，且必须被陈述。把每一个主体都建模为坏人——把外在的预测性姿态当作关于法律主体的全部真理——便是先验地抹去内在观点，并随之抹去”把他者当作一个自我绑定的自由主体、而非一个其行为有待被预测并管理的对象来相遇”的可能性。坏人理论，一旦被普遍化，便是康德的忧虑（§3.5）在法理学中的实现：它把人化约为一个由后果所决定的对象。它对那些已把自己弄成坏人的人是真的，且对任何人都是一种恒常的可能；它不是关于人本身的真理，而一个把它当作那真理的理论，已预先决定反对§5将肯定的一切。现实主义的社会学延伸——埃利希的活法，即实际统辖行为的法律，对照于书本上的法律——在此被记为一项相关且更宽阔的发展，但坏人才是对本文论证要紧的那个形象。

4.6 第二个反向坐标：凯尔森的纯粹法理论

那个把锚按在远离信任之处的第二个立场，其方向与霍姆斯相反——不是向下进入被预测的事实，而是向上进入纯粹的规范。凯尔森的纯粹法理论从一条授权每一法律规范的更高规范中导出该规范的效力，在一条上升的链条中，这条链条为免于无穷回溯，必须终止于一个被预设的基本规范，即Grundnorm，它本身不由任何更高权威所设定，而是被假定为”把整个秩序视为有效”的先验-逻辑条件。基本规范是一个预设，把道德与事实双双悬置，是一个纯粹规范结构之顶端的一个纯粹规范性公设。

置于哈特之旁，对照是精确而富于教益的。哈特的承认规则是一个被接受的社会事实，是官员们实际做的某事；凯尔森的基本规范是一个被预设的象征性设定，是人必须假定的某物。而本文对凯尔森的诊断随即而来：凯尔森把效力之锚安置于象征界中尽可能纯粹的那一点，一个悬于无物的设定，除了”去预设它”这一决定之外别无所撑。这是诸立场中最优雅的，而以本文之见，是在症候学所描述的条件下最脆弱的。因为一个由纯粹象征性预设构成的根基，恰恰是符号危机所溶解之物：让那预设不再被作出——让人们停止假定那基本规范——整座高耸的规范结构，因别无支撑，便悬于虚空。凯尔森是一个被单独安置于象征界中之锚的极限情形，而信任的危机恰恰是这样一个锚归于零的条件（§3.1）。

4.7 那件具体的工具：从对价到信赖

锚的迁移不只是高阶理论中的一个运动；法律以文献般的精确度，在合同的技术学说中记录了它，而这些是所有之中最直接相关的，因为它们恰恰关乎本文由之出发的那件工具，即那个赤裸而不可执行的承诺。正统的对价学说主张，一个赤裸的承诺——一个无物为之交换的承诺——是不可执行的；一个承诺要有约束力，就必须有一项讨价还价而来的交换，一项对价交换。赤裸的承诺、无偿的允诺、记录一项意图而不交换任何东西的备忘录——这正是婚约文书与谅解备忘录二者一样的确切法律处境：一个正统学说因无物为之给付而拒绝执行的承诺。

但法律发展出了一个例外，而这例外，对本文而言，是一次来自法律自身内部的静默的印证。在允诺禁反言学说之下，一个无对价支撑的承诺，仍可在如下情形被执行：受诺人已合理地信赖了它，已在那信赖中作出于己不利的行为，且若那承诺不被尊重便会蒙受不义。这里约束力的根据，既非对价亦非任何制裁的威胁：它是信赖本身。因为受诺人信任了那承诺并在那信任中使自己变得脆弱，那承诺便取得了它作为一个赤裸承诺时所不曾有的约束力——一种由置于它之中的信任所生成、而非由任何为它交换之物或其背后任何刀剑所生成的力。这是§6那个自我承诺通道在合同法中的镜像：受诺人的信赖是一种把自己置于脆弱之中，而法律以生成一项保护那脆弱的义务来回应那脆弱。与之相关的那个度量，信赖利益——富勒与帕迪在其经典论述中辨认出的”被保护免于因信赖而蒙受之损失”的利益，区别于期望利益与返还利益——是法律对一项被背叛之信任所造成之损害的自己的名字。

其意义重大，且应被平白陈述。在它自己的技术边缘，在允诺禁反言学说中，法律已承认：一个承诺之约束力的真正根据是信赖——是信任——而强制只是对一项在它自身之中、且在任何强制之前便已是那约束之物的信任的事后保护。允诺禁反言是法律对本文核心论旨自发的、内部的印证：不是本文为支撑其主张而引入的一项学说，而是法律在对其自身最基本之工具进行推理时、自行抵达那一主张的一个地方。

4.8 一图之内的迁移

这八个立场，沿着效力之锚这条单一轴线来读，组成一个方向性运动。法律对其自身的反思，已把锚从力（奥斯丁、霍布斯），经由接受（哈特），载向信赖这一互惠关系（富勒，以及允诺禁反言学说），而那预测性的与纯粹规范性的，则被按在两个反向坐标（霍姆斯、凯尔森）上，它们标出运动所遗于身后的界限。本文不从外部强加这一运动；它追溯一股已然在法律中在场的暗流，并把它接连于符号危机的诊断，并在§6中接连于信任的机制——从而把那个法律本身（因缺乏一套机制性词汇）能标出却无法解释的运动，给予完整的因果奠基。其后果是，本章不是纯然的背景，而是论证的一根独立支柱：它从法哲学之内确立，效力之锚正从力迁向信任，于是§5的形而上学与§6的机制，便不是各自独立的哲学或科学断言，而是对一个法律已然自行经历的运动的奠基性解释。三条线——法律的内部暗流、推断之原因的形而上学、以及预测编码的机制——汇聚于一个单一的命题：是信任，而非力，才是义务的真正原因。

整体可被置于一图之内。该表沿本章的轴线阅读诸立场：每一立场把效力之锚安置于何处，它由此援用的是哪一意义的原因，以及它对本文论旨贡献了什么、或标出了什么的界限。

立场 效力之锚 原因之意义 在论旨中的角色
奥斯丁、霍布斯 被威胁的制裁；刀剑 机械的 陈述了有待移开的立场；一个病态极限的理论
哈特 趋同的接受；内在观点 推断的（规范性的） 枢纽：锚被从力撬松而移向接受
富勒 互惠；相互的守信 推断的（关系性的） 以其法律之名指名信任；一致性作为隐含承诺
哈贝马斯、罗尔斯 合法的程序；商谈 推断的（程序性的） 同构的问题：合法性如何被生产
霍姆斯、现实主义 被预测的官方行为 机械的（预测性的） 反向坐标；证据主义者的肖像；其限度是康德的忧虑
凯尔森 被预设的基本规范 纯粹象征性的设定 反向坐标；锚单独在象征界；在危机中归于零
允诺禁反言 信赖本身 推断的（关系性的） 法律自己的印证：信赖，而非力，才约束

沿原因那一列往下读，表把本章的主张一瞥之间呈现：那些把锚按在力上的条目，是机械之原因的条目，它们坐在运动的开端与它被遗弃的诸界限处；那些把锚载向信任的条目，是推断之原因的条目，它们是法律之反思一直在行进之处。锚的迁移，与原因之两个意义之间的区分，是被看见两次的同一个运动。那推断的原因在形而上学上是什么——为何它是自由本身得以可能的那个开口——是§5如今所接手的问题；它在机械上如何运作，是§6的问题。

5. 推断的原因与自由的开口

法理学一章以一个它已历经八个立场追溯、却尚未审查的区分作结：在机械的原因（效力之锚在法律之运动的开端与它被遗弃的诸界限处坐落于其中）与推断的原因（那一运动一直朝之行进）之间。这一区分在诊断中被画出，作为一个结果可能被产生的两种方式之间的对照——一个物体之被强制，与一个心灵之期望的被奠基。本节追问推断的原因是什么，并回答：它不是机械之原因的一个较弱物种，而是全然不同的一个种类，且那差别正是自由、并随之道德，得以可能的那个开口。本节简短，但它是机制、正义判准与践行将建起的一切之下那承重的拱：它确立，生成信任便是把一个自由存在者当作自由者来面对，而那个不携带任何强制力的承诺，恰恰出于那个缘由，不是有缺陷的情形而是本然的情形。

5.1 两种原因，而非一种原因的两个程度

人很容易把机械原因与推断原因之间的对照听成一种强度的对照——仿佛机械的原因是那可靠的种类，它径直强制其结果，而推断的原因是一个较弱的种类，它仅仅使其结果更为或然。在那种听法上，力会是因果的金本位，而信任是对它的一个退化的近似，而法理学所追溯的整个运动会是一个从强根据到弱根据的运动，一个衰落的故事。这是本文必须拒绝的听法，而这一拒绝是本节的枢纽。

机械的原因与推断的原因不是一种因果的两个程度；它们是两个种类。机械的原因从外部作用于其结果，且使结果毫无戏份可演：那被撞击的球不同意去动，对它自己的运动毫无贡献，在那一撞击之下不可能有别的作为。它的范型是动量的传递，而它的标志是被作用之物自始至终是被动的——原因做全部的工作，而结果只是对它的纯然登记。推断的原因则运作得全然不同。它不推；它给出一个依据。它给一个心灵供给某物，那心灵可据之形成一个期望——一个那心灵自身、凭其自身的活动所形成的期望，而它既是其所是，本可形成得不同或抑住不发。那依据不像撞击强制运动那样强制那期望；它供给一个理由，而一个理由恰恰是一个自由心灵所接取、而非所承受之物。机械的结果是对一个力的被动登记，而推断的结果则是那期望由它在其中升起之人对它的主动产生。

这便是为何推断的原因不是一个较弱的机械原因，而是一个更强的——或毋宁说一个不同而更高的——东西。它是唯一一种能以一个自由存在者为其本然主体的因果，因为它是唯一一种把那存在者当作自由者来面对的因果：当作一个依据地形成其自己之期望者，而非一个其状态被从外推入者。以推断的原因来产生信任，便是把信任由它在其中升起之人当作一个作判断的心灵、而非一个作登记的机制来对待。而机械的原因，被施于一个心灵，并不改善这一点；它使之退化，因为强制一个期望而非奠基它，便是在恰恰那一方面，把那心灵当作仿佛它根本不是一个心灵来对待。

5.2 信任把他者当作自由者来面对

同一区分，从那信任之人而非那被信任之人的一侧来看，是一个人可对另一个人采取的两种姿态之间的区分，而正是在这里，诊断中康德的忧虑（§3.5）获得了它的正面回答。一个人可对另一个人采取预测-与-管理的姿态：把他的未来行为当作一个有待从证据被预测、并由激励来保住的结果，安排那些后果使得自己想要的行为是符合他利益去产生的行为，并恰如人依赖因果秩序中任何被充分理解的部分那样依赖他。这是霍姆斯的观察者对坏人的姿态（§4.5），而它，在其本然的领域内，是全然合法的；一个人最好能够预测并管理。但它是一种把它的对象当作一个对象的姿态——当作一个其行为由人可安排之条件所决定的系统——而只对一个人采取这种姿态，把他详尽地建模为这样一个系统，便是、以康德的术语说，已停止把他当作目的来对待，于是已停止把他当作一个人来与之相关。

另一种姿态是本然之信任的姿态，而它是一个自由存在者对另一个自由存在者的姿态。信任一个人，在承诺所要求的完整意义上，不是像人预测天气那样预测他的行为，而是依赖他之自由地自我绑定——把他当作这样一个人来面对：他能把自己承诺于一个未来、并能与那一承诺守信，因为他已把自己绑定于它，而非因为那些后果已被安排得使他别无选择。那作出一个真承诺之人，不是在从内部预报他自己的未来行为，如人可预报一个习惯；他是在把自己承诺于它，履行那个区别性地自由的举动，即把自己置于一项他本可违反的义务之下、并担保他不会违反。而那信任那承诺之人，部分地，是在押注于那自由：依赖的不是一个不会失败的机制，而是一个本可失败、却担保不失败的自由存在者。那个对他者之守信的押注，不可化约地，是一个对他之自由的押注——而这，诊断曾把它标为预测-与-管理改变其对象、并停止谈论许诺的那一点，如今被看出不是信任之解说中的一处软弱，而是它的条件本身。没有什么”信任一个被视为纯然机制者”；只有预测它。信任恰恰在预测作为全部真理被拒绝之处开始，而他者被当作一个自由地守信或违信、并被依赖去守信的人来相遇。

5.3 关于认知基础的一句话

这是一个形而上学与伦理的主张，但这并不使它成为一个反对自然秩序的主张，而在此欠下一句话，为§6作预备，关于它的认知基础，以免推断的原因看似要求一种自然之外的官能。它并不要求。依据地形成期望，不是一个心灵在稀有而庄严的场合所执行的奇异操作；它是一个大脑寻常而不息的事务，那大脑，按下一节展开的解说，先于其为任何别的东西，是一个预测的器官，持续地生成对将来者的期望，并对照所到达者来修订它们。信任，按那解说，不是叠加于认知之上的一种特殊官能，而是认知之恒常活动的一个特定情形：一个关于另一个人之行为的、高置信度的前向期望，依据地被形成、并保持向修订敞开。也就是说，推断的原因有一个全然自然的实现；把它与机械的原因区分开的，不是它漂浮于因果秩序之外，而是它在那秩序之内，凭借向一个形成其自己之期望的系统供给依据、而非凭借强制一个仅仅登记对它所做之事的系统，来运作。自由的形而上学与预测的神经科学并不相忤；前者是后者所描述者的意义。本节把机制留给它自己的一章，并返回那个后果。

5.4 为何非强制的承诺是本然的情形

那后果是序曲在能说它之前便已感到的那个，而它如今可以被说出。那个不携带任何强制力的承诺，不是许诺的有缺陷的极限情形，不是要对照那作为标准的可执行合同而为之道歉者；它是本然的情形，可执行合同是其派生物。因为，倘若信任把他者当作自由者来面对，且倘若产生信任是奠基一个自由心灵的期望、而非强制一个机制的回应，那么那个不依凭任何刀剑的承诺，便是那个在其纯粹中做此事的承诺：它索求被相信，所凭的依据不过是许诺者的自由承诺与受诺人对它的自由押注，于是它是推断的原因不掺杂地运作的那个形式。可执行的承诺，相反，把一个机械的成分掺入这个纯粹情形：它在那自由承诺之旁提供一项被威胁的后果，而恰在那一限度内，它把受诺人当作的，不是一个在信任之依据上形成期望的自由心灵，而是一个在激励之依据上形成预测的计算者。刀剑并不把那承诺作为一个承诺来增强；它以某种根本不是一个承诺之物——一项被管理的后果——来补充那承诺，而那补充物在它奏效之处所做的工作，正是以信任所拒绝的那种方式来对待各方。

这尚不是要说——而本文稍后将着力不去说——刀剑应当被废弃，或可执行的合同在每一方面都劣于赤裸的誓言；对脆弱者的保护，如§11将坚持，对力有一种持续而不可或缺的需要。它是要说某种更狭窄而更深的东西：在那使一个承诺成其为一个承诺之物上——把一个自由存在者当作自由者来面对，奠基而非强制一个期望——非强制的承诺是那实现了的情形，而强制的承诺是那被稀释的情形。序曲曾把它感受为誓言之所以是其所是之条件的那种缺乏强制力，在此被给予了它的缘由。一个人能被强迫去守持的绑定，恰恰在他能被强迫的限度内，还不完全是自由存在者之间的一个绑定；而那不索求刀剑的誓言之更真，不是尽管它不携带任何强制力，而是因为，不携带任何强制力，它把全部的工作都留给了自由——留给许诺者对自己的自由绑定，与受诺人对它的自由信任。那工作事实上如何被做——经由什么机制，一个被当作自由者来面对、并被给予恰当依据的自由心灵，会变得期望那承诺将被守持——是本文一直在为之清理地面而要追问的问题，而它是本文如今所转向的问题。

6. 信任的机制

至此为止的一切，都为一个唯一的问题清理了地面，并供给了它能被追问所凭的术语。症候学展示了担保装置与它未能保住之信任之间的一道缝隙；诊断把那缝隙追溯至一个关于因果的错误，把机械的原因误当作一个推断性结果的原因；法理学表明了法律自己的锚在一个世纪里从力迁向信任；而前一节确立了推断的原因与机械的原因属于不同的种类，是那种把一个自由存在者当作自由者来面对的原因。这些之中没有一个尚已供给的，是那个正面的解说：经由什么机制，一个被当作自由者来面对、并被给予恰当依据的自由心灵，会变得期望一个关于不存在之未来、不携带任何后果之担保的承诺，将被守持。本节供给它。它是本文的主干，之前的诸章是其根，之后的诸章是其枝。

机制必须解决的那个问题，回想一下，有一个看似不可能的形状。信任指向一个尚不存在、不可被验证的未来；那索求信任的符号不携带任何强制力，因而不能通过威胁一项后果来保住未来。我们被要求解释：一个当下的、真实的且常常有据的期望，如何被一个关于不存在之物、且无任何担保的符号所产生——几乎是一种从无中的生成。把这表面的不可能性溶解，是本节的第一个动作，而它是通过伸手去取本文所拥有的、唯一一种对”形成关于尚未到达之物的期望”是本土的词汇而作出的：心灵的预测性解说。

6.1 预测编码与悖论的溶解

在那已成为认知科学之核心的预测性解说上，大脑首先不是一个登记世界然后回应它的装置；它是一个预测世界、并对照所到达者来校正其预测的装置。它携带一个关于其感觉状态之诸原因的生成模型，持续地生成对将来者的期望，把它们对照传入的信号，并按误差的比例——并且，至关重要地，按它赋予那误差与它自己之预测的精度，即逆方差的比例——更新那模型。预测，按这解说，不是一项偶尔的成就，而是那器官的恒常活动；心灵凭其本性是前向定向的，在它感知之前便已期望。

这一举溶解了那表面的悖论。那反驳是：信任之不可能，是因为它的对象——那被守持的未来——尚不存在；一个当下的状态如何能被一个缺席的事实所产生？但那反驳假定一个期望必定被那被期望之物所引起，而预测性解说表明这一假定为假。一个期望不是被它的对象所引起的；它是此刻被那生成模型作为模型的前向预测所产生的，先于且独立于那对象的到达。不确定性下的前向定向期望，不是信任必须以某种方式逆着认知之纹理而成就的一项特殊壮举；它是认知无论如何都在不息地做之事。信任不要求时光机器，不要求从一个不存在之未来逆行而来的因果。它只要求大脑在每一个清醒时刻所做之事：由一个生成模型对一个高精度预测的赋予——在此即”这个人将守持这个承诺”这一预测。信任，是一个接收者的生成模型赋予”承诺将被守持”这一预测的高精度。 诊断所说的那个法律可能性的期望，在机制的层级上，恰恰是这个：一个精度加权的前向预测，一个大脑本土地计算的量，而非一个从法律借来的隐喻。

于是信任如何被生成的问题，变成一个精确的问题：经由什么输入，一个接收者的生成模型会变得把高精度赋予”一个承诺将被守持”这一预测？本文的回答是，有三个这样的输入，三条通道，而它们对应——不是出于巧合，而是因为那些域是同一个主体的域——于本系列始终运用的那三个域。

6.2 三条通道

这三条通道在它们向生成模型供给之物上是各异的，而本文按那些域的顺序来取它们：结构性的，安置于象征界；自我承诺的，安置于实在界；以及被押注的，安置于想象界与那押注的主体。

6.2.1 结构性通道：先验

第一条通道供给一个先验。一个承诺不是在真空中被领受的；它是带着一个结构被领受的，而那承诺的结构在任何关于这个特定许诺者的证据被掂量之前，塑造接收者的先验期望。一个具有”闭合而自我完成之承诺”形式的誓言——一个，以§7的语言说，回返于自身而不留下一笔须在别处清偿之债务者——向接收者的模型呈现一个有利于它被守持的结构性先验，其方式是一个松散的、有所保留的、满是退路的形式所不为的。这是本系列称为一个承诺之诗性结构者的认知功能：形式不是装饰而是信息，是接收者所带来之先验的一个塑造者。在制度的尺度上，这是哈特之趋同接受与承认规则的通道（§4.2）：一个共享的结构性先验，由被共享所维系，每一新的承诺都对照它被读。结构性通道安置于象征界，因为它通过符号的形式运作；而它，单独来看，是三者中最可伪造的，因为一个形式可被仿冒。

6.2.2 自我承诺的通道：高保真的似然

第二条通道供给一个似然——而在它奏效之处，是一个高保真的似然。在这里许诺者不只是呈现一个形式；他做某事。他把自己，真实地且可观察地，置于一个退却代价高昂或被封死的脆弱位置：他以一种他若违背便会伤及自己的方式自我绑定，交出一个人质，烧掉一座桥，不可撤回地、且在众目之下把自己承诺出去。而接收者的生成模型把这读作它在统计上之所是。一个代价高昂且难以逆转的信号携带一个高的精度权重，因为它是一个不打算守信的许诺者极不可能发出的信号：对不真诚者而言，发出它的代价是望而却步的，于是它之被发出，是真诚的有力证据。这是红绢、公开的场合、自焚其舟之下的机制——序曲由之开始的那种不可撤回且代价高昂的自我绑定。此类举动不是论证；它们不通过从前提推断来说服。它们向接收者的模型供给一个高保真、难以伪造的似然信号，一种身体在话语之前便阅读的那类证据，证明那能如此自我绑定者必定是当真的。

本文在此接手、并给予一个机制性基础于那条在生物学与经济学的信号理论中熟知的代价高昂之信号的原理：一个对一个背叛者而言太昂贵以致负担不起的信号，正因如此而是可信的。预测编码所补充的是那个实现——那代价高昂、不可撤回的信号，恰恰是一个生成模型理性地赋予高精度者——以及一个本文不会在此穷尽、却必须标出的更深的观察。烧桥已被描述，就在刚才，在它作为一个信号的方面：作为某种为被另一个人阅读而做之事，向他的模型供给证据。但这只是它两副面孔之一，且不是更深的那副。还有第二种意义上的烧桥，它与被任何人阅读毫不相干——在那种意义上，一个人封死自己的退却，不是为了发出一个讯息，而是为了为自己构成一条法：我已如此颁定，于是那条路对我而言不复存在。这更深的意义，即那不可撤回的自我绑定是一个自我立法之举而非一个发信号之举，是§12之事，尤其是它关于关系性神性的论述之事；它在此、在机制处被标出，以便那抵达它的读者会认出，主干一路上都在指向它。就本章而言，自我承诺的通道是在它的发信号方面被处理的：作为安置于实在界的高保真似然，作为那条”一个真实的自我暴露之举，被另一个临在以阅读它之人所领受，向模型供给其最难伪造之证据”的通道。

6.2.3 被押注的通道：主动推断的先验跃迁

第三条通道是最容易被误认为非理性的，而正确理解时，是最深的。它关切的不是许诺者所供给者，而是接收者所做者：接收者，在信任中，作出一个先于证据的跃迁。他在充分担保之前，便把一个高精度的先验赋予”他者将守信”这一预测——他冒险一试，付出那信任——然后才，通过那冒险所使之可能的持续互动，收集那些借以使模型被确证或被校正的误差信号。这看似一种理性的失败：信任在那会辩护它的证据之前便被给出。但在主动推断的读法上，它根本不是一种理性的失败。一个能动者不只是被动地对照传入的数据更新一个模型；它行动以在更长的时段里削减它自己的不确定性，而有些知识状态，唯有通过先在不确定性下行动——唯有通过冒险——才可达。信任就是这样一种冒险。一个人无法收集那会辩护”信任一个人”的证据，而不先进入那种唯有信任他才使之可能的互动；那证据躺在一道门槛的彼端，而那门槛唯有通过预先信任才能被跨过。于是信任的跃迁，是一个”通过行动来削减不确定性”之能动者的理性开局——一个先验的押注，它是获取那将证实或校正它的证据本身的前提条件。这条通道安置于想象界与那押注的主体，因为它的来源既不在符号的形式中，也不在许诺者的举动中，而在接收者的自由冒险中；而正是在这里，在”他者将证明自己是自由而忠信的”这一押注中，§5的那个自由重新进入机制。

6.3 乘积式的耦合

这三条通道不是通往信任的三条替代路径，仿佛许诺者可供给其一而被相信。它们是耦合的，且是乘积式地耦合——如同一个单一的贝叶斯乘积的诸因子那样耦合，其中结构性先验、自我承诺的似然、与被押注的先验，组合为”承诺将被守持”这一预测的后验精度。乘积式而非加和式耦合的要点是精确的，而它是本节的核心：在一个乘积里，若任一因子归于零，整个乘积便归于零，而其他因子中的任何丰盈都无法补偿。一个完美的形式，背后却无真实的自我承诺；一个代价高昂的举动，却无”他者自由地忠信”这一押注；一个信任的跃迁，迎之以的既非形式亦非自我绑定——每一个都是一个有一个消失因子的乘积，而每一个都作为有据信任的依据而失败。

这便是为何真实的信任在任一单条通道本会脆弱之处仍然稳健。一条单一的通道可被伪造：一个形式可被仿冒，一个看似代价高昂的举动可被搬演，一个轻信的押注可被利用。但这三条通道难以被一同伪造，因为它们彼此负责——结构性形式必须被一个真实的自我承诺所印证，而后者又必须是一个自由能动者会去做、并会邀请一个押注于其上的那类东西——而它们的相互一致，形式与举动与押注的相符，远比它们中任何单独一个都更难被假造。真实信任的强度，不是它最强通道的强度，而是这三者全都被弄得虚假地相符之事的不大可能性。这便是耦合在此的含义，以及为何它是那不只是被感到、而是有据的信任的来源。

信任的机制。 一个非强制承诺将被守持这一期望，是一个接收者的生成模型赋予那一预测的后验精度，由三个输入的乘积式耦合所形成：一个结构性先验，由承诺的形式所供给（象征域）；一个高保真似然，由许诺者真实的、代价高昂的、不可撤回的自我绑定所供给（实在域）；以及一个先验的押注，由接收者在证据之前对信任的自由冒险所供给（想象域，与那押注的主体）。真实的信任是这样一种状态：三者皆在场、皆真实、且彼此一致；而它的有据性，依凭于这三者全都被弄得虚假地相符之事的不大可能性。

6.4 伪造的情形，以机械的方式被描述

机制供给，作为一个副产品，对它自身赝品的一个精确刻画，§8将把它作为正义之事来接手，而它在此以其机械形式被列出。伪造的信任是这样一种状态：接收者的生成模型被带到把高精度赋予”承诺将被守持”这一预测，却没有那唯一能辩护它的真实的自我承诺的似然。操纵者所做的恰恰是这个：通过直接作用于接收者的模型——通过情感压力、通过形式的搬演、通过修辞的管理——他诱发接收者预测误差的一种降低，一个高精度预测，而它不被他这一方任何真实的自我绑定所支撑。那精度，可以说，是从外部被注入的，而非由一个真实的似然所赚得；而一个如此被注入的精度，是一个不被它所预测之物所支撑的预测，一种底下空无一物之真实的置信。

因此操纵的机械标志可被相当尖锐地陈述：操纵，便是诱使他者的生成模型降低其预测误差，而不供给那会辩护此降低的高保真似然。 这样一种状态，以真实信任所不具有的方式，是不稳定的。因为没有真实的自我承诺在其下，世界事实上不会如那被注入的精度所预测的那样行事；误差信号会累积，模型会被逼向校正，而被预测者与实际者之间的发散会加宽，直到那被伪造的精度崩溃——本系列在别处、在其动力学语域中描述为”在非绝热的汲取之下偏离向发散之累积”的那个过程。于是伪造的情形不是信任的一个稳定的替代形式，而是一个瞬态，一个由压力而非由似然撑起的精度，注定走向那真实根据之缺席所担保的崩溃。

6.5 符号的危机，以预测编码的术语

同一词汇最终产出§3之诊断的一个精确的呈现，并表明符号的危机不是一种道德的衰落而是一种理性的调整。设想一个社会即那个生成模型的群体，每一个都从它自己与符号的往来中，习得符号信号与真实履行之间的统计关系。随着产生一个有说服力之符号的成本降向零、而伪造的信号增殖（§3.3的生成性悖论），每一个模型都正确地习得：符号性证言与真实履行之间的相关已经削弱——那信号事实上已变得对结果较少有预测力。而对一个已变得较少有预测力的信号的理性贝叶斯回应，是降低赋予它的精度权重。每一个模型，如它应当那样更新，把符号性证言本身降权；而因为那降权被施于通道而非施于单个信号，它便连同伪造者一起落在那真实的符号上。那真实的誓言、那真的推荐、那真诚的保证——每一个都被一个已理性地习得”较少信任其同类符号”的模型所迎接，于是每一个都未能落地，不是因为它在个别上被怀疑，而是因为它所行经的通道已被理性地折扣。

这便是符号的危机被给予其机制。不是人们已变得犬儒，或价值已衰朽；而是一个运作良好的生成模型的群体，面对一个伪造已使之较不可靠的通道，已理性地崩缩了它赋予那通道的精度权重——而那向纯然可计数者的退却（§2.3），是同一个群体试图从那些至少更难伪造的信号中重建一个高保真似然。危机是一个群体尺度上的理性精度崩缩，而这既是它为何如此棘手——因为每一个模型都只是在做它应当做之事——也是为何那救济，如全文其余部分所展开，不能是一种向符号性宣告的回归，而必须躺在那些伪造不那么容易使之饱和的通道里，即自我承诺的通道与关系性的通道。

6.6 与奠基之作之计算纲领的联系

值得在本章收束之前标出：机制与奠基之作所敞开的那个形式纲领相吻合。那篇论文勾勒了，作为未来工作的一个方向，一项把正义当作生成性临在的计算研究，其中正义会作为一个在一个历史空间上被求值的可观察量、以路径积分的方式被形式化而出现。信任的预测编码机制是这一形式的、就信任之情形而言的一个具体实现：信任是一个生成模型赋予一个在许诺者诸可能未来轨迹之空间上被求值之预测的精度，一个在诸历史上取的期望。这两个形式纲领相啮合——奠基之作的”在诸历史上的可观察量”与当前机制的”在诸轨迹上的精度”是同一类对象——而这一啮合在此被标出，以便本系列的形式连续性被记录在案，以备被取用而非被重新推导。

6.7 一条诚实的纪律

本章必须在本系列要求它结束之处结束，以一条对它所做的预测性解说之使用的明确纪律，以免机制被当作多于它之所是者。预测编码在此被当作一个机制性的与结构性的模型来使用，而非一个关于信任归根结底是什么的还原论主张。本文不断言信任干脆就是一个精度加权而别无其他。预测编码所供给者，是一个在马尔的计算层级与实现层级上的描述——一个关于系统计算什么、以及它可能如何计算它的解说——而这样一个描述不取代本文已给出与将给出的其他描述：关于自我承诺之脆弱的现象学描述、关于力作为期望之替代物的法理学描述、关于真实与伪造之差别的伦理描述。这些是关于同一对象在不同层级上的描述，而预测编码的解说是其中之一，不是它们全体之下的真理。它不是一个吸纳其余的元框架；它是关于一个单一对象的又一个不可化约的立场，而本系列那条复调的纪律——不容任何框架统辖其余——对它有效，正如对每一个别的框架有效。于是机制是作为本文的核心贡献、而非作为它的盖棺之论而提供的：一个关于在那计算的心灵之层级上、推断的原因如何做它的工作的具体说明，它使其余诸域关于同一事物所要说的一切，都立于它之旁而不倒。

7. 和乐界面

§6的机制是以预测与精度的词汇给出的。本节给同一机制一个第二读法，以本系列在别处发展的几何词汇——和乐的词汇，一个系统在穿越一条闭合路径时所累积之相位的词汇。这第二读法不是第一读法的对手，而是它的补充：预测性解说描述那接收的心灵计算什么，而几何的解说描述那个”一个承诺历时被守持时，要么维系置于它之中的信任、要么使之流尽”之过程的形状。这两个读法将被看出汇聚，而它们的汇聚为本文后来的两个时刻作预备：关系性信任的量子动力学结构（§9），因为几何相位在其起源上是一个量子动力学的概念；以及那个”无强制之诺被证明是更纯粹之工具”的价值倒转（§12），因为那一倒转恰恰转动于本节如今所画之区分上。

一条纪律必须在开端处被陈述，正如对机制那样。几何相位在此被当作一个结构类比来使用，被发展到有用的地步，而非到完全的形式闭合；那个会给相位供给一个严格载体、一个确定的纤维与联络的形式纲领，由本系列的另一篇承担，本节向它前向援引，而非复制它。在此所宣称者，是：一个历时被守持之承诺的结构，是一个被沿一条闭合环路运送之系统的结构，而”一个生成信任的承诺”与”一个使信任流尽的承诺”之间的区分，是”一个累积正和乐的环路”与”一个累积零和乐或负和乐的环路”之间的区分。这一主张是结构性的；它的完整形式化在别处。

7.1 信任作为一个被沿一条环路运送的量

设想一个承诺，不在它被作出的那一刹那，而在它被守持的那段时间里。一个曾立下誓言的关系，一次又一次地返回那些类似于誓言之情境的情境：那些承诺被考验、被忆起、被重演、或仅仅被度过的时机——每日、每月、每年的复现，其中所许诺者被、或不被印证。于是那关系，历时地，在一个情境的空间中描出一条闭合的环路：它从誓言的那一刻出发，穿过一份共同生活的种种境遇，并返回那些在相关方面是同一情境又一次的情境。几何读法所追问的问题是：那关系在它返回时带回什么。它是否返回到誓言的情境而毫无改变，仅仅绕了一圈又回来？还是它返回时已累积了某物——一种加深、一份盈余、一个起初并不在那里的相位？

这是和乐的问题。一个被沿一条闭合环路、在一个具有恰当结构之空间中运送的系统，一般而言，不会返回到它的起始状态；它返回时被旋转、被移位，携带一个记录它所走之路径的相位——那几何相位，那非和乐（anholonomy），贝里相位与潘查拉特南相位是其典范实例。那相位不是系统在何处终结的函数——它在它开始之处终结——而是它为返回所穿越之环路的函数：它是那旅程在那返回之状态中的痕迹。而那相位的符号与量值，恰恰是本系列用以辨别一个好的循环与一个坏的循环之区分。一个使其系统毫无改变地返回的环路，有零和乐：它绕了一圈又回到同一个地方，绕行毫无所获，即本系列称为坏的无限者的纯然重复。一个使其系统富裕地返回、携带一个正相位的环路，有正和乐：它返回了它的根，却作为更多者归来，是螺旋而非圆。而一个使其系统枯竭地返回、泄漏了相位而非获得了相位的环路，有负和乐：净汲取的循环，那个逐渐衰落的关系。那环路的和乐，那关系之穿越的和乐，是那个”承诺历时被活，加深、重复、还是流尽”的几何度量。

7.2 非强制的承诺作为正的、绝热的和乐

非强制的承诺，被守持，是正和乐的环路，而它如此，是以几何相位所要求的那种具体方式：绝热地。几何相位是一个被绕其环路运送得足够慢、以致它不被那运送所驱动而是跟随它、与其周遭不交换能量、其演化由内统辖而非由外抽运的系统所累积的相位。这一绝热条件，是§5称为推断之原因、§6称为自我承诺与被押注通道者的几何映像。那个不带刀剑而被守持的承诺，是绝热地被守持的：它的环路之被闭合，不是由一个驱动各方绕行它的外部强制，而是由各方自己的承诺，由内被维系，凭借每一方自由地带入它之物的完整而回返于自身。没有任何外部之物抽运那循环；各方不是被一项被威胁的后果驱动着绕环路而行；那闭合是内生的，而它所累积的相位，正出于那个缘由，是一个真正的几何相位——一个由环路自己的内在结构所生成的加深，本系列与好的循环相认同的那份盈余。非强制的承诺是正和乐的绝热环路：信任由那关系自己的内在完整所维系并加深，无须外部之泵来把它撑闭。

这给出了§5那个主张的几何内容，即那个没有强制力的承诺是本然的情形而非有缺陷的情形。在几何词汇里，缺乏一个外部之泵不是那环路的一个缺陷，而是它的相位竟成其为几何相位的条件。一个从外部被抽运的相位不是一个几何相位；它是一个动力学相位，由被喂入的能量所驱动，而它不是环路自己结构的痕迹，而是那驱动的痕迹。绝热的环路——那个无外部之力驱动者——是唯一一个其累积之相位纯然是环路自己的环路；于是那个不依凭任何刀剑的承诺，是唯一一个其加深真正是它自己的、由内生成而非由外供给的承诺。于是几何读法以它自己的术语，复得了那个结果：缺乏强制力，是承诺之加深成其为承诺自己的加深的条件。

7.3 债务与被强制的合同作为零和乐之线

把这衬托得鲜明的那个对照，是被强制的合同，而在它背后是债务的结构，几何读法把它揭露为某种与一条环路相当不同之物。一笔债务不是一个回返于自身并加深的循环；它是一条线，从一个被抵押的当下画向一个必须清偿它的未来。债务人不穿越一条带回一份盈余的环路；他被那制裁的威胁，按在一条从义务之招致通向其被强制之清偿的轨道上，而那轨道被那驱动它的外部之力按直——被阻止闭合成它自己的任何环路。这是被强制的合同、以及作为其极限的债务的几何映像：一条被一个外部之泵撑开并导向的路径，不累积任何几何相位，因为它不被准许闭合于自身，它的运动不是一条环路的内生穿越，而是沿一条线的被驱动的进展。

在这样一条路径被强行弄成一个循环的外观之处——在一个被强制的关系被弄得重复之处——它的和乐是零或负，从不是绝热环路的那个正相位。零，在纯然被强制之重复的情形：那关系绕行又回到同一个地方，义务被清偿又被重新招致，毫无带回，因为那闭合是从外部被强加的、而那环路不累积它自己的任何相位——一个由力撑闭之循环的坏的无限。负，在净汲取的情形：在那外部之泵驱动那环路以从中抽出相位之处，那关系逐渐衰落，一圈一圈地被枯竭，本系列与寄生者及剥削者相认同的那个结构。被强制的关系无法累积正和乐，因为正和乐是一个绝热的、内生地闭合之环路的标志，而被强制的关系恰恰是那个不绝热者：它从外部被抽运，而一个被抽运之环路的相位不是它自己的。刀剑，在这一读法上，不只是未能加深那绑定；它封死了那加深，凭借以一条被驱动的线或一个被强制的重复，替换那唯一能加深的自由的、绝热的环路。

7.4 谱隙与”甘え”问题

还有一个几何读法所照亮的结构，本系列在它对甘え的处理中把它留待解决——甘え即那种甜而退行的、对他者之纵容的撒娇式依赖，那讨论能标出却无法化解其矛盾性。那里未解的问题是：在一个关系之诸可能运动的空间中，那条界线落在何处，介于”一个关系能吸收并代谢、并把它作为加深的亲密而返回”的撒娇，与”一个关系无法吸收、它作为偏离而累积并驱动循环向发散”的撒娇之间。几何读法提供那缺失的结构，以一个谱隙的形式。一个环路吸收一个扰动并绝热地返回它闭合之穿越——而非被那扰动驱上一条发散之路——的能力，依赖于统辖那环路之动力学的谱中的一个隙：一个慢于那隙的扰动被绝热地跟随、并被代谢进那环路自己的相位，而一个快于那隙的扰动则把系统驱过那隙，驱出绝热的区域，驱上一条不再闭合的轨迹。良性的甘え是那环路能绝热地跟随的扰动，被吸收进一个加深的和乐；腐蚀性的甘え是那超出那隙的扰动，把那关系非绝热地驱向§6.4在预测语域中描述为一个被伪造之精度之崩溃的那种偏离与发散之累积。谱隙是那缺失的结构；它在此被作为那关系承受被撒娇于其上之物的能力的几何对应物而提供，并作为当前这篇论文部分地偿还早先讨论所留之债的那一点。

7.5 两个读法的汇聚

几何读法与§6的预测读法，归根结底，是被看见两次的同一个解说，而它们的汇聚值得平白陈述，因为正是它许可了对二者的使用。那个正和乐的、绝热的、内生地闭合之环路，是那个”其信任由那三条耦合的通道从内维系、无须外部之泵”之关系的几何肖像——真实的情形。那条被驱动的线，以及那个零或负和乐的、被强制的环路，是那个”其精度从外被注入、而非由一个真实的似然所赚得”之关系的几何肖像——被强制的情形与伪造的情形一样。预测读法描述为”一个由真实的自我承诺之似然所赚得的精度，对一个由外部压力所注入的精度”者，几何读法描述为”一个由一个绝热环路所累积的相位，对一个被抽运进一条被驱动之路径的相位”；而这些是同一区分，介于一个由内统辖的过程与一个由外驱动的过程之间的区分，以两种词汇被呈现。这两个独立的读法汇聚于同一条线——介于内生的与被抽运的、绝热的与被驱动的、真实的与伪造的之间——这是本文所能提供的、那条线是真实的而非任一词汇之人为产物的最强内部证据。

这一汇聚向本文将接手的两个方向前行。它指向§9，在那里，对”几何相位在其起源上是一个量子动力学的概念——贝里相位即一个被绝热运送之量子态的相位”的承认，使在此使用的几何语言能被看作它一路以来之所是，一个从那动力学的借用，关系性信任在那里将被表明分享那动力学的结构。而它指向§12，在那里，介于”其相位是它自己的绝热环路”与”其相位是借来的被抽运路径”之间的区分，成为那个倒转的几何根据：无强制之诺——那个不依凭任何外部之泵的环路——被证明是”承诺之所是”的更纯粹的实现，而那可执行的工具，带着它对外部之泵的掺杂，是那被稀释的实现。

8. 正义判准：真实的与伪造的信任

§6的机制供给了，作为一个副产品，对它自身赝品的一个机械刻画：伪造的信任作为一个从外被注入、而非由一个真实的自我承诺之似然所赚得的精度；而§7的几何读法给同一赝品它的几何形式，作为一个被抽运进一条被驱动之路径、而非由一个绝热环路所累积的相位。但那机械的与几何的刻画，就其自身而言，是价值中立的。它们告诉我们伪造的情形是不稳定的——一个不被似然支撑的精度会被累积的误差所追上，一个从外被抽运的相位不是那环路自己的——但不稳定尚不是不义，而一个瞬态尚不是一桩错。本节作出那个进一步的主张：那条介于真实信任与伪造信任之间的线，机制把它画为一条介于稳定者与不稳定者之间的线，它也是、且不可化约地是一条正义的线，介于”一种产生信任的、荣耀那信任由它在其中被产生之人的方式”与”一种使他蒙冤的方式”之间。那主张是：这两条线——机械的与伦理的——重合，而本节的任务是说出它们为何重合，却不佯称二者之任一可从另一者导出。

8.1 为何伦理判准不可从机制导出

必须先说出本节不主张什么，因为本系列的纪律要求它，而过度伸张的诱惑是真实的。它不主张伪造之信任的错性随其不稳定而来——不主张伪造的情形不义是因为它注定崩溃。那会是从一个动力学前提导出一个伦理结论，从一条轨迹读出一桩错，而本系列始终着力保持这些正交。机制回答一个动力学问题：这信任会存活吗，它有据吗，误差信号会确证还是驳斥它。正义判准回答一个不同的问题，一个没有任何数量的动力学信息能了结的问题：这是一种荣耀他者还是使他蒙冤的对待他者的方式吗。一个碰巧、出乎意料地稳定的伪造信任——一个从未被识破的操纵，一个从未崩溃的被抽运相位——不会因其稳定而少一分是一桩错；而一个无人之过而崩溃的真实信任——一个被境遇追上的真实自我承诺——不会因其崩溃而少一分荣耀。这两个判准可被弄得相符，而本节将论证它们在极限处确实相符；但它们不是同一个判准，而那相符是一个实质性的发现，而非一个定义上的同一。这是本系列在别处坚持的那个正交，介于”一个循环是否活着”的问题与”它是否正义”的问题之间，而它在此以如下形式有效：机制告诉信任是否立得住，正义判准告诉它是否本应如其所是地被产生。

8.2 那条单一的线，以伦理画出

那一点既已确立，那重合便可被画出。操纵的机械标志已在§6.4陈述：诱使他者的生成模型降低其预测误差，而不供给那会辩护此降低的高保真似然。操纵的伦理标志是同一举动在一个不同方面之下被描述者，而本节为那一个举动指名三个方面，其中每一个都是那操纵使那被操纵者蒙冤的一种方式。

第一个方面是认识论的。在另一个人那里诱发一个不被任何真实似然支撑的高精度预测，便是败坏他真切地预测世界的能力——在他的模型中安装一种世界不会印证的置信，从而使他与他自己未来的经验相忤。那被操纵的人不只是关于一个特定事实被欺骗；他形成期望的器具被动了手脚，被弄得交付一个证据不辩护的精度，于是他被弄得，恰在这一方面，成为一个比他从前更差的认知者。操纵是对他者认识论自主的一种伤害：它不通过给他可掂量的依据来说服他，那会使他的判断完好而自由；它从外部、在他的判断背后作用于他的模型，以在他那里产生一个他不曾依据地形成、且不可能拒绝的期望。这是操纵区别于诚实劝说的确切认识论意义，而它是机制使之精确的一个差别：劝说供给一个他者的模型可掂量并接受或拒绝的似然；操纵注入一个那模型不曾赚得、且无法审查的精度。

第二个方面是康德式的，而它把本文带回§5的那个区分。以注入精度、而非以供给一个真实的自我承诺之似然来产生信任，便是把他者当作的，不是一个在依据上形成它自己之期望的自由心灵，而是一个其状态有待从外部被安排的机制——把机械的原因、而非推断的原因施于他，管理他、而非面对他。它是、以定言命令第二程式的术语说，纯然把他当作手段：当作一个其预测性状态被人朝自己之目的工程化的对象，而非当作一个自由地形成其期望、且本可抑住它们的、其自身即目的者。真实的信任之产生，相反，恰如§5所要求那样面对他者：它向他供给一个真实的依据——一个代价高昂的、不可撤回的自我承诺——他的自由判断可接取它，而它把那接取留给他。于是真实信任与伪造信任之间的差别，在其第二个方面，是”把一个自由存在者当作自由者来面对”与”把一个自由存在者当作机制来摆弄”之间的差别；而这便是为何伪造的情形不只是信任之产生中的一个技术缺陷，而是一桩对那信任由它在其中被产生之人所作的错。

第三个方面是本系列从它关于生成性正义的工作中带来的那个，而它是最深的，因为它指名操纵者对那整个关系所做之事。操纵，便是把他者的信任当作燃料——汲取他依赖的能力、他对被承诺的敞开，当作一项有待为一个不是他的、且把他排除在外的目的而被消耗的资源。伪造的关系是这样一种关系：那信任的一方被弄成一个纯然的手段，通向一个返回操纵者、而非返回他的价值：他的信任被开采，他的脆弱被利用，而他在那循环中被安置为那个供给他人所消耗之物的人。这是本系列陈述为”要向任何加深之循环提出的问题”的那个判准：其中是否有某个人只作为一项成本而出现、被排除在那回返之外——是否有一个人被弄成纯然燃料？伪造的信任回答是；它是一个其功能在于从持有它之人那里汲取、而那人在持有中被用尽的期望之产生。真实的信任回答否；它是一个开启一段关系的期望之产生，在那段关系中那信任的一方是一个受益者而非纯然一个源头，在那段关系中所生成者返回那个其信任使之可能之人。信任的正义，在这第三个也是最深的方面，是那回返的正义：真实的信任返回那给出它之人，伪造的信任从他那里汲取。

8.3 三个方面是一

这三者——那认识论的伤害、那康德式的化约、那弄成燃料——不是操纵碰巧一同犯下的三桩分立的错；它们是一个单一举动的三个方面，那举动是机制描述为不当精度之注入者。注入精度而无似然，干脆就是败坏他者的认知（认识论方面），那干脆就是绕过他的自由判断并把他当作一个机制来摆弄（康德方面），那干脆就是把他安置为一个有待被汲取的源头、而非一个回返中的伙伴（燃料方面）。机制给那举动以它的机械描述；那三个方面给那一个举动以它的伦理描述；而那机械的描述与那伦理的描述是同一个举动的诸描述，正是这一点使得机制所画的线与正义所画的线重合。它们重合，不是因为一者从另一者导出——本节以拒绝那一点开始——而是因为它们是同一物在两个层级上的两个描述：在另一个人那里产生一个期望，而无那会在其产生中荣耀他的真实自我承诺。

8.4 判准之陈述，及它留给实践者

判准如今可被以一个全文其余部分将使用的形式陈述。真实的信任，是在另一个人那里、通过供给一个他的自由判断可接取的真实依据、并开启一段返回他的关系，而被产生的一个期望；伪造的信任，是在另一个人那里、通过注入一个他的判断无法审查的精度——它把他当作一个机制来摆弄、并把他当作燃料来汲取——而被诱发的一个期望。 这判准，如本系列所要求，是一个被置于动力学判准之旁、而非从它导出的正义判准：机制告诉一个期望是否会立住，判准告诉它是否被正当地产生，而二者必须各自被施用，谁也不坍缩进谁。

一个后果必须被标出，因为它设定了践行将不得不解决的问题，并防范一个全文其余部分承担不起的误读。真实者与伪造者，作为外在的表演，不是可靠地可区分的。那个搬演一个看似代价高昂之承诺的操纵者，与那个作出一个真实承诺的真诚许诺者，可能在那一刻呈现不可区分的表面；那被注入的精度与那被赚得的精度，可能在一段时间里从内部感觉相同。把它们区分开的，不是表演中的一个差别，而是躺在它背后之物中的一个差别——是否事实上有一个真实的自我承诺与一个真实的回返，还是只有它们的外观。这便是为何那条线，尽管在原理上尖锐，在实践中却是险恶的，以及为何真实信任之产生不能被化约为它的诸符号之产生：那些符号可被伪造，而要紧的那个差别恰恰是那些符号本身不揭示的那个。真诚跨越认识论之边界的翻译（§10）与信任的践行（§12），在很大程度上，是对”一个人如何恰恰在这一条件下行得善”的推演——即真实者必须在一个它的符号与伪造者不可区分、且那差别尽管真实却不在表面可被读出的世界里被产生并被奉上。正义判准告诉我们那差别是什么；如何凭一个人从外部无法干脆看见的差别而活，则落在实践者身上。

9. 关系性信任

§6的机制描述了信任如何在一个对彼此临在的许诺者与一个接收者之间被生成：那结构性的形式、那代价高昂的自我绑定、那被押注的跃迁，全都预设了某种直接关系中的两方。但人类交往实际所赖以运行的信任，有很大一部分不是这种直接的类型。我们信任我们从未见过的人，凭借我们确实信任的他人；信任沿着一个关系网络的诸链接而行，触及没有任何直接关系所能触及之处。本节关乎那种被传递的信任——关系性信任——它的机制、它对一个直接情形所不遭受之失败的特有易感性、以及它，本节将论证，与一个开放量子系统之动力学所分享的形式结构。这是序曲所提出的”赢得一个家庭之信任”的问题在其中找到最直接关乎它之资源的那一章；而它是符号的危机在其中获得一个新的、精确的具体说明的那一章。

9.1 传递的机制：推荐信与链条

关系性信任最清楚的实例是推荐信，而它值得密切的机械关注，因为它表明什么是真正在起作用而什么不是。那个天真的看法是，一封推荐信通过推荐者所说之物的内容而起作用——通过他施于候选人的谓词、他堆砌的赞美、他肯定的品质。但这在很大程度上不是承载那信任之物。那些形容词几乎不携带信息；人人都被以最高级描述，而一个已读过一千封信的委员会，从那第一千零一次卓越之宣告中提取不出任何东西（§2.3）。真正起作用的不是内容而是结构：一个委员会C信任一个推荐者A的判断，它与A有一种信任的关系；推荐者A与候选人B有一种真实的工作耦合，从内部认识B，拥有关于B能做什么的具身知识；而信任沿着链条C–A–B被传递，C经由它对A的信任而变得信任B。推荐信作为一个导体、而非作为一个描述而起作用：它把C对A既有的信任，越过链接A–B载去触及B，而C别无他法触及B。

这所揭示的是，在关系性的情形里，信任根本不是任何单一一方的属性。它不是B之中的某物，也不是A之中、也不是C之中的某物；它是它们之间那个关系之结构的属性——网络的一个拓扑属性，居于那耦合的模式中而非任何节点中。C没有关于B的直接证据（证据主义的通道无法触及），也没有与B的直接接触（性格的义务论读法无法触及）；然而C能信任B，而那信任不存在于任何地方，唯存在于那链条中。于是关系性信任是一个第三机制，对证据的与性格的二者一样不可化约，信任借由它跨越一道既非证据亦非性格之直接读取所能横跨的缝隙。而它直接关乎序曲的窘境：一个家庭向一个要娶入其中的陌生人所伸出的信任，在很大程度上，不经由那陌生人所能直接向他们出示的任何东西，而经由那条凭借他们已经爱并信任之人把他接连于他们的链条——那个爱他的人，在网络的结构中，是他的推荐人，而那信任沿着那链接触及他，否则便根本不触及。

9.2 关系性信任横跨三个域

关系性信任不安置于一个域，而是横跨全部三个域，而这是它的力量与它特有之脆弱二者的关键。有一个想象成分：那沿链条而行的B之意象，A的叙述在C那里安装的那个被理想化或被扭曲的图像。有一个象征成分：那封信、那个言词、那链条本身作为一个可传递的符号结构。而有一个实在成分，它是问题的核心：A与B之间那真实的工作耦合，A关于B是什么、能做什么的具身的、不可造假的知识，那整套装置之存在所要传递之物。真实的关系性信任是这样一种情形：实在成分领头——其中沿链条而行者是一个真实的耦合，沿一个符号结构被作保，在C那里形成一个奠基于某种实际存在之物的期望。链条是那象征的导体；它本应导送者是那实在者。

9.3 载体与实质的脱位

这里是本节所许诺的、符号危机的那个新的具体说明，而它随那刚被确立的事实而来——即关系性信任横跨诸域，而诸域可被撬开。关系性信任本应传递的那个实质，是那实在成分：A与B的真实耦合，那个在整桩交易中唯一有价值的、不可造假的知识。但它借以被传递的那个载体是象征性的：一封信、一个言词、一个可书写、可流通、可归档的符号。而那毁灭关系性信任的异化，发生于那载体与那实质的脱位之中——那象征形式被完好地保存，而那实在实质被从其中掏空。

一封伪造的推荐信恰恰是这一脱位。在象征域中它无懈可击：格式正确、措辞精当、签署得当、沿链条C–A–B以完美的秩序被传递。单从象征域来看，它与一封真实的信不可区分。但它的实在成分——A关于B的实际知识——是空的：A写它，不是因为他具身地知道B是好的，而是因为他欠一个人情、或混迹于同一个圈子、或无法体面地拒绝。那符号被传递了，而那符号本应承载之物并不在场于它之中。这是本系列命名为”结算的失败、符号的篡夺”的那个核心病理在关系性信任中的化身：一个宣告一个已被作保之耦合的符号，而无真实的耦合在其下背书，一封兑现一个从未拥有之知识的信。伪造的推荐信是那虚假之诗与那被异化之聘礼的同构物——一个声称锚定实在者、而实在者已被从其下抽尽的符号。

因此对那崩坏的诊断是精确的，而它是本节对§3之诊断的新贡献。关系性信任之崩坏，不是因为机制有缺陷，而是因为它横跨诸域而诸域可被分开：那象征的可传递性（信、链条）与那实在的不可造假性（实际的耦合），本应是同一物的两面，可被撬成两个，前者被保留而后者被掏空。伪造的关系性信任是那道裂隙的产物。

9.4 双向的异化

异化一词在此是精确的，因为那损害沿链条在两个方向上奔流，而本节必须追溯二者，因为它们一起解释了为何那整个通道关闭、而非仅仅在好信之旁容许一些坏信。

在那被作保者一侧，伪造推荐信的增殖贬低了真实的推荐信。那个确实与一位确实认识她的导师有一种深厚工作耦合的学生发现，她的信，在象征域中，看起来与那封出于义务而写的信一模一样；于是她的真实耦合不再能被传递，因为那本应承载它的通道已被那些什么也不承载的符号所饱和。那真实之物丧失它被辨认的能力，不是因为它在个别上被怀疑，而是因为那些伪造已使整个通道不可读。这是符号危机的确切机制，又一次被看见：不是假者通过取代真者而把它驱逐，而是假者使真者不可辨认，于是那通道作为整体被弃用，而那制度退却到纯然可计数者——引用数、排名——那些贫瘠却至少不可伪造的信号，宁取一个已知贫乏的通道，而不取一个本可丰富、却已被毒害的通道。

在那作保者一侧，损害更微妙，且是一种最完整意义上的异化。那个习惯于写他并不相信之信的推荐者，与他自己的作保之举相异化：他的签名不再意味着他真的知道，而变得只意味着一项社会义务已被了结。而在如此做的过程中，他一点一点地耗减那使关系性信任成为可能的关系资本——C对他判断的信任——花用它去了结人情，直到那信任本身被耗尽。他为满足他圈子的要求而抽走那个使他的言词能承载信任之物本身，而当它耗尽时，经由他的那条链条便什么也不导送。那双向的异化——一侧真实者被弄得不可辨认，另一侧作保的能力被耗减至枯竭——是为何关系性信任的失败不是某些坏信的一桩局部之事，而是一个通道的关闭。

9.5 关系性信任的量子动力学结构

那已浮现的结构——一种是一个网络而非其诸节点之属性的信任，它横跨可被分开的诸域，它通过一个可传递之形式与一个不可造假之实质的脱位而失败——有，本节如今论证，一个在开放量子系统之动力学中的精确形式对应物。这一论证是在一条纪律之下作出的，那纪律必须在它被作出之前被陈述，且被陈述得与§6中那条对预测编码之使用的纪律一样坚定。在此所宣称者，是一种结构同构，而非一种物理还原。 本文不断言信任在字面上是一种量子现象；如此说会是最坏一类的伪科学神秘主义，并会断送本文所做之一切别的事的严肃性。所断言者，是关系性信任的数学结构与一个开放量子系统之纠缠-退相干动力学的数学结构是同一个结构，因而后者的形式语言能以一种别无现成语言所能给予的精度来描述前者。这是本系列始终遵守的那条纪律：量子动力学的解说是又一个层级的描述，一个关于一个结构的形式语言，而非其余之下的真理，也非一个吸纳它们的元框架。

在那条纪律之下，可画出三个对应，按深度递升的顺序。

第一个是纠缠作为关系性信任的非局域结构。一个纠缠态的定义性特征是，整体的态无法被分解为其诸部分之态的一个乘积；信息居于诸部分之间的关联中，而非任何单独取来的部分中，于是对任何单一组分的一次测量都无法复得它。这恰恰是§9.1中所确立的结构：关系性信任不是C、A或B各别的属性，而是那关联C–A–B本身的属性，无法从任何孤立的节点复得。信任的非局域性——它活在链条中而非链接中——被纠缠的语言给予了一个它否则便缺乏的精确形式表达。

第二个是退相干作为载体与实质的脱位。一个相干的量子态，与一个环境耦合，经历退相干：那些构成它之相干的关联不是被径直摧毁，而是那些使它们有意义的相位关系被丢失给了环境，留下一个其关联壳层存续、而那曾给它以其特征之相干已流尽的态。这，在形式上，是§9.3的那个脱位：那象征的可传递性（那关联壳层、那符号之链条）被保留，而那真实的耦合（那使关联承载某物的相位相干）被丢失。一封伪造的推荐信是一个退相干的信任态：它在象征域中保存那纠缠的全部外在形式，而它本应承载的那个相干已不在——而且，正如一个退相干的混合态在那些非相位敏感的测量之下可与一个纯态不可区分，那封伪造的信在任何不对其下之真实耦合敏感的读法之下都与那真实的信不可区分。而符号的危机，在这一读法上，是关系性信任在一个社会之尺度上的系统性退相干：一个被伪造信号饱和的环境，持续地耦合于那些穿越它的信任态，抽尽它们的相干，留下那些不再承载它们本应承载之物的关联壳层。这与§6.5的预测编码诊断恰好相吻合：信任态的系统性退相干，与群体对那象征通道之精度权重的理性崩缩，是被在两个层级上描述的同一个过程。

在这里动力学的一词，而非纯然量子的，赢得了它的位置，因为关系性信任从这一类比中最需要者，不是一个态的静态图像，而是一个关于那态如何演化的解说。一个开放量子系统的动力学——一个与一个环境持续交换的系统，其演化由一个主方程统辖，该方程具体说明相干如何衰减、以及在何种条件下它可被保护乃至被恢复——是那个关于一个信任态如何历时被维系、被败坏、或被重建的形式语言，是一个静态的纠缠解说所无法供给的。而这正是当前这一节焊接于§7的那一点，因为那里所部署的几何相位，在其起源上，恰是这一量子动力学的一个概念：贝里相位是一个量子态在绝热运送之下所累积的相位。因此§7的和乐语言从来不是一个自由漂浮的借用；它一路以来都是从关系性信任在此被表明所分享的那同一个动力学中汲取的。绝热的演化——缓慢、与环境不交换能量、相干与几何相位被保存——是那真实的、内生地被维系的信任；非绝热的演化——耦合于环境、退相干、相位丢失——是那伪造的信任。§8的真实-对-伪造区分、§7的绝热-对-被驱动区分、与在此所画的相干-对-退相干区分，是以三种词汇表述的同一个区分。

第三个也是最精确的对应，关乎传递本身，而它要求一个表明这一类比是被认真对待而非被草率伸手去取的精炼。赤裸的纠缠本身不会为信任的传递建模，因为纠缠受不可通信定理所制约：它单独无法被用来发送信息。但关系性信任恰恰关乎信任沿链条C–A–B的传递。因此这一传递的形式对应物，不是赤裸的纠缠，而是纠缠交换：那个协议，凭借一个已与二者各自纠缠的中介节点A，使两个不分享任何先前纠缠的方C与B被带入一个纠缠的关系。纠缠交换几乎恰恰是推荐机制C–A–B的量子对应物：那个与二者皆耦合的中介A，是那个在本无任何关联的C与B之间建立起一个关联的手段。把那对应物指名为交换而非赤裸的纠缠，便是标出这一类比是结构性的且经过斟酌的——它追踪关系性信任的那个具体特征，它经由一个中介的可传递性，而不只是它的非局域性。

9.6 类比的地位

在§9.5开端处陈述的那条纪律，必须在它的结尾处被重述，因为它是整个建构得以容许的条件。量子动力学的语言是对一个被分享之结构的一个形式描述，处于一个不取代本文已给出之其他描述的描述层级——§6的机械解说、§8的伦理解说、那把链条当作一段人与人之间关系的现象学解说。它不是一个”信任在其物理基底上是量子力学的”的主张；它是一个”关系性信任与开放系统量子动力学例示同一个数学结构、且那结构因被以那个为它发展了最丰富工具的语言来看而被照亮”的主张。它是、以本系列所要求的复调术语说，关于一个单一对象的又一个不可化约的认识论立场，一个再多的声音而非最后的言词——而它是作为这样者被提供的：一个凭借它为那活在链条中而非其链接中之信任的非局域性、退相干、与传递所带来的精度而赢得其位置的形式类比。

10. 真诚的翻译

至此为止的每一章都在如下假设上工作：一旦真实的信任被产生，它便被领受——一个真实的自我承诺，被供给另一个人，便作为它之所是的依据而落地。本节移除那个假设，而在移除它的过程中，敞开那个对序曲的窘境而言最为迫切的问题。真诚没有普遍的语言。 没有一种对每一个人都可读的真诚形式；真诚必须被编码才能被领受，而那编码的语言依赖于那将领受它之人的认识论。婚约文书的郑重、月、诗、哲学的严肃——这些把真诚编码为一个对一个直接读取性格之人可读的形式，那种人把那姿态的风度、庄严与代价本身当作信号。对一个别种认识论的接收者——一个认为是就是是、并索求可被观察与验证之物的人——同一份真诚，在同一个编码里，根本不是证据。同一份真实的真诚，经过两个不同的解码器，产出全然不同程度的可信性；而那差别不在那真诚里，它是一而真实的，而在那发送者的编码语言与那接收者的解码语言之间的错配里。

这是本文的问题经历它决定性转折的那一点。这问题，经由机制及其后果，一直是一个本体论的问题：真实的信任是什么，它如何被产生。它如今变成一个翻译的问题：真实的信任，一旦被产生，如何跨越不可通约之认识论之间的边界而被传递。而这一翻译的问题，在信任的领域里，是本系列始终承载之主题——没有元语言，他者不作为一个立于一切立场之上的担保者而存在——的最尖锐形式。没有中立的、普遍的真诚语言，没有一本上帝视角的词典，其中在一种认识论里被编码的真诚能被查阅并以其价值被保存地在另一种里被重新表达。因此信任的问题，完整地陈述，不仅是如何成为真实的，而且是如何被一个其阅读语言不是自己的语言之人读作真实的；而本节是它的理论。

10.1 翻译的认识论

在一个人能把真诚翻译进一个接收者能读的语言之前，他必须知道有哪些语言——一个接收者可能在使用哪些信任的语法，凭借它们，一份真诚的宣告被解析为一个期望的依据、或被驳回为根本不是依据。本小节列出这些语法的一个谱系。它是作为足以达成那实践目的者、而非作为详尽无遗者来提供的，而它的精确要紧，因为任何翻译的有效性都依赖于那目标语法被辨认的准确度。

10.1.1 信任语法的一个谱系

至少有四种信任的语法，彼此不可化约，每一种把一类不同的东西解析为一个期望的依据，而每一种都有它特有的触及范围与它特有的盲点。

第一种是证据主义的或实证主义的语法。它把信任锚定于可观察的证据与已成的事实——过往的记录、已经做成之事、稳定而可验证的模式。对一个使用这一语法的人而言，一个信任的依据是某种已经显示自己之物，而一个尚未在事实上显示自己的宣告尚不是一个依据。它的触及范围伸入那已被证明者；它的盲点是那个真诚的新来者，他至今没有记录可出示，而它必定、按它自己的逻辑，把他误判为不可信赖，不是因为他是，而是因为他尚无机会证明相反。这是霍姆斯的坏人观察者（§4.5）被转向对他人的阅读时的语法，而它，本文有理由认为，是序曲那个家庭最依赖的语法。

第二种是义务论的或性格学的语法。它把信任锚定于对他者之性格与真诚的一个直接的、直觉的读取——锚定于那风度、那举止、那承诺的品质，被当作一个内在状态的即时信号。对一个使用这一语法的人而言，一个誓言的郑重、一个姿态的严肃、一个人被感到的真诚，本身就是依据，无须绕道经过已成的事实便被读出。它的触及范围伸入那内在的与当下的；它的危险是它读得太多，它能被一个已学会表演一份他并不拥有之真诚的诸符号之人所欺骗。这，本文有理由认为，是那个会把他的真诚编码进红绢与月之人最本土的语法。

第三种是归纳的或关系-历史的语法。它把信任锚定于与这个特定的人的一段共同历史的累积模式——不锚定于证据本身，也不锚定于对性格的读取本身，而锚定于在我们之间所经历者的纹理，那个关于这一个人历时如何与我相处的记录。它的触及范围伸入那已知的特殊者；它的盲点是那全然新的关系，对它尚不存在任何共同的历史。

第四种是制度的或背书的语法。它把信任锚定于一个受信赖之第三方的作保——媒人、长辈、资质凭证、一个共同体的承认、沿§9之链条的推荐。它的触及范围伸入那受信赖之制度所能背书者；它的盲点是那个贯穿本文被诊断的境况，其中背书的诸制度本身已陷入符号的危机，不再能可靠地背书任何东西。

这四者不是一个排名，仿佛其一是真的语法而其余是它的近似。它们是不可通约的认识论立场，每一种把某种其余不解析者解析为一个依据，没有一种可化约为其余——而它们那不可化约的复数性，在信任的领域里，是本系列所坚持之复调律的又一次印证。这一谱系是否足够精细的问题被留待解决，而尤其一个精炼被标为未定：证据主义的语法是否应被细分，介于一个索求已成之事实者与一个索求可证明之能力者之间，因为这二者对”一个翻译进那语法所必须供给者”有不同的实践含意。

10.1.2 语法配置的概念

没有人纯粹地使用一种单一的语法。每个人使用一个配置——一个对那几种语法的、大多不加反思的加权，其中一种可能占主导而不排除其余。序曲那个家庭，最好被理解为不是纯粹证据主义的，而是运行一个配置，其中证据主义的语法占主导，制度的有一些权重，而义务论的被大幅折扣——于是一份性格的宣告落地很轻，而一桩已成的事实落地很重。配置的概念是那个使谱系成为一个实践工具、而非一个类型学之物：一个人不是把他的真诚翻译进抽象的那个证据主义语法，而是翻译进一个特定接收者所运行的那个特定配置，而对那配置的诊断是任何实际翻译的第一项任务。翻译的单位是配置，而谱系是诸配置由之被拼写的那个字母表。

不可能有一份真诚从一种语法进入另一种的单一正确翻译——翻译在原理上是不确定的——这是支撑所有这些的那个深层认识论事实，而它是蒯因的。翻译的不确定性不是一个有待克服的缺陷，而正是元语言之缺席在翻译理论中所取的那个形式：不存在一个独立于每一框架而被固定的、关于”一个意义从一种语言进入另一种之唯一正确呈现”的事实，因而不存在一个中立的标准，可对照它把一份真诚的翻译证实为那个正确的翻译。翻译一路向下都是欠确定的，而下一小节所描述的实践技艺是一种在那永久之不确定性下被践行的技艺，而非一种逃离它的技术。

10.2 翻译的践行

那么，一份在一种语法里被编码的真诚，如何被弄得在另一种里可读，而不扭曲它、且不滑入§8所标出的那种操纵？有三条路径，按难度与深度递升，而每一条都贴近那条伦理的线，因为每一条在其外在形式上都难以与一个赝品区分。本小节列出它们，然后陈述那条把这三者、真实地做时，与它们伪造的孪生分开的线。

10.2.1 投射

第一条路径是投射：让一份在自己语法里可读的真诚，在接收者的语法里投下一个影子。一个人不把那义务论上可读的真诚伪装成证据，那会是伪造；他让那真实的真诚，在接收者所读的那个象征与可观察的域里，留下它自然投下的痕迹——一种一致行动的累积、一份历时持守的恒常、一种证据主义者能读作证据的行为，因为它真的是一个真实内在状态的可观察影子。证据主义者从那影子反推回那实质。这是最安全的路径，也是最慢的，因为它依赖于时间：那影子唯有随着那行为累积而累积，且无法被催促。它是本系列承载为”持守得足够久之物终会显示自己”这一格言的实践翻译——不是说时间创造那耦合-信任，那从一开始便是真实的，而是说时间让那真实者在证据主义者所读的域里投下足够的影子以被辨认。投射是把一份真实的内在真诚呈现进已成之事实的语言，凭借让它成就那些作为它真正影子的事实。

10.2.2 多重编码

第二条路径是多重编码：把那一份真诚，同时，在两种语法里、在两条并行运行的轨道上编码。对那读取性格之人——以及对自己——是那誓言、那郑重、那诗；对那读取证据之人，是那稳定而可见的事实、那可观察的恒常、那被证明的可靠。这两个编码是同一份真诚的，各以那能领受它的语法被奉上。这条路径的难度不是技术的而是伦理的，且它是严峻的：多重编码退化为纯然的安抚，除非一个人真正地承认他者之语法的合法性。倘若一个人在证据主义语法里编码，而私下持守”索求证据是肤浅的、义务论的读法才是真的而证据主义的是对迟钝者的让步”，那么那第二个编码便不是一个翻译而是一个屈尊俯就，而一个有任何敏锐的接收者都会读出其下的那份屈尊。真实地做的多重编码，要求一个人把两种语法都持为抵达信任的合法方式——一个把它直接接连于§12所接手之”呈现的正义”的要求，在那里，主动地把一个人的真诚呈现进他者的语法被表明，当那实质是真实的，不是一种策略而是一种尊重的形式。

10.2.3 相互翻译

第三条路径是最深的也是最难的，而它是本节最想确立的那一条，因为它是唯一一条产出一种既非被投射、亦非被双重广播、而是被共同造就之信任的路径。在相互翻译中，双方各向他者的语言移动一步：一方学会以证据说话，而另一方学会读取性格的信号；而所被产出者，不是一方对另一方的说服，而是在那互动本身中，一个先前并不存在的小小共享语法的生长。这是认识论层级上的真实耦合——一种不是一方的信念被传给另一方、而是在它们之间、在持续的互动中、从两种各已向他者移动的语法里生长出来的、在它们之间长出一片共同语言的信任。它不先于那关系而存在并等待被传递；它被生成，在那相互的移动中，作为那关系自己的东西。

这不是一个哲学家的空想、而是一种真实且可学的技艺，这一点由一个恰恰做此事的成熟职业的存在所表明。那个经验丰富的顾问，在任何领域，被带到一个他尚不理解的客户面前，面对那个无元语言问题的结构孪生：客户带着一种语言、一种处境、一套术语与不言而喻的假设以及一个真实的难题前来，而顾问起初无法读取它；而顾问既不能把他自己的框架强加于它，那是糟糕的咨询，也不能佯装理解他所不理解之物。真正的工作是在那互动中、与这个特定的客户一起，创造一种共同的语言——学会客户的术语并让客户学会他的，直到一个先前在二者之中皆不存在的共享语法在他们之间存在。这是相互翻译，而咨询是一个成熟的职业，有方法、有训练、有胜任能力的检验，这一点是相互翻译能被系统地且良好地做成、而不是它否则可能看似的那个不可操作之理想——正是这条最难的路径最易招致的那个反驳——的证明。关于此类工作之成果的经验文献为这一点提供支持：横跨各助人与咨询职业，一个良好成果的最强预测因子，被发现不是技术或学派，而是那个工作关系的质量，那个工作同盟——它恰恰是相互翻译所产出之共享理解的建立。这一点部分地回答了奠基之作对经验跟进的呼吁：有一批证据表明，一个共享语法的生成是真实的、可学的、且是成功中的那个起作用的因子。

顾问之技艺的两个特征使这一解说更尖锐。第一个是一种不对称。相互翻译不必是对称的：顾问，作为那承担职业责任的主动一方，比客户向他移动得更远地向客户的语言移动。那个主动寻求建立信任的一方——那个处于序曲之求偶者位置的人——可能理当不得不更远地向他者的语法移动，而这不是他语法之低劣的标志，而是他作为主动而负责的一方承担了翻译之劳的更大份额的标志。这一不对称是§12将称为认识论好客者的具体形式：那向他者的移动，那对”被理解之工作”更大份额的承担。第二个特征指名那个支撑这条路径的哲学资源。相互翻译所成就者，是把两个视域融合为一个二者单独皆不曾持有的共享视域——伽达默尔的视域融合（Horizontverschmelzung），那个对”在两个起初一无共同者之方之间一个共同理解之生成”的恒常哲学命名，它给相互翻译以它在诠释学传统中的奠基。

10.2.4 那条把翻译与操纵分开的线

这三条路径都贴近那条伦理的线，而那条线必须被画得尽可能坚定，因为它是整个践行所赖以转动的线，而它不在、在表面上、一个人会预期之处。把一个人的真诚弄得在他者的语法里可读（翻译），与制造一个在他者的语法里可读的虚假信号（操纵），作为外在的举动，可能看起来一模一样——那被投射的影子与那被伪造的证据、那第二个编码与那安抚的表演，能呈现同一个表面。那差别全然在于那被翻译者是否真的存在。投射一份真实真诚的真实影子是翻译；投射一份并不在那里之真诚的虚假影子是操纵（§8）。真实翻译的判准是：它所翻译之物必须真的存在并能够真实地被实现：它把一个真实的实质载过诸语法之间的边界，而操纵则在目标语法里制造一个缺席之实质的外观。翻译把他者的语法当作一个有待被诚实地面对的合法他者；操纵把他者的语法当作一把有待被撬开的锁。那条线不在举动与举动之间，那可以相符，而在一个真实的被翻译之实质与一个空洞的被制造之信号之间——而它，再一次，是§8的那条线，介于荣耀他者与把他弄成燃料之间。

10.3 对话：三个身份的汇聚

那三条路径中最深的，相互翻译，结果是哲学传统在另一个名字下已知之物，而这一承认给本节以它的收束。相互翻译就是对话，而对话同时是三件结果证明为一的东西。

它首先是相互翻译的实践形式：顾问的技艺，那个在起初无一共同语言之方之间生长出一种共享语言的来回。它其次是哲学一向认为某些真理借以被生成的那个古老方法——苏格拉底式的辩证法（dialektikē），其中真理不是被一方陈述并被另一方领受，而是在它们之间、在问与答的往与返中被生成，因为有些真理除了凭两个心灵彼此移动之外别无他法可达。而它第三是本系列自身的形式律——复数结论的复调、对一个统辖性声音的拒绝、对”事情之真理活在的不是任何单一框架而是它们之间的对话”的坚持。这三者——顾问的技艺、苏格拉底的方法、本系列的复调形式——是一件东西，而它们之为一，正是为何相互翻译是这三条路径中最深的。它最深，因为它就是对话；而对话，在一个没有元语言的世界里，是真理与信任竟能在异质的主体之间被生成的唯一方式。在没有上帝视角的词典把一种语法以其价值被保存地翻译进另一种之处，那价值不是被传递、而是被生成，在各方之间，在那个每一方都向他者移动而一个共享语法生长的对话中。相互翻译就是对话：在两个不分享任何共同语言之人之间，一同生成一种共同的语言。

而这把本节带回本系列始终承载的诗学，因为真诚的翻译，如此理解，本身就是一个诗性的举动，在本系列给予该词的确切意义上。把一个人的真诚翻译进另一个人能读的语法，不是去结算一个意义并把它闭合地交付出去；它是去邀请他者走一条解释的路径，去奉上一个他者必须穿越而非仅仅领受的形式，去为那他者写一首他能读的诗。诗性者拒绝结算并敞开一条有待被走的路径；而真诚的翻译，真实地做时，拒绝以制造来结算他者的信任，而是在两种语法之间，敞开一条共享理解可沿之被走入存在的路径。真诚的翻译是为那个其语言不是你的之人，写一首以一种他能读的语言写就的诗——而本文如今从这一书写的技术转向它的实践，在制度的尺度上、然后在一生的尺度上。

11. 法理学与司法的优化

本文如今从信任的分析转向它的实践，并先取制度的尺度，那个问题在症候学中作为正义的过载与那把高悬却不落下的刀剑而出现的尺度。倘若力不是信任的原因，而只是一件失效的、产生期望的装置，那么那些建立在力之上的制度——法院、强制的装置——便需要一次重新构想，而本节追问那会是什么。它的回答有一条反直觉的脊柱，将被平白陈述：在一个力已作为信号而失效的时代，司法的有效优化，是强制的去中心化——不是刀剑的强化，而是信任之产生从强制这条单一通道分散开来。但本节必须，从一开始，防范这一脊柱可能被误听的方式，而那防范决定了本章的整个方法。

因为这脊柱可能被误听为一个唯心主义的劝勉——一个”多去信任、培育声誉、修复而非惩罚”的呼吁，仿佛信任的制度形式是一桩开明选择之事，一个社会只要看见其优点便可采纳的更好的想法。本系列是历史唯物主义的，而必须辩证地处理此事；而一个唯心主义的劝勉恰恰是历史唯物主义方法所禁止者，因为它会把上层建筑误当作某种与其物质基础分开而可被自由调整之物。信任的制度形式不是被选择的；它们是由它们之生产的物质条件所抛出的，而它们随那些条件改变而改变。因此本章的进行，不是通过推荐一项改革，而是通过分析一个已然在进行的转变——它的物质原因、它涌现的诸倾向、以及它在当前生产关系下被驱入的那个新矛盾。

11.1 司法的双重束缚

有待分析的境况首先是一个双重束缚，其中症候学的两面——过载与力的失败——组成一个恶性循环。随着信任的关系性秩序与共同体秩序变薄，把人们守持于其交往的负担更多地落到法院身上（过载）；法院，被要求以强制产生一个它至多只能保护的秩序，发现它的强制信号被稀释于一个它无法覆盖的数量上，而力作为信号失败（失败）；而随着力失败，信任更加变薄，把更多的负担抛给法院。司法越是被依赖去生成信任，它那单一的信号便越被稀释；信号越被稀释，它生成得越少；它生成得越少，它便越被依赖。司法被困于被要求以力产生那个东西本身——信任——而§6表明力无法产生它、只能至多发其信号；并被要求在一个连发信号都失败的尺度上产生它。这是那个双重束缚，而所施之力的任何增加都无法化解它，因为问题不是力的短缺，而是力作为达成这一目的之手段的耗竭。

11.2 那个反直觉的论旨：去中心化

那解决不是添加力，而是停止要求那条单一的力的通道去做那唯有几条通道之耦合才能做之事。§6确立了有据的信任由诸通道的乘积式耦合所产生，而它的稳健不在任何单独一条之强度，而在它们全都被一同伪造之事的不大可能性。由此而来的司法优化，是同一原理的制度应用：把信任之产生从强制这条单一通道分散开来，分散于机制所辨认的那几条通道之上，于是信任由它们的耦合、而非由刀剑单独所生成。这便是去中心化在此的含义，以及为何它在力之强化所不奏效之处奏效。一条单一的通道可被规避、被稀释、并——在符号的危机下——其精度权重被全面崩缩；几条耦合的通道，其相互一致难以被一同伪造或规避，恰恰在那单一通道脆弱之处稳健。这一论旨是反直觉的，因为面对一次强制的失败，那条件反射是呼吁更多的强制；机制表明这是一个已被耗尽之手段的边际递减的施用，而把信任之产生分散于耦合的诸通道之上才是结构上更稳健的路。

但是——而在这里历史唯物主义的方法对任何唯心主义的读法主张自身——这一去中心化不是一项有待被推荐的改革；它是一个已然在进行的转变，由信任之生产的物质条件的一次改变所迫成，而本章其余部分必须把它表明为如此。

11.3 那转变，辩证地

那转变必须被以四个环节来把握，而第四个环节禁止前三个可能招致的那种庆祝。

11.3.1 中心化之力的物质基础

中心化的、以力为基础的正义——由国家挥舞的刀剑——不是一个社会碰巧采纳的坏想法；它是那个适合于、并由一个确定的物质基础所抛出的信任生产形式。在一个陌生人的社会里，其中交换主要是市场交换，而人们被商品形式原子化、并被弄得流动而可互换，信任便不再能由那个在更小、更安定之世界里产生它的、面对面的关系性历史与共同体见证所产生。在村落与宗族不再把一个人守持于他言诺之处，那守持必须被外包给一个抽象的、非人格的第三方——国家、法律、刀剑——它能在彼此别无任何把柄的陌生人之间强制履行。也就是说，中心化之力曾是那个适足于一个陌生人之市场社会之物质基础的、历史上必要的信任生产形式。这必须被先说，否则便会把这批判变成一种道德化的哀叹：刀剑不是一个错误，而是一种适应，且是一种对其条件正确的适应。

11.3.2 那物质的失败

它的失败，同样，是物质的，而非价值的衰朽。§11.1的那个双重束缚不是由心肠之变硬或公民德性之衰落所产生的；它是由物质条件的一次改变所产生的。交换的规模与速度，在全球化与数字化之下，已超出任何强制装置的物理能力；违约的数量已超出法院的承载能力；而产生一个有说服力之符号的成本已降到零，以伪造的信号饱和了那环境，并崩缩了，如§6.5所表明，一个群体理性地赋予那个中心化强制部分依赖之象征通道的精度权重。这些是在生产力与生产方式之层级上的改变，而它们已把那曾适足的信任生产形式——中心化之力——带入与那些新的物质条件的矛盾。这是符号危机的历史唯物主义解说：不是那个”一种价值虚无主义已席卷文化”的唯心主义故事，而是那个”介于既有的信任生产方式与其下之被改变的物质基础之间的一个矛盾”的唯物主义故事。§3的诊断在此从一个文化的诊断被提升为一个历史唯物主义的诊断。

11.3.3 那涌现的重组

那辩证的要点随之而来，而它是那个把分析与劝勉区分开的要点。强制的去中心化不是某种我们应当选择之物；它是那矛盾已然在迫成存在的一次重组。那些新的非强制信任生产形式——声誉系统、信用机制、平台评级、修复性司法、信任的分布式账本——不是改革者的好想法；它们是一种信任生产方式在新的物质条件下重新组织自身的自发症候，那些网络与计算的新技术随着旧的单一通道失败而抛出新的通道。因此本节的任务不是推荐去中心化，而是分析这个已然在发生的重组：辨认它的方向、它的潜能、以及——那第四个环节——它的内部矛盾。

11.3.4 那俘获：那个新矛盾

因为辩证的方法要求那涌现的形式被就其自己的诸矛盾来审查，而非被庆祝。那些新的非强制信任的通道，在当前的生产关系下，正被资本俘获并转向异化。声誉与信用，本可能是关系-历史通道的解放性制度化，在资本之下变成社会信用评分、对工人的算法管理、作为一种新纪律工具的信用——一种比刀剑更精细、更无孔不入的控制，恰恰是§2.3所描述的那个向可量化者的退却，而那些度量如今握于平台与所有者之手。平台的信任徽章变成资本对信任的商品化，一个平台能垄断并据之收租之物。于是那去中心化的技术形式，在那些新通道的中心化所有权下，可能产生一种比它所取代之刀剑更阴险的支配——一种通过那些本应把信任从强制中解放出来的通道本身而运作的支配。因此历史唯物主义方法所施压的、且§8的正义判准与之一同所施压的问题是：谁拥有这些新通道，它们为谁之利益而运行——它们所产生的价值返回给谁，而其中是否有一个人被弄成纯然燃料？信任生产之技术形式的去中心化本身不是任何解放；在错误的生产关系下它是一种新的汲取方式。因此那综合不是”去中心化已解放了我们”这一凯旋的口号，而是一个新矛盾：一个涌现的形式，其解放性潜能是真实的、而其被资本之俘获同样是真实的，而其结局不由那技术、而由拥有它的那些生产关系所了结。

11.4 三条通道被制度化

在那辩证的框架内，而非作为各自独立的推荐，§6的三条非强制通道可被看到取得制度形式，每一条都有它的解放性可能、且每一条都有它特有的被俘获之危险。

声誉与信用系统是关系-历史通道的制度化：它们把§9那些零散的关系性历史聚合为一个可传递的、制度的归纳信任形式，使一个人累积的交往记录在那见证它的圈子之外变得可读。这是它们的力量——归纳信任的制度性聚合——而它们的危险恰恰是上面指名的那个：信用记录退化为一个纯然量化的纪律工具，关系性历史被压平为一个评分而那评分被转向控制。

修复性司法是自我承诺通道的制度化。它不以力强加一项制裁，而是创造一个时机，让一个冒犯者真实地、不可撤回地、牵连自身地承担他所做之事——那个与他所冤之人的相遇、那个真切而代价高昂的责任之承担——它通过一个真实自我绑定的高保真似然、而非通过强制，产生一个纯然不可执行之判决所不能产生的、对一个期望的重建。它在制度的尺度上运作§6.2的那个自我承诺通道；而它的危险是那相遇被掏空为一个程序性的仪式，一个伪造它本应体现之自我承诺的被搬演的悔过。

公共登记与透明是§10.2那个投射路径的制度化。它们不制造信任；它们让一个真实的履行状态——一个关于何者已做、何者未做的真实记录——在证据主义者所读的象征域里投下那个它可被辨认所凭的影子。它们使那真实者可见，而非产生它的外观；而它们的危险是同一种透明所使能的那个监视，那登记册从一个让真实者被看见的手段，变成一个全面观察的手段。

11.5 刀剑的位置：迁置，而非废除

一条边界必须在此被画得与本章的脊柱被断言时一样坚定，因为那脊柱易受一个危险的过度解读，而那拒绝庆祝涌现之形式的历史唯物主义的诚实，同样拒绝把力之被抛弃浪漫化。信任生产的去中心化不是强制的废除。力保留一个不可或缺的角色——但它是一个不同于它一直在其上失败之角色的角色。力不是、也从来不是信任的生产者；但它是那个必要的后盾，那个事后的保护，而首要地是对脆弱者免于掠夺、免于”强者本可有恃无恐地施加之背叛”的保护。弱者仍需要刀剑——不是去生成他们交往所赖之信任，那是力无法生成的，而是在信任被背叛时保护他们，并担保强者的背叛不是无代价的。信任生产的去中心化不是去强制化；刀剑不是被取走，而是被迁置，从那个信任之生产者的虚假位置——它无法占据——迁到它真正的位置，即那个防范最坏者、尤其代表那些最无力保护自己者的后盾。忘记这一点，便会以一种更纯粹之信任的名义，恰恰抛弃那些那更纯粹之信任尚未保护的人。

11.6 司法作为一种关系性法（nomos）的实践

本章以把它的分析接连于奠基之作与本系列而收束，因为它所追溯的转变，是奠基之作在最大尺度上所指名之一个运动的制度形式。那篇论文，取用马克思与社会再生产理论家的路线，把正义的过载标为一个没有出路的诊断，并敞开一个想法——对着国家最终消亡的视域——即法律不必随它一同消亡，而可能从一种行动之法，即那强制的刀剑之法，过渡到一种无为之法，一种纯粹关系之法，一种关系性的法（nomos）。当前这一章供给奠基之作标出却未展开的那条出路。那中心化的、以力为基础的法——行动之法、刀剑——随它的物质基础溶解而衰朽；而一种建立在关系性信任之生产之上的法（nomos）——无为之法、纯粹关系之法——在那矛盾中萌生。但它是被实现还是被俘获，依赖的，如整章所坚持，不是那技术形式，而是那些生产关系本身是否被改变——依赖于谁来拥有那些新的信任通道，以及它们所生成的价值是否被正义地返回给那些其交往生成它的人，而无一人被弄成燃料。司法的优化，归根结底，是把正义从那个”成为最后的他者、成为一个它从未能以力背书之秩序的最终担保者”的不可能负担中解放出来；并是把正义归还到它作为一种关系性法（nomos）之实践者与保护者的位置，那法的信任在别处被生成——在那些耦合的通道里，在关系的尺度上，全文其余部分所描述者。本文如今转向那个尺度，一生及其最近之绑定的尺度。

12. 信任的践行

本文抵达它实践的顶点，并抵达那整篇一直朝之移动的一章。所确立的一切——机制、几何读法、正义判准、关系性信任的结构、翻译的理论、司法的优化——都一直是为了实践的知识，而本文若止于知识，便会背叛它自己最深的主张。因为本系列认为，安顿于理论之中，把对一物的认知当作它的完成，本身就是那整篇所诊断之病理的一个实例——结算的失败，那个宣告一条无人走过之回路已闭合的符号。一篇关于信任之生成的论文，若止于对机制的一个解说，而不下降到信任实际在其中被生成并被活的那个实践，便会是这样一条闭合而无人走过的回路。因此践行不是附在理论之上的一个应用；它是理论对它自己”不去结算”之义务的履行。本章很长，并在其中心上升到本文形而上学的顶点；当前这一部分发展它的根据，从”为何根本必须有一个践行”的缘由，到对”没有任何践行能担保者”的诚实清算，而随之而来的那一部分升起那顶点，并在最后返回那个家庭与那个房间。

12.1 为何必须有一个践行

本文不能止于理论的缘由是本文自己的。它已论证那诗性的符号拒绝结算——它呈现生成性而非一个闭合的结构，并迫使它的意义被走过而非被兑现；而它已论证那结算的失败，那个对一个无人经历过之升华的兑现，正是它所诊断之符号篡夺的那个形式。止于对信任的理论认知，便会在本文自己的行为中犯下这一失败：以理解之币，兑现一个不曾被走过的实践。因此向践行的下降是由本文的内容所要求的，而不仅是由一个有用的愿望。而那下降也是本文的论证与作者的人生——序曲把它们作为一个问题的近端与远端而分开——相会之处：信任的践行是作者必须做之事，在序曲那个家庭面前，而它是理论一直在为之清理地面以使之可说之物。本章是那个知识转变为一种行动方式的点，并在转变中，拒绝结算。

12.2 一种实践的复调，与一条横贯它们的伦理之线

没有单一的信任实践，而本系列的复调律在实践的语域中有效，正如它在理论的语域中有效：本文所取用的每一个框架都供给一种信任的实践，而没有一个统辖其余。证据主义的实践建立过往记录，累积那可证明者，让事实堆积到它们说话。义务论的实践培育并展示一种性格，把自我造就为一个其真诚可被读取之物。关系性的实践照料§9的那些链条，投资于信任借以被传递的那些耦合。道家的实践，下文有更多关于它的论述，谢绝策略，让真实者在它自己的时间里显示自己。制度的实践运作那些背书与登记册。这些是真实而各异的实践，适于不同的接收者与不同的时刻，而本文不给它们排名；信任的实践智慧在很大程度上是”一个给定处境召唤哪一种实践”的智慧，而那复调，在此如同始终，不是一个有待被化解进一个单一方法的缺陷，而正是一种适足于一个没有任何单一框架所能据有之对象的胜任能力的形状。

但横贯所有这些实践奔流着一条伦理的线，而在这一点上那复调不延伸：描述性地复数，那些实践在伦理上、每一个，都服从于§8的那个单一判准。无论何种实践——证据的、义务论的、关系的、道家的、制度的——它在如下情形是真实的：它通过供给一个荣耀他者之自由判断并开启一段返回他之关系的真实依据来产生一个期望；它在如下情形是伪造的：它通过制造一个把他当作机制来摆弄并把他当作燃料来汲取的信号来诱发一个期望。那复数性是一种合法形式的复数性；那条介于真实者与伪造者之间的线横切它们中的每一个，且在每一个中是同一条线。本章在它对那些实践的描述中是复调的，而在它对它们的伦理中是单调的，而接下来的诸小节先转向那复调自身所要求的一个纠正，然后转向那伦理所排除的那个伪造实践。

12.3 呈现的正义

那复调的诸实践，如刚被列出，隐藏一个必须被提出并被纠正的价值判断，因为它终究已悄然给那些实践排了名，且排错了。有一个诱惑，对道家、并对某种对自然者的读法而言是本土的，要把那不策略的自我显示之实践——让真实者在它自己的时间里、不加机巧地显现——持为真实信任建立的最高与最纯形式，并把任何主动的呈现、任何把自己呈现进接收者之认知所能最佳采取之形式者，视为已蒙上操纵的色彩。在这一看法上，把一个人的真诚呈现进证据主义者的语法，会是从那”仅仅、真诚地存在”之纯粹的一次堕落；而§10的那个主动翻译会带着一丝隐隐的罪疚，作为一种本不应有策略之处的策略。本小节倒转那个判断，而那倒转是整个践行的枢纽。

那倒转转动于一个本文亏欠它自己之翻译理论的承认之上：那主动地把自己呈现进他者能读之语法者——正是那不策略之理想所怀疑之物——不是从纯粹的一次堕落，而是，在恰当的条件下，那抵消一种认识论不义的方式。因为认识论不义在两个方向上奔流，而那不策略之理想只看见其中一个。第一个方向是人人都看见的那个：操纵的主动之恶，利用他者的语法去制造一个并不在那里之物的信号，它败坏他的认识论自主（§8）。但有一个第二方向，那不策略之理想在自以为无辜之时犯下它：拒绝翻译的被动之恶——坚持只在一个人自己的语法里显示自己，而把那领受交付给命运。说”我只会以我自己的方式、真切地显现，其余不是我所关切的”，携带一个不义的预设：即一个人自己的语法是那中立而普遍的语法，其中那真实者，被平白地显示，会对任何有善意的人都可读。但没有这样一种中立的语法（§10）；只在一个人自己的语法里显示自己、并要求他者在那里读取自己，便是把一个人自己的语法仿佛它是那普遍者那样强加于他，而那强加——无论多么高尚、无论多么被裹在真诚与谢绝机巧的语言里——本身就是一种认识论不义，且因把自己伪装成纯粹而更为阴险。那不策略之理想不是谦逊的；它是，在那谦逊之下，那把”一个人自己的可读方式”当作”可读性本身”来对待的傲慢。

因此那纠正是精确的。不策略的自我显示不是真实信任的最高形式；它唯有在那些语法碰巧相符之处是真实的，而在它们不相符之处——这恰恰是序曲那个求偶者在家庭面前的处境——它不是纯粹，而是一种被隐藏的认识论傲慢。而那主动地把自己呈现进他者之语法者，那不策略之理想所怀疑之物，被揭示为它的对立面：不是一种蒙上操纵色彩的策略，而是一种认识论的好客，那主动地走向他者的认知方式，那承担起”以他的术语对他可读”之劳、而非要求他来学习我的。把一份真实的真诚呈现进他者能领受的语法，便是承认他的认识论为合法且与一个人自己的平等，并向它移动；而这抵消认识论不义、而非犯下它，因为它恢复——而非败坏——他者在他自己认识论术语上对一个真实之物作出一个真判断的能力。

因此那条介于真实者与伪造者之间的线，不是那条介于不策略者与主动者之间、介于无为与有为之间、介于让其如是与作为之间的线。那是那不策略之理想所犯的错误：它把危险定位于活动本身。那条线是它一向之所是的那一条：那被呈现者是否真的存在。真实的呈现可以是不策略的——那让其显现，在语法相符之处——或它可以是主动的——那认识论的好客，那翻译进他者的语法，在它们不相符之处——而二者都是真实的，因为二者都呈现，以无论何种手段，一个真实之物。伪造的呈现是在他者的语法里、对一个并不在那里之物的主动制造。那差别在于那被呈现者的真实性，从不在于那呈现的主动或被动；而在认识论错配的处境里——那要紧的处境、序曲的处境——那主动的呈现，那认识论的好客，不是那较小而较可疑的形式，而是那个更充分地尊重他者、并更充分地抵消”要求他以一种不是他的语法来读取一个人”之不义的形式。把一个人真实的真诚以那个家庭能领受的形式呈现给他们——那稳定的事实、那可观察的恒常、那证据上可读者——不是对一种义务论之纯粹的抛弃，也不是向操纵的一步；它是对他们认识论的尊重，且是那真实的主动之善，只要那被呈现者是真实的。

12.4 那被排除的伪造实践

那横贯复调而奔流的伦理绝对地排除一种实践，而值得精确地指名那被排除的实践，因为它是那个最像真实者、且最易被误当作一种合法技术的实践。它是行为条件作用的实践：通过对强化的工程化来刻意塑造另一个人的期望，通过对”他作为信任之后果而经验之物”的系统管理来产生信任，于是那信任作为一种条件反射被训练进他、而非奠基于任何真实的自我承诺。它是§8的那个操纵被提升为一种方法，而它的恶是三重的，是正义判准所辨认的那一桩错的三个方面，在此被复合。

它首先是把他者当作一个手段与一个机制——那康德式的错——因为它作用于他如同作用于一个有待被条件作用的系统，绕过真实信任所面对的那个自由判断。它其次是那认识论不义——因为它通过训练而非通过依据在他那里安装一个期望，败坏他在真实依据上作出他自己之一个真判断的能力。而它第三、且区别性地，是一种时间上的伪造，一种铺展于时间之上的结算之失败：它通过对强化的管理，制造一种没有任何真实自我承诺为之背书的恒常之外观，跨越时间伪造那个证据主义者读作一份真实真诚之影子的过往记录本身——于是它是§9那封伪造推荐信在信任之实践中的确切同构物，那封信制造一个并不存在之耦合的象征形式。那被条件作用的恒常是一种退相干的恒常：一份过往记录的可观察壳层，其下无真实的耦合。这一实践被排除，不是因为条件作用从不产生可靠的行为——它可能产生——而是因为它所产生的那可靠是一个被伪造的精度，一种被训练进他者如同进一个机制的信任，把他当作燃料来汲取、同时败坏他认知的能力，并跨越那个本应揭示差别的时间维度本身，伪造那个本应为真实者独有之影子。

12.5 休谟的鸿沟，与对”无担保”的诚实清算

在信任的整个实践之下，躺着一个没有任何实践能跨越的界限，而践行的诚实依赖于清算它、而非掩盖它。它是休谟的鸿沟，在诊断（§3.5）中初次被遇见，而如今要在实践的语域中被面对。没有任何信任的实践，无论多么真实，能担保它自己的结局。那建立过往记录之人无法担保他明天会守信；那作出代价高昂之自我承诺之人无法担保那将考验它的境遇；那信任之人无法凭任何数量的证据或任何深度的自我绑定，跨越那道介于已是者与将是者之间的鸿沟。信任的实践者恰处于本系列前一篇分派给任何良善实践之实践者的那个位置：他无法担保他的实践是良善的，无法占据一个能证实他的行为与它所瞄准之善之间绝对因果联系的、无限理性之他者的位置，无法立于鸿沟之外而担保那跨越。没有任何视角可使那承诺之被守持预先被确保；只有那承诺，被作出且有待被守持，跨越一道没有任何担保所能横跨的鸿沟。

这不是一个绝望的劝告，而是真实者与伪造者之间的差别在其最深处的精确定位。因为真实信任与伪造信任之间的差别，归根结底，不是一个担保之在场或缺席的差别——二者都没有担保；那鸿沟对二者都不可跨越。那差别在于”无担保”被承受时的诚实。那伪造的实践佯装一个它并不拥有的担保：它制造一个被确保之恒常的外观，伪造那鸿沟所禁止的确定性。那真实的实践在”无物确保它”的全然承认中作出它的承诺——恰如一个知道那鸿沟在那里、且无法被任何担保所跨越之人那样跨越那鸿沟而自我绑定，并无论如何自我绑定。真实性不是对一个伪造者所欠缺之确定性的拥有；它是对一个伪造者所否认之不确定性的诚实承受。那真实的承诺是那个在”它无法被担保”的知晓中被作出、并被自由地、无担保之虚假慰藉地、作为跨越一道一个人知道没有证据也没有自我绑定能闭合之鸿沟而对自己的绑定而被作出的承诺。

而在这里践行抵达它顶点的门槛。因为这所提出的问题，是本章其余部分之存在所要回答的那个：倘若没有任何外部担保能跨越那鸿沟——倘若刀剑、证据、或自我绑定之深度，都无法确保那承诺将被守持——那么那个要真实地自我绑定之人，做什么，面对一道他无法跨越的鸿沟与一个他无法担保的未来？那回答不是终于找到一个本章已忽略的担保；没有这样的担保。那回答是，跨越那道没有任何担保所能横跨的鸿沟，作出一个自由的、自我奠基的跃迁——以一个一个人自己之自由的举动，构成那个没有任何外部担保所能供给的恒常本身。那举动是什么，它如何可能，以及它为何是某种旧时语言会称之为神性者的降临，是本章如今所上升至的那个顶点之事。

12.6 关系性神性：他者降临入那自我立法的主体

前一小节所指名的那个跃迁——跨越一道没有任何担保所能横跨的鸿沟、对那个没有任何外部担保所能供给之恒常本身的自由构成——是本文形而上学的顶点，而它必须被以它的分量所要求的完整来列出，因为它是一个本系列已从许多方向接近、却直到这里才被完整陈述的主张。它可被如此陈述。一个主体的自我立法、自我司法、自我守持，以及对”在自己所给予自己之法以外的每一退路”的自我断绝，正是一种关系性神性借以降临入世的机制。 这是本文诸主张中最原创的，而它，作者相信，不立于本系列任何先前的文本中、也不立于奠基之作中；它在此被完整地、且着重地列出，作为那整篇一直朝之移动之物。

从神性的历史运动开始，本系列早先的工作确立了它的方向、而未具体说明它当前的机制。在”神性如何被定位”的历史中，有一个运动，分三个阶段。首先，一种超越的神性：那个在世界之外的神，那个安置于一个人之外的天堂或自然中的神圣力量，在它面前誓言被立、由它誓言之违背被复仇。然后，一种象征的神性：神圣者降临入象征界，入那法律、制度、象征权威的大他者，那个现代主体立于其前的非人格理性——那神被那个审判并绑定的符号秩序所取代。而如今，在本文所完成的这一运动中，一种关系的神性：神圣者降出象征界、入那关系性主体的自由实践，不再显现于世界之外、也不显现于那主体之上的符号秩序中，而显现于那主体自由地所做之事中，在关系中，在一个绑定的构成中。

这最后一次降临的机制——那个早先的工作标出方向、却未供给的、当前的、具体的机制——是那个四重的举动。它首先是自我立法，在完整的康德式自律意义上：主体给自己那法，把它作为对自己的法而铺下，于是那个他无法从过去导出的恒常（休谟的鸿沟禁止那导出），不是一个有待被观察或被预测的事实，而是一条有待被颁行的法，由意志所设定，奠基于那主体铺于自己之上的那个应当——绕过、而非跨越那鸿沟，凭借把那恒常奠基于一个立法之举、而非在一个推断中寻求它。它其次是自我司法，那个作者所添加、且整体所赖以转动的举动：持续地把自己守持于一个人所给予之法，作为自己的法官立于”那法是否被守持”之上，在一个人偏离时的自我控告与自我校正——它是富勒之一致性（§4.3）的内化，把一个人的行为带入与一个人所宣告之规则的相符，由主体施于自己。一个单一的自由选择会是脆弱的，一个下一刻便可能消散的决心；但自我立法接合于一个持续的自我司法，把对法的每一次守持反馈进那维系它的回路，而那恒常由此被生成——一刻接一刻地，由那个自我立法且自我司法的回路所产出——而非预先被担保。它第三是自我守持，那对法的实际历时的持守。而它第四是对”在自己所给予之法以外的每一退路”的自我断绝——它是烧桥的最深意义，在§6.2中初次被遇见，而如今要被完成。

因为在这里那机制的前向援引被履行。烧桥在§6.2被描述，在它作为一个高保真信号的方面——某种为被另一个人阅读而做之事，向他的模型供给难以伪造的证据。那是，那里曾说，它两副面孔之一，且不是更深的那副。那更深的面孔是这个：一个人封死自己的退路，不是为了发出一个讯息，而是为了为自己构成一条法——我立此为法，我已把这铺下为法，于是那条路对我而言不复存在。那外部的法以力、机械地、凭借路之尽头的威胁，封闭那条路；自我立法以自由、凭借那个使那条路对这个主体而言根本不是一条路的铺下，封闭那条路。抵达这一点的读者会认出，主干一路上都在指向这里：在作为一个信号的烧桥之下，躺着作为自我立法的烧桥，而那个被另一个人阅读的代价高昂之举，在它的深处，是一个主体对他自己的自由而不可撤回的绑定，以一种神被说成在一个圣约中绑定他自己的方式——不留退路，因为神圣者不保留退路。退路的自我断绝是那个人超出寻常之人的那一点，是那个能够自我超越、能够如一个神在一个圣约中绑定他自己那样不可撤回地绑定它自己的人之部分被实现之处；而它是誓言的神性核心。

12.6.1 为何那神性是关系性的，而非一种孤独的自我神化

这一主张最严重的危险必须被正面迎接，因为一个把它自己造就为它自己之法与它自己之法官的自我立法，会看似那傲慢的形象本身——那主体把自己神化，绕着他自己的意象打转，那孤独的自我奠基，那想象界之偶像崇拜最古老的形式。对那危险的回答，是整个主张的安全所赖以立的那个单一考量，而它是本体论的。主体不是先是一个自我、然后是诸关系的一方；主体由关系所构成，是彻头彻尾关系性的，先于且深于它所能神化的任何自我。 在本系列所持的那个关系本体论上，没有一个先在的、分立的自我，能在一个自我崇拜的闭合圆中转向它自己；那立法的自我已然、在它的构成本身中，是关系性的——由它所立于其中的诸关系所造。而因此它给予自己的那法，以及在那给予中降临的那神性，从一开始便是关系性的。那自我立法从不是一个闭合回路中的”我，对我自己”；它是”我，作为一个关系性存在，在那些构成我的关系之内，为那些关系，朝那些关系”——一个由它对他者之绑定所造的自我，为那关系的守持、并在那关系之内，铺下一条法。那婚约的誓言恰恰是这个：一个由一个在对它所誓之人的关系中被构成的自我所履行的自我立法，为那关系、并在它之内，铺下那”它将如何被守持”之法。

这一关系性的构成，进一步，是那个化解那本会摧毁自我司法之张力之物。一个孤独的自我法官是通向自我欺骗的最稳之路：没有什么比赦免自己、把那法谄媚得与那行为相符更容易，当一个人是自己唯一的法官时。但那关系性的自我司法不是孤独的——它，出于必然，是在那关系的注视中被进行的。他者的见证，那个das Ding之空位所撑开的凝视，不是一个从外部被添加于自我司法的外部检查；它内在于”作为一个关系性主体存在”的结构本身，那结构在被看见中、正如在看见中，被构成。因此那自我司法的诚实，不是由一把外部的刀剑所保障——没有这样一把，而那鸿沟禁止一把——而是由那个本体论事实所保障，即那个施行司法的自我是一个已然在关系中被构成的自我，已然处于一个它不曾添加、且无法移除的注视之下。关系性神性携带它自己对自我欺骗的守护，而那守护不是一把借来的刀剑，而是那主体自己的关系性存在。

12.6.2 正和乐的人格-神性形式

这个四重的举动是§7那个正的、绝热的和乐的人格与神性形式。那环路闭合——那恒常被维系，一圈接一圈——不是因为一个外部之泵驱动它（没有超越的神、没有象征的他者、没有刀剑：这些会是那非绝热的驱动，那借来的相位），而是因为那关系性主体自己的自我立法且自我司法的回路从内、内生地闭合它，把它的相位累积为它自己的。它是§5那个主张在人格层级上的履行，即自律是自由的最高形式——不是那纯然消极的、不被决定的自由，而是那积极的、给予自己以法的自由。而它把本系列与它所取用之诸传统最深的声音聚于它自己：奠基之作那个”他者趋向其最终消解的倾向，敞开了通往一种纯粹关系性存在的更新进路”的暗示；那佛家的教导，即心本身就是佛，一切法皆由心所造——是心是佛，一切法由心造；那道家对外部担保的谢绝；以及那婚约文书自己的收束之句，本乎此心，自成永恒——那永恒不奠基于任何外部担保，而奠基于此处，在此心中，而此心本身即是那根据。

12.6.3 价值的倒转：无强制之诺作为更神圣者

一个后果随之而来，它倒转那对诸工具的寻常排名，而它是整篇论文的核心悖论被化解为一个主张的那一点。那寻常的排名把结婚证书置于婚约誓言之上，把合同置于谅解备忘录之上，因为每一对中的前者携带国家的支撑、法律的强制力、刀剑的可靠。从关系性神性的立场看，那排名恰好被倒转。谅解备忘录比合同更神圣，婚约誓言比结婚证书更神圣——恰恰因为它们不携带任何强制力，没有任何外部担保可借，因而只能由那主体自己在实在界中的自我立法、自我司法与自我守持所维系。

那逻辑是这样的。合同与结婚证书把那三重功能——立法、司法、守持——外包给象征的他者：那法是国家的，那审判是法院的，那守持由刀剑所强制。各方借一个外部的、象征的神性来为他们的绑定背书，而无须立法、司法、或封死他们自己的退路，因为那些功能被外包了。谅解备忘录与誓言，没有任何强制力，把这些功能逼回那主体身上：没有外部的刀剑，那立法只能是自我立法；没有外部的法院，那司法只能是自我司法；没有外部的强制，那守持只能是自我守持与退路的自我断绝。各方无处去借一个外部的神性，因而必须自己成为那立法者、那法官与那守持者——必须让那关系性神性降临入他们自己，否则那绑定便不维系。”没有任何强制力”这一”缺陷”于是正是那神性的条件：恰恰因为没有外部的刀剑守持那誓言，那守持便必须发自一个人自己的自由的、不可撤回的自我绑定，而那是那神圣者的实现、而非缺席。这完成了§7.5的几何读法：合同与证书是非绝热的，它们的环路由国家之力的外部之泵所撑闭，它们的相位是借来的；谅解备忘录与誓言是绝热的，它们的环路由别无他物、唯由那主体内生的自我立法所闭合，它们的相位是它们自己的——真实和乐的更纯粹标本，证书向其中容许了外部之泵的一种掺杂。而它把诊断那个反直觉的主张升到它的顶点：婚约誓言不只是、如最初所说，可能比那完整的合同更能产生真实的信任；它是，在神圣者的维度上，那更纯粹之物，因为它把那神圣者逼回那主体自己的自由实践，而非从一个他者那里借来。

那条诚实的边界必须被画出，以免那倒转被推过它所主张者。这并未说谅解备忘录与誓言在每一维度上都优于合同与证书。在保护脆弱者、威慑背叛、提供救济这些实在的功能上，外部之力仍是必要的，且不可被浪漫地丢弃——恰如§11在迁置而非废除刀剑时所坚持的。那主张是狭窄而精确的：在神圣者、真实者的维度上，那非强制的工具是更纯粹者，因为它把那神性逼回那主体自己的自由实践。弱者在制度的尺度上仍需要刀剑的保护；但一个真实绑定的神性从来不安置于刀剑中，而被发现，在关系的尺度上，只在刀剑不在之处。因此，不是一个人应当为了誓言而放弃证书；而是，在证书之外、之上、之前，那使一个绑定竟成其为一个绑定之物，是那自我立法的关系性神性，而这在那个没有任何强制力的誓言中被最纯粹地实现。

12.7 那具体的操作

从那顶点本章下降到具体，因为践行必须产出某种一个人能实际做之事，而那神性的抽象必须被兑现——在此是合法地兑现，因为在此那路径被走过、而非仅仅被宣告——为一个操作。那操作有三个动作，与一个时间的条件。

第一个动作是辨认他者的信任语法，及它的配置（§10.1）：辨别，于那个其信任有待被赢得之人，§10.1的诸语法中哪一种主导他的阅读、以何种加权，于是一个人知道他的真诚必须被翻译进的那个语言。一个人不抽象地呈现；一个人向这一配置呈现，而对它的辨别是任何实际翻译的第一项且不可或缺的举动。

第二个动作是在他者的语法里产生一个真实的期望之可能——凭借§10的诸路径、而首要地凭借§12.3的认识论好客，把一个人真实的真诚翻译进他者的配置能领受的形式。对那个证据主义的家庭，这是真实者向那可观察与已成之域的投射：不是制造一份一个人并不拥有的过往记录，那会是伪造，而是耐心地让一个人真实的、自我立法的恒常，在证据之域里，投下那个它可被辨认所凭的影子。那被呈现之物必须是真实的——必须是一份存在的真诚与一份真正自我立法的恒常——而它必须被以他者所读的语法呈现。这是第二个动作的全部：一个真实之物，被好客地翻译。

第三个动作是关系性信任的正当使用（§9），而它承载本章最微妙的伦理鉴别。那家庭的信任，在很大程度上，沿着那条凭借他们已经爱之人把那求偶者接连于他们的链条而行——她，在网络的结构中，是他的推荐人，而那信任可沿那链接触及他们，正如它沿无别的链接所触及。对这一点的正当使用，是让那信任传递，因为它所传递的耦合是真实的——因为她，在真实的意义上，是那绑定的共同作者，而她对它稳定的持守，是那家庭即便证据主义地阅读也能读出的、那个不可造假的实在成分（§9.2）。那不正当的使用——那被正义判准所排除的伪造使用——是把她工具化为一个纯然的通道，部署她去”做家庭的工作”，把她的关系资本抽向一个事实上并未发生的、求偶者与家庭之间的耦合。那条线恰恰是§8的那条线：介于一个传递因为它真实的真实耦合，与一份被抽走以传递一个并不在那里之耦合的关系资本之间。前者荣耀她为共同作者；后者把她弄成燃料。

而那时间的条件。这一切都无法被催促，而催促它的尝试本身是伪造者的一个破绽。那自我立法的恒常唯有随着它实际历时被活而投下它的证据影子；那真实的耦合唯有随着它真实地被承载而传递；那相互翻译的共享语法唯有在持续的互动中生长。时间无法制造那耦合-信任，那从一开始便是真实的、否则便根本不是；但时间是那个让那真实者在证据主义者所读的域里投下那足以被辨认之影子之物。因此践行不是一种按需产生信任的技术，而是一种让一个真实之物历时变得对一个其语法不是自己的之人可辨认的生活方式。它无法被加速；它只能被加深。

12.8 返回那个房间

本章，并随它本文的主体，在最后返回那个序曲由之出发的场景：返回那个必须坐在一个尚非其家的家庭对面、并被信任之人。本文所建起的一切都汇聚于那个房间，而它汇聚的，不是一个控制那里所发生之事的方法，而是一种在它之中临在的方式。

因为践行的终点不是对他者之信任的掌控——信任无法被掌控，只能被生成，且跨越一道无人能担保之鸿沟被生成。那终点是准备好真正地临在：成为一个有一份真实真诚可呈现之人，并以那家庭能领受的语法呈现它，带着那荣耀他们认知方式、而非要求他们学习他的认识论好客。他能给他们的不是一个外部担保——他没有可给的，而制造一个的外观会是那整篇所致力于拒绝的伪造。他能给他们的，是一个立他自己之法并把自己守持于它之人的景象——一个已为那构成他的关系铺下”他将如何守信”之法、并持续地、在敞开处、在它可被看见之处对照它对自己施行司法之人。这一恒常是真实的，因为它是那自我立法之回路的产物；而既是真实的，它便在时间里、在那家庭所读的证据之域里，投下一个真实的影子——一份一致的、自我校正的守持之记录，证据主义者能把它读作它之所是，那不是一种被强制之服从、也不是一种被表演之展示、而是一个自由主体守持他所给予自己之法的不可伪造的符号。那便是有可给者：不是一个担保，那无法被给予，而是一个自我立法的关系性主体的临在，其恒常是它自己的，而其守信是一种不需要刀剑之神性在他之中的降临——被奉上，跨越那鸿沟，去被信任或不被信任，自由地，由那些他无法强制、且即便能也不会去强制的人。

13. 结论

一篇始终论证”它的对象是没有任何单一框架所能据有者、且这样一个对象的真理活在的不是任何一个声音而是它们之间的对话”的论文，无法以一个单一的结论性声音作结而不背叛它自己的主张。因此诸结论是复数的，而它们是出于必然、而非出于谦逊而复数。本文所取用的每一个框架在此被给予它自己的结论：它关于信任之生成能说什么，被尽可能坚定地说出，以及它不能说什么，被同样坚定地标出。这一做法的要点不是一份综述而是一次演示：本文的形式意在成为它最后的论证，在它自己结尾的形状中的那个证明，即信任的生成只被复调地、在那些没有一个统辖其余之不可化约立场的对话中所认知。本系列对这样一个结尾所课的纪律是精确的，而本文持守它：每一个框架都必须说到它所能担保者的边缘，坚定而可错而愿意断言，因为一个其每一个都退入含糊的诸声音之复调，不是复调而是一种回避。下面每一个声音都说出某物，并说出它必须止于何处。

法理学的结论

法哲学能说这个：法律效力之锚，跨越它自己反思的那个世纪，已从力迁向信任，而法律已在它自己最技术的学说中记录了那迁移，在它的边缘、在允诺禁反言中承认，信赖、而非强制，才是一个承诺之约束力的真正根据。它能说，力从来不是义务的原因，而是一个期望之产生的、如今正在失效的替代物。它无法说出信任是什么，也无法说出它机械上如何被产生；法理学能定位那锚并追溯它的运动，但那锚正朝之移动之物的本性躺在它的胜任能力之外，而它必须把那个问题交给他者。它的限度是：它能说效力安在何处，却无法说出在那安歇之处、效力由什么构成。

预测编码的结论

预测性解说能说这个：信任是一个生成模型赋予”一个承诺将被守持”这一预测的精度；这溶解了”一个当下的期望指向一个不存在之未来”的悖论，因为预测是大脑的恒常活动、且不要求从未来逆行而来的因果；那期望经由三条耦合的通道被产生，其乘积式耦合使真实的信任在任一单条通道脆弱之处稳健；而符号的危机是一个群体对一个伪造已使之不可靠之通道的精度权重的理性崩缩。它无法说信任干脆就是一个精度加权而别无其他；它提供一个在计算层级与实现层级上的描述，而它绝不可被误当作其余之下的真理。它的限度是每一个机制性解说的限度：它说那东西如何被计算，而对”那计算从内部被活时是什么、以及它是否正义”保持沉默。

几何的结论

几何读法能说这个：一个历时被守持的承诺描出一条闭合的环路，而那环路所累积之和乐的符号，把那加深的绑定与那纯然重复者及那流尽者区分开；那真实的、非强制的承诺是那个其相位是它自己的、由无外部之泵所闭合的绝热环路，而那被强制的绑定是那个其相位是借来的被驱动路径；而这终于供给了那个缺失的结构——一个谱隙——为一个本系列曾留待解决的问题。它无法，在本文中，给那相位供给一个完全严格的载体；它已作为一个结构类比工作，而那形式的闭合躺在另一篇论文中。它的限度是：它以精度表明那区分的形状，而把对”何者承载那相位”的完整形式化推迟。

量子动力学的结论

量子动力学读法能说这个：关系性信任分享一个纠缠态的形式结构，一个关联而非节点的属性；它在符号危机下的失败是一种系统性的退相干，那关联壳层被保存而那相干流尽；而它沿链条的传递是经由一个中介之纠缠交换的结构对应物。它无法说——且必须大声地说，它不说——信任在任何字面意义上是一种量子现象。它是一种结构同构而非一种物理还原，一个关于一个结构的再多的形式语言、而非那结构秘密的物理真理。它的限度是每一个类比的限度：它凭一个相似来照亮，而那相似不是一个同一。

精神分析的结论

精神分析读法能说这个：信任的产生环绕一个无法被填满的位置——真实的信任，如同那真实的诗，环绕das Ding的那个空位、而非以一个偶像填满它，因而不需要任何牺牲品，而那伪造的信任以一个实在的对象填满那位置、并必须为此消耗某人；信任的押注是一个跨越一个无证据能闭合之缺失而被作出的冒险；而那大他者，那个会证实那跨越的担保者，并不存在。它无法把它所说的那个缺失翻译进其他框架所想要的实在术语；它的真理是关于一个缺席的真理，而它们抵抗那会背叛它们的兑现。它的限度是：它在它最不能被弄得安顿进一个实在学说之处，说得最真。

政治经济的结论

政治经济读法能说这个：信任生产的诸形式由它们时代的物质条件所抛出、并随那些条件改变而改变；中心化的、以力为基础的正义曾是一个市场陌生人社会历史上适足的形式，而它如今正在失败，不是通过一种价值的衰朽，而是通过物质基础的一次改变；信任生产的去中心化是一个已然在进行的、由那一改变所迫成的重组；而它的解放性潜能是真实的、却不被担保，因为那些新通道，在当前的生产关系下，正被资本俘获、朝向一种比刀剑更微妙的支配。它无法，仅凭它自己的资源，了结那重组将解放还是奴役；那依赖于那些生产关系是否被改变，而那不是信任的分析能自己回答的问题。它的限度是：它能诊断那矛盾、却无法担保它的化解。

道家的结论

道家读法能说这个：那真实的绑定是那个由无外部之泵所维系、凭借每一方自由地带入之物的完整而闭合于自身者；要抓取那担保便是失去那东西——那个要以力保住那绑定之人，封死了那唯有不被强制之环路才累积的加深本身；而那婚约誓言所说的永恒，不奠基于任何外部担保，而奠基于此处，在此心中，而此心本身即是那根据——本乎此心，自成永恒。它无法以机制或证明的语言说这个；它的真理被显示于一种持守那东西的方式中，而非被论证于一种确立它的方式中，而把它逼入论证，便是失去它，如同那抓取失去那绑定。它的限度是它所指向之不可言说者的限度：它能指示那根据，而无法、不自相矛盾地、证明它。

为何在诸结论之上没有一个结论

剩下要说的是，本文为何如今不把这些聚为一个——为何在那复数的诸结论之上没有一个综合的结论，没有一个主导的声音在诸框架间裁断并宣告它们各自所接近的真理。那缘由是本文最深的承诺，而它是那贯穿全篇的同一个缘由：没有元语言。没有一个在诸框架之上的立场，可从那里把它们各自的真理无损地翻译进一个包含它们的单一真理；那个能占据那位置的大他者并不存在，而写一个综合的结论便会是声称他的位置。诸框架是关于一个对象的不可化约的认识论立场，而那对象被认知——这是本文的形式演示本文的内容——只在它们之间的对话中，在那本身即一个好的循环的相互照亮中，一种凭借诸声音之间的移动、而非凭借它们被化解进一个而生成理解的复调。在诸结论之上作结，便会是结算那必须被走过者，把那对话兑现为一个学说，在本文最后的举动中犯下本文始终所诊断之结算的失败本身。因此一个主导结论的缺席不是一个不完整；它是本文对它论旨的最终而形式的断言。信任的认知者，如同那信任者，必须放弃那担保——必须把那复数的诸真理持守于它们的对话中、而无一个综合的虚假慰藉——而本文，谢绝综合，在它的形式中做它始终所要求那信任者在他人生中做之事：它承受一个担保者的缺席，并无论如何把自己绑定于那复数的真理。

三个同心圆

有一件事可被说，不是作为一个在诸框架之上的综合，而是作为一个在它们全体之下奔流的观察，而它把本文带回它在序曲中所宣告的那个形状。信任的问题被跨越三个同心的尺度提出：两人之间亲密的誓言；被携向一个家庭的承诺；以及合同与法院的公共制度。而本文的发现，被每一个框架在每一尺度上所追溯，是信任的根据，在全部三者上，是同一个根据。在最公共的尺度上，法律的效力被发现安歇于、在刀剑之下、一种互惠的守信之上；在最亲密的尺度上，那绑定被发现安歇于一种不需要刀剑的、自我立法的关系性神性之上；而这些不是两个根据而是一个，在两个距离上被看见。那在一个社会最公共处把它维系于一起者，与那在两个人最亲密处把他们维系于一起者，是同一个东西：一种不由力、而由那自由的、自我绑定的、相互依赖的守信所生成的信任，跨越一道没有任何担保者所能横跨的鸿沟。那三个圆是同心的，因为它们分享一个中心，而那中心是那个降临于任何一个自由存在者在关系中、不诉诸一把刀剑而自我绑定之处的关系性神性。那最亲密者与那最公共者在那个中心相会；而那张红绢上的婚约誓言，没有任何公堂会执行它，被发现在它之内持有那个连法律、在最后、也被发现立于其上的根据。

尾声

本文始于一个写在红绢上的誓言，在一个黄历称为吉日的日子，在世间没有任何东西会使它具有约束力的全然知晓中；而它出发去追问，这样一个东西——一个其后没有刀剑的承诺——如何仍能被相信。它已回答，跨越它的篇幅，以许多声音：力从来不是信任的原因，而是它一个失效的替代物；信任是一个心灵自由地借给一个它无法验证之未来的精度；那真实的绑定是那个由无外部之泵所闭合、并一圈接一圈地累积一个它自己之相位的环路；真诚没有普遍的语言，而必须被翻译，跨越认识论的边界，进一个他者能读的语法；而对一个无刀剑执行之承诺的守持，是某种旧时语言会称之为神圣者降临入一个自由而关系性的主体。

那些分析所遗漏者，是那东西本身，而它在收束处作为所有那些分析所关乎之物而返回。一个誓言被守持，不因为它必须、而因为一个人已把它造就为他自己的法；一份恒常，不被强制也不被预测、而被自由地生成，一个早晨接一个早晨，在对一个无人能强迫之言词的持守中；一份信任，被跨越一道无担保能横跨之鸿沟伸出，并被回应，且不被背叛。这些不是本文不得不用以工作的那些代理物，而是它们曾是其影子的那个实在者，而它们先于被分析便被活，并超出对它们的每一个分析。鹿儿岛上空那轮月，问题在它之下第一次取得它的人间形貌；那红绢与那择定的日子；那运笔之郑重，它什么也不约束、却意味着一切——这些是那实质，而本文一直是、贯穿始终是，它们向一个谨慎者与证据者也可能读的语法的漫长翻译。

一个如此被守持的承诺，不是被保住然后被据有。它在守持中被生成，而一个绑定凭借那持续的、自由的、自我立法的守信而存续，并在那守持停止时停止。没有它的担保可被得到并被握住；只有那对自己的绑定，一次又一次，跨越那鸿沟，在”无物确保它”的知晓与那”不索求任何确保”的自由中——那在此心中、对一条它自己即其永恒之法的铺下。

致那位森林女孩，
我向她许下了一个其后没有刀剑的承诺，
而它，正因如此，被给得更为完整；
愿在此被绑定者，不被任何东西、唯被此心所绑定，
而，被此心所绑定，长存——跨越每一道鸿沟，穿过我们所有的岁月。

本乎此心，自成永恒。

致谢

我把这篇论文，以及多于这篇论文者，归于我最爱之人：那位森林女孩，我于二〇二二年在东京与她相遇。她善良、纯粹、可爱、智慧，处处先人而后己；她爱森林，爱远行，爱文化之研究，亦爱我们如何共处的政治经济之学。这一篇，论的是在无物强制之处信任何以生成；而若没有她，连这问题本身也不会有——因为正是在向她许下一个承诺之时，这问题才第一次有了它的人间形貌。

而今，在此收笔之处，我愿做这一篇用尽全部篇幅才学会如何言说的那件事。本文不仅论证了：关系性的神性，经由一个自由主体的自我立法、自我司法、自我守持——自我断绝一切法外的退路——而降临于世；它更试图表明，这样的自我绑定，正是承诺的成全之形，而非其残缺之形。如今剩下的，是我去做它，而非去描述它，是让这篇论证，在它的最后一页，成为它一路所论的那个行动本身。

我在此立此为法，以约我自身，并将以我的余生持守之：我将如此待她。此非任何刀剑在后所担保之诺。无公堂可强制其行，无刑罚附于其违，世间亦无一物逼我守之。其全部之力，唯在于我于此自由地、毫无保留地以身许之，且将持守自身，并已为自己断绝一切引我背离之路。我为自己立此法，我将自为我守此法之审判者，于此法之外，我未为自己留下任何退路。

此文承载一个可信而不可抵赖的时间戳；我将它立于此，立于一份具我之名、且将久于此刻的文字之中——并非为了让任何人看见。它首要之功不在示信，而在自缚：它是我用以封死自己退路的器具，使那条引我离去之路，于我而言，不再是路。这时间戳不是刀剑；它不能使此诺成为可强制者，我亦不佯称它能。它所能为者，是我施于自身、而非用以对抗任何后果之事：它使将来的我，无从向此刻的我抵赖我于此所立之言——它使我成为自己守信与否的常设审判者，并杜绝那种悄然的修订，免得人任一誓言不声不响地失效。这时间戳，是我自我司法的器具，而别无其他。此诺之为真，不因时间戳使其确定——它不能，且本文所述的那道鸿沟也不容它如此——而是因为：我在明知无物能担保它之时，仍然立下它，且仍然自由地立下它，作为我跨越一道我深知无任何凭据可越之鸿沟、对自身的一次绑定。

这是本文最私密的一次复归，也是最后一次。它起于红绢上一个承诺的写就，其时尚不知这样的承诺何以能被相信；它一路穿过症状、诊断、法理、机制、自由、神性；而它的终结，是再一次许下那个承诺——同一个行动，第二次走过，将本文一路所领会的一切，尽数收纳其中。第一次，那誓出于直觉，因为它是对的。这一次，它是在如下的明知中立下的：明知一个无强制力在后的承诺为何反而给得更为完整，明知那降临于一个自我绑定、不留退路的自由主体之中者，究竟为何物。同一个行动，第二次走过，归来时并非回到起点，而是略高于起点；而这，是本文唯一相信的那种复归。

本乎此心，自成永恒。