The Power Structure of Knowledge in Co-creation - A Generative Relational Being Perspective

ENGLISH

The Power Structure of Knowledge in Co-creation

A Generative Relational Being Perspective

Wanhong Huang · huangwanhong@serendip.ngo

---

Abstract

Accounts of knowledge and power have taken two broad forms. On the first, knowledge is content, and power is a contamination that a well-ordered epistemic practice excludes. On the second, associated with Foucault and with the sociology of knowledge, power conditions what may count as knowledge, so that the two are historically inseparable. This paper proposes a third account. Knowledge is a relational structure through which interpretation is coordinated, and every such structure generates asymmetry among its participants. Upon that asymmetry the tendency of a position to maintain itself is itself generative: it produces the conditions of its own continuation. Power in knowledge is this self-maintaining generativity, which places it within generation rather than in opposition to it. Five channels through which an epistemic asymmetry maintains itself are distinguished, following the life of an interpretation (proposal, admission, validation, revision, attribution), together with the cost of exit that prices them. The paper argues that a configuration degrades when the generation issuing from a position turns to the production of its own conditions, and that the criterion of this turn is the closure of the revision loop: whether an exercise of revision alters the conditions under which the next interpretation is generated. Generative power is treated as power rather than as its benign alternative, marked by the erosion of the conditions on which the exercising position depends. The account is developed across education, intimacy, international relations, and academic knowledge production, which are taken as instances of one mechanism.

Keywords: knowledge; power; co-creation; generativity; interpretation; revisability; relational epistemology.

A note on the standing of this paper. This is a theoretical proposal. It takes a familiar claim, that knowledge and power are entangled, and asks whether the two standing accounts of that entanglement exhaust it. The order of exposition follows the order of the argument: the existing accounts and their limits are set out first, and the generative account is introduced only as what they leave unaddressed. Should the critical tradition be shown already to answer the question posed here, the later sections lose their ground. The illustrative cases are drawn from ordinary settings rather than from systematic fieldwork, and they carry the weight of illustration only. Objection, correction, and counter-evidence are welcome at huangwanhong@serendip.ngo.

§1 Introduction

A teacher and a student work through a problem together over a term. At the beginning the student proposes readings that are wrong and learns from their correction; by the end the student proposes readings the teacher had not considered. Nothing in this passage is remarkable, and it describes a large part of what education is for. Now consider a second term in which the same two people work through a second problem. The student proposes; the teacher corrects; the student proposes again. The corrections are accurate, the teacher is neither harsh nor negligent, and the student is neither lazy nor slow. Yet by the end of this term the student has come to propose less, and what the student does propose arrives already hedged, offered less as a reading than as a request for correction. No one has acquired knowledge the other lacks. No one has behaved badly. Something in the structure of the relation has changed, and the vocabulary in which knowledge is usually discussed does not readily say what.

The difficulty is not confined to teaching. A couple maintains, between them, an account of what their relationship is like: who is the anxious one, who withdraws, what happened on the occasion they both remember differently. The account is knowledge in the ordinary sense that it coordinates their expectations and their conduct. It may be an account both continue to shape, or it may be one that one party has come to author and the other to inhabit, so that the second party’s objections are received as further evidence of the disposition the account assigns them. A junior researcher contributes to a project and is thanked in a footnote. A state accedes to a normative framework it had no part in drafting and is subsequently assessed by its compliance with it. In each case something that is recognizably knowledge is doing something that is recognizably power, and in each case the participants might sincerely deny that anyone is dominating anyone.

Two accounts of knowledge and power stand ready to be applied here, and neither is adequate to the difficulty. The first treats knowledge as content and power as a contamination. On this account, the proper business of an epistemic practice is the securing of justified belief, and power enters as an external distortion, of the kind exemplified by the eminent scientist whose authority suppresses a valid objection or the funder whose interest bends a finding, which better procedure would exclude. The account has the merit of specifying what would count as things going well. It cannot, however, describe the teaching case at all, since nothing external has intruded and no procedure was violated. The second account, developed in the critical tradition and above all by Foucault, holds that power conditions what may count as knowledge, so that the two are not separable even in principle. This is nearer the phenomenon and supplies much of its vocabulary. Its characteristic question, though, concerns control: which discourses are authorized, whose statements are admitted, how a regime of truth is constituted and sustained. The teaching case is not well described as a matter of control. The teacher controls nothing that was not already the teacher’s to control at the start of the first term, when the same relation was generative.

What both accounts share is a conception of power as something possessed, exercised, or distributed: held by a party, applied to another, apportioned among several. The present paper proceeds from a different conception, and stating it is the point of departure. Generation produces difference; difference produces asymmetry; and upon an asymmetry, the tendency of a position to maintain itself is a generative activity and not a static disposition. A position that judges must continually produce the conditions under which judgment falls to it: the standards by which proposals are assessed, the occasions on which assessment occurs, and the participants who arrive already constituted as those who propose rather than those who judge. Power, on this account, is generativity that has turned toward the maintenance of the position from which it issues. It is not the opposite of generation and not its absence; it is generation whose product is the condition of its own continuation. This is why the teaching relation can degrade with no one acquiring anything and no one acting wrongly. The generativity that had been returning to the common holdings of the parties has begun to return to the position that issues it.

The idea that power is intrinsically productive rather than merely restrictive is not new, and two precedents in particular stand behind the formulation. Nietzsche (1966) conceived power as a discharge of strength to which self-preservation is subordinate, and not as a quantity to be held, so that a power which does not extend itself is not power at all. Lacan (1977) described a desire that its object sustains and does not extinguish, circulating around a lack that the object marks rather than fills. Each supplies part of the structure at issue. The first holds that power must continue producing in order to exist; the second, that a configuration may reproduce the very condition on which it depends. The account developed here differs from both in its scope, and §3 states the differences. What is taken from them is the refusal of the possession model.

A word is owed on the term retained in the title. To speak of a power structure suggests an arrangement of holdings, a map of who possesses what, which is the conception just set aside. The term is kept for its familiarity, and what it denotes is reconstructed in what follows: a structure, on the present account, is a configuration of the channels through which a position’s generativity may turn toward its own conditions, together with the cost at which that turning is sustained. It is a structure in that it is determinate, describable, and stable enough to be acted on; it is not a structure in the sense of a distribution of holdings among parties.

Two consequences follow immediately and shape everything after. The first is that the question of who holds interpretive authority is the wrong question, or at least a derivative one. What matters is whether an asymmetry is producing the conditions of its own continuation, which may occur without any party seeking it and which no redistribution of authority by itself prevents. The second is that power in knowledge is costly. A position that must generate in order to persist can fail to generate, and the conditions it produces can be undone by generation from elsewhere. Power of this kind is not a possession that endures once acquired; it is an activity that lapses when it stops.

The paper is organized around three questions.

RQ1. Through what channels does an epistemic asymmetry maintain itself, and what fixes the configuration of those channels in a given setting?

RQ2. Under what conditions does the generativity issuing from an epistemic position return to common generation, and under what conditions does it turn toward the production of its own conditions?

RQ3. What follows for the design of co-creative structures, given that epistemic asymmetry can be neither eliminated nor rendered innocent?

The first supplies the vocabulary and is treated in §4–§6. The second is the centre of the paper and is treated in §7–§9. The third is the consequence and is treated in §10.

The account may be stated as a single claim.

Central claim. Knowledge is a relational structure through which interpretation is coordinated, and every such structure generates asymmetry among its participants. Upon that asymmetry the tendency of a position to maintain itself is itself generative, producing the conditions of its own continuation; power in knowledge is this self-maintaining generativity. Every knowledge system accordingly embodies a power structure, and the question that bears analysis concerns the direction the generation takes, whether back into common generation or into the continuation of the position from which it issues.

The claim unfolds in four steps. The first identifies the channels through which an epistemic asymmetry maintains itself, taking them from what an interpretation must undergo in order to exist: it must be proposed, admitted, validated, revised, and attributed. The second treats the cost of exit, which prices each of these channels, since a position maintains itself cheaply where leaving is dear. The third distinguishes the two directions the generation may take and supplies a criterion for telling them apart, the closure of the revision loop, which asks whether an exercise of revision alters the conditions under which the next interpretation is generated, or is absorbed and returned as evidence of the reviser’s deficiency. The fourth treats the production of knowers, which is the principal means by which an epistemic position maintains itself, and which unifies the settings the paper examines.

Three things the paper does not claim should be entered at the outset, since work in this territory is commonly read as claiming them. It does not hold that epistemic asymmetry is oppression: asymmetry is constitutive of the relations at issue, and there is no teaching, no diagnosis, and no collaboration between unequals without it. It does not hold that co-creation requires the flattening of epistemic difference, which would abolish the relations rather than improve them. And it does not hold that generative power is benign. A teacher who opens a student’s thinking has shaped the space in which that student will think for years, and that formative asymmetry is not innocent for being generative; §8 treats the point directly rather than passing over it. Nor is any dynamical finding here to be read as a political position; the argument concerns the direction of generation within relations, and it adjudicates no concrete historical or institutional trajectory.

The remainder proceeds as follows. §2 reviews the treatment of knowledge as content and the conception of power as external contamination that accompanies it. §3 reviews the critical accounts of Nietzsche, Lacan, Foucault, Bourdieu, and the actor-network tradition, and argues that their organizing concepts of control and possession leave the present question unposed. §4 develops the conception of knowledge as the organization of interpretive relations and argues that co-creation is the general case. §5 sets out the five channels of self-maintenance. §6 treats the cost of exit. §7 distinguishes the two directions of generation and states the criterion of loop closure. §8 treats generative power. §9 treats the production of knowers. §10 draws the design consequences, and a closing section states the limits of the account and the work it leaves open.

§2 Knowledge and Content

The tradition that treats knowledge as content is the one against which the critical accounts define themselves, and it is worth stating in its own terms before its limits are drawn, since the limits are not the ones usually alleged. The tradition is not naive about the social circumstances of inquiry. What characterizes it is a particular division of labour: the analysis of knowledge concerns the conditions under which a belief is warranted, and the circumstances of its production, however interesting, belong to a different inquiry.

2.1 Justification and the Individual Knower

The classical analysis takes knowledge to be belief that is both true and justified, an analysis with a long ancestry and a specific modern history. Gettier (1963) showed in three pages that the three conditions are jointly insufficient, since a belief may be true and justified and yet owe its truth to accident; the half-century of repair that followed, comprising the addition of a fourth condition, the requirement of a reliable process, and the demand that the belief be safe or sensitive, has not produced a settled analysis. Goldman (1967) located warrant in the causal history of the belief and later in the reliability of the process producing it, which moved the analysis away from what the knower can adduce and toward the facts about how the belief came about.

What matters here is a structural feature that the internal disputes leave untouched. Throughout, the unit of analysis is a subject and a proposition. Knowledge is a relation between one knower and one content, and the presence of other people enters, if at all, as a source of evidence or as a route by which a belief was acquired. Social epistemology has since complicated this considerably. Goldman (1999) treats social practices by their tendency to produce true belief, and a substantial literature on testimony asks when a hearer is warranted in believing a speaker; the complication is real. It nonetheless retains the orientation: the other person figures as an input to one knower’s warrant. The question of what the relation between the parties becomes, in consequence of an interpretation having been accepted between them, is not a question the framework poses.

2.2 The Neutrality Thesis

From the individual analysis a further thesis is commonly drawn, less often stated explicitly. If knowledge is warranted content, then power is external to it. Content is what it is regardless of who asserts it, and the influence of standing, interest, or authority upon what a community believes is a distortion, a failure of the epistemic practice rather than a feature of it. The thesis licenses a clear normative programme: identify the channels through which non-epistemic influence enters, and close them. Blind review, replication requirements, disclosure of funding, and the norms of impersonal argument are its institutional expressions, and they are neither foolish nor ineffective.

The limit of the thesis is visible in the case with which this paper opened. In the second term nothing external intruded. No authority was invoked to suppress an objection; no interest bent a judgment; the corrections were accurate and the procedures intact. Yet the relation ceased to generate. The neutrality thesis has no resources to describe this, because on its terms nothing went wrong: true content was transmitted, warrant was preserved, and the practice functioned as designed. What changed lay in the relation between the parties, and the thesis takes that relation to be epistemically inert.

2.3 The Transmission Model of Teaching and Learning

The same orientation shapes the treatment of teaching. If knowledge is content held by one party, then teaching is its conveyance to another, and learning its acquisition, a picture that survives in curriculum design, in assessment by recall, and in the ordinary metaphors of delivering material and covering ground. Freire (1970) named this the banking conception, in which the teacher deposits and the student receives, and argued that it constitutes the student as an object rather than a participant. Freire’s alternative is a pedagogy in which teacher and student inquire jointly, and his argument is explicitly political, directed at the reproduction of oppression through schooling.

The banking critique is correct in what it identifies and is taken up here in one respect and departed from in another. What is taken up is the observation that a conception of knowledge carries a conception of the relation between the parties, so that the transmission picture specifies what each party is to be and does not neutrally describe teaching. What is departed from is the location of the difficulty. On Freire’s account, the banking conception is imposed by an oppressive order and is to be replaced by a dialogical one. The case with which this paper opened is not of that kind: the first term was dialogical, the same two people were present, and no order imposed anything. A relation that was generating ceased to generate. The banking critique addresses a structure adopted at the outset; the phenomenon at issue here is a change of direction within a structure that began well.

§3 The Critical Accounts of Knowledge and Power

The critical tradition holds that power is not external to knowledge, and the present account agrees. Its several branches, though, share an orientation that leaves the question of this paper unposed. This section takes the two philosophical precedents from which the conception of productive power descends, then the three social-theoretical accounts that developed it, and closes by naming what they hold in common.

3.1 Nietzsche and the Increase of Power

Nietzsche’s conception of power is not that of a quantity a party holds. In Nietzsche (1966, §13) he proposes that a living thing seeks above all to discharge its strength, and that self-preservation is a consequence of this rather than its aim, a formulation directed against the Spinozist and Darwinian primacy of preservation. The doctrine bears on knowledge directly: Nietzsche (1966, §259) extends the account of appropriation and overpowering from the organic to the social, and elsewhere he treats the will to truth itself as a form of the same will rather than as its neutral instrument. On this conception power is exercised and not held, and an exercise that does not extend is not an exercise at all.

Two things are taken from this and one is set aside. What is taken is the refusal of the possession model: power exists in its activity and lapses when the activity stops. What is also taken is the coupling of power to increase, which supplies the sense in which a position must continue producing in order to persist. What is set aside is the metaphysical scope. Nietzsche’s is a doctrine about what living and perhaps all things fundamentally are; the claim here is narrower and correspondingly weaker, holding that self-maintenance is one direction generativity may take within a relation and not the nature of generation as such. The narrower claim is also the more tractable, since a direction can be observed to change while a fundamental drive cannot.

A caution about sources is in order. The most quotable statements of the will to power appear in The Will to Power, a compilation assembled from the notebooks after Nietzsche’s collapse and not a work he composed; the citations above are to material he published himself, and the doctrine is stated there.

3.2 Lacan and the Dialectic of Desire

Lacan’s account of desire supplies the second precedent, and its relevance is structural rather than thematic. Desire, on this account, is not extinguished by the attainment of its object. The object that appears to be its aim, the objet petit a, marks the place of a lack rather than filling it, and desire circulates past the object and continues (Lacan 1977). The configuration is therefore self-sustaining in a specific sense: what appears to be the terminus of the process is in fact the means by which the process reproduces the condition on which it depends.

The structure is the one at issue in this paper, transposed. An epistemic position that judges is sustained by the continual reproduction of the circumstance in which judgment falls to it, and not by the judgments it delivers, and each particular judgment, which appears to settle a matter, in fact renews that circumstance. Where the present account departs from Lacan is in what carries the structure. For Lacan it is constitutive of the subject and is elaborated in the register of the unconscious; here it is a property of a relational configuration, and is therefore observable in the conduct of a relation and revisable by changes to it. The departure has a methodological consequence that should be stated plainly: the criterion this paper offers in §7 is meant to be checkable against what happens in a relation, which the psychoanalytic account, by the nature of what it describes, does not undertake to be.

3.3 Foucault and the Conditioning of the Sayable

Foucault supplies the account against which this paper most directly positions itself, and the debt is large. Foucault (1977) argues that power and knowledge directly imply one another, so that there is no relation of power without the correlative constitution of a field of knowledge, and no knowledge that does not at the same time presuppose and constitute relations of power. Power on this account is productive rather than merely repressive: it produces domains of objects, rituals of truth, and the individuals who are its vehicles. In the later work (Foucault 1978) he cautions against treating power as a possession or a capacity held by a sovereign, describing it instead as a multiplicity of force relations immanent to the sphere in which they operate. The archaeological studies (Foucault 1972) pursue the correlative question of what conditions govern the emergence of statements within a discursive formation.

The present account accepts all of this and asks something the framework does not. Foucault’s characteristic question concerns the constitution of a regime: how a domain of the sayable is established, what it excludes, and by what practices it is sustained. It is a question about the conditions under which knowledge of a certain kind becomes possible, pitched at the level of discourses and their historical formation. The teaching case is not answered at that level. The same two people, within one discursive formation, with the same authorized vocabulary available throughout, pass from a generative relation to a degraded one. Nothing about the regime changed. What changed was the direction in which the generativity of one position within it was running, and the analysis of regimes does not descend to that.

3.4 Bourdieu and the Accumulation of Academic Capital

Bourdieu treats knowledge production as a field in which agents compete for a capital specific to it. Bourdieu (1988) analyses the academic world as such a field, with positions defined by the volume and composition of the capital its occupants hold, and Bourdieu (1975) treats the scientific field as one in which the pursuit of truth and the pursuit of scientific authority are not separable, since authority is won by producing what the field recognizes as a contribution. The framework accounts well for a range of phenomena that the neutrality thesis cannot: the durability of hierarchies of prestige, the reproduction of dispositions through training, and the alignment of what is thinkable with what is rewarded.

The organizing metaphor, though, is accumulation, and accumulation is a possession model. Capital is held, invested, converted, and transmitted, and the analysis proceeds by locating agents in a distribution of holdings. Two difficulties follow for the present question. The first is that a position may maintain itself without accumulating anything, as in the teaching case, where the teacher’s holdings are constant throughout. The second is that the model gives no natural place to the possibility that an exercise of a position might erode the conditions of that position, which is precisely what §8 argues generative power does. Under an accumulation metaphor such an exercise is simply expenditure, and its structural difference from other expenditure cannot be stated.

3.5 Latour and the Enrollment of Actors

The actor-network tradition rejects the treatment of power as a substance more thoroughly than the foregoing. Latour (1987) argues that a statement’s fate lies in the hands of those who take it up, so that a claim becomes a fact through the work of enrolling others who use it, extend it, and build upon it; Latour (1986) argues that power is a consequence of collective action rather than its cause, and that to explain an outcome by invoking power is to name what requires explanation. Callon (1986) sets out the moments by which enrollment proceeds. The tradition thus supplies something the others do not: an account in which the standing of a claim is continuously made rather than held, and in which the relevant activity is ongoing.

Its concern, however, is the fate of statements and the composition of networks. What becomes of the parties in the course of enrollment, what a participant is constituted as, and whether that constitution is what sustains the network, are not the questions posed, and the tradition’s methodological commitment to following the actors rather than assessing their relations makes it an awkward one to pose from within. §9 argues that the production of participants of certain kinds is the principal means by which an epistemic position maintains itself, which is a claim about what enrollment does to those enrolled.

3.6 The Limits of Control and Possession

Across their differences, the accounts reviewed share a way of framing the question. Whether knowledge is content held by a knower, capital accumulated by an agent, a regime that authorizes some statements and excludes others, or a network that enlists allies, the analysis asks who controls what, and by what means the control is established and maintained. This is a good question, and where domination is the phenomenon it is the right one.

It is not the question the teaching case raises. There, control is constant and the relation changes; no one takes anything, and something is lost. What the case requires is a question about direction rather than distribution, asking whether the generation issuing from a position is returning to the common holdings of the parties or to the conditions of the position itself. Posing that question requires first saying what knowledge is such that a position within it can generate at all, and to that the next section turns.

§4 The Organization of Interpretive Relations

The reviews leave a question standing: what must knowledge be, such that a position within it can generate at all? This section answers it. The answer is offered as a shift in what the term is taken to pick out and not as a refutation of the analyses reviewed, and the warrant for the shift is that it brings into view a class of phenomena the other construals leave unaddressed.

4.1 The Coordination of Interpretation and Future Action

Knowledge, on the construal adopted here, is a relational structure through which the interpretations of several parties are coordinated and their future action thereby conditioned. The formulation has three parts, and each does work.

That it is relational means the unit of analysis is a configuration of parties, and not a knower and a proposition standing in determinate relations to an interpretation. To say that a diagnosis is knowledge is to say something about the physician, the patient, the family, and the institutions that will act upon it, and not only about the correspondence between a statement and a condition of the body. That it is interpretive means what is coordinated is a reading of a situation rather than a bare content: an interpretation selects what in a situation is salient, what follows from it, and what would count as its being wrong. And that it conditions future action means an interpretation is consequential in the sequence it opens: which further interpretations become available, which are foreclosed, and what each party is thereafter positioned to do.

This construal does not deny that a proposition may be true or false, nor that a belief may be warranted or unwarranted. Those questions remain, and the analyses reviewed in §2 address them. What the construal denies is that the truth of the content exhausts what knowledge is doing among the parties who hold it. A true diagnosis and a false one are alike in reorganizing a family; a correct correction and a mistaken one are alike in positioning one party as the corrector.

4.2 The Immediacy of Relational Effect

The construal carries a consequence that distinguishes it from the critical accounts as well as from the classical ones. On the critical accounts, the relation between knowledge and power is genetic or conditional: power shapes what comes to count as knowledge, and knowledge in turn sustains the arrangements that produced it. The relation is real but mediated, running through the historical constitution of a discourse or the accumulated positions of a field. On the present construal the effect is immediate in a specific sense. Capacities are allocated among the parties in virtue of an interpretation’s acceptance, and the allocation is a feature of that acceptance and not a subsequent application of it. The allocation reaches a party when the interpretation does, which for parties absent at its acceptance may be later; what is denied is that the reallocation waits upon the interpretation being acted on.

Consider the physician’s statement that a patient has a particular condition. The statement, once accepted, does not merely describe. It establishes who is entitled to say what the patient’s sensations signify, what the patient’s own report of them counts as, whether evidence or symptom or noncompliance, which parties are to be consulted, what the family is to expect, and how a subsequent disagreement will be classified. None of this waits upon the statement being acted on. It obtains in virtue of the statement being accepted as knowledge, since being accepted as knowledge is being granted this coordinating force.

The teacher’s statement that a proposed reading is mistaken has the same character. It corrects a reading; and in the same motion it renews the standing of the party who corrects, specifies the register in which the student’s next proposal will be received, and settles what the student’s disagreement, should it come, would be an instance of. A parent’s account of why a child is crying does likewise, and so does one partner’s account of what a recurring difficulty between them is really about.

Claim 1. An accepted interpretation allocates capacities among the parties in virtue of its acceptance and not in virtue of its later application, and begins thereupon to generate the conditions of its continued acceptance. The relational effect of knowledge belongs to its constitution.

The second half of the claim is the one that matters for what follows, and it should be read carefully. To say that an accepted interpretation generates the conditions of its continued acceptance is not to say that the interpretation is self-confirming in the sense of being unfalsifiable. It is to say that acceptance establishes an arrangement of who assesses, on what occasions, and by what standards, and that this arrangement is what the next interpretation will encounter. The diagnosis establishes that subsequent reports of sensation will be received as symptoms; the correction establishes that subsequent proposals will be received as candidates for correction. Whether the arrangement so established persists is a further question, and it is the question of §7.

4.3 The Generality of Co-creation

If knowledge is the coordination of interpretation among parties, then the settings in which knowledge is co-created are not confined to organized inquiry. Four are treated in this paper, and they are selected because they differ in almost every respect except the one at issue.

In education, teacher and student generate an account of a subject matter and, inseparably, an account of the student’s competence at it. The second is generated whether or not anyone attends to it, since every assessment of a proposal is also a placement of the proposer.

In intimate relations, two parties generate an account of what their relationship is, what each of them is like within it, and what their shared history was. This account is knowledge in the full sense used here: it coordinates interpretation, conditions future action, and settles what a subsequent disagreement will be taken to be about. It is also, characteristically, generated without any of the procedural safeguards that organized inquiry supplies.

In international relations, states and other actors generate accounts of what a situation is, what a norm requires, and which parties are competent to say. The literature on norm emergence describes how such accounts spread and become taken for granted (Finnemore & Sikkink 1998), and the constructivist tradition holds that the structures of international politics are constituted by shared understandings rather than given by material distribution alone (Wendt 1999). What the present account adds is a question these do not pose: what happens to the parties in consequence of a shared understanding being established, and whether the arrangement so established continues to admit revision from those it positions.

In academic knowledge production, researchers generate accounts of a domain together with accounts of who is competent within it. The apparatus of review, citation, and authorship makes the second explicit in a way the other three settings do not, which is why it furnishes the clearest evidence, and why it has attracted most of the theoretical attention reviewed in §3.

The claim entered here is that these are instances of one structure rather than analogies drawn between distinct ones. Each involves parties coordinating interpretation; in each, the coordination allocates capacities immediately; in each, an established arrangement conditions what the next interpretation will meet. Academic production is accordingly one instance among four, and the fact that its arrangements are documented does not make it the paradigm case. The clinical, domestic, and diplomatic instances are less documented and not less structured.

Claim 2. Co-creation is the general case of knowledge and organized inquiry a special case of it. Education, intimate relation, international practice, and academic production instantiate a single structure, the coordination of interpretation among parties with its immediate allocation of capacities, and differ in the configuration of that structure rather than in kind.

What that configuration consists of has so far been left unspecified. Capacities have been said to be allocated, without a statement of which capacities there are or why those. Supplying the specification is the work of the next section, and the specification is to be derived rather than enumerated: the channels are to follow from what an interpretation must undergo in order to exist at all.

§5 The Life of an Interpretation

An interpretation, to exist as knowledge among parties, must undergo a definite course. It must be put forward by someone; it must be taken up rather than passed over; it must be found to hold; it must be alterable in the light of what follows; and it must be assigned to a source. Each stage is a point at which the parties stand in an asymmetric relation to one another, and each is accordingly a channel through which an asymmetry may maintain itself. The list that follows is derived from this course and is closed by it: the stages are five because the course of an interpretation has five, and a sixth would require a stage in the life of an interpretation that has not been named.

A word on what is claimed by the derivation. It is not claimed that the five are logically necessary in the sense that no coherent alternative could be framed, nor that they are the only vocabulary in which the phenomena could be described. What is claimed is that they follow from a minimal description of what must happen for an interpretation to be operative among parties, that they are jointly sufficient for the cases this paper treats, and that the enumerations offered in the adjacent literatures, which typically proceed by listing salient asymmetries as they are noticed, lack a principle by which to be completed or closed. A reader who would add a sixth is invited to name the stage it corresponds to.

5.1 Proposal

An interpretation must first be put forward. The asymmetry at this stage concerns who may advance a reading of a situation, and in what register a reading advanced by each party arrives. The register matters as much as the entitlement. A student may be formally free to propose while every proposal arrives as a candidate for correction; a patient may describe a sensation while the description arrives as material awaiting interpretation rather than as an interpretation itself. The distinction between advancing a reading and supplying raw material for another’s reading is the substance of this channel, and it is not settled by permission.

An asymmetry maintains itself here by conditioning the register in advance. Where one party’s contributions are habitually received as data, the receiving position is renewed by each such reception, and the contributing party learns the register and supplies contributions fitted to it. Neither party need intend this. It is sufficient that reception be consistent.

5.2 Admission

A proposal put forward must be taken up. The asymmetry at this stage concerns which proposals enter consideration at all: not which are accepted, but which are treated as having been made. A proposal may be met with silence, with a change of subject, with an acknowledgment that does not engage, or with the observation that it is not the present concern; in each case it has been put forward and has not been admitted.

This channel is the most easily overlooked, because non-admission leaves little trace. A rejected proposal is on the record as rejected; a proposal never taken up is on the record as nothing. The consequence is that a configuration can be highly restrictive at this stage while appearing open at the next, since the proposals that reach validation are precisely those already filtered. An asymmetry maintains itself here with the least expenditure of any channel, for it need not argue.

5.3 Validation

An admitted proposal must be found to hold or not to hold. The asymmetry at this stage is the one most readily recognized as epistemic authority: who determines whether an interpretation stands, and by what standards. Two features of it bear on the argument.

The first is that standards and their application are not separable in practice. A configuration in which standards are explicit and public differs from one in which they reside in the judgment of a party, and the difference lies in the first exercising its asymmetry through something the other party can inspect and invoke.

The second is that validation, more than the other channels, produces a durable trace. A validated interpretation is thereafter available as an established point from which further interpretation proceeds, and the position that validated it is thereby renewed. This is the channel in which self-maintenance is most direct: to validate is to reproduce the circumstance in which validation falls to one.

5.4 Revision

An established interpretation must be alterable in the light of what follows. The asymmetry at this stage concerns who may reopen what has been settled, and, more consequentially, what an attempt at reopening is received as.

The second half is where the substance lies, and §7 is devoted to it. It is nearly universal that revision is formally available: the student may object, the patient may seek a second opinion, the junior author may query the framing, the partner may say that the account of the recurring difficulty is wrong. What varies is whether the attempt alters the arrangement or is received within it, taken as evidence of the objector’s limited grasp, the patient’s anxiety, the junior’s inexperience, or the partner’s characteristic difficulty. In the latter case revision has been exercised and the arrangement has been renewed by the exercise. This is the channel in which an asymmetry maintains itself most powerfully, precisely because the maintenance operates through the mechanism nominally provided to check it.

5.5 Attribution

An interpretation must be assigned to a source. The asymmetry at this stage concerns whose contribution is recorded as a contribution, and it is the channel with the clearest institutional expression: authorship, order of authorship, acknowledgment, citation. It has less formal expression in the other settings and operates there no less, as in the household where one party’s suggestion becomes the household’s decision without remainder, or the negotiation where a proposal is adopted under another delegation’s name.

Attribution is consequential beyond credit, because it conditions the register at the next round of proposal. A party whose contributions are recorded arrives at the next proposal as one who has contributed; a party whose contributions are absorbed arrives as one who has not. The channel therefore closes back onto the first, and an asymmetry maintaining itself through attribution does so by shaping what each party will be at the next occasion.

5.6 The Configuration of the Five Channels

The five channels are not a hierarchy and not a sequence of increasing severity. They are distinct points at which an asymmetry may be maintained, and a given setting configures them independently. The seminar that admits freely and validates narrowly differs from the one that admits narrowly and validates by public standards; the clinic that attributes carefully to the patient’s own report differs from the one that does not, whatever their similarity at validation. The configuration is the object of analysis, and the following table records the form each channel takes in the four settings.

	Education	Intimate relation	International practice	Academic production
Proposal	Register in which a student’s reading arrives	Standing to say what the relation is like	Standing to say what a situation is	Standing to frame a research problem
Admission	Which questions are treated as questions	Which grievances are taken up	Which items reach an agenda	Which submissions are sent for review
Validation	Assessment and its standards	Whose account of a shared event holds	Determination of compliance	Peer review and its criteria
Revision	Reception of a challenge to a correction	Reception of an objection to the account	Reopening of a settled norm	Reception of a challenge to an established result
Attribution	Recognition of a student’s contribution	Recording of whose suggestion it was	Whose proposal a provision is under	Authorship, order, acknowledgment

The table is offered as an analytical grid rather than as a finding. Its use is to make configurations comparable: two settings that appear similar in their formal arrangements may be found to differ at a particular channel, and a setting that appears open may be found restrictive at admission while unrestrictive elsewhere.

Claim 3. The channels through which an epistemic asymmetry maintains itself are five, namely proposal, admission, validation, revision, and attribution, and are closed by the course an interpretation undergoes to be operative among parties, on the minimal description defended above. Settings differ in the configuration of these channels rather than in their number, and a configuration may be restrictive at one channel while unrestrictive at others.

Two questions are left open by the enumeration. The first is what determines the cost at which an asymmetry maintains itself through these channels, which is not given by the channels themselves; a configuration identical at all five may be readily borne in one setting and not in another. The second is what distinguishes a configuration whose generativity returns to the common holdings of the parties from one whose generativity returns to itself, which the channels record but do not explain. The first is treated next; the second in §7.

§6 Exit and the Price of Epistemic Capacity

The five channels record how an asymmetry is maintained; they do not record what the maintenance costs. Two configurations identical at every channel may nonetheless differ, and the variable that differentiates them lies outside the channels: the cost, to each party, of leaving the relation altogether. This section treats that cost. It is not a sixth channel, since exit is not a stage in the life of an interpretation; it is the condition under which the five operate, and it prices them.

6.1 The Cost of Leaving

Hirschman (1970) distinguished two responses available to a member of a declining organization: exit, the withdrawal of participation, and voice, the attempt to change matters from within. His central observation for present purposes is that the two interact. The availability of exit conditions the character and effectiveness of voice, and an organization from which departure is easy faces a different discipline from one whose members are held. Hirschman’s concern was the responsiveness of firms and organizations to decline. The observation transfers to the epistemic case with a modification, and the modification is the point.

Where leaving is dear, an epistemic position maintains itself at low cost. It need not persuade, need not deliver, need not continue to generate anything the other party finds worth staying for, because the other party’s remaining is not a judgment on what the position supplies. Where leaving is cheap, the same position must maintain itself expensively: it persists only so long as it continues to generate something that holds the parties, and a lapse in generation is answered by departure. Recalling that power on this account is generativity turned toward the conditions of its own continuation, the cost of exit sets the price of that turning. A high cost of exit is a subsidy.

The formulation yields a claim that is stronger than it first appears. It is not that captivity makes an epistemic asymmetry oppressive; it is that captivity permits an asymmetry to persist without generating. The same physician, the same standards, the same manner produce a different structure in a setting where the patient may consult another and in a setting where the patient may not, and the difference lies in what the conduct must accomplish to be sustained, and not in the physician’s conduct at any of the five channels.

6.2 Captive Configurations in Intimacy and Organization

Intimate relations are the setting in which the cost of exit is highest and least legible, since the cost is not principally material. To leave is to lose a shared history, a joint account of oneself, and the interpretive resources built up over years; and the account of the relation held between the parties is itself among the things that would be lost. A party who has come to be positioned as the one who misunderstands does not readily leave, in part because leaving would require the very interpretive standing the arrangement has withheld. This is the structure that distinguishes an ordinary asymmetric relation from one that has become closed: the conjunction of an unfavourable configuration with a cost of exit that removes the pressure to improve it.

Organizational settings furnish more legible instances. A doctoral student whose supervisor is the field’s principal figure at a particular institution faces an exit cost measured in years and in the standing that letters of reference confer; a junior researcher on a visa tied to an employer faces one measured in the right to remain. In both, the configuration of the five channels may be no different from that facing a student or researcher with alternatives, and the structure is different, because nothing in it must be earned.

The point is not confined to the party with less capacity. Exit costs run in both directions, and a position that would be costly to vacate, such as a supervisor whose funding depends on a student’s completion or a partner whose account of the relation depends on the other’s corroboration, is a position under pressure to generate. Where both parties would find leaving costly, the configuration is held in place from both sides, which is the condition under which the longest-lived and both the most generative and the most degraded relations are found.

6.3 Asymmetric Alliance and the Absence of an Outside

The international case exhibits a variant worth distinguishing, since it bears on what the analysis can claim beyond the interpersonal. A state within an asymmetric security arrangement, or within a monetary or trading order it did not design, faces an exit cost that is high and in some instances undefined: there may be no outside to depart to. Where a normative framework has become the medium in which claims of any kind are made intelligible, withdrawal from it forfeits the standing to make claims at all and reaches no alternative.

This is the limiting case of a costly exit, and it shows what the variable does at its extreme. A framework with no outside maintains itself without generating, since the parties it positions have no comparison against which its performance could be assessed and no destination that departure would reach. Nothing in this entails that such a framework is malign, and the analysis does not license that inference: a framework may be both inescapable and generative, and the absence of an outside says only that its generativity is not enforced by the possibility of departure. What it must instead be enforced by is the subject of §7.

A caution is owed on the transfer of a concept developed for individuals and firms to states. Exit for a state is not the decision of a unitary agent, and the costs fall unevenly on domestic constituencies; the case is offered as an instance of the structural variable and not as a claim about how states in fact decide.

Claim 4. The cost of leaving prices the five channels. Where exit is dear, an epistemic position maintains itself without needing to generate anything that would hold the parties; where exit is cheap, the same position persists only in so far as it continues to generate. A high cost of exit is accordingly a subsidy to whatever asymmetry the configuration already contains, and not an additional asymmetry.

The claim identifies what enforces generation from outside a relation. It leaves untreated the question of what may enforce it from within, which becomes the whole question wherever exit is dear, which is to say in the intimate relations, the formative educational relations, and the encompassing international frameworks that matter most. That question is the centre of the paper, and the next section takes it up.

§7 Absorption and the Closure of the Revision Loop

The question left standing is what enforces generation from within a relation. This section answers it, and the answer requires first correcting a picture that the preceding sections may have suggested. The correction is that absorption and self-maintenance are two directions available to a single generativity, and not two forces contending, one benign and one malign. Once that is stated, a criterion can be given for telling them apart, and the criterion is the paper’s principal analytical proposal.

7.1 Two Directions of One Generativity

Where an asymmetry exists, the position it defines generates. It produces standards, occasions, distinctions, and participants suited to it, and it must produce them continually, since a position that stops producing them ceases to be a position. The question is where the product goes.

It may return to the common holdings of the parties. A teacher who validates a student’s reading produces, in that act, a standard; and the standard may be one the student can thereafter apply, contest, and extend, in which case what the validation generated has entered the common holdings of the two parties and is available to the student at the next proposal. Or it may return to the position. The same validation may produce a standard that resides in the teacher’s judgment, is applied case by case, and is not available to the student except as something to be submitted to, in which case what the validation generated is the renewed circumstance in which validation falls to the teacher.

Little distinguishes these two at the level of the act. Both are validations, both may be accurate, both may be delivered with equal care. They differ in the destination of what they generate, and the difference accumulates. It is for this reason that the case with which this paper opened involves no acquisition and no misconduct. The teacher of the second term is not doing anything the teacher of the first term did not do. The generativity that had been running into the common holdings is running into the conditions of the teaching position, and the student is encountering, at each proposal, an arrangement that the previous exchange renewed.

The term absorption names the first direction: the asymmetry generated by a relation is taken up into the common generation of the parties, so that what would otherwise consolidate a position becomes a holding both may use. Self-maintaining generativity, in the sense given in §1, names the second. Absorption is not the suppression of an asymmetry and not its equalization. The teacher remains the more competent party throughout, and the asymmetry is undiminished; what changes is that its generativity is continually returned. Degradation, correspondingly, is the turning of this direction and not the appearance of something new, and it requires no agent.

7.2 The Insufficiency of Available Revision

A natural proposal at this point is that what keeps a relation generative is revisability: an asymmetry is tolerable so long as it remains open to challenge, and degrades when it becomes unchallengeable. The proposal is close to correct and, as ordinarily stated, nearly empty.

It is nearly empty because revision is almost always formally available. The student may object; the patient may seek another opinion; the junior author may query the framing; the partner may say the account is wrong; the state may enter a reservation. Configurations that permit no challenge whatever are rare, and they are not the interesting cases. The interesting cases are those in which challenge is permitted, is exercised, and changes nothing, for the reason that it was received and not that it was overruled.

Consider the child told that a proposed course is being forbidden for the child’s own good, who replies that the reason is not the real reason. The reply is a challenge, it was permitted, and it was made. It is received as evidence of the immaturity that warranted the original judgment. The arrangement has not been shaken by the challenge; it has been fed by it, since the challenge supplied a fresh occasion for the judgment to be exercised and confirmed. The partner told that they are being defensive, who replies that they are not being defensive, meets the same structure. So does the student whose objection to a correction is received as not yet understanding the correction.

The structure is worth naming precisely, because it is what makes availability the wrong criterion. In each case the party challenging is required to produce something, whether an objection, a reason, or a denial, and the arrangement is configured so that whatever is produced counts as further material for the position being challenged. The challenger cannot decline to respond, since silence is also received as confirmation. Revision has been exercised, and the exercise has renewed what it was directed against.

7.3 The Open Loop and the Closed Loop

What the foregoing indicates is that the criterion must concern the effect of an exercise of revision rather than its availability. It may be stated as follows.

The criterion of loop closure. An exercise of revision closes the loop when it alters the conditions under which the next interpretation is generated: the standards by which proposals are assessed, the register in which they are received, the occasions on which assessment occurs, or the standing of the parties to assess. It leaves the loop open when it is received within those conditions and returned as material confirming them.

Three features of the criterion should be noted. The first is that it concerns conditions rather than outcomes. A challenge that is considered seriously and found to be mistaken may close the loop, provided that the grounds on which it was found mistaken are thereafter available to the challenger; a challenge that is conceded on the particular while the standing to challenge is left untouched may leave the loop open. Being right is neither necessary nor sufficient.

The second is that the criterion is checkable in principle, which was the ground on which §3 distinguished this account from the psychoanalytic one. What it asks after is whether the terms of the next exchange differ from the terms of the last, and this is a matter of observable conduct: whether a standard invoked once becomes a standard the other party may invoke, whether a concession granted becomes a precedent that may be cited, whether an objection admitted alters what the objector may thereafter propose without preface. The checking is not trivial and the paper offers no procedure for it; the claim is that the criterion refers to something that could be looked for.

The third is that closure admits of degree and is a property of a sequence rather than of an act. No single exchange settles whether a relation’s revision loop is closed, since the effect of an exercise appears in what follows it. The criterion is applied to a run of exchanges, and what it detects is a tendency.

7.4 Degradation Without Acquisition

The account now assembled explains the phenomenon with which the paper began, and it is worth setting out the explanation explicitly, since it is the principal evidence for the account.

In the first term, the teacher’s corrections closed the loop. Each correction made available to the student a standard the student could thereafter apply: the ground on which the answer was assessed, and not the answer alone, so that the student arrived at the next proposal with more to propose from, and the teacher arrived at it facing a party whose proposals required a different reception. The asymmetry was undiminished and its generativity was returning to the common holdings of the parties.

In the second term the loop opened. The corrections remained accurate and the manner unchanged; what altered was that the ground of assessment stopped travelling. Each correction settled a case and renewed the circumstance in which the teacher settles cases, and the student, encountering this, learned the register, which is the point at which the hedged proposal appears, since a proposal offered as a request for correction is a proposal fitted to the arrangement it will meet. Nothing was acquired by either party. No procedure was violated. The direction of the generation turned, and the arrangement thereafter reproduced itself through the student’s own adaptation to it.

Claim 5. Absorption and self-maintenance are two directions of a single generativity rather than two opposed forces. A configuration absorbs the asymmetry it produces when the generativity issuing from a position returns to the common holdings of the parties, and degrades when that generativity returns to the conditions of the position itself. Degradation accordingly requires neither acquisition nor misconduct, and its criterion is the closure of the revision loop.

One implication should be drawn out, as it bears on the design consequences of §10. If degradation requires no agent, then no configuration is secured by the good intentions of the party the asymmetry favours, and none is secured by the formal availability of challenge. What secures a configuration is that the exercise of revision within it has effects that persist, which is a property of the arrangement and not of anyone’s disposition. A generous teacher whose corrections do not travel maintains a position as surely as an ungenerous one, and rather more effectively, since the generosity conceals the closure.

§8 Generative Power

The account so far has treated the five channels, the price at which they are held, and the criterion distinguishing the two directions a position’s generativity may take. It has not yet named what a position does when its generativity runs the other way. That is the subject of this section, and the concept it introduces is the one on which the paper stakes its contribution. The concept is also the one most liable to be misread, and the misreading is the reason for the section’s final part.

8.1 The Expansion of the Space of Proposal

The five channels are stages in the life of a given interpretation. A position may act upon them, and everything treated so far has concerned such action. A position may also act upon something prior: the space from which interpretations may be drawn at all: what may be treated as a candidate reading of a situation, what may be raised as a question about it, and what would count as a consideration bearing on it.

This is a distinct kind of exercise, and the distinction is not one of degree. To validate a proposal is to act within a space of proposals already constituted. To alter what may be proposed is to act on the constitution of that space. A teacher who marks a reading correct operates in the first register. A teacher who shows that a question the student took to be closed is open, or that a consideration the student took to be irrelevant bears on the matter, operates in the second: the student thereafter has available a range of proposals that were not previously available, and this availability does not depend on the teacher’s further participation.

Call the capacity to act in this second register generative power. It is exercised in every setting the paper treats. A supervisor who reframes what a field’s central difficulty is has expanded what a student may work on. A partner who articulates a difficulty in terms neither had used before has made available to both a way of describing their relation that neither previously had. A delegation that introduces a category into a negotiation, supplying a way of characterizing a dispute that the parties had not been able to characterize, has expanded what may thereafter be proposed by any of them. In each case something has been made possible rather than settled.

8.2 The Power in Generative Power

The temptation at this point is to treat generative power as the benign form of power and to organize the paper’s normative content around promoting it. That treatment must be refused, and refusing it is the burden of this subsection.

Generative power is power in the full sense the paper has given the term, and it is the most consequential form of it. What it does is shape the space from which a party’s future interpretations will be drawn, and a space, once shaped, conditions everything drawn from it. The student whose sense of what constitutes an interesting problem was formed by a supervisor will pose problems bearing that formation for years, in settings the supervisor will never see, and will pose them as their own. The partner who acquired a vocabulary for the relation from the other will describe the relation in that vocabulary, including when describing what is wrong with the other. A delegation whose category is adopted has shaped what subsequent parties, including parties not yet present, can say.

The formative asymmetry is greater than any exercised at the five channels, not less, and it is greater in proportion to its success. A correction is bounded: it settles one case and can be revisited. A reshaped space of proposal is unbounded and is characteristically invisible to the party it forms, since what has been made available appears as how the matter stands and not as someone’s contribution. That a formation is generative does not make it innocent, and the party formed has no vantage from which to assess a formation they think with.

Claim 6. Generative power is power. It acts on the space from which interpretations may be drawn rather than on the fate of a given interpretation, and its effects are accordingly more extensive, longer-lived, and less visible to the party they form than any exercise at the five channels. It is not the benign alternative to power and confers no innocence on the position exercising it.

8.3 The Criterion of Reciprocal Capacity

If generative power is not distinguished by innocence, it must be distinguished by something structural, and it is. The mark of a generative exercise is that it erodes the conditions on which the exercising position depends.

The position of the corrector depends on there being a party who requires correction, and each correction that settles a case without making the ground of assessment available renews that dependence. An exercise of generative power runs the other way: the teacher who makes available a range of proposals the student may draw from, and the grounds on which proposals are assessed, has produced a party who requires the teacher less. The exercise consumes its own condition. Where such exercise continues, the position of the teacher becomes progressively less necessary to the student’s proposing, and the relation either terminates in the student’s independence or continues on terms that must be constituted afresh.

This supplies the structural criterion, and it should be stated in a form that does not collapse into the earlier one. Loop closure, in §7, concerned what happens to a given exercise of revision. The present criterion concerns what happens to the exercising position across a run of exercises: whether its necessity is being renewed or consumed. The two are related, since a configuration whose revision loops close will tend to consume the necessity of the validating position, and they are not identical, since a position may make grounds available while remaining necessary for reasons of its own, and a position may be consumed by circumstances having nothing to do with revision. Absorption, in the sense of §7, stands to both as the wider notion: a configuration absorbs when the generativity issuing from a position returns to the common holdings, and the consumption of a position’s necessity is one consequence such a return may have, though not the only one and not a condition of it.

The criterion is also what makes generative power costly, which is why it is not the default. A position maintains itself cheaply by settling cases; it expends itself by making available the grounds on which cases are settled. Nothing in the account suggests that the expending is more natural than the maintaining. The account suggests the reverse: self-maintenance is what a position does when nothing counteracts it, and generative exercise is what must be sustained against that tendency.

8.4 The Weight of Formative Asymmetry

Two consequences follow that the account should state rather than pass over, since both are uncomfortable and both are entailed.

The first concerns gratitude. A party formed by a generative exercise characteristically regards the formation as a gift and the former as a benefactor, and the regard is warranted, since something was made available that would not otherwise have been. It is also, structurally, the residue of an asymmetry so successful that it is no longer perceived as one. The student’s sense that a supervisor opened the field is both true and the trace of a formation the student cannot assess from within. Nothing here suggests the gratitude is misplaced. What is suggested is that its presence is not evidence that the formation was benign, since the formation is precisely what would produce the standpoint from which it is evaluated.

The second concerns the limits of the reciprocal-capacity criterion. A position may erode its own necessity in one respect while establishing it in another: the supervisor who makes a student independent in method may thereby make the student dependent on a way of seeing what is worth attending to, and the second dependence is deeper and less legible than the first. The criterion is therefore to be applied with a specification of the respect in which necessity is being consumed, and a global judgment that a relation is generative is not available from it. What the criterion yields is local: in this respect, the position is expending itself; in that respect, it is not.

The account therefore does not deliver a class of relations certified as generative. It delivers a question that may be asked of a relation, in a specified respect, at a given time, asking whether the generativity issuing from a position is consuming or renewing that position’s necessity, together with an expectation that the answer will differ across respects and change over time. This is less than a criterion of good relations, and it is what the analysis supports.

§9 The Production of Knowers

The account has treated a position’s generativity as producing standards, occasions, and distinctions. It produces something further, and the further thing is the principal means by which an epistemic position maintains itself: it produces participants of determinate kinds. This section sets out that mechanism, which is what makes the four settings instances of one structure rather than parallel illustrations.

9.1 The Formation of Epistemic Position

A configuration of the five channels does not encounter parties already constituted as those who propose and those who assess. It constitutes them, by the accumulation of receptions. A party whose contributions are consistently received as material rather than as readings arrives at the next occasion as one who supplies material; a party whose objections are consistently received as instances of a disposition arrives as one who has that disposition. The constitution operates through the ordinary functioning of the channels, requires no intention, and proceeds without being declared.

Its efficacy lies in the party’s own adaptation. A student who has learned the register in which proposals are received will hedge before proposing, and the hedging is competence and not compliance: it is a correct reading of the arrangement, and a proposal offered without it would be received worse. What the arrangement has produced is a party whose skilled participation reproduces it, and not a party who submits to it. This is why a configuration can persist without maintenance from the position it favours. The other party maintains it, competently, and at the level of what they perceive themselves to be rather than what they are told to do.

The mechanism accordingly answers a question left open in §7: how a direction of generation, once turned, sustains itself. It sustains itself because it produces parties for whom that direction is the shape of the relation, and the fifth channel closes the circuit, since attribution having recorded who contributed, each party arrives at the next proposal as what the record made them.

9.2 The Student, the Junior Author, the Norm-Taker

The mechanism may be observed in the settings the paper treats, and its instances are not analogies.

In education it produces the student who asks poorly. Such a student is not one who knows less; the knowledge deficit is real and is not what is at issue. What is at issue is that the student has been constituted as one whose questions are occasions for correction rather than contributions, and, having been so constituted, produces questions fitted to that reception, which are narrower, hedged, and aimed at securing a correction rather than opening a matter. An observer encountering the student at this stage will find the constitution confirmed by the evidence, since the evidence is what the constitution produces.

In academic production it produces the junior author who is an acknowledgment. The mechanism runs through attribution, and its effect is at proposal: a researcher whose contributions have been recorded as assistance arrives at the next collaboration as an assistant, is assigned work of the kind assistants are assigned, and produces contributions that are correctly described as assistance. The record and the reality converge, and the convergence is the mechanism working rather than evidence that the record was accurate.

In intimate relation it produces the party who misunderstands. An account of the relation that assigns to one party a characteristic failure of understanding will, if sustained, produce a party who checks their understanding before advancing it, defers on contested matters of shared history, and treats their own reading of an episode as the one requiring verification, which is what a party who misunderstands would do.

In international practice it produces the norm-taker. A state positioned as a recipient of standards drafted elsewhere develops the institutional apparatus of a recipient: offices for compliance and reporting rather than for drafting, expertise in demonstrating conformity rather than in framing what conformity should consist of. Having developed these, it participates as a taker, and its participation is competent.

Fricker (2007) names a proximate phenomenon and the relation to it should be stated precisely. Fricker’s testimonial injustice occurs when a hearer, owing to prejudice, accords a speaker less credibility than they deserve, and she treats the harm as a wrong done to a person in their capacity as a knower, including, in what she calls the harm’s persistent effects, the possibility that a speaker so treated may lose confidence and cease to develop their epistemic capacities. The mechanism described here overlaps with that account and differs in its source. Fricker’s cases turn on identity prejudice operating through a hearer’s credibility judgments, and her diagnosis is correspondingly about stereotype and the ethics of the hearer. The production described here requires no prejudice and no identity marker: it follows from the configuration of the channels alone, and would occur between parties who bore no prejudice toward one another whatever. The two mechanisms may compound, and where they do the effects are severe; the present claim is that the structural mechanism operates in their absence.

9.3 One Mechanism Across Four Domains

The four instances share a form. In each, a configuration produces a party of a kind; the party’s competent participation reproduces the configuration; and the reproduction furnishes evidence that the party is of that kind. What distinguishes the instances is the channel through which the production principally runs, whether reception at proposal in education, attribution in academic production, validation of a shared account in intimate relation, or admission and standard-setting in international practice, and not the mechanism.

This is the ground for the claim entered in §4 that these settings instantiate one structure. They are not similar in their content, their scale, their duration, or their formality. They coincide in that each is a coordination of interpretation among parties, that each configuration allocates capacities immediately, and that each configuration, sustained, produces participants whose participation sustains it.

Claim 7. An epistemic position maintains itself principally by producing participants of determinate kinds. The production operates through the accumulated reception of contributions, requires no intention and no prejudice, and is completed by the produced party’s own competent participation, which reproduces the configuration and furnishes the evidence that the party is as the configuration constituted them.

The claim carries a consequence for what a remedy would have to be. A configuration that has produced its participants is not corrected by an adjustment at a single channel, since the participants will meet the adjustment as the parties the configuration made them. The student granted the floor will use it to request a correction, and the junior author granted authorship will contribute what an assistant contributes. What such a configuration requires is treated next.

§10 The Design of Generative Knowledge Structures

The analysis carries consequences for the arrangement of settings in which knowledge is co-created, and this section draws them. It is constrained at the outset by §9: a configuration that has produced its participants is not corrected by an adjustment at a single channel, since the participants meet the adjustment as the parties the configuration made them. What follows is accordingly a small number of conditions and not a list of measures, each stated with the mechanism by which it works and the ways in which it is defeated.

A prior word on the standing of the whole. The account does not aim at the elimination of epistemic asymmetry, which would abolish the relations at issue rather than improve them, and it does not certify any arrangement as generative, for the reason given in §8: what the criteria yield are local judgments in specified respects. The conditions below are directions in which a configuration may be pressed, and each has costs that a designer must weigh against what it secures.

10.1 Loop Closure and Design

The first condition follows from §7: an exercise of revision should alter the conditions under which the next interpretation is generated. What makes this a design condition rather than an exhortation is that the alteration can be built into an arrangement rather than left to the disposition of the party who receives the challenge.

The general mechanism is that the grounds of an assessment travel with it. Where a validation is accompanied by the standard on which it rests, stated so that the other party can thereafter invoke it, the exercise has added to the common holdings; where it is delivered as a judgment, it has renewed the circumstance in which judgments are delivered. This is the difference, in teaching, between a correction and a correction whose ground the student may thereafter apply without the teacher; in review, between a rejection and a rejection whose criteria constrain the reviewer’s next assessment as much as they explain this one; in an intimate relation, between settling what happened and articulating what would count as evidence about what happened, which either party may subsequently produce.

Three defeats of the condition should be named, since each is common and each preserves its appearance. The first is the standard stated so that it cannot be applied by anyone else, resting on a discernment the assessing party possesses and does not transfer. The second is the concession granted on a particular while the standing to challenge is left untouched, which §7 noted leaves the loop open. The third is the recorded objection: an arrangement that logs challenges without any provision for their bearing on subsequent assessment converts revision into documentation, and the log becomes evidence that revision is available.

10.2 The Governance of Exit Cost

The second condition follows from §6. Since a high cost of exit subsidises whatever asymmetry a configuration contains, the deliberate reduction of exit cost is among the most effective interventions available, and it operates without requiring anything of the parties.

Its instruments are familiar and their effect is now specifiable. Multiple supervisors, portable funding, external examiners, the right to seek a second opinion, employment not tied to a single sponsor, and forums in which a claim may be pursued outside the relation that generated it all reduce the price of departure. Each thereby withdraws the subsidy: a position that could formerly persist without generating must now generate something that holds the party, or lose them. The condition is unusual among those given here in that it works upon the least tractable cases, since it does not require the favoured position’s cooperation.

Two limits bound it. Where the relation is one whose value depends on its not being readily exited, as formative relations characteristically are and intimate ones definitionally are, reducing exit cost past a point degrades what the relation is for, and the condition is not to be pursued without regard to that. And the international case of §6 shows that some frameworks have no outside, so that exit cost cannot be reduced at all. Where it cannot, the whole burden falls on loop closure, which is the substantive reason the criterion of §7 matters most precisely where this condition is unavailable.

10.3 The Revisability of Attribution

The third condition follows from the circuit identified in §5 and §9. Attribution conditions the register at the next proposal, and a record once made becomes the ground on which the next contribution is assigned. An arrangement in which attribution is settled once and thereafter fixed is therefore not merely unjust in the ordinary sense; it is the mechanism by which a produced party is held in production.

What the condition requires is that the record be alterable in the light of what a contribution turns out to have been. Its instruments include contribution statements that specify what was done rather than assigning ranks, provision for the reassessment of contributions when their significance becomes clear later, and, in the settings that lack formal attribution, the practice of naming whose suggestion a decision followed. In domestic and diplomatic settings this last is nearly the whole of what is available, and it is not trivial: a household or a delegation in which the origin of an adopted proposal is regularly stated maintains a record that a party may draw on at the next occasion.

10.4 Curricula, Review Processes, Households, Treaty Regimes

The conditions bear differently on the settings the paper treats, and the differences are instructive as to what design can reach.

A curriculum is unusual in that its designer controls the configuration in advance and the parties arrive unproduced. Loop closure can be built into assessment by requiring that criteria be stated in a form students may apply to one another’s work and to the assessor’s; attribution can be made explicit where collaborative work is undertaken. The characteristic defeat is the criterion that is published and applied by discernment.

A review process has the most developed instruments and the highest exit cost in specialised fields, where the set of competent reviewers is small. Loop closure is served by criteria that bind the reviewer’s subsequent assessments, by published reports, and by the availability of an appeal that reaches someone other than the original assessor; the characteristic defeat is the appeal returned to the party appealed against.

A household has no formal instruments and the highest stakes at validation, since what is validated is an account of the parties themselves. The available conditions are the articulation of what would count as evidence, which makes assessment something either party may perform, and the naming of whose suggestion was adopted. The characteristic defeat is the concession on the particular that leaves the account untouched.

A treaty regime has exit costs that are often irreducible and sometimes undefined, which throws the burden onto loop closure and attribution. What is available is the reopening of settled provisions on a schedule rather than on the initiative of a party who would bear the cost of raising them, and the recording of which delegation’s proposal a provision followed. The characteristic defeat is the review conference whose outcome cannot alter the terms under which the next review is conducted.

Claim 8. Three conditions bear on whether a co-creative configuration continues to absorb the asymmetry it generates: that the grounds of an assessment travel with it, so that an exercise of revision alters the conditions of the next interpretation; that the cost of exit be reduced where the relation permits, since a high cost subsidises whatever asymmetry the configuration contains; and that attribution remain alterable, since the record conditions the register at the next proposal. None certifies a configuration as generative, and each is defeated by measures that preserve its appearance.

The last part of the claim is the one to be insisted on. Each condition has a form that satisfies its description and defeats its purpose: the standard that cannot be applied by another, the appeal returned to the party appealed against, the log of objections with no provision for their effect. An arrangement exhibiting the instruments is not thereby shown to be generative, and the criteria of §7–§8 are what the instruments are to be assessed against, rather than the reverse.

§11 Limits and Open Work

The account rests on claims that could be wrong, and setting them out as such is part of setting it down. This section states what the paper has argued, what it would take for the argument to fail, and what the argument does not reach.

11.1 What the Account Claims

The paper has proposed that knowledge is a relational structure through which interpretation is coordinated, that every such structure generates asymmetry, and that upon an asymmetry the tendency of a position to maintain itself is generative, producing the standards, occasions, and participants that constitute the conditions of its own continuation. Power in knowledge is this self-maintaining generativity, which places it within generation rather than opposed to it, and makes it costly, lapsing, and defeasible rather than possessed.

From this three analytical instruments follow. The five channels of §5 specify where an asymmetry maintains itself, derived from the course an interpretation must undergo. The cost of exit of §6 specifies the price at which it does so. The criterion of loop closure of §7 distinguishes the direction in which a position’s generativity runs, and the criterion of reciprocal capacity of §8 distinguishes whether an exercise consumes or renews the position’s necessity. The production of knowers in §9 is the mechanism by which a direction, once turned, sustains itself.

11.2 What Would Defeat the Account

Several things would. If the critical tradition already answers the question posed here, and the analysis of regimes, fields, or networks does descend to the direction of generation within a relation whose configuration is unchanged, then §3 has misread it and the paper has no ground. If the enumeration of five channels admits a sixth that answers to a stage in the life of an interpretation, the derivation fails and the enumeration reverts to the ad hoc lists it was meant to improve on. If cases can be produced in which a relation’s revision loops demonstrably close and the relation nonetheless degrades, or in which they demonstrably open and it does not, the central criterion is not doing the work claimed for it.

The last of these is the most consequential and the least tested, and the paper’s principal weakness lies there. The criterion of loop closure was said in §7 to be checkable in principle, and no procedure for checking it was supplied. What would be required is a specification of what counts as an alteration in the conditions of the next interpretation, robust enough that two observers of the same sequence of exchanges would classify it alike. Until that exists, the criterion organizes description without licensing measurement, and the applications in §10 rest on it more heavily than it can presently bear.

11.3 What the Account Does Not Reach

The illustrative cases throughout are constructed rather than observed. They were chosen for their structural clarity and they carry the weight of illustration only; none is evidence that the configurations described are common, and the paper offers no estimate of how often relations of the kinds treated exhibit them. Fieldwork in any of the four settings could establish that the mechanism, though coherent, is rare.

The four settings were selected for their dissimilarity, and the argument that they instantiate one structure is correspondingly a structural argument rather than an empirical finding. It would be strengthened by the demonstration that the channels can be specified operationally in each, and embarrassed by settings in which the five cannot be identified.

The treatment of the international case is the most exposed. Exit for a state is not the decision of a unitary agent, the costs of departure fall unevenly on domestic constituencies, and the transfer of a vocabulary developed for interpersonal relations to collective actors requires justification the paper has not supplied beyond the structural analogy. The case is offered as an instance of the variable and not as a contribution to the theory of international politics.

The account is descriptive in a further respect that limits what §10 can claim. Having refused to certify configurations as generative, and having conceded in §8 that the criteria yield only local judgments in specified respects, the paper cannot say of any actual arrangement that it is well-ordered. What it can say is where to look and what would count as the relevant evidence.

11.4 Place Within the Programme

The line of analysis belongs to a broader effort to treat being, value, and governance as generated in and through relation. The parent framework, generative relational being, concerns how relation generates being; a companion concern, generative relational governance, treats how a generated relation is governed once it exists. The present paper occupies the intersection: knowledge is the medium in which the parties to a relation coordinate their interpretations, and the configuration of that medium is at once what the relation generates and what governs its further generation. That the same structure appears in the account of relational degradation developed elsewhere in the programme, where a distinction collapses and a symbol executes the collapse, is not incidental, and the correspondence between the closure of a revision loop and the mechanisms described there is work this paper leaves to be done properly.

11.5 Further Work

Three lines follow. The first is the operationalisation of loop closure, without which the account remains a way of describing cases rather than a means of assessing them. The second is empirical work in a single setting, of which the review process is the most tractable, since its channels are documented and its exercises of revision leave records. The third is the relation between the two criteria the paper offers, loop closure and reciprocal capacity, which were distinguished in §8 and whose connection was there left as a tendency rather than established. A formal treatment of a configuration and its trajectories would bear on this, and is not attempted here.

---

A note on method. No formal apparatus has been used to state what prose states plainly here, and the abstention is deliberate. A configuration written as a vector of weighted capacities would decorate the argument without deriving anything from it, and would borrow a rigor the argument has not earned. The mechanisms admit two honest presentations: clear prose, or a genuine connection to the language of dynamical systems, in which a configuration would answer to a point in a space of arrangements, a closed revision loop to a perturbation that displaces it, an open one to a perturbation the arrangement absorbs without displacement, and a produced knower to a fixed point of the configuration’s own dynamics. The second is not attempted here. What is claimed is stated in the first.

References

Bourdieu, Pierre (1975). The Specificity of the Scientific Field and the Social Conditions of the Progress of Reason. Social Science Information, 14(6), 19–47.

Bourdieu, Pierre (1988). Homo Academicus. Trans. Peter Collier. Stanford: Stanford University Press. [French orig. 1984.]

Callon, Michel (1986). Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay. In John Law (ed.), Power, Action and Belief: A New Sociology of Knowledge? London: Routledge, 196–233.

Finnemore, Martha, and Kathryn Sikkink (1998). International Norm Dynamics and Political Change. International Organization, 52(4), 887–917.

Fricker, Miranda (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Foucault, Michel (1972). The Archaeology of Knowledge. Trans. A. M. Sheridan Smith. New York: Pantheon Books. [French orig. 1969.]

Foucault, Michel (1977). Discipline and Punish: The Birth of the Prison. Trans. Alan Sheridan. New York: Pantheon Books. [French orig. 1975.]

Foucault, Michel (1978). The History of Sexuality, Volume I: An Introduction. Trans. Robert Hurley. New York: Pantheon Books. [French orig. 1976.]

Freire, Paulo (1970). Pedagogy of the Oppressed. Trans. Myra Bergman Ramos. New York: Herder and Herder.

Gettier, Edmund L. (1963). Is Justified True Belief Knowledge? Analysis, 23(6), 121–123.

Hirschman, Albert O. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press.

Goldman, Alvin I. (1967). A Causal Theory of Knowing. The Journal of Philosophy, 64(12), 357–372.

Goldman, Alvin I. (1999). Knowledge in a Social World. Oxford: Oxford University Press.

Lacan, Jacques (1977). The Four Fundamental Concepts of Psycho-Analysis. Ed. Jacques-Alain Miller, trans. Alan Sheridan. London: Hogarth Press.

Latour, Bruno (1986). The Powers of Association. In John Law (ed.), Power, Action and Belief: A New Sociology of Knowledge? London: Routledge, 264–280.

Latour, Bruno (1987). Science in Action: How to Follow Scientists and Engineers through Society. Cambridge, MA: Harvard University Press.

Nietzsche, Friedrich (1966). Beyond Good and Evil: Prelude to a Philosophy of the Future. Trans. Walter Kaufmann. New York: Vintage Books. [German orig. 1886.]

Wendt, Alexander (1999). Social Theory of International Politics. Cambridge: Cambridge University Press.

中文

共创中知识的权力结构

一种生成性关系存在的视角

黄万宏 · huangwanhong@serendip.ngo

---

摘要

关于知识与权力的诸说明采取了两种宽泛的形式。在第一种上，知识是内容，而权力是一种污染，一套良序的认识实践把它排除在外。在第二种上，与福柯以及知识社会学相联系，权力条件化什么可算作知识，从而两者在历史上不可分。本文提出第三种说明。知识是一个诠释经由它被协调的关系结构，而每一个这样的结构都在其参与者之间生成不对称。在那一不对称之上，一个位置维持自身的倾向本身是生成性的：它产生它自己延续的诸条件。知识中的权力就是这一自我维持的生成性，这把它置于生成之内、而非同生成相对立。本文区分了一个认识不对称维持自身所经由的五个通道，追随一个诠释的生命历程（提出、纳入、认证、修正、归属），连同为它们定价的退出成本。本文论证一个配置在从一个位置发出的生成转向对其自身诸条件的生产时败坏，而这一转向的判据是修正环路的闭合：一次修正的施行是否改变下一个诠释被生成于其下的诸条件。生成性权力被当作权力、而非当作它的良性替代者来处理，以施行位置所依赖之诸条件的侵蚀为标志。该说明跨教育、亲密关系、国际关系与学术知识生产而展开，这些被当作同一个机制的诸实例。

关键词： 知识；权力；共创；生成性；诠释；可修正性；关系认识论。

关于本文定位的一则说明。 这是一个理论提议。它取一个熟悉的主张，即知识与权力相纠缠，并问关于那一纠缠的两种现存说明是否穷尽了它。阐述的次序追随论证的次序：现存的诸说明及其界限先被铺陈，而生成性的说明只作为它们所留下未加处理者被引入。倘若能证明批判传统已然回答此处所提的问题，后面的诸节便失去它们的立足之处。诸例证案例取自寻常情境、而非取自系统的田野工作，且它们只承载例证的分量。反对、更正与反证，欢迎寄至 huangwanhong@serendip.ngo。

§1 引言

一位教师与一名学生用一个学期一起通盘处理一个问题。开始时，学生提出错误的读解，并从对它们的更正中学习；到末尾，学生提出教师未曾考虑过的读解。这一过程中没有任何东西不寻常，而它描述了教育所为之目的的一大部分。现在考虑第二个学期，其中同样的两个人通盘处理第二个问题。学生提出；教师更正；学生再次提出。诸更正是准确的，教师既不严厉也不疏忽，而学生既不懒惰也不迟钝。然而到这一学期末尾，学生渐渐提出得更少，而学生确实提出的东西已然带着防备到来，与其说作为一个读解、不如说作为一个对更正的请求而被给出。没有人获得了对方所缺的知识。没有人举止不当。这一关系的结构中有某样东西改变了，而知识通常在其中被讨论的语汇不容易说出那是什么。

这一困难不限于教学。一对伴侣在他们之间维持一份关于他们的关系是怎样的说明：谁是焦虑的那个，谁在退缩，在那个他们记得不同的场合发生了什么。这份说明在寻常意义上是知识，即它协调他们的诸预期与他们的行为。它可能是一份两者都继续塑造的说明，也可能是一份一方渐渐编撰而另一方渐渐栖居其中的说明，从而第二方的诸反对被接收为该说明所指派给他们之性情的进一步证据。一名初级研究者对一个项目有所贡献，而在一个脚注中被致谢。一个国家加入一个它无份参与起草的规范框架，并随后由它对该框架的遵从被评估。在每一情形中，某样可辨认地是知识的东西在做某样可辨认地是权力的事，而在每一情形中，参与者们可能真诚地否认任何人在支配任何人。

关于知识与权力的两种说明随时可被应用于此，而两者都不足以对付这一困难。第一种把知识当作内容、把权力当作一种污染。在这一说明上，一套认识实践的本分是确保被辩护的信念，而权力作为一种外部扭曲进入，属于以下这种：其权威压制一个有效反对的著名科学家，或其利益扭曲一个发现的资助者，而更好的程序会把它排除。这一说明有其可取之处，即界定什么会算作事情进展顺利。然而它根本无从描述教学的情形，因为没有任何外部之物闯入，也没有任何程序被违反。第二种说明，在批判传统中、并首要地由福柯所发展，主张权力条件化什么可算作知识，从而两者即便在原则上也不可分。这更靠近该现象，并供出它的大部分语汇。然而它特征性的问题关乎控制：哪些话语被授权，谁的陈述被纳入，一个真理体制如何被构成并被维系。教学的情形不宜被描述为一件控制的事。教师控制的东西，没有一样不是在第一学期开始时、当同一关系是生成性的时候便已归教师控制的。

两种说明所共享的，是一个把权力当作某样被占有、被施行或被分配之物的构想：由一方持有，被应用于另一方，在数方之间分派。本文从一个不同的构想出发，而陈述它就是出发点。生成产生差异；差异产生不对称；而在一个不对称之上，一个位置维持自身的倾向是一种生成性的活动、而非一种静态的性情。一个进行判断的位置，必须持续地产生判断落于它的诸条件：诸提议由之被评估的诸标准、评估于其上发生的诸场合，以及那些已然被构成为提出者而非判断者的参与者。权力，在这一说明上，是已转向对它从中发出之位置的维持的生成性。它不是生成的反面，也不是它的缺席；它是其产物为它自己延续之条件的生成。这正是教学关系能在没有人获得任何东西、没有人行动有误的情况下败坏的缘由。曾一直返回各方共同持有之物的生成性，已开始返回发出它的那个位置。

权力在内在意义上是生产性的、而非仅仅是限制性的，这一想法并不新，而尤其有两个先例立于这一表述之后。尼采（1966）把权力设想为一种力量的释放，自我保存从属于它，而非设想为一个要被持有的量，从而一种不扩展自身的权力根本不是权力。拉康（1977）描述了一种其对象维系而不熄灭的欲望，围绕一个该对象所标记、而非所填满的匮乏而循环。每一个都供出成问题之结构的一部分。第一个主张权力必须持续生产才能存在；第二个主张一个配置可能再生产它所依赖的那个条件本身。此处所发展的说明在其范围上同两者都有别，而§3陈述这些差别。从它们取来的，是对占有模型的拒绝。

有一个词需要就标题中保留的术语作出交代。谈论一个权力结构暗示一种持有之物的安排，一幅谁占有什么的地图，而这正是方才被搁置的构想。这一术语因其熟悉而被保留，而它所指称的东西在下文中被重构：一个结构，在当下的说明上，是一个位置的生成性可转向其自身诸条件所经由之诸通道的一个配置，连同那一转向被维系的成本。它是一个结构，就它是确定的、可描述的，且稳定得足以被作用于其上而言；它不是一个结构，在各方之间持有之物之分配的意义上。

两个后果立即随之而来，并塑造此后的一切。第一个是，谁持有诠释权威这个问题是错误的问题，或至少是一个派生的问题。要紧的是一个不对称是否正在产生它自己延续的诸条件，而这可能在没有任何一方寻求它的情况下发生，且没有任何权威的重新分配本身能阻止它。第二个是，知识中的权力是有代价的。一个必须生成才能持续的位置能够未能生成，而它所产生的诸条件能够被来自别处的生成所撤销。这种权力不是一份一经获得便存续的占有物；它是一种一旦停止便失效的活动。

本文围绕三个问题组织。

RQ1。 一个认识不对称经由什么通道维持自身，而什么在一个给定情境中固定那些通道的配置？

RQ2。 在什么条件下，从一个认识位置发出的生成性返回共同的生成，而在什么条件下，它转向对其自身诸条件的生产？

RQ3。 鉴于认识不对称既不能被消除也不能被弄得无辜，对于共创性结构的设计，有什么随之而来？

第一个供出语汇，在§4–§6中被处理。第二个是本文的中心，在§7–§9中被处理。第三个是后果，在§10中被处理。

该说明可被陈述为单一的一个主张。

中心主张。 知识是一个诠释经由它被协调的关系结构，而每一个这样的结构都在其参与者之间生成不对称。在那一不对称之上，一个位置维持自身的倾向本身是生成性的，产生它自己延续的诸条件；知识中的权力就是这一自我维持的生成性。因而每一个知识系统都体现一个权力结构，而值得分析的问题关乎生成所取的方向，是返回共同的生成、还是进入它从中发出之位置的延续。

该主张分四步展开。第一步辨认一个认识不对称维持自身所经由的诸通道，从一个诠释为了存在必须经历的东西中取来它们：它必须被提出、被纳入、被认证、被修正，并被归属。第二步处理退出成本，它为这些通道中的每一个定价，因为一个位置在离开代价高昂之处廉价地维持自身。第三步区分生成可取的两个方向，并供出一条把它们区分开来的判据，即修正环路的闭合，它问一次修正的施行是否改变下一个诠释被生成于其下的诸条件，还是被吸收并作为修正者之缺陷的证据被返回。第四步处理知者的生产，这是一个认识位置维持自身的主要手段，也是统一本文所考察诸情境的东西。

有三样本文并不主张的东西应在开篇引入，因为这一领域中的工作通常被读作在主张它们。它不主张认识不对称是压迫：不对称是成问题之诸关系的构成性成分，而没有它便没有教学、没有诊断，也没有不平等者之间的协作。它不主张共创要求认识差异的抹平，那会废除这些关系、而非改善它们。而它不主张生成性权力是良性的。一位打开一名学生思考的教师，已塑造了那名学生将在其中思考多年的空间，而那一形塑性的不对称并不因其是生成性的便无辜；§8直接处理这一点、而非略过它。此处任何动态的发现也不应被读作一个政治立场；论证关乎诸关系之内生成的方向，而它不裁决任何具体的历史或制度轨迹。

其余部分如下推进。§2评述把知识当作内容的处理，以及伴随它的把权力当作外部污染的构想。§3评述尼采、拉康、福柯、布迪厄与行动者网络传统的批判性诸说明，并论证它们控制与占有的组织性概念留下本文的问题未被提出。§4发展把知识当作诠释关系之组织的构想，并论证共创是一般情形。§5铺陈自我维持的五个通道。§6处理退出成本。§7区分生成的两个方向，并陈述环路闭合的判据。§8处理生成性权力。§9处理知者的生产。§10引出设计后果，而一个结尾节陈述该说明的界限及它留作悬而未决的工作。

§2 知识与内容

把知识当作内容的传统，正是批判性诸说明借以界定自身的那个传统，而在划出它的界限之前，值得以它自己的术语陈述它，因为那些界限不是通常被指称的那些。该传统对探究的社会境况并非天真。刻画它的，是一种特定的分工：对知识的分析关乎一个信念被担保于其下的诸条件，而它的生产境况，无论多么有趣，属于一门不同的探究。

2.1 辩护与个体知者

经典的分析把知识当作既真又被辩护的信念，一种有着长久渊源与特定现代历史的分析。葛梯尔（1963）在三页中显示这三个条件是共同不充分的，因为一个信念可能既真又被辩护、而其真却归于偶然；随之而来的半个世纪的修补，包括加上第四个条件、可靠过程的要求，以及信念须是安全的或敏感的这一要求，并未产生一个已定的分析。戈德曼（1967）把担保定位在该信念的因果历史中，其后定位在产生它的过程之可靠性中，这把分析从知者所能援引的东西挪开，而挪向关于该信念如何得以形成的诸事实。

此处要紧的是一个内部争议未加触动的结构性特征。自始至终，分析的单位是一个主体与一个命题。知识是一个知者与一个内容之间的关系，而其他人的在场若说进入，也是作为一个证据来源，或作为一个信念被习得所经由的路径。社会认识论此后把这一点复杂化甚多。戈德曼（1999）按社会实践产生真信念的倾向来处理它们，而一大批关于证言的文献问一个听者在什么时候被担保去相信一个说者；这一复杂化是真实的。它尽管如此仍保留那一取向：他人作为一个知者之担保的一项输入而出现。各方之间的关系在一个诠释已在他们之间被接受的后果中变成什么，不是该框架所提出的问题。

2.2 中立性论题

从个体的分析中，一个进一步的论题通常被引出，而较少被明确陈述。倘若知识是被担保的内容，那么权力对它是外部的。内容是它之所是，无论谁断言它，而地位、利益或权威对一个共同体所相信之物的影响是一种扭曲，是认识实践的一个失败、而非它的一个特征。这一论题许可一个清晰的规范纲领：辨认非认识性影响经由之进入的诸通道，并关闭它们。盲审、复现要求、资助的披露，以及非个人论证的诸规范，是它的制度表达，而它们既不愚蠢也不无效。

这一论题的界限在本文开篇的情形中可见。在第二学期，没有任何外部之物闯入。没有权威被援引来压制一个反对；没有利益扭曲一个判断；诸更正是准确的，而诸程序完好。然而这一关系停止了生成。中立性论题没有资源来描述这一点，因为在它的术语上没有任何东西出错：真的内容被传递，担保被保存，而实践如设计地运作。所改变的东西在于各方之间的关系，而该论题把那一关系当作在认识上惰性的。

2.3 教与学的传递模型

同一取向塑造对教学的处理。倘若知识是一方所持有的内容，那么教学是它对另一方的传送，而学习是它的习得，一幅在课程设计中、在以回忆为形式的评估中，以及在传递材料、覆盖进度的寻常隐喻中存续的图景。弗雷勒（1970）把这命名为储蓄式构想，其中教师存入而学生接收，并论证它把学生构成为一个客体、而非一个参与者。弗雷勒的替代方案是一种教师与学生共同探究的教育学，而他的论证明确是政治性的，指向压迫经由学校教育的再生产。

储蓄式批判在它所辨认的东西上是正确的，且在一个方面被此处采纳、在另一个方面被背离。被采纳的，是这一观察：一个关于知识的构想携带一个关于各方之间关系的构想，从而传递图景界定每一方应是什么，而不中立地描述教学。被背离的，是困难的定位。在弗雷勒的说明上，储蓄式构想由一个压迫性的秩序所强加，且要被一个对话式的构想所替换。本文开篇的情形不是那种：第一学期是对话式的，同样的两个人在场，而没有任何秩序强加任何东西。一个曾在生成的关系停止了生成。储蓄式批判处理一个开始时便被采纳的结构；此处成问题的现象是一个开局良好之结构内部的一次方向改变。

§3 关于知识与权力的批判性诸说明

批判传统主张权力对知识不是外部的，而当下的说明同意。然而它的数个分支共享一个留下本文之问题未被提出的取向。本节先取生产性权力的构想由之下传的两个哲学先例，再取发展了它的三个社会理论说明，并以命名它们所共有之物结束。

3.1 尼采与权力的增加

尼采关于权力的构想不是一方所持有之一个量的构想。在尼采（1966，§13）中他提议一个活物首要地寻求释放它的力量，而自我保存是这一点的一个后果、而非它的目的，一个针对斯宾诺莎式与达尔文式的保存首要性的表述。这一学说直接关乎知识：尼采（1966，§259）把占取与压倒的说明从有机领域扩展到社会领域，而在别处他把求真意志本身当作同一意志的一种形式、而非它的中立工具。在这一构想上，权力被施行而非被持有，而一次不扩展的施行根本不是一次施行。

有两样东西从此被取来，而一样被搁置。被取来的，是对占有模型的拒绝：权力存在于它的活动中，并在活动停止时失效。同样被取来的，是权力与增加的耦合，它供出一个位置必须持续生产才能持续的那一意涵。被搁置的，是那形而上学的范围。尼采的是一个关于活物、或许一切物根本上是什么的学说；此处的主张更狭窄、并相应地更弱，主张自我维持是生成性在一个关系之内可取的一个方向、而非生成之为生成的本性。较狭窄的主张也更易处理，因为一个方向能被观察到改变，而一个根本的驱力不能。

一则关于来源的告诫是适宜的。求权力意志最可引述的诸陈述出现在《求权力意志》中，那是一部在尼采崩溃后从笔记中被集结而成、而非他所撰写的编纂物；上面的诸引用是他自己发表的材料，而该学说在那里被陈述。

3.2 拉康与欲望的辩证法

拉康关于欲望的说明供出第二个先例，而它的相关性是结构性的、而非主题性的。欲望，在这一说明上，不因其对象的获得而被熄灭。看似是它目标的那个对象，即小对象 a，标记一个匮乏的位置、而非填满它，而欲望绕过该对象而循环并继续（拉康 1977）。因而该配置在一个特定意义上是自我维系的：看似是该过程终点的东西，实际上是该过程借以再生产它所依赖之条件的手段。

该结构就是本文所成问题的那个，经过转置。一个进行判断的认识位置，由判断落于它之境况的持续再生产所维系，而非由它所交付的诸判断所维系，而每一个特定的判断，看似了结一件事，实际上更新那一境况。当下的说明同拉康分道之处，在于什么承载该结构。对拉康而言它是主体的构成性成分，并在无意识的层面被展开；此处它是一个关系配置的一条性质，因而在一个关系的行为中可观察，且可由对它的诸改变加以修正。这一分道有一个应被直白陈述的方法论后果：本文在§7中所提供的判据，意在可对照一个关系中所发生之事而被查核，而精神分析的说明，就它所描述之物的本性而言，并不承诺是可查核的。

3.3 福柯与可说者的条件化

福柯供出本文最直接地相对于其定位自身的说明，而这份亏欠是巨大的。福柯（1977）论证权力与知识直接相互蕴含，从而没有一个权力关系不伴随一个知识领域的相关构成，也没有任何知识不同时预设并构成诸权力关系。权力在这一说明上是生产性的、而非仅仅是压抑性的：它产生诸客体域、诸真理仪式，以及作为它之载体的诸个体。在后期工作（福柯 1978）中他告诫不要把权力当作一个主权者所持有的占有物或能力，而是把它描述为内在于其运作之领域的诸力量关系的一个多重体。考古学诸研究（福柯 1972）追寻那个相关的问题，即什么条件支配诸陈述在一个话语构型之内的出现。

当下的说明接受这一切，并问某样该框架不问的东西。福柯特征性的问题关乎一个体制的构成：一个可说者的领域如何被确立，它排除什么，以及由什么实践被维系。这是一个关于某种知识变得可能于其下之诸条件的问题，被安置在诸话语及其历史构型的层面。教学的情形不在那个层面被回答。同样的两个人，在一个话语构型之内，自始至终有同样被授权的语汇可用，从一个生成性的关系转到一个败坏的关系。关于该体制没有任何东西改变。所改变的，是其内一个位置的生成性所奔向的方向，而对诸体制的分析不下降到那里。

3.4 布迪厄与学术资本的累积

布迪厄把知识生产当作一个场域，其中诸行动者为一种它所特有的资本竞争。布迪厄（1988）把学术世界分析为这样一个场域，其中诸位置由其占据者所持有之资本的量与构成来界定，而布迪厄（1975）把科学场域当作一个其中对真理的追求与对科学权威的追求不可分的场域，因为权威由生产该场域承认为一份贡献之物而被赢得。该框架很好地说明了一系列中立性论题无法说明的现象：声望等级的持久性、诸性情经由训练的再生产，以及可思之物与被奖赏之物的对齐。

然而那组织性的隐喻是累积，而累积是一个占有模型。资本被持有、被投资、被转换、被传递，而分析借把诸行动者定位在一个持有之物的分配中而进行。就当下的问题而言有两个困难随之而来。第一个是，一个位置可能在不累积任何东西的情况下维持自身，一如在教学的情形中，其中教师的持有之物自始至终恒定。第二个是，该模型不给以下可能性任何自然的位置：一个位置的一次施行可能侵蚀那个位置的诸条件，而这恰恰是§8论证生成性权力所做之事。在一个累积隐喻之下，这样一次施行只是支出，而它同其他支出的结构差异无法被陈述。

3.5 拉图尔与行动者的招募

行动者网络传统比前述更彻底地拒绝把权力当作一种实体的处理。拉图尔（1987）论证一个陈述的命运在于那些采纳它的人手中，从而一个主张经由招募其他使用它、扩展它并在其上建构的人的工作而成为一个事实；拉图尔（1986）论证权力是集体行动的一个后果、而非它的原因，而借援引权力来解释一个结果，是在命名那要求解释的东西。卡隆（1986）铺陈招募借之进行的诸环节。因而该传统供出某样其他传统所不供出的东西：一个其中一个主张的地位被持续地造成、而非被持有，而其中相关的活动是持续进行的说明。

然而它的关切是诸陈述的命运与诸网络的构成。诸方在招募的过程中变成什么，一个参与者被构成为什么，以及那一构成是否是维系该网络的东西，不是所提出的问题，而该传统追随行动者、而非评估他们诸关系的方法论承诺，使这一问题从内部难以提出。§9论证某些种类之参与者的生产是一个认识位置维持自身的主要手段，而这是一个关于招募对被招募者做了什么的主张。

3.6 控制与占有的界限

跨其诸差异，所评述的诸说明共享一种给问题定框架的方式。无论知识是一个知者所持有的内容、一个行动者所累积的资本、一个授权某些陈述并排除另一些的体制，还是一个招纳同盟的网络，分析问谁控制什么，以及控制由什么手段被确立并被维持。这是一个好问题，而在支配是那个现象之处它是那个正确的问题。

它不是教学的情形所引起的问题。在那里，控制恒定而关系改变；没有人取走任何东西，而某样东西丢失了。该情形所要求的，是一个关于方向、而非分配的问题，问从一个位置发出的生成是正返回各方共同持有之物、还是返回该位置自身的诸条件。提出那个问题要求先说出知识是什么、以致其内一个位置根本能生成，而下一节转向那一点。

§4 诠释关系的组织

诸评述留下一个悬而未决的问题：知识必须是什么、以致其内一个位置根本能生成？本节回答它。这一回答是作为对该术语被当作挑出什么的一个转移、而非作为对所评述诸分析的一个驳斥而提供的，而这一转移的担保是它把一类其他构想所留下未加处理的现象带入视野。

4.1 诠释与未来行动的协调

知识，在此处所采纳的构想上，是数方的诸诠释经由它被协调、而它们的未来行动由此被条件化的一个关系结构。这一表述有三个部分，而每一个都做工。

它是关系性的，意味着分析的单位是一个诸方的配置，而不是一个知者与一个命题相对于一个诠释处在确定的诸关系中。说一个诊断是知识，就是就医师、患者、家庭，以及将作用于它之诸制度说某样东西，而不仅仅就一个陈述同身体的一个状况之间的对应说某样东西。它是诠释性的，意味着被协调的是对一个情境的一个读解、而非一个赤裸的内容：一个诠释选取一个情境中什么是突出的、什么从它随之而来，以及什么会算作它出错。而它条件化未来行动，意味着一个诠释在它所开启的序列中是有后果的：哪些进一步的诠释变得可用、哪些被封闭，以及每一方此后被定位去做什么。

这一构想不否认一个命题可能真或假，也不否认一个信念可能被担保或未被担保。那些问题依然存在，而§2中所评述的诸分析处理它们。该构想所否认的，是内容的真穷尽了知识在持有它的各方之间所做之事。一个真的诊断与一个假的诊断在重组一个家庭上无别；一个正确的更正与一个弄错的更正在把一方定位为更正者上无别。

4.2 关系效果的即时性

该构想携带一个把它同批判性诸说明、也同经典诸说明区分开来的后果。在批判性诸说明上，知识与权力之间的关系是发生学的或条件性的：权力塑造什么渐渐算作知识，而知识反过来维系产生了它的诸安排。这一关系是真实的、但被中介的，经由一个话语的历史构成或一个场域累积起来的诸位置而运行。在当下的构想上，该效果在一个特定意义上是即时的。诸能力凭一个诠释的被接受而在各方之间被分配，而这一分配是那一被接受的一个特征、而非它的一次后续应用。分配在诠释抵达一方时抵达那一方，对在它被接受时缺席的各方而言这可能更晚；所否认的，是这一重新分配等候诠释被作用于其上。

考虑医师关于一个患者有一个特定状况的陈述。该陈述，一经被接受，不仅仅描述。它确立谁有资格说该患者的诸感觉意味什么，患者自己对它们的报告算作什么，是证据、症状还是不遵从，哪些方要被咨询，家庭要预期什么，以及一个后续的分歧将如何被分类。这一切没有一样等候该陈述被作用于其上。它凭该陈述被接受为知识而成立，因为被接受为知识就是被授予这一协调之力。

教师关于一个被提出的读解是弄错的这个陈述，有同样的特征。它更正一个读解；而在同一动作中它更新进行更正的一方的地位，界定学生的下一个提议将被接收于其中的语域，并了结学生的分歧，若它到来，会是什么的一个实例。一位家长关于一个孩子为什么在哭的说明如此，而一方伴侣关于他们之间一个反复出现的困难实际上关乎什么的说明亦如此。

主张 1。 一个被接受的诠释凭其被接受、而非凭其后来的应用，在各方之间分配诸能力，并由此开始生成它继续被接受的诸条件。知识的关系效果属于它的构成。

该主张的后一半是就下文而言要紧的那一半，而它应被仔细地读。说一个被接受的诠释生成它继续被接受的诸条件，不是说该诠释在不可证伪的意义上是自我确证的。它是说，被接受确立一种关于谁评估、在什么场合、由什么标准的安排，而这一安排就是下一个诠释将遭遇的东西。该诊断确立后续的感觉报告将被接收为症状；该更正确立后续的诸提议将被接收为更正的候选。如此被确立的安排是否持续，是一个进一步的问题，而它是§7的问题。

4.3 共创的一般性

倘若知识是诸方之间诠释的协调，那么知识被共创于其中的诸情境不限于有组织的探究。本文处理四个，而选取它们是因为它们在几乎每一个方面都有别，除了成问题的那一个。

在教育中，教师与学生生成一份关于一门学科内容的说明，并且不可分地，一份关于学生在它上之能力的说明。第二份被生成，无论是否有人留意它，因为每一次对一个提议的评估也是对提出者的一次安置。

在亲密关系中，两方生成一份关于他们的关系是什么、他们各自在其内是怎样的，以及他们的共同历史曾是什么的说明。这份说明在此处所用的完整意义上是知识：它协调诠释、条件化未来行动，并了结一个后续的分歧将被当作关乎什么。它也，特征性地，在没有有组织的探究所供出的任何程序性保障的情况下被生成。

在国际关系中，诸国家与其他行动者生成关于一个情境是什么、一个规范要求什么，以及哪些方有能力去说的诸说明。关于规范出现的文献描述这类说明如何扩散并变得被视为理所当然（菲尼莫尔与辛金克 1998），而建构主义传统主张国际政治的诸结构由共享的诸理解所构成、而非仅由物质分配所给定（温特 1999）。当下的说明所添加的，是这些所不提出的一个问题：在一个共享理解被确立的后果中各方发生什么，以及如此被确立的安排是否继续容许来自它所定位者的修正。

在学术知识生产中，诸研究者生成关于一个领域的诸说明，连同关于谁在其内有能力的诸说明。评审、引用与作者身份的装置，以其他三个情境所不具的方式使第二份变得明确，这正是它供出最清晰证据的缘由，也是它吸引了§3中所评述之大部分理论注意的缘由。

此处所引入的主张是，这些是同一个结构的诸实例、而非在诸不同结构之间画出的诸类比。每一个都涉及诸方协调诠释；在每一个中，协调即时地分配诸能力；在每一个中，一个被确立的安排条件化下一个诠释将遭遇什么。因而学术生产是四个之中的一个实例，而它的诸安排被记录在案这一事实并不使它成为范例情形。临床、家庭与外交的诸实例较少被记录、而并不较少被结构化。

主张 2。 共创是知识的一般情形，而有组织的探究是它的一个特殊情形。教育、亲密关系、国际实践与学术生产实例化单一的一个结构，即诸方之间诠释的协调连同其对诸能力的即时分配，而它们在那一结构的配置上、而非在种类上有别。

那一配置由什么构成，迄今被留作未加界定。诸能力已被说成被分配，而没有一个关于有哪些能力、或为什么是那些的陈述。供出这一界定是下一节的工作，而该界定要被推导、而非被列举：诸通道要从一个诠释为了根本存在而必须经历的东西中随之而来。

§5 一个诠释的生命历程

一个诠释，为了作为知识存在于诸方之间，必须经历一个确定的历程。它必须被某人提出；它必须被采纳、而非被略过；它必须被发现成立；它必须在随之而来者的照映下可被更改；而它必须被指派给一个来源。每一个阶段都是各方相对于彼此处在一个不对称关系中的一个点，而每一个因而是一个不对称可维持自身所经由的通道。随之而来的这份名单从这一历程被推导，并由它闭合：诸阶段是五个，因为一个诠释的历程有五个，而一个第六个会要求一个诠释生命中一个尚未被命名的阶段。

关于这一推导所主张之物的一个词。并不主张这五个在没有连贯的替代方案能被构想这一意义上是逻辑必然的，也不主张它们是这些现象能被描述于其中的唯一语汇。所主张的，是它们从一个关于为使一个诠释在诸方之间运作而必须发生之事的极简描述中随之而来，它们对本文所处理的诸情形是共同充分的，而相邻诸文献所提供的诸列举，通常借在突出的诸不对称被注意到时把它们列出而进行，缺乏一条凭之被完成或被闭合的原则。一个愿添加一个第六个的读者，被邀请去命名它所对应的阶段。

5.1 提出

一个诠释必须先被提出。这一阶段的不对称关乎谁可推进对一个情境的一个读解，以及由每一方推进的一个读解在什么语域中到来。语域与资格一样要紧。一名学生可能在形式上自由地提议，而每一个提议都作为更正的候选到来；一名患者可能描述一种感觉，而该描述作为等候诠释的材料、而非作为一个诠释本身到来。推进一个读解与为另一方的读解供出原材料之间的区分，是这一通道的实质，而它不由许可来了结。

一个不对称在此借事先条件化语域而维持自身。当一方的诸贡献习惯性地被接收为数据时，接收的位置被每一次这样的接收所更新，而贡献的一方学会该语域，并供出适配于它的诸贡献。任何一方都无需意欲这一点。接收前后一贯便已足够。

5.2 纳入

一个被提出的提议必须被采纳。这一阶段的不对称关乎哪些提议根本进入考虑：不是哪些被接受，而是哪些被当作已被作出。一个提议可能被以沉默、以一个话题的转变、以一个不予参与的承认，或以它不是当前关切这一评述所接遇；在每一情形中它已被提出而未被纳入。

这一通道是最易被忽略的，因为不纳入留下很少痕迹。一个被拒绝的提议作为被拒绝而在案；一个从未被采纳的提议作为无物而在案。其后果是，一个配置能在这一阶段高度限制性、而在下一阶段显得开放，因为抵达认证的诸提议恰恰是那些已被过滤的。一个不对称在此以任何通道中最少的支出维持自身，因为它无需论辩。

5.3 认证

一个被纳入的提议必须被发现成立或不成立。这一阶段的不对称是最容易被辨认为认识权威的那个：谁决定一个诠释是否立得住，以及由什么标准。它的两个特征关乎论证。

第一是，标准与它们的应用在实践中不可分。一个其中标准是明确且公开的配置，同一个其中它们驻留在一方判断中的配置有别，而这一差别在于第一个经由某样另一方能检视并援引之物来施行它的不对称。

第二是，认证比其他通道更甚地产生一个持久的痕迹。一个被认证的诠释此后作为一个进一步诠释由之进行的既立点而可用，而认证了它的位置由此被更新。这是自我维持最直接的通道：认证就是再生产认证落于人之境况。

5.4 修正

一个既立的诠释必须在随之而来者的照映下可被更改。这一阶段的不对称关乎谁可重开已被了结之事，而且，更有后果地，一次重开的尝试被接收为什么。

后一半正是实质所在之处，而§7专门处理它。修正在形式上可用，这几乎是普遍的：学生可以反对，患者可以寻求第二意见，初级作者可以质疑那个定框，伴侣可以说关于那个反复困难的说明是错的。所变化的，是那尝试改变该安排、还是被接收于它之内，被当作反对者有限领会、患者焦虑、初级者缺乏经验，或伴侣特征性困难的证据。在后一种情况下，修正已被施行，而该安排已被这一施行所更新。这是一个不对称最强有力地维持自身的通道，恰恰因为这一维持经由名义上被提供来制衡它的那个机制运作。

5.5 归属

一个诠释必须被指派给一个来源。这一阶段的不对称关乎谁的贡献被记录为一份贡献，而它是有着最清晰制度表达的通道：作者身份、作者顺序、致谢、引用。它在其他情境中有较少的形式表达，而在那里运作得同样不少，一如在家庭中一方的建议无余数地成为家庭的决定，或在谈判中一个提议以另一个代表团的名义被采纳。

归属的后果超出功劳，因为它条件化下一轮提出时的语域。一方其贡献被记录，则在下一次提出时作为一个已有贡献者到来；一方其贡献被吸收，则作为一个未有贡献者到来。因而该通道闭合回到第一个通道之上，而一个经由归属维持自身的不对称，借塑造每一方在下一场合将是什么而如此做。

5.6 五个通道的配置

五个通道不是一个层级，也不是一个严重性递增的序列。它们是一个不对称可被维持的诸不同点，而一个给定情境独立地配置它们。自由地纳入而狭窄地认证的研讨班，同狭窄地纳入而按公开标准认证的研讨班有别；仔细地归属于患者自己报告的诊所，同不这样做的诊所有别，无论它们在认证上多么相似。配置是分析的对象，而下面的表格记录每一个通道在四个情境中所采取的形式。

	教育	亲密关系	国际实践	学术生产
提出	一名学生的读解在其中到来的语域	说这关系是怎样的地位	说一个情境是什么的地位	界定一个研究问题的地位
纳入	哪些问题被当作问题	哪些委屈被采纳	哪些事项抵达议程	哪些投稿被送去评审
认证	评估及其标准	谁对一个共享事件的说明成立	遵从的判定	同行评审及其判据
修正	对一个更正之挑战的接收	对该说明之一个反对的接收	一个已定规范的重开	对一个既立结果之挑战的接收
归属	对一名学生贡献的认可	对那是谁的建议的记录	一项条款在谁的提议之下	作者身份、顺序、致谢

该表格是作为一个分析格网、而非作为一个发现而提供的。它的用处是使诸配置可比较：两个在其形式安排上显得相似的情境，可能被发现在一个特定通道上有别，而一个显得开放的情境，可能被发现在纳入上限制性、而在别处不限制性。

主张 3。 一个认识不对称维持自身所经由的通道是五个，即提出、纳入、认证、修正与归属，并由一个诠释为在诸方之间运作而经历的历程所闭合，在上面所辩护的极简描述上。诸情境在这些通道的配置上、而非在它们的数目上有别，而一个配置可能在一个通道上限制性、而在其他通道上不限制性。

这一列举留下两个悬而未决的问题。第一个是，什么决定一个不对称经由这些通道维持自身的成本，而这不由通道本身给出；一个在所有五个上都相同的配置，可能在一个情境中容易承受、而在另一个中不容易。第二个是，什么把一个其生成性返回各方共同持有之物的配置，同一个其生成性返回自身的配置区分开来，而诸通道记录它、却不解释它。第一个被接下来处理；第二个在§7中。

§6 退出与认识能力的价格

五个通道记录一个不对称如何被维持；它们不记录这一维持耗费什么。两个在每一个通道上都相同的配置尽管如此可能有别，而把它们区分开来的变量在诸通道之外：对每一方而言，全然离开这一关系的成本。本节处理那一成本。它不是一个第六通道，因为退出不是一个诠释生命中的一个阶段；它是五个通道运作于其下的条件，而它为它们定价。

6.1 离开的成本

赫希曼（1970）区分了一个衰落中的组织的成员可用的两种回应：退出，即参与的撤回，以及呼声，即从内部改变事态的尝试。就当下的目的而言他的核心观察是，两者交互。退出的可获得性条件化呼声的特征与有效性，而一个离开容易的组织面对一种不同于其成员被拴住之组织的纪律。赫希曼的关切是诸公司与诸组织对衰落的响应性。这一观察带着一个修改转移到认识的情形，而那修改正是要点。

当离开代价高昂时，一个认识位置以低成本维持自身。它无需说服、无需交付、无需持续生成任何另一方发现值得留下之物，因为另一方的留下不是对该位置所供出之物的一个判断。当离开代价低廉时，同一位置必须昂贵地维持自身：它只在它持续生成某样拴住各方之物的期间持续，而生成的一次失效被以离去回应。回想在这一说明上权力是转向其自身延续之诸条件的生成性，退出成本设定那一转向的价格。一个高的退出成本是一份补贴。

这一表述得出一个比它初看更强的主张。不是被囚禁使一个认识不对称成为压迫性的；而是被囚禁容许一个不对称在不生成的情况下持续。同一位医师、同样的标准、同样的态度，在一个患者可以咨询另一位的情境中，与在一个患者不可以的情境中，产生一个不同的结构，而这一差别在于该行为为被维系必须成就什么，而不在于该医师在五个通道中任何一个上的行为。

6.2 亲密关系与组织中的被囚配置

亲密关系是退出成本最高且最不可辨认的情境，因为该成本主要不是物质的。离开就是失去一份共享的历史、一份关于自己的共同说明，以及多年间构筑起来的诸诠释资源；而各方之间所持有的关于这关系的说明，本身在会被失去的东西之中。一方已渐渐被定位为误解的那一个，不容易离开，部分因为离开会要求那一安排恰恰已扣留的诠释地位。这正是把一个寻常的不对称关系同一个已变得闭合的关系区分开来的结构：一个不利配置与一个移除改善它之压力的退出成本的合取。

组织情境提供更可辨认的诸实例。一名博士生其导师是该场域在一个特定机构的首要人物，面对一个以年计、并以推荐信所赋予之地位计的退出成本；一名初级研究者其签证系于一个雇主，面对一个以居留权计的退出成本。在两者中，五个通道的配置可能同一名有替代选择的学生或研究者所面对的没有不同，而结构是不同的，因为其中没有任何东西必须被挣得。

这一点不限于能力较少的一方。退出成本双向而行，而一个空出来会代价高昂的位置，例如一位其资助取决于一名学生完成的导师，或一方其关于这关系的说明取决于另一方佐证的伴侣，是一个处在生成之压力下的位置。当两方都会发现离开代价高昂时，该配置从两侧被固定就位，而这正是最长寿的、并且既最生成性又最败坏的诸关系被发现的条件。

6.3 不对称同盟与一个外部的缺席

国际的情形展现一个值得区分的变体，因为它关乎该分析能在人际之外主张什么。一个处在一个不对称安全安排之内，或处在一个它未曾设计之货币或贸易秩序之内的国家，面对一个高的、并在某些实例中未定义的退出成本：可能没有一个外部可离去到。当一个规范框架已变成任何种类的主张被弄得可理解于其中的媒介时，从它撤出便丧失根本作出主张的地位，且抵达不到替代方案。

这是一个代价高昂之退出的极限情形，而它显示该变量在其极端处做什么。一个没有外部的框架在不生成的情况下维持自身，因为它所定位的各方没有一个其表现可对照被评估的比较，也没有一个离去会抵达的目的地。这其中没有任何东西蕴含这样一个框架是恶性的，而该分析不许可那一推断：一个框架可能既不可逃脱又是生成性的，而一个外部的缺席只是说它的生成性不由离去的可能性所强制执行。它必须转而由什么所强制执行，是§7的主题。

关于把一个为个人与公司发展的概念转移到诸国家的做法，一则告诫是应有的。一个国家的退出不是一个单一行动者的决定，而离去的成本不均匀地落在诸国内选民身上；该情形是作为那个结构变量的一个实例、而非作为一个关于诸国家实际上如何决定的主张而提供的。

主张 4。 离开的成本为五个通道定价。当退出代价高昂时，一个认识位置在无需生成任何会拴住各方之物的情况下维持自身；当退出代价低廉时，同一位置只在它持续生成的限度内持续。因而一个高的退出成本是对该配置已然含有之任何不对称的一份补贴，而非一个额外的不对称。

该主张辨认从一个关系之外强制执行生成的东西。它把什么可从内部强制执行它这个问题留作未加处理，而这个问题在退出代价高昂之处成为全部的问题，也就是说在最要紧的诸亲密关系、诸形塑性教育关系与诸包容性国际框架中。那个问题是本文的中心，而下一节接手它。

§7 吸收与修正环路的闭合

留作悬而未决的问题是什么从一个关系之内强制执行生成。本节回答它，而这一回答要求先更正前面诸节可能已暗示的一幅图景。这一更正是，吸收与自我维持是一个单一生成性可取的两个方向、而非两种相争的力量、一善一恶。这一点一经陈述，一条判据便可被给出以把它们区分开来，而这一判据是本文主要的分析提议。

7.1 一个生成性的两个方向

当一个不对称存在时，它所界定的位置生成。它产生适配于它的诸标准、诸场合、诸区分与诸参与者，而它必须持续地产生它们，因为一个停止产生它们的位置不再是一个位置。问题是那产物去往何处。

它可能返回各方共同持有之物。一位认证一名学生读解的教师，在那一动作中产生一个标准；而该标准可能是一个学生此后能应用、能争辩并能扩展的标准，在那种情况下认证所生成的东西已进入两方共同持有之物，并在下一次提出时对学生可用。或者它可能返回该位置。同一次认证可能产生一个驻留在教师判断中、被逐案应用、且除了作为一个要被服从之物外对学生不可用的标准，在那种情况下认证所生成的东西是认证落于教师之被更新的境况。

在动作的层面上，几乎没有什么把这两者区分开来。两者都是认证，两者都可能准确，两者都可能以同等的用心被交付。它们在它们所生成之物的目的地上有别，而这一差别累积。正因这一缘由，本文开篇的情形不涉及任何获得、也不涉及任何不当行为。第二学期的教师并没有在做任何第一学期的教师未曾做的事。曾一直奔入共同持有之物的生成性，正奔入教学位置的诸条件，而学生在每一次提出时遭遇一个前一次交换所更新的安排。

吸收这一术语命名第一个方向：一个关系所生成的不对称被采纳进各方的共同生成，从而本会巩固一个位置之物成为两者都可用的一份持有。自我维持的生成性，在§1所给的意义上，命名第二个。吸收不是一个不对称的压制，也不是它的均等化。教师自始至终仍是较有能力的一方，而不对称未被削减；所改变的，是它的生成性被持续地返回。相应地，败坏是这一方向的转向、而非某样新东西的出现，而它无需一个行动者。

7.2 可获得之修正的不充分

此处一个自然的提议是，使一个关系保持生成性的，是可修正性：一个不对称只要保持对挑战开放便可容忍，而在它变得不可挑战时败坏。这一提议接近正确，而如通常所陈述的，几近空洞。

它几近空洞，因为修正几乎总在形式上可用。学生可以反对；患者可以寻求另一个意见；初级作者可以质疑那个定框；伴侣可以说该说明是错的；国家可以提出一项保留。全然不容许任何挑战的诸配置是罕见的，而它们不是有趣的情形。有趣的情形是那些其中挑战被容许、被施行、而什么也不改变的情形，缘由是它被接收、而非它被驳回。

考虑一个被告知一个被提出的做法为其自身之好而正被禁止的孩子，该孩子回答说那个理由不是真正的理由。这一回答是一个挑战，它被容许，而它被作出。它被接收为担保了原初判断的那种不成熟的证据。该安排未被这一挑战撼动；它被它所喂养，因为该挑战供出一个新的场合供那个判断被施行并被确证。被告知他们在防御、而回答说他们没有在防御的伴侣，遇到同一结构。其对一个更正的反对被接收为尚未理解该更正的学生亦然。

这一结构值得被精确地命名，因为它正是使可获得性成为错误判据的东西。在每一情形中，进行挑战的一方被要求产生某样东西，无论是一个反对、一个理由还是一个否认，而该安排被配置得使无论什么被产生都算作被挑战之位置的进一步材料。挑战者无法拒绝回应，因为沉默也被接收为确证。修正已被施行，而这一施行已更新它所针对的东西。

7.3 开环与闭环

前述所指示的是，该判据必须关乎一次修正之施行的效果、而非它的可获得性。它可被陈述如下。

环路闭合的判据。 一次修正的施行闭合环路，当它改变下一个诠释被生成于其下的诸条件时：诸提议由之被评估的诸标准、它们被接收于其中的语域、评估于其上发生的诸场合，或诸方去评估的地位。它使环路保持敞开，当它被接收于那些条件之内、并作为确证它们的材料被返回时。

该判据的三个特征应予注意。第一是，它关乎条件、而非结果。一个被认真考虑并被发现弄错的挑战可能闭合环路，只要它被发现弄错所依据的诸根据此后对挑战者可用；一个在特定之点上被让步、而挑战的地位被留作未加触动的挑战，可能使环路保持敞开。正确既非必要也非充分。

第二是，该判据在原则上是可查核的，而这正是§3把这一说明同精神分析的说明区分开来所依据的根据。它所探问的，是下一次交换的诸条款是否有别于上一次的诸条款，而这是一件可观察之行为的事：一个曾被援引的标准是否成为另一方可援引的标准，一个被授予的让步是否成为一个可被援引的先例，一个被承认的反对是否改变反对者此后可无需前言便提出的东西。这一查核并不平凡，而本文不提供关于它的任何程序；所主张的是该判据指涉某样可被寻找之物。

第三是，闭合容许程度，且是一个序列、而非一个动作的性质。没有单一的交换了结一个关系的修正环路是否闭合，因为一次施行的效果出现在随其而来者中。该判据被应用于一连串交换，而它所侦测的是一种倾向。

7.4 无获得的败坏

现已集结的说明解释本文开篇的现象，而值得明确地铺陈这一解释，因为它是该说明的主要证据。

在第一学期，教师的诸更正闭合了环路。每一次更正使一个学生此后能应用的标准对学生可用：该答案被评估所依据的根据、而非该答案本身，从而学生带着更多可供提出之物抵达下一次提出，而教师抵达它时面对一个其提议要求一种不同接收的一方。不对称未被削减，而它的生成性正返回各方共同持有之物。

在第二学期，环路敞开了。诸更正保持准确，而态度不变；所更改的，是评估的根据停止了行进。每一次更正了结一个案例，并更新教师了结案例之境况，而学生，遭遇这一点，学会了那个语域，而这正是那带防备的提议出现之点，因为一个作为对更正之请求而被给出的提议，是一个适配于它将遇到之安排的提议。任何一方都没有获得任何东西。没有任何程序被违反。生成的方向转了向，而该安排此后经由学生自己对它的适应而再生产自身。

主张 5。 吸收与自我维持是一个单一生成性的两个方向、而非两种相对立的力量。一个配置吸收它所产生的不对称，当从一个位置发出的生成性返回各方共同持有之物时，而在那一生成性返回该位置自身诸条件时败坏。因而败坏既不要求获得也不要求不当行为，而它的判据是修正环路的闭合。

有一个蕴含应被引出，因为它关乎§10的设计后果。倘若败坏无需一个行动者，那么没有任何配置由不对称所偏爱之一方的善意所确保，也没有任何配置由挑战的形式可获得性所确保。确保一个配置的，是其内修正的施行有持续的效果，而这是该安排、而非任何人性情的一条性质。一位其更正不行进的慷慨教师，与一位不慷慨的教师同样确定地维持一个位置，且相当更有效地，因为那慷慨遮蔽了闭合。

§8 生成性权力

迄今的说明已处理五个通道、它们被持有的价格，以及区分一个位置之生成性可取之两个方向的判据。它尚未命名一个位置在其生成性奔向另一方向时做什么。那是本节的主题，而它所引入的概念是本文押注其贡献于其上的那个。该概念也是最易被误读的那个，而那一误读正是本节最后一部分的缘由。

8.1 提议空间的扩展

五个通道是一个给定诠释生命中的诸阶段。一个位置可作用于它们，而迄今所处理的一切都关乎这样的作用。一个位置也可作用于某样在先之物：诸诠释根本可从中被抽取的那个空间：什么可被当作对一个情境的一个候选读解、什么可作为关于它的一个问题被提起，以及什么会算作一个关乎它的考虑。

这是一种不同种类的施行，而这一区分不是一个程度的区分。认证一个提议就是在一个已被构成的提议空间之内行动。更改什么可被提议就是作用于那个空间的构成。一位把一个读解标记为正确的教师在第一个语域中运作。一位显示一个学生当作已封闭的问题是敞开的、或一个学生当作不相关的考虑关乎此事的教师，在第二个语域中运作：学生此后有一系列此前不可用的提议可用，而这一可用性不取决于教师的进一步参与。

把在这第二个语域中行动的能力称为生成性权力。它在本文所处理的每一个情境中被施行。一位重新定框一个场域的中心困难是什么的导师，已扩展了一名学生可从事之物。一方以两者此前都未曾用过的术语表述一个困难的伴侣，已使一种描述他们关系的方式对两者可用，而两者此前都不具有它。一个把一个范畴引入一场谈判、供出一种刻画各方此前未能刻画之争端的方式的代表团，已扩展了此后可由他们中任何一方提议之物。在每一情形中，某样东西被弄得可能、而非被了结。

8.2 生成性权力中的权力

此处的诱惑是把生成性权力当作权力的良性形式，并围绕促进它来组织本文的规范内容。那一处理必须被拒绝，而拒绝它是本小节的负担。

生成性权力是本文赋予该术语之完整意义上的权力，而它是它最有后果的形式。它所做的，是塑造一方的未来诠释将从中被抽取的那个空间，而一个空间，一经被塑造，条件化从它抽取的一切。一名学生其关于什么构成一个有趣问题的感觉由一位导师所形塑，将多年地提出带着那一形塑的诸问题，在那位导师永不会看到的诸情境中，并将作为他们自己的问题提出它们。一方从另一方习得了一套关于这关系之语汇的伴侣，将以那套语汇描述这关系，包括在描述另一方有什么不对时。一个其范畴被采纳的代表团，已塑造后续各方，包括尚未在场的各方，能说什么。

那形塑性的不对称大于、而非小于在五个通道上被施行的任何不对称，而它与它的成功成比例地更大。一次更正是有界的：它了结一个案例，且能被重访。一个被重塑的提议空间是无界的，且特征性地对它所形塑的一方不可见，因为已被弄得可用之物显现为事情如何是、而非某人的贡献。一个形塑是生成性的，这并不使它无辜，而被形塑的一方没有一个可从中评估一个他们用以思考之形塑的立足点。

主张 6。 生成性权力是权力。它作用于诸诠释可从中被抽取的那个空间、而非作用于一个给定诠释的命运，而它的诸效果因而比在五个通道上的任何施行都更广泛、更长寿，且对它们所形塑的一方更不可见。它不是权力的良性替代者，且不给施行它的位置赋予任何无辜。

8.3 互惠能力的判据

倘若生成性权力不由无辜来区分，它必定由某样结构性之物来区分，而它确是如此。一次生成性施行的标志是它侵蚀施行位置所依赖的诸条件。

更正者的位置取决于有一方要求更正，而每一次了结一个案例、却不使评估的根据可用的更正，都更新那一依赖。一次生成性权力的施行奔向另一方向：一位使一系列学生可从中抽取的提议、以及诸提议被评估所依据之根据可用的教师，已产生一个更少要求教师的一方。这一施行消耗它自己的条件。当这样的施行继续时，教师的位置对学生的提出变得逐渐更不必要，而这一关系或以学生的独立终止，或以必须被重新构成的诸条款继续。

这供出那个结构性判据，而它应被以一个不坍缩进较早那个的形式陈述。环路闭合，在§7中，关乎一个给定的修正施行发生什么。当下的判据关乎施行位置跨一连串施行发生什么：它的必要性是被更新还是被消耗。两者相关，因为一个其修正环路闭合的配置将倾向于消耗认证位置的必要性，而它们不相同，因为一个位置可能使诸根据可用、而出于它自己的诸缘由仍是必要的，而一个位置可能被同修正毫无关系的诸境况所消耗。吸收，在§7的意义上，相对于两者是那个更宽的概念：一个配置吸收，当从一个位置发出的生成性返回共同持有之物时，而一个位置之必要性的消耗是这样一次返回可能有的一个后果，尽管不是唯一的一个、也不是它的一个条件。

该判据也正是使生成性权力有代价的东西，而这正是它不是默认的缘由。一个位置借了结案例而廉价地维持自身；它借使案例被了结所依据的诸根据可用而耗费自身。该说明中没有任何东西暗示这一耗费比这一维持更自然。该说明暗示相反：自我维持是一个位置在没有任何东西抵消它时所做之事，而生成性施行是必须对抗那一倾向而被维系之物。

8.4 形塑性不对称的分量

有两个后果随之而来，该说明应陈述、而非略过它们，因为两者都令人不适、且两者都被蕴含。

第一个关乎感激。一方由一次生成性施行所形塑，特征性地把那一形塑视作一份礼物、把形塑者视作一位恩人，而这一看法是有担保的，因为某样本不会存在之物被弄得可用。它也，在结构上，是一个成功到不再被察觉为不对称之不对称的残余。学生关于一位导师打开了那个场域的感觉，既是真的、又是学生无法从内部评估之一个形塑的痕迹。此处没有任何东西暗示那感激是错置的。所暗示的，是它的在场不是那一形塑曾是良性的证据，因为那一形塑恰恰是会产生它被评估所从出之立足点的东西。

第二个关乎互惠能力判据的界限。一个位置可能在一个方面侵蚀它自己的必要性、而在另一个方面确立它：一位使一名学生在方法上独立的导师，可能由此使该学生依赖于一种看什么值得留意的方式，而第二种依赖比第一种更深、且更不可辨认。因而该判据要带着对必要性正被消耗于其中之方面的一个界定被应用，而一个关于一个关系是生成性的整体判断，从它是不可得的。该判据所得出的是局部的：在这一方面，该位置正耗费自身；在那一方面，它没有。

因而该说明不交付一类被认证为生成性的关系。它交付一个可就一个关系、在一个被界定的方面、在一个给定时间被提出的问题，问从一个位置发出的生成性是正消耗还是正更新那个位置的必要性，连同一个该答案将跨诸方面有别并随时间改变的预期。这少于一个好关系的判据，而它正是该分析所支撑的。

§9 知者的生产

该说明已把一个位置的生成性当作产生诸标准、诸场合与诸区分。它产生某样进一步之物，而那进一步之物是一个认识位置维持自身的主要手段：它产生确定种类的诸参与者。本节铺陈那一机制，而它正是使四个情境成为同一个结构的诸实例、而非平行诸例证的东西。

9.1 认识位置的形成

五个通道的一个配置不遭遇已然被构成为提出者与评估者的诸方。它借接收的累积而构成它们。一方其贡献一贯地被接收为材料、而非被接收为读解，在下一场合作为一个供出材料的一方到来；一方其反对一贯地被接收为一种性情的诸实例，作为一个有那种性情的一方到来。这一构成经由诸通道的寻常运作而运作，无需意图，且在不被宣告的情况下进行。

它的效力在于该方自己的适应。一名已学会诸提议被接收于其中之语域的学生，将在提议之前带防备，而这带防备是能力、而非顺从：它是对该安排的一个正确读解，而一个没有它而被给出的提议会被更糟地接收。该安排所产生的，是一方其娴熟的参与再生产它，而不是一方服从它。这正是一个配置能在没有来自它所偏爱之位置之维持的情况下持续的缘由。另一方维持它，娴熟地，且在他们把自己察觉为什么、而非他们被告知去做什么的层面上。

因而该机制回答一个在§7中留作悬而未决的问题：一个生成的方向，一经转向，如何维系自身。它维系自身，因为它产生对之而言那一方向就是该关系之形状的诸方，而第五个通道闭合那个回路，因为归属已记录谁贡献之后，每一方在下一次提出时作为那记录所使他们成为者到来。

9.2 学生、初级作者、规范接受者

该机制可在本文所处理的诸情境中被观察，而它的诸实例不是类比。

在教育中它产生提问糟糕的学生。这样一名学生不是一名知道得更少的学生；知识赤字是真实的，而它不是成问题之物。成问题的是，该学生已被构成为一名其问题是更正的场合、而非贡献的学生，而在被如此构成之后，产生适配于那一接收的诸问题，那些问题更狭窄、带防备，且旨在确保一个更正、而非打开一件事。一个在这一阶段遭遇该学生的观察者，将发现那一构成被证据所确证，因为该证据正是那一构成所产生之物。

在学术生产中它产生作为一个致谢的初级作者。该机制经由归属运行，而它的效果在提出处：一名研究者其贡献已被记录为协助，在下一次协作时作为一个助手到来，被指派助手被指派之种类的工作，并产生被正确地描述为协助的诸贡献。记录与实情汇合，而这一汇合是该机制在起作用、而非该记录曾准确的证据。

在亲密关系中它产生进行误解的一方。一份把一种特征性的理解失败指派给一方的关于这关系的说明，若被维系，将产生一方，该方在推进其理解之前查核它，在关于共享历史的有争议事项上顺从，并把他们自己对一个插曲的读解当作那个要求验证的读解，而这正是一个进行误解的一方会做的。

在国际实践中它产生规范接受者。一个被定位为在别处起草之诸标准之接受者的国家，发展一个接受者的制度装置：用于遵从与报告、而非用于起草的诸办公室，展示合规、而非定框合规应由什么构成的专长。在发展了这些之后，它作为一个接受者参与，而它的参与是娴熟的。

弗里克（2007）命名一个邻近的现象，而与它的关系应被精确地陈述。弗里克的证言不义发生在一个听者因偏见而给予一个说者少于他们应得之可信度时，而她把这一伤害当作对一个人在其作为知者之能力中所做的一件错事，包括，在她称为该伤害之持续效果之物中，一个被如此对待的说者可能失去信心并停止发展其认识能力的可能性。此处所描述的机制同那一说明重叠，而在其来源上有别。弗里克的诸情形转在经由一个听者的可信度判断运作的身份偏见上，而她的诊断相应地关乎刻板印象与听者的伦理。此处所描述的生产无需偏见、也无需身份标记：它仅从诸通道的配置随之而来，而会发生在彼此不怀任何偏见的诸方之间。这两个机制可能复合，而在它们复合之处其效果是严重的；当下的主张是那个结构性机制在它们的缺席中运作。

9.3 一个机制跨四个领域

四个实例共享一个形式。在每一个中，一个配置产生一个种类的一方；该方娴熟的参与再生产该配置；而这一再生产供出该方是那个种类的证据。把这些实例区分开来的，是生产主要经由之运行的通道，无论是教育中提出处的接收、学术生产中的归属、亲密关系中对一个共享说明的认证，还是国际实践中的纳入与标准设定，而不是那个机制。

这是§4中所引入之主张的根据，即这些情境实例化一个结构。它们在其内容、其规模、其持续时间或其形式性上不相似。它们相合于此：每一个都是诸方之间诠释的一个协调，每一个配置即时地分配诸能力，而每一个配置，被维系，产生其参与维系它的诸参与者。

主张 7。 一个认识位置主要借产生确定种类的诸参与者而维持自身。这一生产经由诸贡献的累积接收而运作，无需意图、也无需偏见，且由被产生之一方自己娴熟的参与所完成，那一参与再生产该配置并供出该方正如该配置所构成他们的证据。

该主张携带一个关于一个补救会必须是什么的后果。一个已产生其诸参与者的配置，不由一个单一通道上的一次调整来更正，因为诸参与者将作为该配置所使他们成为的诸方遇到那一调整。被给予发言权的学生将用它来请求一个更正，而被给予作者身份的初级作者将贡献一个助手所贡献之物。这样一个配置要求什么，被接下来处理。

§10 生成性知识结构的设计

该分析为知识被共创于其中之诸情境的安排携带诸后果，而本节引出它们。它开篇即被§9约束：一个已产生其诸参与者的配置，不由一个单一通道上的一次调整来更正，因为诸参与者作为该配置所使他们成为的诸方遇到那一调整。因而随之而来的是少数几个条件、而非一份措施清单，每一个都带着它借以起作用的机制以及它被挫败的诸方式而陈述。

关于整体定位的一个先行之词。该说明不以认识不对称的消除为目标，那会废除成问题的诸关系、而非改善它们，而它不把任何安排认证为生成性的，缘由在§8中给出：诸判据所得出的是在被界定诸方面的局部判断。下面的诸条件是一个配置可被推向的诸方向，而每一个都有一个设计者必须对照它所确保之物加以权衡的诸代价。

10.1 环路闭合与设计

第一个条件随§7而来：一次修正的施行应改变下一个诠释被生成于其下的诸条件。使这成为一个设计条件、而非一个劝勉的，是那一改变能被内建进一个安排、而非被留给接收挑战之一方的性情。

那一般的机制是，一次评估的诸根据随它行进。当一次认证伴随它所依据的标准、被陈述得使另一方此后能援引它时，这一施行已增添于共同持有之物；当它作为一个判断被交付时，它已更新诸判断被交付之境况。这是，在教学中，一次更正与一次学生此后可无需教师便应用其根据之更正之间的差别；在评审中，一次拒绝与一次其判据约束评审者的下一次评估、一如它们解释这一次之拒绝之间的差别；在一个亲密关系中，了结发生了什么与表述什么会算作关于发生了什么的证据、任一方可随后产生它之间的差别。

该条件的三种挫败应予命名，因为每一种都常见、且每一种都保存它的外观。第一是被陈述得使它无法被任何别人应用的标准，依据评估之一方所具有、且不转移的一种识别力。第二是在一个特定之点上被授予、而挑战的地位被留作未加触动的让步，§7指出这使环路保持敞开。第三是被记录的反对：一个记录诸挑战、而没有关于它们关乎后续评估之任何规定的安排，把修正转换为存档，而那记录成为修正可用的证据。

10.2 退出成本的治理

第二个条件随§6而来。既然一个高的退出成本补贴一个配置所含有的任何不对称，退出成本的蓄意降低是可用的最有效干预之一，而它在不要求各方任何东西的情况下运作。

它的诸工具是熟悉的，而它们的效果如今是可界定的。多位导师、可携带的资助、外部考官、寻求第二意见的权利、不系于单一赞助者的雇佣，以及一个主张可在生成它之关系之外被追求的诸论坛，全都降低离去的价格。每一个由此撤回那份补贴：一个从前能在不生成的情况下持续的位置，如今必须生成某样拴住该方之物，否则失去他们。该条件在此处所给的诸条件中不寻常，因为它作用于最难处理的诸情形，缘由是它不要求被偏爱位置的合作。

有两条界限界定它。当这关系是一个其价值取决于它不易被退出的关系时，一如诸形塑性关系特征性地是、而诸亲密关系定义上是，把退出成本降低过一个点便败坏这关系所为之目的，而该条件不应不顾及那一点而被追求。而§6的国际情形显示某些框架没有外部，从而退出成本根本无法被降低。当它无法被降低时，全部负担落在环路闭合上，而这正是§7的判据恰恰在此条件不可用之处最要紧的实质缘由。

10.3 归属的可修正性

第三个条件随§5与§9中所辨认的那个回路而来。归属条件化下一次提出时的语域，而一个一经作出的记录成为下一份贡献被指派所依据的根据。因而一个其中归属被一次了结、并此后固定的安排，不仅仅在寻常意义上是不义的；它是一个被产生之方被保持在生产中所经由的机制。

该条件所要求的，是那记录在一份贡献结果证明曾是什么的照映下可被更改。它的诸工具包括界定什么被做、而非指派等级的贡献陈述，关于在诸贡献的意义后来变得清晰时重新评估它们的规定，以及，在缺乏形式归属的诸情境中，命名一个决定追随了谁之建议的做法。在家庭与外交诸情境中，这最后一项几乎是可用之物的全部，而它并不平凡：一个其中一个被采纳提议的来源被定期陈述的家庭或代表团，维持一份一方可在下一场合汲取的记录。

10.4 课程、评审流程、家庭、条约体制

诸条件对本文所处理的诸情境有着不同的关涉，而这些差异就设计能触及什么而言富有教益。

一个课程不寻常之处在于它的设计者事先控制该配置，而诸方以未被生产的姿态到来。环路闭合能被内建进评估，借要求诸判据被陈述为一种学生可应用于彼此工作、并应用于评估者工作的形式；归属能在协作工作被承担之处被弄得明确。特征性的挫败是被发表并由识别力应用的判据。

一个评审流程有最发达的诸工具，并在专业化的诸场域中有最高的退出成本，其中有能力的评审者的集合是小的。环路闭合由约束评审者后续评估的诸判据、由被发表的报告，以及由一个抵达原初评估者以外某人之申诉的可获得性所服务；特征性的挫败是被返回给被申诉之方的申诉。

一个家庭没有形式的诸工具，且在认证处有最高的利害，因为被认证的是一份关于各方自身的说明。可用的诸条件是对什么会算作证据的表述，它使评估成为任一方可执行之物，以及对谁的建议被采纳的命名。特征性的挫败是在特定之点上、而使该说明未加触动的让步。

一个条约体制有着常常不可降低、且有时未定义的退出成本，这把负担抛到环路闭合与归属之上。可用的是按一个时间表、而非按一个会承担提起它们之成本之一方的倡议来重开已定的诸条款，以及对一项条款追随了哪个代表团之提议的记录。特征性的挫败是其结果无法改变下一次评审被进行于其下之条款的评审会议。

主张 8。 三个条件关乎一个共创性配置是否继续吸收它所生成的不对称：一次评估的诸根据随它行进，从而一次修正的施行改变下一个诠释的诸条件；退出成本在这关系容许之处被降低，因为一个高的成本补贴该配置所含有的任何不对称；以及归属保持可更改，因为那记录条件化下一次提出时的语域。没有一个把一个配置认证为生成性的，而每一个都被保存它外观的诸措施所挫败。

该主张的最后一部分是要被坚持的那一部分。每一个条件都有一个满足它的描述、而挫败它的目的的形式：无法被另一个人应用的标准、被返回给被申诉之方的申诉、没有关于其效果之规定的反对记录。一个展现诸工具的安排并不因此被显示为生成性的，而§7–§8的诸判据正是诸工具要被对照评估之物、而非反过来。

§11 界限与开放的工作

该说明依据一些可能是错的主张，而如此地把它们铺陈出来，是把它放下的一部分。本节陈述本文已论证了什么、要什么该论证才会失败，以及该论证够不到什么。

11.1 该说明所主张的

本文已提议知识是一个诠释经由它被协调的关系结构，每一个这样的结构都生成不对称，而在一个不对称之上一个位置维持自身的倾向是生成性的，产生构成它自己延续之诸条件的诸标准、诸场合与诸参与者。知识中的权力就是这一自我维持的生成性，这把它置于生成之内、而非同它相对立，并使它有代价、失效且可被击败，而非被占有。

从这一点三个分析工具随之而来。§5的五个通道界定一个不对称维持自身之处，从一个诠释必须经历的历程被推导。§6的退出成本界定它如此做所处的价格。§7的环路闭合判据区分一个位置的生成性所奔向的方向，而§8的互惠能力判据区分一次施行是消耗还是更新该位置的必要性。§9中知者的生产是一个方向、一经转向、维系自身所经由的机制。

11.2 什么会击败该说明

有几样东西会。倘若批判传统已然回答此处所提的问题，而对诸体制、诸场域或诸网络的分析确实下降到一个其配置不变之关系之内生成的方向，那么§3已误读了它，而本文没有立足之处。倘若五个通道的列举容许一个对应于一个诠释生命中一个阶段的第六个，那么该推导失败，而该列举退回到它本意要改进的那些临时名单。倘若能产生一些情形，其中一个关系的修正环路可证明地闭合、而这关系尽管如此败坏，或其中它们可证明地敞开、而它没有败坏，那么那个中心判据没有在做被为它所主张的工作。

这些之中的最后一个是最有后果的、且最少被检验的，而本文的主要弱点在于那里。环路闭合的判据在§7中被说成在原则上可查核，而没有关于查核它的任何程序被供出。所会被要求的，是对什么算作下一个诠释之诸条件中一次改变的一个界定，稳健得足以使同一连串交换的两个观察者对它作同样的分类。在那存在之前，该判据组织描述、而不许可测量，而§10中的诸应用比它当前所能承受的更沉重地依据它。

11.3 该说明够不到什么

自始至终的诸例证案例是构造的、而非被观察的。它们因其结构清晰而被选取，而它们只承载例证的分量；没有一个是所描述的诸配置常见的证据，而本文不提供关于所处理种类之诸关系多么频繁地展现它们的任何估计。四个情境中任何一个中的田野工作，都能确立该机制，虽然融贯、却是罕见的。

四个情境因其不相似而被选取，而它们实例化一个结构的论证相应地是一个结构性论证、而非一个经验发现。它会由诸通道能在每一个中被可操作地界定这一证明而被加强，而由五个通道无法被辨认的诸情境而被难堪。

对国际情形的处理是最暴露的。一个国家的退出不是一个单一行动者的决定，离去的成本不均匀地落在诸国内选民身上，而把一套为人际关系发展的语汇转移到集体行动者，要求本文在结构类比之外未曾供出的辩护。该情形是作为那个变量的一个实例、而非作为对国际政治理论的一个贡献而提供的。

该说明在一个进一步的方面是描述性的，而这限制§10能主张什么。在拒绝了把诸配置认证为生成性的、并在§8中承认了诸判据只得出在被界定诸方面的局部判断之后，本文无法就任何实际的安排说它是良序的。它所能说的，是去哪里看以及什么会算作相关的证据。

11.4 在纲领中的位置

这一分析路线属于一个把存在、价值与治理当作在关系中并经由关系被生成的更宽努力。母框架，生成性关系存在，关乎关系如何生成存在；一个相伴的关切，生成性关系治理，处理一个被生成的关系一经存在如何被治理。当下这篇论文占据那个交叉点：知识是一个关系的各方协调他们诠释的媒介，而那一媒介的配置既是该关系所生成之物、又是治理它进一步生成之物。同一个结构出现在纲领别处所发展的关系败坏说明中，其中一个区分坍缩而一个符号执行那一坍缩，这不是偶然的，而修正环路的闭合同那里所描述之诸机制之间的对应，是本文留待被恰当地做的工作。

11.5 进一步的工作

有三条路线随之而来。第一是环路闭合的可操作化，没有它该说明仍是一种描述诸情形的方式、而非一个评估它们的手段。第二是单一情境中的经验工作，其中评审流程是最易处理的，因为它的诸通道被记录在案，而它的诸修正施行留下记录。第三是本文所提供之两条判据、环路闭合与互惠能力之间的关系，它们在§8中被区分，而它们的连接在那里被留作一种倾向、而非被确立。一个关于一个配置及其诸轨迹的形式处理会关乎这一点，而此处不予尝试。

---

关于方法的一则说明。 没有形式装置被用来陈述此处散文所直白陈述之物，而这一克制是刻意的。一个被写作一个加权能力向量的配置，会装点该论证而不从它推导任何东西，并会借来一份该论证未曾挣得的严格。这些机制容许两种诚实的呈现：清晰的散文，或一个同动力系统之语言的真正连接，其中一个配置会答对一个诸安排之空间中的一个点，一个闭合的修正环路答对一个使它位移的扰动，一个敞开的环路答对一个该安排在不位移的情况下吸收的扰动，而一个被产生的知者答对该配置自身动态的一个不动点。第二种在此不予尝试。所主张的，在第一种中被陈述。

参考文献

Bourdieu, Pierre (1975). The Specificity of the Scientific Field and the Social Conditions of the Progress of Reason. Social Science Information, 14(6), 19–47.

Bourdieu, Pierre (1988). Homo Academicus. Trans. Peter Collier. Stanford: Stanford University Press. [French orig. 1984.]

Callon, Michel (1986). Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay. In John Law (ed.), Power, Action and Belief: A New Sociology of Knowledge? London: Routledge, 196–233.

Finnemore, Martha, and Kathryn Sikkink (1998). International Norm Dynamics and Political Change. International Organization, 52(4), 887–917.

Fricker, Miranda (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Foucault, Michel (1972). The Archaeology of Knowledge. Trans. A. M. Sheridan Smith. New York: Pantheon Books. [French orig. 1969.]

Foucault, Michel (1977). Discipline and Punish: The Birth of the Prison. Trans. Alan Sheridan. New York: Pantheon Books. [French orig. 1975.]

Foucault, Michel (1978). The History of Sexuality, Volume I: An Introduction. Trans. Robert Hurley. New York: Pantheon Books. [French orig. 1976.]

Freire, Paulo (1970). Pedagogy of the Oppressed. Trans. Myra Bergman Ramos. New York: Herder and Herder.

Gettier, Edmund L. (1963). Is Justified True Belief Knowledge? Analysis, 23(6), 121–123.

Hirschman, Albert O. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press.

Goldman, Alvin I. (1967). A Causal Theory of Knowing. The Journal of Philosophy, 64(12), 357–372.

Goldman, Alvin I. (1999). Knowledge in a Social World. Oxford: Oxford University Press.

Lacan, Jacques (1977). The Four Fundamental Concepts of Psycho-Analysis. Ed. Jacques-Alain Miller, trans. Alan Sheridan. London: Hogarth Press.

Latour, Bruno (1986). The Powers of Association. In John Law (ed.), Power, Action and Belief: A New Sociology of Knowledge? London: Routledge, 264–280.

Latour, Bruno (1987). Science in Action: How to Follow Scientists and Engineers through Society. Cambridge, MA: Harvard University Press.

Nietzsche, Friedrich (1966). Beyond Good and Evil: Prelude to a Philosophy of the Future. Trans. Walter Kaufmann. New York: Vintage Books. [German orig. 1886.]

Wendt, Alexander (1999). Social Theory of International Politics. Cambridge: Cambridge University Press.