Generative Obstruction - Toward a Minimal Theory of Conflict in Generative Relational Being

ENGLISH

Generative Obstruction

Toward a Minimal Theory of Conflict in Generative Relational Being

Wanhong Huang · huangwanhong@serendip.ngo

---

Abstract

Epicurus located happiness in the absence of two disturbances: pain of the body and disturbance of the soul. The distinction is structural before it is hedonic, separating what impedes a system from within from what impedes it in its coupling with others, and this paper retains the structure while dropping the hedonic content, asking what obstructs generation in place of what disturbs tranquility. Existing theories of conflict begin from incompatible primitives: power, interests, values, identity, institutions. Each explains real phenomena, and each imports an ontological starting point the others do not share. This paper proposes a single primitive explanatory object, generative obstruction, defined as any interaction that diminishes the ongoing generative capacity of a constrained generative system, and treats conflict as the manifestation of such obstruction in place of a primitive. Power, resource competition, misinterpretation, unmet demand, and symbolic commitment then appear as mechanisms producing a common phenomenon, in place of competing definitions of conflict. The paper’s substantive result concerns the reference trajectories that attributions of obstruction require: since no ontologically privileged trajectory is available to a relationally constituted subject, such trajectories are admissible as working models held for coordination, and an argument is given that a model treated as accurate degrades the channel by which it would be corrected, so that its rigidity bears on obstruction where its content does not. A constructed case exhibits the argument. The framework’s limits are stated at length: obstruction is defined by a partial order that leaves an important class of cases unclassified, and attributions of obstruction are licensed from within a relation while remaining unlicensed from outside it, which restricts the account more severely than its exposition might suggest.

Keywords: generative obstruction; conflict; generative relational being; reference trajectory; revisability; symbolic closure.

A note on the standing of this paper. This is a foundational paper, offered in place of a theory of conflict resolution. It proposes one primitive explanatory object and works out what follows from admitting only that object; the discipline it attempts is subtractive, and much of its labour consists in declining concepts that a theory of conflict might be expected to carry, among them justice, value, fairness, and alienation. Its central result is a claim about the rigidity of reference models, argued from the framework’s own resources and exhibited on a constructed case in place of an empirical one. Several of its commitments are left as open questions where the resources to settle them were unavailable, and these are collected in the final two sections in place of being distributed as qualifications. Should the primitive be shown to do no work that existing primitives do not already do, the paper loses its ground. Objection, correction, and counter-evidence are welcome at huangwanhong@serendip.ngo.

§1 Introduction

Epicurus defines happiness negatively, and by two terms. It consists in aponia, the absence of pain in the body, and ataraxia, the absence of disturbance in the soul. The definition is ordinarily read as hedonic doctrine and assessed as such: whether the absence of disturbance is a plausible account of the good life, whether the tranquil life is the choiceworthy one. Read that way it belongs to the history of ethics and bears little on the theory of conflict.

There is another way to read it. Set aside the hedonic content and attend only to the structure, and what the definition supplies is a partition of impediment into two loci. Bodily pain impedes a system in the operation of its own processes; it is an obstruction internal to the system, requiring no second party. Disturbance of the soul, in the Epicurean treatment, arises characteristically from what stands between a person and others or between a person and the order of things: from fear of the gods, fear of death, the demands of political entanglement, the disturbances that attend living among others whose purposes are not one’s own. It is an obstruction that presupposes a coupling. What Epicurus thereby distinguishes is two places where trouble occurs, in preference to two intensities of one trouble: within the system and in the system’s relation to what it is coupled with.

This paper takes the partition and discards the rest. Three things go. The hedonic content goes, since nothing here suggests that the absence of disturbance is the human good, and the argument is indifferent to whether it is. The terminus goes: Epicurean happiness is a state to be arrived at and maintained, whereas what follows concerns processes that continue in place of states that settle. And the negative definition goes as a definition, though it is retained as a heuristic, since the concern below falls on what obstruction is and how it propagates, a positive question about a mechanism, leaving aside the negative question about a condition. What survives the subtraction is the two-locus structure, and it survives in transposed form. Where Epicurus asks what disturbs tranquility, this paper asks what obstructs generation; and where he distinguishes disturbance of the body from disturbance of the soul, this paper will distinguish obstruction of a system’s own generative dynamics from obstruction of its capacity to develop those dynamics under relational coupling.

The transposition earns its place, and it is worth being explicit about what it buys. A theory that introduced an internal and relational distinction on its own authority would be introducing a second primitive, and would owe an argument for the partition. Taking the distinction from a place where it was drawn for other reasons entirely, and drawn before there was any theory of generation to serve, is some evidence that the seam lies in the phenomena in preference to the theory. The partition is inherited; only the terms in which it is drawn are new.

Conflict is among the best-theorized objects in the social sciences, and the difficulty lies in the form the explanations take, leaving scarcity aside. Realism begins from power and reads conflict as what follows from its distribution. Liberal institutionalism begins from institutions and reads conflict as what their absence or weakness permits. Materialist accounts begin from the relations of production; constructivist accounts from identity and its formation; normative accounts from disagreement about value. Each of these explains real cases, and explains some of them better than its rivals do. What they do not share is a common object. Each begins from a different primitive, and a primitive differs from a variable: it is what the theory declines to explain in order to explain everything else. When theories differ in their primitives they do not disagree in the ordinary sense, since they are not making incompatible claims about the same object. They are constituting different objects.

The consequence is that the accumulated body of conflict theory does not accumulate. Findings within a tradition compound; findings across traditions are difficult even to place beside one another, because the entity whose behaviour is being explained is differently constituted in each. A finding about power asymmetry and a finding about identity formation may both be sound and still fail to bear on each other, for want of a common object to which both could be referred, leaving data aside.

This suggests a question that is prior to the contest among the traditions and does not require adjudicating it. Suppose power, institutions, identity, and value amount to competing accounts of how conflict is produced, in place of competing answers to the question of what conflict is. Then they would not be rivals at all. They would be mechanisms, and mechanisms can coexist without embarrassment: several may operate at once, they may compose, one may dominate in one case and another in another. What such a reading requires is a specification of the phenomenon they severally produce, an object more primitive than any of them, in terms of which each can be described as a route to a common outcome. Supplying such an object is the work this paper attempts.

The analogy from medicine is instructive and worth stating plainly, since it fixes the kind of move being made. Fever escapes classification by pathogen. It is characterized physiologically, as a phenomenon, and viruses, bacteria, autoimmune processes, and malignancies are then understood as distinct mechanisms capable of producing it. That architecture achieves something beyond a compromise among competing theories of fever; it is what made the competing theories into complementary ones. The proposal here is architecturally the same, with conflict in the place of fever and generative obstruction in the place of the physiological characterization.

The primitive proposed is generative obstruction: any interaction that diminishes the ongoing generative capacity of a constrained generative system. The terms are developed in §3 and §4; what matters at the outset is the shape of the commitment.

Two features of that shape distinguish it from the primitives it would replace. The first is its dynamical character. Power distribution, institutional density, and identity configuration are all properties a situation has at a time, and a theory built on them evaluates situations at times. Generative capacity is a property of a process, namely what a system can go on generating, and it cannot be read off an instantaneous configuration. Obstruction is accordingly path-dependent: an interaction that diminishes capacity now may enlarge it later, and the same distribution of any static quantity may or may not be obstructing depending on the trajectory it belongs to. This follows as a consequence of the primitive, requiring no additional assumption, and §4 derives it as such.

The second is that it is relational without being reducible to relations between fixed parties. The systems whose generativity is at issue are constituted within the couplings in which they stand; they do not enter relations already formed. This commitment, longstanding in the framework of Generative Relational Being, is what makes the paper’s central difficulty acute. Any attribution of obstruction requires a reference, since to say that $A$ obstructed $B$ is to say that something $B$ would otherwise have generated was foreclosed, and a relationally constituted subject has no privileged baseline to serve as that reference. The trajectory the subject would have followed in isolation is unavailable as a reference, since the isolation it presupposes was never the subject’s condition. The difficulty is taken up directly in §7, and the paper’s substantive result is what emerges from it: reference trajectories are admissible as working models in place of facts, and the obstruction-relevant variable is whether a model remains open to revision by the party it models, leaving aside how far it departs from what it models. The answer is partial. It licenses models held by parties to a relation and fails to license models held by observers outside one, which restricts the framework’s reach considerably and is stated as such in §9.

A theory that explains everything within its domain has usually failed to specify a domain. This paper states its scope in preference to maximizing it, and two limitations are entered at the outset because they shape what follows. The first concerns mechanisms: no exhaustive taxonomy of the mechanisms producing obstruction is offered or attempted. Several are discussed in §5; others certainly exist; the framework’s claim is about the phenomenon they produce, not about the completeness of any list of producers. The second concerns the reach of the primitive itself. Not every conflict is a manifestation of generative obstruction. Two parties may want incompatible things while neither diminishes the other’s generative capacity, and the resulting conflict is real, is sometimes severe, and lies outside what this framework explains. In place of absorbing such cases by stipulation, which remains available in every instance and secures nothing, §5 supplies a criterion for telling whether a given conflict falls within the framework’s reach. A theory that says where it does not apply is stronger than one that cannot be shown not to apply.

Normative questions are declined entirely. Justice, fairness, value, ethics, and happiness require a further layer and are not treated here. §9 says why, and notes one consequence of the paper’s own result: that once reference trajectories are understood as non-ontological, deviation from them loses whatever normative weight it appeared to carry, and unrevisability is left doing the work.

The remainder proceeds as follows. §2 surveys the existing explanatory primitives and states the reduction of each to a mechanism. §3 sets out constrained generative dynamics: generativity as process, constraint, and generative capacity. §4 defines generative obstruction, partitions it by locus following the reading of Epicurus above, and derives its modal properties, including path-dependence. §5 treats conflict as the manifestation of obstruction, arrays the mechanisms without privilege among them, and gives the scope criterion by way of the case of unmet demand. §6 isolates the one property that earns symbolic mechanisms a separate treatment, that they make obstruction self-reproducing. §7 takes up the problem of the reference trajectory, gives the engineering reading, and states the rigidity claim. §8 demonstrates that claim on a constructed case in which content and disagreement are held fixed and only revisability varies. §9 states what the paper does not attempt, and §10 what it leaves open.

§2 The Reduction of Existing Primitives to Mechanisms

The traditions surveyed here each explain a class of cases and do so well. The aim of this section is to ask what follows from reading each as an account of production, where the tradition itself offers an account of constitution. The reduction proposed takes a uniform form: where a tradition holds that conflict is a certain configuration, the reading offered here holds that the configuration is one route by which generative capacity comes to be diminished.

Power is the most developed case and the most instructive. The realist account locates conflict in the distribution of capability: parties in an anarchic setting attend to relative position, and conflict follows from the incentives that position creates. The account predicts a great deal. What it leaves unstated is why a distribution should matter at all, and the answer available when the question is pressed is telling. An adverse distribution matters because of what the stronger party can then do to the weaker, which is to say, because of the range of futures it can foreclose. The appeal is already to something beyond distribution. Distribution serves as the mechanism; foreclosure of possible futures is what the mechanism produces. This also explains a known weakness. A capability differential between parties whose developments intersect little forecloses correspondingly little, and such cases are among those realism handles worst.

Interests and resource competition admit the same reading. Two parties requiring the same finite resource conflict, on the standard account, because the resource resists division. Scarcity alone leaves the matter open. What makes a resource contest conflictual is that one party’s securing the resource removes something the other required in order to continue developing, and the contest grows less conflictual in proportion as substitutes exist, as timescales permit sequential access, or as the resource proves dispensable to either party’s continuation. Resource competition thus operates as a mechanism whose obstructive force varies with facts about generative dependence, and those facts concern the parties in place of the resource.

Institutions enter the same way with the sign reversed. The institutionalist observes that conflict is suppressed where institutions are dense and erupts where they are thin, explaining this by information, commitment, and the shadow of the future. Described in the vocabulary proposed here, institutions stabilize the conditions under which parties can go on developing in each other’s presence: they make foreclosure less likely and less permanent. Their absence leaves obstruction ungoverned. This accounts for the imperfection of institutional density as a predictor of peace, since an institution that stabilizes one party’s development at the cost of another’s suppresses conflict only partially, and may relocate it. The literature’s persistent difficulty with unequal institutional orders is a symptom of measuring density in place of what density is supposed to secure.

Identity is the case where the reduction is least obvious and most worth making. The constructivist holds that parties act on the basis of who they take themselves and others to be, and that conflict follows from identities constituted in mutual opposition. Read as a mechanism, identity fixes what a party takes its own continuation to require and what it takes another party’s continuation to threaten. Identities constituted in opposition are those under which each party’s development is understood to require the other’s containment. Whether such an understanding is accurate remains a separate question, and its capacity to remain fully operative while inaccurate gives a first indication of something §6 treats at length: mechanisms that fix interpretations stand in a distinctive relation to the persistence of obstruction.

Value disagreement completes the survey. Parties committed to incompatible normative orders are held to conflict in virtue of the incompatibility. Normative disagreement is common and mostly inert. Two parties may hold irreconcilable views about the good and leave each other’s development untouched, provided neither requires the other’s conversion or submission as a condition of its own continuation. Value disagreement becomes conflict once joined to such a requirement, and the requirement carries the weight. This explains why the difference between deep pluralism and sectarian conflict fails to track the depth of the disagreement.

A common structure runs through the five cases. Each tradition names something real; in each, what renders the named thing conflictual is a further fact about the foreclosure of one party’s continuation by another. The traditions differ in what they name and agree, tacitly, on what makes it matter. That tacit agreement is what the present proposal makes explicit.

Claim 1. Power, resource competition, institutional configuration, identity, and value disagreement admit a uniform reading as mechanisms producing a common phenomenon. In each case, what renders the named configuration conflictual is a further fact about the foreclosure of a party’s ongoing development, and this further fact is shared across all five.

Two consequences follow. The traditions become complementary, since mechanisms compose where definitions exclude one another. And a burden is incurred: the common phenomenon requires specification in terms that presuppose none of the mechanisms. Should generative obstruction prove definable only by appeal to power, or to interests, or to identity, the reduction is circular and the exercise yields little. Discharging that burden is the work of the two sections that follow, which is why they remain silent about conflict.

§3 The Apparatus: Generativity, Constraint, and Capacity

This section states the minimal apparatus, concerning systems and their development. It introduces no notion of conflict, obstruction, or relation between parties, all of which enter in §4. The order is required by the burden just incurred, which demands that the primitive be specifiable independently of the mechanisms it unifies.

Generativity comes first. A system is generative in the sense used here if it produces structure absent from it before, where that structure becomes the basis for what it produces next. The emphasis falls on the second clause. Change alone leaves generativity undetermined: a system that oscillates traverses novel states while its capacity to produce remains fixed, and a system perturbed from equilibrium and returning to it ends where it began. What distinguishes generation is that its products enter the conditions of subsequent production, so that the system’s own history constrains and enables what it can go on to make. A research programme generates in this sense, since results become the instruments and the questions of later results. A developing organism does, a language does, an institution does. The common feature lies in the incorporation of what has been produced into the productive capacity itself.

One qualification belongs here, since the condition just stated admits a reading stronger than the one intended. Incorporation covers a range of cases. At one end sits strict accumulation, where products are retained and compound; at the other sits mere conditioning, where what was produced shapes what can follow while leaving no persistent trace of itself. A conversation generates in the second sense: what has been said conditions what can be said next, and little of it is retained. Both fall within the scope of the term, and the condition requires only that production be historically conditioned by production, allowing accumulation as one case among several. The weaker reading is the operative one throughout, and §4 relies on nothing stronger.

Generativity is accordingly a property of processes and holds of states only derivatively. A state admits description independent of how it was reached; a generative capacity resists such description, since what a system can produce next depends on what it has already produced and on the manner of the producing. Two systems occupying identical states may differ considerably in generative capacity, having arrived by different routes and thereby accumulated different instruments, foreclosed different options, and acquired different dependencies. This point carries considerable weight later, and it is worth marking as a feature of generation itself, stated here in advance of any claim about conflict.

Constraint enters next. Generative processes of the kind at issue operate under conditions determining which productions are available to them, and these conditions divide into several kinds. They are material, in that resources and energy are finite; structural, in that what has already been produced forecloses some paths as it opens others; temporal, in that production takes time and certain sequences resist compression; and relational, in that a system coupled to others develops under conditions those others partly set. Constraint operates as the condition of generation. An unconstrained process is formless, since constraint is what makes some productions available in preference to others and thereby gives the process a direction. The term constrained generative system accordingly marks no special case. It marks the general one.

Generative capacity is the third and final term: what a system can go on generating from where it now stands, given its history and its constraints. Three features of the notion require comment, each bearing weight later.

Capacity is forward-looking and therefore inferred in place of observed. What is observed is what a system has produced; what capacity concerns is what it could produce, and the inference between them is defeasible. A system may have produced little while retaining large capacity, and may have produced much while exhausting the conditions that made the production possible. This gap between record and capacity is the reason the notion is needed, since a theory attending only to what had been produced would lack the means to distinguish a system that had been prevented from one that had yet to act.

Capacity resists representation as a scalar. It approximates a space of available continuations, and it changes in ways a single number captures poorly. It contracts, when continuations formerly available cease to be. It expands, when new instruments or couplings open continuations that did not exist. And it deforms without contracting or expanding in extent, when the available continuations are exchanged for others of comparable range and different character. The third possibility is easily overlooked and matters in §4, since a system whose capacity has been redirected wholesale may register no loss on any measure of extent while having been changed in the respect that counts.

Capacity is relationally conditioned, and typically to a considerable degree. A system’s available continuations depend on the couplings it stands in: what a research programme can generate depends on its collaborators and readers, what a state can generate depends on its neighbours, what a person can generate depends on those they live among. This states a condition of capacity and stops short of any claim about obstruction, since a coupling may enlarge capacity as readily as reduce it, and §4 requires the enlarging case as much as the reducing one.

Three terms are now in place: generativity, constraint, and generative capacity. None makes reference to power, interest, institution, identity, or value. None makes reference to conflict, and none yet involves more than one system at a time. The burden incurred at the close of §2 is discharged to this extent, that the vocabulary in which obstruction will be defined stands independent of the mechanisms obstruction is meant to unify.

§4 Generative Obstruction: Definition, Loci, and Modal Properties

The apparatus of §3 concerned single systems. Obstruction requires two, and it is introduced here in three steps: the definition, the partition by locus inherited from the reading of Epicurus in §1, and the modal properties that follow from the definition together with the account of capacity already given.

4.1 The Ordering on Continuation Spaces

§3 held that generative capacity resists representation as a scalar and approximates a space of available continuations. A definition of obstruction as diminution of capacity accordingly requires an account of what diminution amounts to on such a space, and the account must be given before the definition is stated.

The resource available is a partial order. Where the continuations available to a system under one condition are among those available under another, capacity has diminished in a sense requiring no further apparatus: write $G’ \preceq G$ where the continuations in $G’$ are contained in those of $G$. Contraction, as described in §3, is the case where this holds strictly, and expansion the converse.

The order is partial in the strict sense, and the incomparable cases are those §3 named deformation: a continuation space exchanged for another of comparable extent and different character stands in no containment relation to its predecessor. Neither $G’ \preceq G$ nor $G \preceq G’$ holds, and the definition below therefore fails to classify such cases.

This is a limitation of the present formulation and is entered as such. Two responses are available and neither is adopted here. A measure on continuation spaces would order the incomparable cases by extent, at the cost of treating spaces of equal extent as equivalent, which is precisely what the deformation mode was introduced to deny. An evaluative ordering would rank continuations by their worth, which requires the normative apparatus §9 declines. The framework as it stands orders the comparable cases and registers the incomparable ones as a distinct phenomenon requiring treatment it does not supply.

4.2 The Definition and Its Reference Problem

Definition 1. An interaction constitutes a generative obstruction of a system when the continuations available to that system under the interaction are properly contained in those available otherwise: $G’ \prec G$.

Four features of the definition deserve comment before it is put to work.

Its scope is the comparable cases. Deformation of a continuation space, where no containment relation holds, falls outside Definition 1 and is registered in §4.4 as a distinct mode. This costs the framework the cases the tradition treats under the heading of alienation, which §9 takes up, and the cost is preferred to an ordering the framework cannot support.

It is comparative in form. To say that continuations have been foreclosed is to compare those available under the interaction with those available otherwise, and the comparison presupposes a specification of the otherwise. This is the definition’s principal difficulty. §7 takes it up and, as will be seen there, supplies a partial answer in place of a complete one. The problem is stated here and left standing, since the intervening sections require the definition in its provisional form.

It concerns capacity in place of production. A system prevented from producing what it was about to produce has been obstructed; a system that produces nothing while its capacity stands intact has not. The distinction follows from the account of capacity in §3 and matters for cases where the two come apart. Rest, deliberate abstention, and the fallow period of a research programme are cases where production falls while capacity holds or grows. Treating these as obstruction would be an error, and the definition avoids it by attaching to capacity.

It is silent about intention. An interaction obstructs when it diminishes capacity, whatever the intentions behind it. A party may obstruct another while intending assistance, and may intend obstruction while producing none. This silence is deliberate and has a cost, since it removes from the theory’s reach a distinction that moral assessment requires. The cost is accepted here as the price of keeping the primitive mechanical, and §9 returns to what it forecloses.

4.3 The Two Loci

The partition inherited in §1 divides obstruction by its place of operation. Transposed into the vocabulary of §3, the two loci are these.

Internal obstruction operates on a system’s own generative dynamics. Its instances include the depletion of the material conditions of production, the disorganization of the system’s informational processes, and the exhaustion of the capacity to sustain the process at all. What characterizes the locus is that the diminution requires no second system for its description. A system suffering internal obstruction has been impaired in the operation of its own processes, and the impairment would be describable were the system considered alone.

Relational obstruction operates on a system’s capacity to develop its dynamics under coupling. Its instances include lock-in, where a coupling fixes a system’s development along a path it can no longer leave; the imposition of a rigid reference model, treated at length in §7; and institutional configuration that stabilizes one party’s development at another’s expense. What characterizes the locus is that the diminution requires the coupling for its description. A system suffering relational obstruction retains its own processes in working order and finds the conditions of their development set by another.

Two remarks about the partition. The first concerns its independence from the mechanism/phenomenon distinction of §2. Locus concerns where obstruction operates; mechanism concerns what produces it. A single mechanism may produce obstruction at either locus: a resource contest that starves a party produces internal obstruction, and the same contest resolved by an agreement that fixes one party’s development produces relational obstruction. The two distinctions cut across each other and neither reduces to the other.

The second concerns the traffic between the loci, which runs in both directions and which the partition should be read to permit. Sustained relational obstruction produces internal obstruction, as a system whose development has been fixed by another loses, over time, the internal resources that development sustained. Internal obstruction shapes the couplings a system can enter, and a depleted system is more readily subjected to relational obstruction than a flourishing one. The partition marks two places where obstruction operates, leaving the passage between them open. Nothing here requires that a given episode be assigned exclusively to one locus.

The distribution of the paper’s attention between the loci is uneven, and the reason is worth stating. Internal obstruction is the better-studied of the two and is treated in several mature literatures under other names. The relational locus is where this framework has something to add, and it is where the paper’s substantive result falls. The remaining sections concern it almost exclusively.

4.4 Modal Properties

The properties collected here follow from Definition 1 together with the account of capacity in §3. They are stated as properties of the primitive, and they are the resources on which the later sections draw.

Extent. Obstruction may be local, closing a portion of a system’s available continuations while leaving the remainder intact, or global, diminishing capacity across the range. The distinction admits degrees and turns on how much of the continuation space the diminution reaches.

Duration. Obstruction may be temporary, with capacity restored as the interaction lapses, or persistent, with capacity remaining diminished after the interaction has ceased. Persistence requires explanation, since an interaction that has ceased explains nothing about what follows it. The explanation is the concern of §6.

Reversibility. Obstruction may be reversible, where the foreclosed continuations remain recoverable, or irreversible, where their foreclosure has entered the conditions of subsequent production and cannot be undone by removing the interaction. Reversibility and duration are distinct: an obstruction may persist while remaining reversible in principle, and may be brief while foreclosing something permanently.

Character, and the limit of the definition here. A system’s continuation space may be deformed, its continuations exchanged for others of comparable extent and different character. Definition 1 does not reach these cases, since the partial order supplies no relation between the spaces, and this section registers them as a phenomenon standing outside the primitive in place of a mode within it.

The registration is worth making for two reasons. Such cases occur, and a framework silent about them would be silent about a substantial class of what its subject matter contains. And they are where the concerns the tradition treats under the heading of alienation would fall, so that the framework’s inability to reach them locates a gap precisely in place of leaving it diffuse. What can be said of such cases within the present resources is that a change of character has occurred, which requires no account of what the space ought to have contained. What cannot be said is whether the change constitutes obstruction, and §9 enters this as an open question.

Path-dependence. This property does the most work later, and it follows from the definition together with §3 in a way worth setting out step by step.

Definition 1 makes obstruction a diminution of generative capacity. §3 established that generative capacity depends on a system’s history, since what a system can produce next depends on what it has already produced and on the manner of the producing. Combining the two: whether a given interaction diminishes capacity depends on the history in which the interaction falls. The same interaction, occurring at two points in a system’s history, diminishes capacity at one and enlarges it at the other, since the continuations it closes and opens are determined by what the system has already produced. A constraint imposed on a system that has yet to develop the instruments the constraint presupposes forecloses much; the same constraint imposed after those instruments exist forecloses little and may open a path that was unavailable before.

Two consequences follow, and both bear on how obstruction can be assessed.

The first concerns instantaneous assessment. Whether an interaction is obstructive resists being read from a description of the situation at the moment of the interaction, since the assessment draws on the history such a description omits. Two situations identical in their instantaneous descriptions may differ in whether a given interaction obstructs, the difference lying in their histories. This is what separates the present primitive from primitives of the static kind. Power distribution, institutional density, and identity configuration are properties a description at a time supplies; whether capacity has been diminished stands outside what such a description reaches.

The second concerns the assessment of intervals. Since obstruction at one moment may enlarge capacity later, and enlargement at one moment may foreclose later, an assessment restricted to a single moment risks misreading the interval it falls in. The unit of assessment is accordingly the trajectory, and what is assessed is the course of obstruction across it. Nothing in this requires that trajectories be compared to some preferred trajectory, and §7 gives reasons for doubting that such a comparison is available. What is available is the course of a system’s capacity across the interval, and that is what the assessment concerns.

Claim 2. Generative obstruction is path-dependent, and its path-dependence follows from the definition together with the historical conditioning of generative capacity, requiring no further assumption. Whether an interaction diminishes capacity depends on the history in which it falls; the unit of assessment is accordingly the trajectory in place of the moment.

The promise entered in §1 is discharged to this extent. Path-dependence was there described as a consequence of the primitive in place of an addition to it, and the derivation above is the consequence in question. What the derivation requires from §3 is the weak reading of historical conditioning defended there, that production be conditioned by production; the strong reading, requiring accumulation, is unnecessary.

§5 Conflict as the Manifestation of Obstruction, and the Limits of That Reading

The primitive is now specified, and the reduction proposed in §2 can be stated in the vocabulary of §4. This section does three things: it states the relation between obstruction and conflict, it arrays the mechanisms without privilege among them, and it supplies the criterion by which a given conflict is assigned to the framework’s reach or placed outside it. The third is the section’s principal work.

5.1 The Relation Between Obstruction and Conflict

Conflict enters the framework as a derived phenomenon. The relation may be stated as follows.

Definition 2. Two systems stand in mutual generative obstruction when each diminishes the ongoing generative capacity of the other. Conflict, on the reading proposed here, is the manifestation of generative obstruction that has become relationally significant between the parties to it.

Two clauses in the definition require comment, and the second carries a difficulty.

Mutuality is one condition among the possible ones, and its status should be marked. Obstruction runs in one direction readily: a party may diminish another’s capacity while its own capacity stands unaffected, and the asymmetric case is common. What the asymmetric case yields is domination in place of conflict, and the difference is worth preserving. A conflict has two parties acting upon each other; a domination has one party acting and another undergoing. The framework registers both and describes them differently, which is a point in its favour, since theories built on power distribution have a known difficulty distinguishing the two.

Relational significance is the second clause, and it is doing more work than its brevity suggests. Not every mutual diminution of capacity manifests as conflict. Two parties may each slightly reduce the other’s continuations, register the reduction little, and continue.

One candidate threshold is available within the framework’s resources and is offered provisionally. Obstruction has become relationally significant where it has entered the parties’ working models of each other: where each party’s model of the other includes the obstruction as a term, and each party’s conduct is conditioned by that term. This makes the threshold a fact about the models in place of a fact about the intensity of the obstruction, which has the advantage of connecting Definition 2 to the apparatus §7 develops, and the disadvantage of resting on the same working models whose standing that section finds limited.

The candidate is offered without confidence. It may prove too permissive, since models absorb a great deal that organizes no relation, and the threshold would then admit conflicts that are none. What can be said in mitigation is that the vagueness attaches to the manifestation in place of the primitive: whether continuations have been foreclosed admits sharper treatment than whether the foreclosure has become significant, and the sharper question is the one the framework is built on. Supplying a defensible threshold is entered in §10.

5.2 The Mechanisms, Without Privilege

The mechanisms surveyed in §2 now take their place as routes to obstruction. Resource competition produces obstruction when one party’s securing of a resource removes a condition of the other’s continuation. Power asymmetry produces it when capability differential is converted into foreclosure. Institutional configuration produces it when arrangements stabilize one party’s development at another’s expense, and suppresses it when they stabilize the conditions of development for both. Misinterpretation produces it when a party acts on a construal of the other that leads it to foreclose what it took to be already closed. Unmet demand produces it under conditions the following subsection specifies. Symbolic and normative commitment produces it when the commitment requires the other’s conversion or containment as a condition of the committed party’s own continuation.

Two observations about this list. The mechanisms compose: several may operate at once, and the obstruction they jointly produce may exceed what each produces alone. And the list is open. Others exist, and the framework’s claim concerns the phenomenon these produce, leaving the enumeration incomplete by design. A theory of mechanisms would require the enumeration; a theory of the phenomenon they produce does not.

5.3 The Scope Criterion, Tested on Unmet Demand

The most pressing objection to a framework of this kind is that conflicts arise which involve no obstruction, and the objection is correct. The case of unmet demand states it sharply, and working through that case yields the criterion this subsection supplies.

Consider two parties, where one wants something the other declines to provide. The frustration is real and the conflict that follows is real. Where is the obstruction? Two answers are available and one of them is a trap.

The trap is to locate the obstruction in the frustration itself: to hold that an unsatisfied want diminishes the wanting party’s generative capacity, and that the conflict is therefore obstructive after all. The move is available in every case, since wanting something and lacking it is the condition of wanting anything at all. Taken seriously it makes every unmet want an obstruction, and a concept that applies to everything within its domain discriminates nothing within it. The move also relocates a conflict that is manifestly between two parties into the interior of one of them, which sits poorly with a framework whose commitments are relational. This answer is declined.

The second answer accepts the case as it stands and asks what would have to be true for obstruction to be present. Two conditions emerge, and either suffices.

The first is that satisfying the demand would foreclose the other party’s continuations. Where a demand can be met only at the cost of the other’s capacity to go on developing, the demand’s satisfaction constitutes relational obstruction of the party who meets it, and the conflict falls within the framework’s reach. This condition is about the content of the demand and what meeting it would cost.

The second is that the demand’s frustration measurably degrades the wanting party’s own capacity. This is the internal locus, and the requirement of measurability is what keeps it from collapsing into the trap. What is required is that capacity be diminished in a way describable independent of the frustration itself: continuations that were available and have ceased to be, instruments that were in use and are no longer, processes that were operating and have failed. A frustration that leaves the wanting party’s capacity intact fails the condition, however sharply it is felt.

Where neither condition holds, the conflict falls outside the framework. Such conflicts occur. Two parties may want incompatible things, may argue over the incompatibility, may find the disagreement painful and durable, and may leave each other’s generative capacity untouched throughout. The framework describes what such a conflict is not and leaves what it is to another account.

Claim 3. A conflict falls within the reach of the present framework when satisfying one party’s demand would foreclose the other’s continuations, or when the demand’s frustration degrades the wanting party’s capacity in a manner describable independent of the frustration. Conflicts satisfying neither condition are real and lie outside what this framework explains.

Three remarks on the criterion.

The criterion concerns scope and carries no evaluative content. A conflict inside the framework’s reach is no worse and no more urgent than one outside it for that reason alone. What the criterion determines is which questions the framework can answer about a given case.

The criterion falls short of self-applying. Whether satisfying a demand would foreclose a party’s continuations, and whether a frustration has degraded capacity independent of itself, are questions about the case that require investigation. The criterion states what to investigate and leaves the investigating undone.

The criterion inherits the reference problem left standing in §4. Both conditions turn on comparisons of capacity, and comparisons of capacity presuppose a specification of what capacity would otherwise have been. The criterion is therefore provisional in the same respect that Definition 1 is, and the provisionality is discharged, so far as it can be, in §7.

The observation with which this subsection began deserves a final word, since a reader may take it as a concession. A framework that declines to explain a class of conflicts might seem weaker than one that explains them all. The reverse holds. A framework that absorbs every case does so by a stipulation available to competing frameworks equally, and having absorbed everything it distinguishes little and forbids little. Stating where a framework fails to apply is what makes its applications informative, and supplying a criterion by which the failure can be recognized in a particular case is what makes the statement more than a disclaimer.

5.4 Value as Orientation of Generative Direction

The demand case established a pattern: a candidate that appears to require its own primitive, examined, resolves into a mechanism operating on capacity. Value admits the same treatment, and the treatment is worth setting out, since value is the candidate most often granted primitive standing in accounts of conflict.

The natural view holds that parties committed to incompatible values conflict in virtue of the incompatibility, and §2 entered a first qualification: normative disagreement is common and mostly inert, becoming conflictual once joined to a requirement that the other convert or submit. That qualification located the conflict in the requirement. A second and more informative account is available, and it concerns what value does to a generative process.

Values leave generation intact and operate on its direction. A system committed to a value continues to generate, with its trajectory biased toward some regions of its possibility space in preference to others. Writing $\phi$ for a system’s generative dynamics and $\phi_v$ for those dynamics under a value commitment, the relation is one of orientation:

$$\phi ;\longmapsto; \phi_v .$$

The mapping preserves generativity and redistributes it. What the system can produce remains what it could produce; which of those productions it moves toward has been reweighted.

This places value within a mode already defined. Section §4 distinguished contraction of a continuation space from deformation of it, where deformation exchanges available continuations for others of comparable range and different character. Value operates in the deformation mode: a value commitment closes no continuation outright and biases the trajectory among those that remain. The correspondence was unforced, the mode having been introduced for other reasons, and it supplies value a place in the framework requiring no new apparatus.

The account earns its keep in the case of coupled systems, and intimacy furnishes the clearest instance. Two parties in a close coupling are frequently described as having different values, and the description is accurate at the level of what is said. What it names dynamically admits sharper statement. One party’s commitments orient the coupled system toward a region characterized by career, exploration, and mobility; the other’s orient it toward a region characterized by family, stability, and locality. Each orientation is coherent, each leaves the other party’s capacity formally intact, and neither is an error.

The difficulty lies in the coupling. Where two systems generate jointly, they generate along one trajectory, and one trajectory admits one orientation at a time. The parties are accordingly negotiating the direction of a shared generative process in place of coordinating their separate ones, and disagreement about values is disagreement about the future dynamics of the system they jointly constitute.

Claim 4. Value commitments operate as mechanisms of orientation: they preserve a system’s generative capacity and bias its trajectory toward particular regions of its possibility space. Where two systems generate jointly, their commitments bear on one trajectory, and value disagreement is accordingly disagreement about the direction of a shared generative process.

Three remarks, the first concerning what the claim withholds.

The orientation is descriptive throughout. What Claim 4 attributes to a value commitment is the direction it does impose, leaving aside any direction a coupling ought to follow. The distinction matters, since a preferred direction attributed to the coupling itself would reintroduce the specification §7 finds unavailable, and the reintroduction would be subject to the same objection.

The claim predicts a contrast case, and the prediction requires a criterion for jointness that the framework supplies only partially. Generation is joint, in the sense at issue, where the parties’ continuation spaces are coupled such that a continuation taken by one removes continuations from the other: their trajectories are constrained to one path in the respect the orientations concern. Generation is parallel where the coupling leaves each party’s continuations available whatever the other takes.

The criterion admits degrees and resists sharp application, since coupling is rarely total and rarely absent. Section §7.1 holds that parties are constituted within their couplings, from which it follows that generation between coupled parties is joint to some degree in every case, and the contrast is accordingly one of degree in place of kind. What the claim predicts, so qualified, is that value disagreement bears on conflict in proportion to the degree of jointness, and that parties whose continuations remain mutually available can hold divergent orientations with correspondingly less conflict. This accounts for a familiar observation, that value differences tolerable between colleagues become acute between parties whose futures are jointly generated, and it is weaker than a categorical contrast would be.

The claim connects to what follows. Value commitments are symbolic in the sense §6 specifies, and they close in the manner described there: a commitment conditions which interactions are undertaken, and the interactions so conditioned supply the experience by which the commitment is assessed. Value conflicts accordingly persist beyond the conditions producing them, which distinguishes them from conflicts of the demand kind and explains their characteristic duration.

§6 Symbolic Closure and the Persistence of Obstruction

Symbolic mechanisms appeared in §5 alongside the others, and the treatment given there was the correct one: a commitment to a normative or symbolic order produces obstruction under specifiable conditions, and it takes its place in the list without privilege. This section asks what warrants extending that treatment, and answers with a single property. The property is narrow, and the narrowness is the point, since a broader claim would install symbols as a second primitive and undo the architecture of §2.

6.1 The Property at Issue

The modal properties of §4 included duration, and a remark was entered there which is now discharged. Obstruction that persists after the interaction producing it has ceased requires explanation, since an interaction that has ended explains nothing about what follows it. Where obstruction outlasts its cause, something has taken over the work of producing it.

Symbolic mechanisms are distinguished by their capacity to do exactly this.

Claim 5. Symbolic mechanisms are those by which obstruction becomes self-reproducing. By fixing an interpretation, they regulate the subsequent interactions that would otherwise revise it, and obstruction so fixed persists beyond the conditions that produced it.

The claim concerns a mode of persistence and reaches no further. It says nothing about symbols constituting subjects, mediating experience, or structuring the social field, all of which may be true and none of which this paper requires. What it says is that a mechanism which fixes interpretations occupies a distinctive position with respect to the duration property, and the position is what earns the section.

6.2 The Mechanism of Closure

The structure by which fixing produces persistence can be set out in steps.

An interpretation, once held, conditions what a party attends to and how it construes what it finds. This much is unremarkable and holds of interpretations generally. What matters for closure is the case where the interpretation conditions the very interactions that would supply evidence against it.

Consider a party holding the interpretation that another is hostile. The interpretation conditions its conduct: it withholds, it prepares, it declines the openings it would otherwise take. That conduct conditions the other’s conduct in turn, and the other, meeting withholding and preparation, withholds and prepares in return. The first party now finds its interpretation confirmed by conduct its own interpretation helped produce. The interaction that might have revised the interpretation has been shaped by the interpretation into an interaction that confirms it.

Two features of this structure deserve marking.

The loop is closed with respect to evidence. Interactions continue to occur and continue to be interpreted, and what would count as disconfirming has been made improbable by the interpretation’s own operation. The party receives evidence that its own conduct has arranged, which differs from refusing evidence. This distinguishes closure from mere stubbornness, and it is why closure resists the remedies appropriate to stubbornness.

The loop is indifferent to the accuracy of what it fixes. An interpretation that was accurate when formed closes in the same manner as one that was mistaken, and the mistaken case is the more consequential, since the loop supplies the confirmation the interpretation lacked. This connects to the observation entered in §2 about identity: identities constituted in opposition remain fully operative while inaccurate, and closure is the structure that permits it.

6.3 Symbolic Mechanisms Among the Others

The relation between this section and §5 requires precision, since the extension granted here is easily overread.

Symbolic mechanisms remain mechanisms. They produce obstruction in the manner the other mechanisms do, and Claim 5 adds nothing to their productive role. What Claim 5 concerns is what happens after production: the obstruction they produce has a characteristic relation to duration, and the characteristic relation is what the section describes.

The extension is available to other mechanisms under conditions. Closure is a structural property, and a non-symbolic mechanism can acquire it where its operation likewise conditions the interactions that would revise its effects. An institutional arrangement that determines who is admitted to the deliberations by which the arrangement might be changed exhibits closure in the same structural sense. What distinguishes symbolic mechanisms is that closure is characteristic of them in place of incidental, since fixing an interpretation is what they do.

The property is a matter of degree. Interpretations are more and less firmly fixed, loops more and less tightly closed, and the openings by which revision might enter more and less numerous. Claim 5 concerns a tendency and admits the graded reading throughout.

6.4 What This Prepares

Closure is stated here as a general property of interpretation-fixing mechanisms. Its principal application in this paper is narrower and arrives in §7.

The reference models required by Definition 1 are interpretations of one party’s generative dynamics held by another. They are accordingly subject to closure in the manner described: a model held by $A$ of $B$’s dynamics conditions $A$’s conduct toward $B$, which conditions $B$’s conduct, which supplies $A$ with the evidence by which the model is assessed. A model closed in this fashion has ceased to be revisable by the party it models, and the argument of §7 is that this loss of revisability, in place of the model’s content, is what bears on obstruction.

Claim 5 supplies the mechanism for that argument. Closure explains how a reference model becomes unrevisable, and §7 states what follows once it has.

§7 The Reference Problem and the Rigidity of Working Models

Definition 1 was left standing with a difficulty attached, and §5 inherited it twice over. The difficulty is that obstruction is comparative: a diminution of capacity is a diminution against what capacity would otherwise have been, and the comparison requires a specification of the otherwise. This section takes up the difficulty, argues that the specification cannot be supplied in the form the difficulty seems to demand, and states what follows from the failure.

7.1 The Unavailability of a Privileged Trajectory

The natural specification is a trajectory the system would have followed absent the interaction: what the party would have generated had the obstruction not occurred. Two arguments tell against it, and they are independent.

The first is Hume’s. To speak of the trajectory a system would have followed, where that trajectory is doing normative work, is to move from a description of what is to a claim about what ought to have been. The move requires a warrant, and the descriptive facts alone supply none. A system’s history, its constitution, and its constraints determine what it did and what it can do; they leave undetermined what it should have done. A specification of the otherwise that carried normative weight would need such a warrant, and none is available from the resources this framework admits.

The second argument is internal to the framework and, for present purposes, the more decisive. Section §3 established that generative capacity is relationally conditioned: a system’s available continuations depend on the couplings it stands in. Generative Relational Being takes this further, holding that the systems in question are constituted within their couplings in place of entering couplings already formed. On this commitment the counterfactual required by the natural specification is malformed. The trajectory the party would have followed absent its couplings is unavailable for specification, since the subtraction the counterfactual requires removes the party whose trajectory was to be specified. The operation dissolves the thing it was meant to isolate.

The two arguments converge on the same result and are worth keeping distinct, since the first is a general constraint on inference and the second a consequence of this framework’s own commitments. A framework with individualist commitments would face the first alone.

One consequence should be registered immediately. The concept of alienation, in the form that defines it as deviation from what a subject would authentically have become, requires exactly the specification shown unavailable. This paper accordingly declines the concept, and §4 declined it already by treating deformation of a continuation space as a change in character requiring no account of what the space ought to have contained.

7.2 The Engineering Reading

The unavailability of a privileged trajectory suggests two responses, and one is a false dichotomy. It might be concluded that trajectories cannot be spoken of at all, and that Definition 1 fails for want of the comparison it requires. It might alternatively be concluded that some privileged trajectory must exist after all, and the arguments above be resisted. A third position is available, and it is the one taken here.

Treat the reference trajectory as an engineering construct carrying no ontological commitment.

The move has precedent in several mature frameworks, and the precedents share a structure. In control theory, a reference trajectory is the path selected relative to a control objective; nothing in the theory holds that the system ought to follow it in any sense beyond the objective’s own. In model predictive control, the reference is revised online as the horizon advances. In Bayesian inference, priors are working commitments held for their role in subsequent updating. In predictive coding, the brain operates as though it possessed a generative model of its environment, and the model is assessed by its performance in prediction and action in place of its correspondence to the environment’s ultimate structure.

What these share is that the model is an instrument. It is held for what it enables, evaluated by whether it enables it, and revised when it fails to. No claim about the world’s structure is entered, and none is needed.

Applied here: write $\hat{\gamma}$ for a working model of a system’s generative dynamics, held by a party for the purpose of coordination. The circumflex marks the model’s status: this is a provisional construct subject to revision in place of a fact about the system it models. The model changes as interaction proceeds, and its changing is part of the dynamics under study in place of a departure from a standard external to them.

Two questions follow, and the paper’s result lies in the second.

7.3 The Justification Constraint

The first question is what licenses the construct at all. An instrument is justified by the objective it serves, and the objective must be specified, since an instrument justified by an unspecified objective is justified by nothing.

The specification available here is narrow. A reference model is required because attributions of obstruction require a reference: to say that $A$ diminished $B$’s capacity is to say that something available to $B$ ceased to be available, and this states a comparison. Absent some working model of what was available, the attribution has no content. The model is accordingly the minimal apparatus that makes the primitive operational.

The narrowness of this justification is what keeps the position stable, and the point deserves emphasis because the position is easily destabilized. Should the justification be extended, so that the model serves the party’s flourishing, or its authentic development, or its proper end, the ontological commitment declined in §7.1 returns under another name. The model would then be answerable to a standard beyond coordination, and the standard would be precisely what the Humean argument forbids. The engineering reading holds only while its objective remains operational.

This partially discharges the debt entered in §4. Definition 1’s comparison is licensed by a working model, and the working model is licensed by its role in making the comparison operational. What the account does not supply is a guarantee that any particular model is the right one, and no such guarantee is available.

A further limitation requires statement, since it bears on how far the framework reaches and has been understated in the exposition so far.

The reference model licensed above is one held by a party to a relation, for the purpose of coordinating within it. An attribution of obstruction made by an observer outside the relation requires a reference model held by that observer, and the justification given above fails to license it. The observer coordinates with neither party and therefore lacks the objective by which a party’s model was justified. An observer’s model is answerable to nothing the account has specified.

Two consequences follow, and they narrow the framework’s claims.

Attributions of obstruction are, on the resources assembled here, well-formed from within a relation and ill-formed from outside it. A party may say that its continuations have been foreclosed, holding a working model of what was available; a second party may say the same; an observer saying it on their behalf is holding a model the framework cannot license. This is a considerable restriction and it is accepted.

The restriction bears on §8 and on the empirical programme of §10. A comparison of obstruction across two conditions, made by a theorist, is an observer’s attribution. What the construction of §8 can establish is accordingly narrower than a demonstration that obstruction differed: it can establish that the parties’ own attributions would differ, which is what §8 claims. Whether an observer’s attribution can be licensed at all is left open, and §10 enters it as the framework’s most pressing unresolved question.

What can be supplied, within the relation, is a condition on how models are held, and that is the subject of the following subsection.

7.4 Rigidity as the Obstruction-Relevant Variable

The second question is what distinguishes a working model held well from one held badly. Here the analogy with predictive coding requires a qualification, and the qualification yields the paper’s result.

In predictive coding the model is a single system’s model of an environment. The environment does not hold a model of the system, and it does not respond to being modelled. The case at issue differs in both respects. Each party holds a model of the other; each party is modelled by the other; and each is capable of revising the other’s model by acting in ways the model failed to anticipate. The models are mutually referential and the modelled parties are agents.

This difference has a consequence. In the single-system case, a model that fails is corrected by the environment, which supplies disconfirming evidence indifferently. In the mutual case, whether a model is corrected depends on whether the modelling party remains open to correction by the party modelled, and this openness is a variable of the relation.

Section §6 supplies the mechanism by which openness is lost. A model held by $A$ of $B$’s dynamics conditions $A$’s conduct toward $B$; $A$’s conduct conditions $B$’s conduct; and $B$’s conditioned conduct is what $A$ receives as evidence bearing on the model. Where the loop closes, the model is no longer answerable to $B$, and $B$’s capacity to revise it has been removed.

Whether closure degrades the correction channel admits an argument in place of stipulation, and the argument runs as follows.

Let $A$ hold a model of $B$ that $A$ treats as accurate. Corrections issued by $B$ are received by $A$ against that model, and the model determines how they are construed. Where the model is treated as accurate, a correction indicating that it is inaccurate is anomalous, and $A$ has two construals available: that the model erred, or that $B$’s conduct is inconsistent. Treating the model as accurate is what makes the second construal available; a party holding its model provisionally has only the first.

Section §3 established that generation is historically conditioned, so that what a system produces enters the conditions of its subsequent production. Applied to $B$’s corrective conduct: corrections met with the second construal condition $B$’s subsequent corrective conduct, and they condition it downward, since a correction construed as inconsistency achieves less than the correction intended and costs $B$ more than it returns. The rate of correction accordingly falls.

The consequence for $A$’s model is that the evidence bearing on it diminishes in the specific respect that would have disturbed it. $A$’s model is thereby held against a thinner correction channel, and the conditions that permitted the second construal are strengthened. This is the closure of §6 arising in the particular case of a model held of another party.

The argument requires that the model be treated as accurate in place of provisionally, and this is what the engineering reading of §7.2 was constructed to prevent. The two halves of this section are accordingly connected: the engineering reading is what keeps the second construal unavailable, and abandoning it is what permits closure to begin.

Claim 6. A working model of another party’s generative dynamics is admissible as an instrument while it remains revisable by the party it models. Where such a model comes to be treated as accurate in place of provisional, the argument above indicates that the correction channel degrades and the model tends toward unrevisability. What bears on obstruction, on this account, is the rigidity with which a model is held in place of the content the model asserts.

Four consequences follow, and they are the section’s substance.

The variable is a property of the relation. Rigidity concerns how a model is held and by whom it can be revised, both of which are facts about the coupling between the parties. It is accordingly a relational-locus phenomenon in the sense of §4, and it takes two parties to arise: a model unrevisable by the party it models presupposes a party it models.

Deviation loses its weight. Once the reference model is understood as non-ontological, a party’s departure from what another’s model of it predicted is model error and nothing further. The departure carries no normative significance, and the framework lacks resources with which to supply it any. This is what the declining of alienation amounts to in operational terms.

The claim is falsifiable. It predicts that unrevisable models are attended by rising obstruction, and that the relation holds with the content of the models controlled. A case in which models are held rigidly and obstruction fails to rise would tell against it, as would a case in which obstruction rises with revisability maintained. Section §8 constructs the comparison the prediction requires.

The claim connects the paper’s two later concerns. Symbolic closure, treated in §6 as a general property of interpretation-fixing mechanisms, is the mechanism by which rigidity arises; rigidity, treated here, is what closure produces when what is closed is a model of another party. The two sections describe one phenomenon at different levels of generality.

§8 A Constructed Case - Rigidity with Content Held Fixed

Claim 6 asserts that rigidity bears on obstruction and that content does not. Establishing this requires a comparison in which content is held fixed and rigidity varies, and the comparison is constructed here for that purpose.

8.1 The Standing of the Case

The case that follows is constructed in place of observed, and the difference in standing should be stated before the case is given.

Its function is to isolate a variable. A theoretical claim of the form “$X$ bears on $Y$ with $Z$ controlled” requires that $Z$ be controlled, and controlling $Z$ in observed material is difficult where the material was generated by processes indifferent to the theorist’s purposes. A constructed case supplies the control by stipulation. What it supplies in control it forgoes in evidential weight, and it establishes nothing about frequency, magnitude, or occurrence.

The case accordingly illustrates the argument of §7.4 in place of establishing it. That argument derived the degradation of the correction channel from the treatment of a model as accurate together with the historical conditioning of §3; the construction below exhibits the derivation in a particular setting, with the content of the models held fixed so that the role of rigidity is visible. What the construction adds is legibility. What it withholds is evidence, and whether the mechanism operates outside the construction is an empirical question §10 enters as such.

One further restriction applies, entered in §7.3. Attributions of obstruction are licensed from within a relation and unlicensed from outside it. The comparison below is accordingly stated in terms of what the parties’ own attributions would be, in place of a claim about obstruction made on their behalf.

The reader’s test is accordingly narrow. It is whether the two conditions below differ in one respect alone. Should they differ in a second, the case fails to isolate what it claims to isolate, and the objection is decisive.

8.2 The Construction

Two parties, $A$ and $B$, stand in a coupling within which they generate jointly: a shared undertaking whose trajectory depends on both. One decision before them requires that its timing be settled.

Each holds a working model of the other’s generative dynamics in the sense of §7. $A$’s model of $B$: $B$ is establishing a body of work whose early phase admits little interruption, and $B$ requires roughly three years in place before a change of circumstance is viable. $B$’s model of $A$: $A$’s work is portable, and $A$ can absorb a change of circumstance at short notice.

Three features of these models should be marked, since the construction depends on them.

They are substantively reasonable. Each is the sort of model a party would form on the evidence available, and neither is careless. Nothing in the case turns on either party being a poor modeller.

They concern generative dynamics in place of preferences. $A$’s model states what $B$’s work requires, leaving aside what $B$ wants. This distinguishes the case from a disagreement about aims and locates it where Claim 6 operates.

They are inaccurate in some measure, and both are. $B$’s third year is in fact the flexible one and the first two are not; $A$’s portability is real in respect of equipment and unreal in respect of the collaborators $A$ works among. The inaccuracy is stipulated because it is required: a comparison in which one condition contains error and the other does not would vary two things.

8.3 The Two Conditions

Condition I: models revisable.

$B$ states that the third year is the flexible one. $A$ receives this and adjusts the model accordingly. $A$ states that portability holds for equipment and fails for collaborators. $B$ receives this and adjusts in turn. Each correction is occasioned by conduct or statement the existing model failed to anticipate, and each is incorporated.

The disagreement about timing persists. It may persist for a considerable period, since the adjusted models place the parties’ constraints in tension and the tension is real. What the parties possess throughout is a model of the other that tracks the other’s dynamics within the tolerance their coordination requires. Neither is planning against a fiction. Obstruction remains low: the continuations available to each are affected by the coupling as any coupling affects them, and neither party’s capacity is diminished by the other’s model of it.

Condition II: one model unrevisable.

The models are as before, in content and in accuracy. $A$’s model of $B$ has closed in the sense of §6: it conditions $A$’s conduct toward $B$, and the conduct so conditioned supplies the evidence by which $A$ assesses the model.

The sequence proceeds as §7.4 indicates. $B$’s constraints develop, as constraints do, and $A$ continues to plan against the model as originally formed. $A$’s proposals are accordingly mistimed relative to $B$’s actual position. $B$ corrects, as in Condition I. $A$ receives the correction against a model on which the correction should not have been required, and construes it as inconsistency in $B$ in place of error in the model. $B$, meeting this construal, corrects less often, since correction is met with a construal that penalizes correcting. $B$’s dynamics become less legible to $A$ in consequence, and $A$’s model, now meeting less that would disturb it, is held with increased confidence.

What has been diminished is $B$’s capacity to develop under this coupling. $B$’s continuations are shaped by a model $B$ can no longer affect, and the shaping proceeds by $A$’s conduct, which the model conditions. This is relational obstruction in the sense of §4, and it accumulates in the manner §6 describes, persisting beyond the conditions that produced it.

8.4 What the Comparison Shows

The conditions are identical in the content of both models, in the accuracy of both models, and in the substance of the disagreement. They differ in whether one model is held provisionally or as accurate, and in the revisability that follows. The obstruction each party would attribute is low in the first condition and accumulating in the second.

Three competing readings require exclusion, since each would explain the difference without recourse to rigidity.

The difference is error. It might be held that $A$ was simply wrong about $B$, and that being wrong accounts for what follows. The reading fails on the construction. $A$ is wrong in Condition I to the same degree and in the same respects, and no obstruction accumulates there. Error occurs in both conditions and is absorbed in one of them. What differs is whether the error is correctable by the party it concerns, and this is rigidity in place of error.

The difference is disagreement. It might be held that the parties disagreed and that disagreement produced the obstruction. The reading fails likewise. The disagreement about timing is present in both conditions and persists in Condition I without obstruction. Removing the rigidity from Condition II leaves the disagreement standing and workable.

The difference was stipulated. It might be held that the construction builds in the feedback that produces its result, so that the loop is observed because it was assumed. This objection would be decisive against a construction offered as evidence, and it is answered by the standing entered in §8.1. The feedback is derived in §7.4 from the treatment of a model as accurate together with the historical conditioning of generation, and the construction exhibits the derivation in place of assuming it. What remains stipulated is the setting, which is what a constructed case supplies.

The difference is bad faith. It might be held that $A$ in Condition II is acting in bad faith, and that the case describes a moral failing in place of a structural one. The construction excludes this. $A$’s construal of $B$’s corrections as inconsistency follows from the model $A$ holds: on that model, the corrections should not have been necessary, and a party who takes its model to be accurate has grounds for the construal. The mechanism operates without ill intent, which is consistent with the silence about intention entered at Definition 1 and is a point in favour of the claim in place of against it.

8.5 Scale

The construction has been stated with two parties whose coupling is close, and a question arises about what the argument requires of that setting.

The argument requires two systems generating jointly, each holding a working model of the other’s dynamics, with the models bearing on conduct. Nothing in it requires that the systems be individuals. States hold models of other states’ developmental constraints and act on them; institutions hold models of the populations they administer; research communities hold models of the programmes they assess. In each, the model conditions conduct, the conduct conditions what the modelled party does, and the loop closes or remains open.

This is a substantive claim in place of an assumption, and it is defended by the structure of the case: the construction refers to properties of the modelling relation throughout, and refers nowhere to properties peculiar to persons. Should the claim fail, the failure would consist in some property of the modelling relation that holds between persons and fails between collectives, and identifying such a property would be a contribution the framework would need to absorb. Section §10 enters the question.

The scale-invariance claim carries a consequence worth stating. If it holds, the case of §8 bears on the international setting with which §2 began, and the realist observation that states misread one another’s constraints becomes an instance of the mechanism described here in place of a separate phenomenon. Establishing this would require the comparison of the present framework with power-based accounts that this paper defers, and §10 names it as the first item of subsequent work.

§9 What the Framework Declines and What the Declining Costs

A framework built by subtraction owes an account of what it removed. This section states what has been declined, why, and at what cost, since a limitation entered without its cost is a limitation understated.

9.1 The Normative Layer

Justice, fairness, value in its evaluative sense, ethics, and happiness have been absent throughout, and their absence is deliberate. The reason is architectural. Each would enter as a further primitive, and a framework attempting a minimal primitive base cannot admit further primitives while the first remains undischarged. The order proposed is that obstruction be worked out first, and that normative questions be raised on that basis by whoever finds the basis sound.

The cost is considerable and should be stated plainly. The framework, as it stands, distinguishes obstruction from its absence and declines to rank obstructions. It supplies no ground for holding that one party’s foreclosure of another’s continuations is objectionable. It supplies no ground for preferring less obstruction to more. A reader who asks what follows practically from the account receives, at this stage, the answer that nothing follows practically, since practical conclusions require premises the account withholds.

One consequence of the paper’s own argument sharpens this. Section §7 established that reference trajectories carry no ontological standing, from which it follows that a party’s departure from another’s model of it is model error carrying no normative weight. Whatever normative purchase the notion of a trajectory once seemed to offer is thereby withdrawn. What survives as a candidate for normative treatment is unrevisability, which is a property of a relation and would require its own argument. The framework identifies the candidate and stops.

9.2 Intention

Definition 1 attaches to what an interaction does to capacity and remains silent about what the interacting party intended. The silence keeps the primitive mechanical, and §8 showed the silence earning its place, since the case turns on a mechanism operating without ill intent.

The cost is that moral assessment requires the distinction the framework withholds. Harm inflicted knowingly and harm inflicted in the course of reasonable conduct differ in ways any account of responsibility must register, and the present account registers neither. A framework wishing to support such assessment would need to reintroduce intention as a further term, and the reintroduction would carry consequences for the primitive that this paper has not examined.

9.3 Alienation

The concept has been declined twice, at §4 and §7, and for the same reason: in the form that defines it as departure from what a subject would authentically have become, it requires a specification of the authentic that the arguments of §7 find unavailable.

A second exclusion operates alongside the first and was entered in §4. Cases of the alienation kind involve deformation of a continuation space, and deformation stands outside the partial order on which Definition 1 rests. Such cases are therefore excluded twice: once for want of the authentic baseline, and once for want of any relation between the spaces the order can supply.

What has been retained is the description. A system whose continuations have been exchanged for others of comparable extent has undergone a change of character, and this is statable without reference to what the continuations ought to have been. Whether such a change constitutes obstruction is left open by the framework, and the openness is a gap in place of a position.

What has been forgone is the critical force the concept carried. To say that an arrangement alienates is to condemn it; to say that it deforms a continuation space is to describe it. The description is available to the framework and the condemnation is not, and a reader who valued the concept for its critical force will find the substitute thin. The reply available is that critical force resting on an unavailable specification is force borrowed against an account nobody has supplied, and that identifying the debt is worth something. Whether it is worth what it costs is a fair question and is left open.

9.4 Attribution from Outside a Relation

Section §7.3 found that the account licenses reference models held by parties to a relation and fails to license models held by observers outside it. This restricts the framework more severely than the exposition of §4 through §5 suggested, and the restriction is stated here in full.

Attributions of obstruction are, on the resources assembled, well-formed from within a relation. A party holding a working model of what was available to it may say that continuations were foreclosed. An observer saying so on that party’s behalf holds a model the account cannot license, since the observer coordinates with no one and the justification of §7.3 was a justification by coordination.

What this costs is most of what a social-scientific framework is ordinarily wanted for. Third-party assessment, comparison across cases, and aggregate claims about obstruction in a population are all observer attributions. The framework in its present state supports none of them, and a reader whose interest was in such assessment should regard the account as not yet serving it.

Two mitigations are available and neither is developed here. An observer’s model might be licensed by a coordination objective of the observer’s own, in which case attributions would be relative to that objective and would require its statement. Or an observer might report the parties’ attributions in place of making one, which is available and considerably weaker. Which of these can be made to work is entered in §10 as the framework’s most pressing open question.

9.5 The Vagueness of Relational Significance

Definition 2 converts obstruction into conflict by way of relational significance, and §5 offered a candidate threshold without confidence. The candidate rests on the parties’ working models, which §7 finds to be of limited standing, and it may prove too permissive. What can be said in mitigation was said there: the vagueness attaches to the manifestation and leaves the primitive alone, so that the framework’s sharper question is the one it is built on. What cannot be said is that the vagueness is harmless. A framework whose central derived concept has an imprecise threshold will yield disputed classifications in the cases where classification matters most.

9.6 The Enumeration of Mechanisms

The mechanisms treated in §5 number seven, and their selection was governed by their standing in the existing literature in place of any principle internal to the framework. Others exist. Honour, grievance, and the emotions have claims to inclusion and have been omitted, and no argument has been offered that the seven are jointly exhaustive or individually irreducible.

This is a real gap and its shape should be understood. The framework’s claim concerns the phenomenon these mechanisms produce, which leaves the enumeration incomplete by design. What the incompleteness costs is any account of how mechanisms relate to one another: whether some reduce to others, whether they compose in regular ways, whether the list admits a principled closure. These are questions the framework poses and does not answer.

§10 Directions for Subsequent Work

The items below are ordered by their bearing on the framework’s standing, beginning with those on which it is most exposed.

10.1 The Licensing of Observer Attributions

This is the framework’s most pressing unresolved question, and its resolution conditions most of what follows.

Section §7.3 licensed reference models by their role in coordination, which restricts well-formed attributions of obstruction to parties within a relation. Whether an observer’s attribution can be licensed at all is open. Three routes suggest themselves. An observer might be granted a coordination objective of its own, relative to which its model would be licensed and its attributions indexed. An observer might be restricted to reporting the parties’ attributions, which is available at once and yields a considerably weaker framework. Or the justification of §7.3 might be replaced by one that does not run through coordination, in which case the argument of that subsection would require reworking throughout.

Until one of these is developed, the framework’s claims should be read as claims about what parties to a relation can say of it.

10.2 Extension of the Ordering

Definition 1 rests on a partial order, and the incomparable cases stand outside it. Whether the order can be extended to reach the deformation cases without importing evaluative commitments is open. What would be required is a relation between continuation spaces of comparable extent and different character that is neither a measure of size nor a ranking by worth. Whether such a relation exists is a question about the structure of continuation spaces, and it is prior to the formalization described below.

10.3 Comparison with Power-Based Accounts

The framework’s claim to have gained something over the traditions surveyed in §2 has been argued and remains untested. What would test it is a case that a power-based account explains well, treated in both vocabularies, with the differences in what each account sees recorded. The case would need to be historical in place of constructed, since what is at issue is explanatory purchase on material neither framework arranged.

Section §8 indicates where such a case would be found. If the scale-invariance argument holds, the misreading of one state’s developmental constraints by another is an instance of the rigidity mechanism, and cases in which such misreading persisted through changing capability distributions would separate the accounts. A power-based account attributes the persistence to the distribution; the present account attributes it to closure of the models, which is a distinct claim admitting distinct evidence.

10.4 Empirical Test of the Rigidity Claim

Claim 6 is falsifiable and untested. Testing it requires the comparison of §8 in observed material: cases in which the content of parties’ models of one another is comparable and their revisability differs. The prediction is that obstruction accumulates with unrevisability and remains low with revisability maintained.

Two difficulties attend the test and should be anticipated. Revisability is a dispositional property and is observed through its exercise, so that a model never challenged supplies no evidence of whether it would have yielded. And obstruction requires the comparison of capacities, which §7 showed to rest on working models held by the observer as much as by the parties. Neither difficulty is peculiar to this framework, and neither is dissolved by noting that.

10.5 Formalization

The apparatus has been kept in prose throughout, and the abstention was deliberate: a formalism introduced before the objects are stable decorates the argument in place of advancing it. The objects are now stated with sufficient precision that formalization is a reasonable next step, and three questions would organize it. What structure represents a continuation space, given that it admits contraction, expansion, and deformation, and that deformation preserves extent. How does obstruction propagate and accumulate along a trajectory, given its path-dependence. And what conditions on a relation correspond to the revisability of a model held within it.

The connection to dynamical systems is the natural one to pursue, with generative dynamics answering to a flow, continuation spaces to reachable sets, and closure to the formation of an invariant region. Whether these correspondences survive contact with the phenomena is the question such work would settle.

10.6 The Formation and Dissolution of Closure

Section §6 described closure and left its dynamics unexamined. What conditions bring a loop to close, and what conditions open one that has closed, are questions the framework poses and requires other resources to answer. The second is of particular interest, since closure is described here as self-reproducing, and the mechanisms by which self-reproducing structures come apart are less obvious than those by which they form.

10.7 Scale-Invariance

Section §8 claimed that the rigidity mechanism holds wherever two systems generate jointly while modelling one another, and defended the claim structurally. The claim is open to refutation by the identification of a property of the modelling relation that holds between persons and fails between collectives, and such a property may exist. Collectives model through procedures, and procedures may close in ways individuals do not, or resist closing in ways individuals cannot.

10.8 The Threshold of Relational Significance

Definition 2’s threshold was offered provisionally in §5 and may prove too permissive. What is wanted is a condition distinguishing obstruction that organizes a relation from obstruction the parties absorb, stated in terms the framework can support and tested against cases where the classification is disputed.

10.9 The Normative Layer

Whatever normative theory is built on this basis would begin from the candidate §9 identified: unrevisability, as a property of relations, is what remains after trajectories have been deprived of normative standing. Why a relation’s openness to revision by its parties should carry normative weight is a question this framework poses and lacks the resources to answer. It is the question a second paper would take up.

§11 Closing

Thesis. Conflict is proposed as the manifestation of generative obstruction within constrained relational dynamics. The present work refrains from exhausting the mechanisms through which such obstruction arises, aiming to establish a minimal foundation on which mathematical, symbolic, and normative theories may be built.

What the paper has added to that proposal is one substantive result and a clearer view of what the proposal costs. The result: attributions of obstruction require reference models; reference models admit no ontological standing; a model treated as accurate degrades the channel by which it would be corrected; and what bears on obstruction is accordingly the rigidity with which such models are held in place of the content they assert. The cost: the primitive is defined by a partial order that leaves deformation cases unclassified, and attributions are licensed from within a relation while remaining unlicensed from outside one. Both are entered in §10 as open, and the second conditions how far the rest should be read.

中文

生成性阻碍

通往一种关于生成性关系存在中冲突的极简理论

黄万宏 · huangwanhong@serendip.ngo

---

摘要

伊壁鸠鲁把幸福安置在两种扰动的缺席之中：身体之痛与灵魂之乱。这一区分在成为享乐学的说法之前，先是结构性的，它把从内部阻挠一个系统的东西与在其同他者的耦合中阻挠它的东西分开；本文保留这一结构而舍弃其享乐学内容，所问的是什么阻碍了生成，而非什么扰乱了宁静。既有的冲突理论从互不相容的原初对象出发：权力、利益、价值、身份、制度。每一种都解释了真实的现象，每一种都引入了其他理论并不共享的本体论起点。本文提出单一的原初解释对象，即生成性阻碍，定义为任何削弱一个受约束的生成性系统持续生成能力的互动，并把冲突当作此种阻碍的显现，而非当作一个原初对象。权力、资源竞争、误解、未获满足的需求以及象征性承诺，于是显现为产生同一现象的机制，而非对冲突相互竞争的定义。本文的实质结果关乎阻碍的归属所要求的参照轨迹：由于对一个由关系构成的主体而言并无本体论上享有特权的轨迹可用，此类轨迹只能作为为协调而持有的工作模型而被接纳，并且本文给出一个论证：一个被当作准确的模型会败坏本可用来矫正它的通道，因而是其刚性而非其内容与阻碍相关。一个构造出的案例展示了这一论证。本文对框架的界限作了充分陈述：阻碍由一个偏序定义，而该偏序把一类重要的情形留作未分类；并且阻碍的归属在关系之内获得许可，而在关系之外则未获许可，这一点对该说法的限制比其阐述所暗示的更为严厉。

关键词： 生成性阻碍；冲突；生成性关系存在；参照轨迹；可修正性；象征性闭合。

关于本文定位的一则说明。这是一篇基础性论文，提供的是一个替代物，而非一套冲突化解理论。它提出一个原初解释对象，并推演出只接纳该对象所能得出的东西；它所尝试的规训是减法式的，其大部分工作在于拒绝一套冲突理论本可被期待携带的诸多概念，其中包括正义、价值、公平与异化。它的核心结果是一个关于参照模型之刚性的主张，从框架自身的资源中论证出来，并在一个构造出的案例而非经验案例上加以展示。它的若干承诺，在无从落实的地方被留作悬而未决的问题，这些问题被汇集在最后两节，而非作为限定条件散布各处。倘若能证明该原初对象所做的工作既有原初对象已经在做，本文便失去了立足之处。反对、更正与反证，欢迎寄至 huangwanhong@serendip.ngo。

§1 引言

伊壁鸠鲁以否定的方式、并借两个词项来定义幸福。它由aponia，即身体之痛的缺席，与ataraxia，即灵魂之乱的缺席，构成。这一定义通常被当作享乐学教义来读，并作为享乐学教义来评估：宁的缺席是否是对好生活的可信说明，宁静的生活是否是值得选择的生活。如此读来，它属于伦理学史，而与冲突理论关系不大。

还有另一种读法。搁置享乐学内容，只留意其结构，那么这一定义所提供的，是把阻挠划分为两个所在。身体之痛在一个系统自身诸过程的运作中阻挠它；它是系统内部的阻挠，无需第二方。灵魂之乱，在伊壁鸠鲁的处理中，特征性地起于横亘在一个人与他者之间、或一个人与万物秩序之间的东西：起于对诸神的畏惧、对死亡的畏惧、政治纠缠的诸般要求，以及在其目的与自己不同的他者之间生活所带来的扰动。它是一种预设了耦合的阻挠。伊壁鸠鲁由此所区分的，是麻烦发生的两个所在，而非同一麻烦的两种强度：系统之内，与系统同其所耦合之物的关系之中。

本文取这一划分而弃其余。有三样东西被舍去。享乐学内容被舍去，因为这里没有任何东西表明扰动的缺席就是人之善，而本论证对它是否如此漠不关心。终点被舍去：伊壁鸠鲁式的幸福是一种要被抵达并维持的状态，而下文所关切的是持续进行的诸过程，而非会安定下来的诸状态。否定式的定义作为定义被舍去，尽管作为一种启发被保留，因为下文的关切落在阻碍是什么以及它如何传播之上，这是一个关于机制的肯定式问题，而把关于某种状况的否定式问题搁在一边。经此减法而存留下来的，是两所在的结构，而它以转置的形式存留。伊壁鸠鲁所问的是什么扰乱宁静，本文所问的是什么阻碍生成；他所区分的是身体之乱与灵魂之乱，本文将区分的是对一个系统自身生成性动态的阻碍，与对其在关系耦合下发展这些动态之能力的阻碍。

这一转置有其价值，值得把它所换来的东西说清楚。一套凭自身权威引入内部与关系之区分的理论，其实是在引入第二个原初对象，并欠下一个为该划分辩护的论证。从一个出于全然别样的理由、并且早在有任何生成理论可供服务之前就已划出该区分的地方取来这一区分，这在某种程度上是一个证据，表明这条缝在于现象而非在于理论。这一划分是承继而来的；新的只是划分它所用的词项。

冲突是社会科学中被理论化得最充分的对象之一，而困难在于诸解释所采取的形式，把稀缺搁在一边。现实主义从权力出发，把冲突读作其分布所引出的东西。自由制度主义从制度出发，把冲突读作其缺席或孱弱所容许的东西。唯物主义的说法从生产关系出发；建构主义的说法从身份及其形成出发；规范性的说法从关于价值的分歧出发。这些说法各自解释真实的案例，并且对其中某些案例解释得比其对手更好。它们所不共享的，是一个共同的对象。每一种都从一个不同的原初对象出发，而原初对象不同于变量：它是理论为了解释其余一切而拒绝去解释的东西。当诸理论在其原初对象上有别时，它们并不在通常意义上相互不合，因为它们并不在就同一对象作出互不相容的主张。它们是在构成不同的对象。

其结果是，累积起来的那一大批冲突理论并不累积。一个传统之内的发现会复利式地累加；跨传统的发现则连并置都难，因为其行为被解释的那个实体，在每一种传统中都被以不同方式构成。一个关于权力不对称的发现和一个关于身份形成的发现，可能都是可靠的，却仍然彼此不相干，因为缺一个可供两者归指的共同对象，把数据搁在一边。

这提示了一个问题，它先于诸传统之间的争胜，且无需去裁决它。假设权力、制度、身份与价值，等于是关于冲突如何被生产出来的相互竞争的说明，而非关于冲突是什么这一问题的相互竞争的答案。那么它们根本就不是对手。它们会是机制，而机制可以毫不尴尬地共存：数种机制可以同时运作，它们可以相互复合，一种可以在一个案例中占主导而另一种在另一个案例中占主导。这样一种读法所要求的，是对它们各自所产生的那个现象的界定，一个比它们中任何一个都更原初的对象，凭它每一种机制都可被描述为通往一个共同结果的一条路径。供出这样一个对象，是本文所尝试的工作。

来自医学的类比富有启发，值得直白说出，因为它固定住了这里所作动作的种类。发烧无法按病原体来分类。它在生理学上、作为一种现象被刻画，而病毒、细菌、自身免疫过程与恶性肿瘤，随后被理解为能够产生它的诸种不同机制。那一架构所成就的，超出了相互竞争的发烧理论之间的一个折中；它是使相互竞争的诸理论成为相互补充的诸理论的东西。这里的提议在架构上是同样的，冲突处在发烧的位置，生成性阻碍处在生理学刻画的位置。

所提议的原初对象是生成性阻碍：任何削弱一个受约束的生成性系统持续生成能力的互动。这些词项在§3与§4中展开；一开始要紧的是这一承诺的形状。

该形状的两个特征把它同它要取代的诸原初对象区分开来。第一是它的动态特性。权力分布、制度密度与身份配置，都是一个情境在某个时刻所具有的性质，而建立在它们之上的理论是在评估诸时刻上的诸情境。生成能力是一个过程的性质，即一个系统能够继续生成什么，而它无法从一个瞬时配置读出。阻碍因而是路径依赖的：一个此刻削弱能力的互动可能在此后扩大它，而任何静态量的同一分布，可能是阻碍性的也可能不是，取决于它所属的轨迹。这作为该原初对象的一个后果而随之而来，无需附加任何假设，§4即以此推导出它。

第二是它是关系性的，却又不可化约为固定各方之间的关系。其生成性成为问题的那些系统，是在它们所处的耦合之内被构成的；它们并非以业已成形的姿态进入关系。这一承诺，在生成性关系存在的框架中由来已久，正是使本文核心困难变得尖锐的东西。任何对阻碍的归属都要求一个参照，因为说 $A$ 阻碍了 $B$，就是说 $B$ 本会生成的某样东西被封闭掉了，而一个由关系构成的主体并无享有特权的基线可充当那个参照。该主体在孤立中本会遵循的轨迹作为参照不可用，因为它所预设的那种孤立从来不是该主体的处境。这一困难在§7中被直接处理，而本文的实质结果正是从中浮现出来的：参照轨迹作为工作模型而非事实被接纳，而与阻碍相关的变量，是一个模型是否仍对它所建模的那一方保持开放、可被修正，把它偏离其所建模者多远搁在一边。答案是局部的。它许可关系中各方所持的模型，而不许可关系之外的观察者所持的模型，这大幅收窄了框架的触及范围，并在§9中如实陈述。

一套在其领域之内解释一切的理论，通常已然未能界定一个领域。本文陈述其范围，而非将其最大化，且开篇即引入两条限制，因为它们塑造着下文。第一条关乎机制：本文不提供、也不尝试对产生阻碍的诸机制作穷尽的分类。§5讨论了数种；其他机制无疑存在；框架的主张关乎它们所产生的现象，而非任何一份生产者名单的完备性。第二条关乎该原初对象自身的触及范围。并非每一场冲突都是生成性阻碍的显现。两方可能想要互不相容的东西，而其中任何一方都未削弱另一方的生成能力，由此产生的冲突是真实的，有时是严重的，并且落在本框架所解释的东西之外。与其凭规定把这类案例吸收进来（这在每一个案例中都可行，却什么也保证不了），§5供出一条判据，用来判断一场给定的冲突是否落在框架的触及范围之内。一套能说出自己在哪里不适用的框架，强于一套无法被证明不适用的框架。

规范性问题被完全拒绝。正义、公平、价值、伦理与幸福，都要求一个进一步的层次，在此不予处理。§9说明缘由，并指出本文自身结果的一个后果：一旦参照轨迹被理解为非本体论的，偏离它们便失去了它们看似携带的任何规范性分量，而由不可修正性来承担这份工作。

其余部分如下推进。§2概览既有的诸解释性原初对象，并陈述把每一个化约为一种机制的做法。§3铺陈受约束的生成性动态：作为过程的生成性、约束，以及生成能力。§4定义生成性阻碍，依照上文对伊壁鸠鲁的读法按所在划分它，并推导其模态性质，包括路径依赖。§5把冲突当作阻碍的显现来处理，把诸机制不加特权地排列开来，并借未获满足之需求的案例给出范围判据。§6分离出使象征性机制值得单独处理的那一条性质，即它们使阻碍自我再生产。§7接手参照轨迹的问题，给出工程学读法，并陈述刚性主张。§8在一个把内容与分歧固定不变、只让可修正性变化的构造案例上，演示该主张。§9陈述本文不予尝试的东西，§10陈述它留作悬而未决的东西。

§2 把既有原初对象化约为机制

这里所概览的诸传统，每一个都解释一类案例，且解释得很好。本节的目的，是问一问：把每一个都读作一种关于生产的说明，会得出什么，而该传统本身提供的是一种关于构成的说明。所提议的化约采取一致的形式：某个传统主张冲突就是某种配置，这里所提供的读法则主张，该配置是生成能力得以被削弱的一条路径。

权力是最成熟的案例，也是最富教益的。现实主义的说明把冲突定位在能力的分布之中：处在无政府状态中的各方留意相对位置，而冲突随该位置所造成的激励而来。这一说明预测甚多。它未加言明的，是为什么一个分布竟应当要紧，而当这一问题被追问时可得的答案很能说明问题。一个不利的分布之所以要紧，是因为较强的一方由此能对较弱的一方做什么，也就是说，因为它能封闭掉的诸种未来的范围。这一诉求已然诉诸分布之外的某样东西。分布充当机制；对可能诸未来的封闭是该机制所产生的东西。这也解释了一个已知的弱点。发展彼此交叠甚少的各方之间的能力落差，所封闭的相应甚少，而这类案例正属于现实主义处理得最糟的那些。

利益与资源竞争容许同样的读法。按标准说明，两方需要同一有限资源便发生冲突，因为该资源抗拒分割。稀缺本身把此事留作悬而未决。使一场资源争夺具有冲突性的，是一方对该资源的获取取走了另一方为继续发展所需要的某样东西，而这场争夺的冲突性会随以下情况成比例地减弱：随着替代品的存在、随着时间尺度容许顺次取用，或随着该资源被证明对任一方的延续都可有可无。资源竞争因而作为一种机制运作，其阻碍之力随关于生成性依赖的诸事实而变，而那些事实关乎各方，而非关乎该资源。

制度以同样的方式、只是符号相反地进入。制度主义者观察到，冲突在制度稠密处被压制、在制度稀薄处爆发，并以信息、承诺与未来之阴影来解释这一点。用这里所提议的语汇来描述，制度稳定住各方得以在彼此在场的情况下继续发展的诸条件：它们使封闭更不可能、更不持久。它们的缺席使阻碍无所管辖。这说明了制度密度作为和平之预测指标的不完善，因为一个以牺牲另一方发展为代价来稳定一方发展的制度，只是局部地压制冲突，并可能把冲突移置。文献在处理不平等制度秩序上持久的困难，是测量密度而非测量密度本应确保之物的一个症候。

身份是化约最不显然、也最值得去作的案例。建构主义者主张，各方依据它们把自己和他者当作何人来行动，而冲突随在相互对立中被构成的诸身份而来。读作一种机制，身份固定住一方把自己的延续当作需要什么、把另一方的延续当作威胁着什么。在对立中被构成的诸身份，是那种在其之下各方发展被理解为需要遏制对方的身份。这样一种理解是否准确，仍是一个另外的问题，而它在不准确时仍能保持完全运作的能力，给出了对§6详加处理之物的第一个提示：固定诠释的诸机制，同阻碍的持续处在一种独特的关系之中。

价值分歧完成了这一概览。承诺于互不相容的规范秩序的各方，被主张凭那不相容而冲突。规范性分歧是常见的，而且大多是惰性的。两方可能持有关于善的不可调和的看法，而彼此的发展毫发无损，只要任何一方都不把对方的皈依或屈服当作自身延续的条件。价值分歧一旦与这样一种要求相接，便成为冲突，而是那要求承担着分量。这解释了为什么深层多元与派系冲突之间的差别，未能追踪分歧的深度。

一个共同的结构贯穿这五个案例。每一个传统都命名了某样真实的东西；在每一个之中，使被命名之物具有冲突性的，是一个进一步的事实，关于一方对另一方延续的封闭。诸传统在它们所命名之物上有别，却心照不宣地在什么使之要紧上一致。那一心照不宣的一致，正是本提议所要挑明的东西。

主张 1。 权力、资源竞争、制度配置、身份与价值分歧，容许一种一致的读法：作为产生同一现象的诸机制。在每一情形中，使被命名的配置具有冲突性的，是一个进一步的事实，关于对一方持续发展的封闭，而这一进一步的事实为所有五者共享。

由此得出两个后果。诸传统成为相互补充的，因为机制在诸定义相互排斥之处可以复合。而一份负担被承担下来：那个共同的现象要求以不预设任何机制的词项来界定。倘若生成性阻碍被证明只能借诉诸权力、或利益、或身份来定义，那么该化约便是循环的，而这一工作所得甚微。清偿那份负担，是随后两节的工作，这正是它们对冲突保持沉默的缘由。

§3 装置：生成性、约束与能力

本节陈述极简的装置，关乎系统及其发展。它不引入任何关于冲突、阻碍或各方之间关系的概念，这一切都在§4中进入。这一次序由方才承担下来的负担所要求，它要求该原初对象可独立于它所统一的诸机制而被界定。

生成性居首。一个系统在此处所用的意义上是生成性的，如果它产生此前不在其中的结构，且该结构成为它接下来所产生之物的基础。重点落在第二个从句上。变化本身把生成性留作未定：一个振荡的系统穿行于诸新颖状态，而其生产之能力保持固定；一个被从平衡态扰动又返回其上的系统，止于它开始之处。使生成有别于此的，是它的诸产物进入后续生产的诸条件，从而系统自身的历史约束并使能它可以接着造出什么。一个研究纲领在此意义上生成，因为结果成为后续结果的工具与问题。一个发育中的有机体如此，一门语言如此，一个制度如此。共同的特征在于把已被产生之物纳入生产能力本身。

一则限定属于此处，因为方才所陈述的条件容许一种强于本意的读法。纳入涵盖一个范围内的诸情形。一端坐落着严格的累积，其中诸产物被保留并复利；另一端坐落着单纯的条件化，其中被产生之物塑造着接下来能有什么，却不留下自身任何持存的痕迹。一场对话在第二种意义上生成：已被说出的话条件化着接下来能说什么，而其中留存下来的甚少。两者都落在该词项的范围之内，而该条件只要求生产在历史上被生产所条件化，把累积容许为诸情形中的一种。较弱的读法自始至终是那个起作用的读法，而§4不依赖任何更强的东西。

生成性因而是诸过程的性质，只以派生的方式属于诸状态。一个状态容许独立于它如何被达到的描述；一份生成能力抗拒这样的描述，因为一个系统接下来能生产什么，取决于它已经生产了什么以及生产的方式。占据同一状态的两个系统，可能在生成能力上相差甚多，因为它们经由不同的路径抵达，从而累积了不同的工具、封闭了不同的选项、习得了不同的依赖。这一点在后文承载可观的分量，值得作为生成本身的一个特征加以标记，在此先于任何关于冲突的主张而陈述出来。

约束随之进入。这里成为问题的那种生成性过程，在诸条件之下运作，这些条件决定哪些生产对它们可用，而这些条件分为几类。它们是物质的，就资源与能量有限而言；是结构的，就已被产生之物封闭一些路径又开启另一些而言；是时间的，就生产需要时间且某些序列抗拒压缩而言；是关系的，就一个耦合于他者的系统在那些他者部分设定的条件下发展而言。约束作为生成的条件而运作。一个不受约束的过程是无形的，因为约束正是使一些生产而非另一些成为可用、从而给过程一个方向的东西。因此，受约束的生成性系统这一词项标记的不是任何特殊情形。它标记的是那个一般情形。

生成能力是第三个也是最后一个词项：一个系统从它此刻所立之处、鉴于其历史与其约束，能够继续生成什么。该概念的三个特征需要评说，每一个都在后文承载分量。

能力是前瞻的，因而是被推断而非被观察的。被观察的是一个系统已经生产了什么；能力所关乎的是它能够生产什么，而两者之间的推断是可被击败的。一个系统可能生产甚少而保有大量能力，也可能生产甚多而耗尽了使那生产成为可能的诸条件。记录与能力之间的这一裂隙，正是需要该概念的缘由，因为一套只留意已被生产之物的理论，将缺乏手段去区分一个被阻止的系统与一个尚未行动的系统。

能力抗拒被表示为一个标量。它近似于一个可用延续的空间，而它以一个单一数字拙劣捕捉的方式变化。它收缩，当此前可用的诸延续不再可用之时。它扩张，当新的工具或耦合开启了此前不存在的诸延续之时。它变形而在幅度上既不收缩也不扩张，当可用的诸延续被换成另一些范围相当而特性不同的延续之时。第三种可能容易被忽略，而它在§4中要紧，因为一个能力被整体重定向的系统，可能在任何幅度量度上都不记录任何损失，而在那要紧的方面已被改变。

能力是被关系条件化的，且通常在相当大的程度上如此。一个系统的可用延续取决于它所处的诸耦合：一个研究纲领能生成什么取决于它的合作者与读者，一个国家能生成什么取决于它的邻邦，一个人能生成什么取决于他所生活于其间的那些人。这陈述的是能力的一个条件，且止步于任何关于阻碍的主张，因为一个耦合可能既容易扩大能力也容易缩减它，而§4既需要扩大的情形也同样需要缩减的情形。

三个词项现已就位：生成性、约束与生成能力。没有一个指涉权力、利益、制度、身份或价值。没有一个指涉冲突，也没有一个尚且涉及一次多于一个系统。§2结尾所承担的负担被清偿到如此程度：阻碍将在其中被定义的那套语汇，独立于阻碍本应统一的诸机制而立。

§4 生成性阻碍：定义、所在与模态性质

§3的装置关乎单个系统。阻碍要求两个，它在此分三步引入：定义、依照§1对伊壁鸠鲁的读法按所在的划分，以及由定义连同已给出的能力说明而随之而来的模态性质。

4.1 延续空间上的序

§3主张生成能力抗拒被表示为一个标量，而近似于一个可用延续的空间。把阻碍定义为能力的削减，因而要求一个关于削减在这样一个空间上意味着什么的说明，而该说明必须在定义被陈述之前给出。

可用的资源是一个偏序。当一个系统在一个条件下可用的诸延续，处于它在另一个条件下可用的诸延续之中时，能力便在一种无需进一步装置的意义上被削减了：当 $G’$ 中的诸延续被包含于 $G$ 的诸延续之中时，记为 $G’ \preceq G$。如§3所述的收缩，是这一关系严格成立的情形，而扩张是其反面。

该序在严格意义上是偏的，而不可比较的诸情形正是§3所命名的变形：一个被换成另一个范围相当而特性不同的延续空间，同它的前身不处在任何包含关系之中。$G’ \preceq G$ 与 $G \preceq G’$ 都不成立，因而下面的定义未能对这类情形加以分类。

这是本表述形式的一个局限，并如实引入。有两种回应可用，而这里都不采取。延续空间上的一个测度会按幅度把不可比较的诸情形排序，代价是把幅度相等的诸空间当作等价，而这恰恰是变形这一模态被引入去否认的东西。一个评价性的序会按诸延续的价值给它们排序，而这要求§9所拒绝的规范性装置。就目前之姿态，框架把可比较的诸情形排序，而把不可比较的诸情形登记为一种独特的、要求它并未供出之处理的现象。

4.2 定义及其参照问题

定义 1。 一次互动构成对一个系统的生成性阻碍，当该系统在该互动之下可用的诸延续，真包含于它以别样方式可用的诸延续之中时：$G’ \prec G$。

在把这一定义投入使用之前，它的四个特征值得评说。

它的范围是可比较的诸情形。一个延续空间的变形，其中不成立任何包含关系，落在定义 1 之外，并在§4.4中作为一种独特的模态被登记。这使框架付出了传统在异化名目下所处理的那些情形的代价，§9接手此事，而这一代价被优先于一个框架无从支撑的序。

它在形式上是比较的。说诸延续已被封闭，就是把该互动之下可用的与以别样方式可用的加以比较，而这一比较预设了对那个别样方式的界定。这是该定义的主要困难。§7接手它，且如在那里将看到的，供出的是一个局部答案而非一个完整答案。这一问题在此陈述并留其悬而未决，因为中间的诸节需要该定义的临时形式。

它关乎能力而非生产。一个被阻止去生产它正要生产之物的系统已被阻碍；一个在其能力完好无损的情况下什么也不生产的系统则未被阻碍。这一区分随§3的能力说明而来，并在两者相分离的情形中要紧。休息、有意的克制，以及一个研究纲领的休耕期，都是生产下降而能力保持或增长的情形。把这些当作阻碍将是一个错误，而该定义借附着于能力而避免它。

它对意图保持沉默。一次互动在它削减能力时便阻碍，无论其背后的意图为何。一方可能在意欲协助的同时阻碍另一方，也可能在意欲阻碍的同时什么也没产生。这一沉默是刻意的，且有其代价，因为它把道德评估所要求的一个区分从理论的触及范围中移除。这一代价在此被接受，作为使该原初对象保持机械性的价钱，而§9回到它所封闭的东西。

4.3 两个所在

§1所承继的划分，按阻碍运作的所在把它划分。转置进§3的语汇，两个所在如下。

内部阻碍作用于一个系统自身的生成性动态。它的诸实例包括生产之物质条件的耗竭、系统信息过程的失序，以及维持该过程本身之能力的枯竭。刻画这一所在的，是该削减无需第二个系统即可描述。一个遭受内部阻碍的系统，在其自身诸过程的运作中已受损，而该损伤在把该系统单独考虑时便是可描述的。

关系阻碍作用于一个系统在耦合下发展其动态的能力。它的诸实例包括锁定，其中一个耦合把一个系统的发展固定在一条它再也无法离开的路径上；一个刚性参照模型的施加，在§7中详加处理；以及以牺牲一方为代价来稳定另一方发展的制度配置。刻画这一所在的，是该削减需要那个耦合才能被描述。一个遭受关系阻碍的系统保有其自身诸过程处于可运作的状态，而发现其诸过程发展的诸条件由另一方设定。

关于该划分的两点评说。第一点关乎它同§2机制／现象之区分的独立性。所在关乎阻碍在何处运作；机制关乎什么产生它。单一机制可能在任一所在产生阻碍：一场使一方饥馁的资源争夺产生内部阻碍，而同一场争夺以一个固定一方发展的协议来化解，则产生关系阻碍。这两个区分彼此横切，而其中任何一个都不化约为另一个。

第二点关乎两所在之间的往来，它双向而行，且该划分应被读作容许它。持续的关系阻碍产生内部阻碍，因为一个其发展已被另一方固定的系统，随时间流逝而失去那发展曾维系的诸内部资源。内部阻碍塑造一个系统能进入的诸耦合，而一个被耗竭的系统比一个繁盛的系统更容易被施以关系阻碍。该划分标记阻碍运作的两个所在，而把两者之间的通道留作敞开。这里没有任何东西要求一个给定的插曲被排他地归于某一个所在。

本文在两所在之间注意力分配的不均衡，其缘由值得陈述。内部阻碍是两者中被研究得更充分的一个，并在数门成熟文献中以别的名目被处理。关系所在是本框架有所补益之处，也是本文实质结果所落之处。其余诸节几乎排他地关乎它。

4.4 模态性质

此处所汇集的诸性质，由定义 1 连同§3的能力说明而随之而来。它们作为该原初对象的诸性质而陈述，且是后续诸节所汲取的资源。

幅度。 阻碍可以是局部的，封闭一个系统可用延续的一部分而使其余部分完好，也可以是全局的，跨该范围削减能力。这一区分容许程度，且取决于该削减触及延续空间的多少。

持续时间。 阻碍可以是暂时的，能力随该互动的失效而恢复，也可以是持存的，能力在该互动已停止后仍保持被削减。持存要求解释，因为一个已停止的互动对随其而来的东西什么也解释不了。这一解释是§6所关切的。

可逆性。 阻碍可以是可逆的，其中被封闭的诸延续仍可恢复，也可以是不可逆的，其中它们的封闭已进入后续生产的诸条件，且无法借移除该互动而被撤销。可逆性与持续时间是有别的：一个阻碍可能在原则上仍可逆而持存，也可能短暂而永久地封闭某样东西。

特性，以及定义在此处的界限。 一个系统的延续空间可能被变形，其诸延续被换成另一些范围相当而特性不同的延续。定义 1 触及不到这些情形，因为该偏序供不出诸空间之间的任何关系，而本节把它们登记为一种立于该原初对象之外的现象，而非其中的一种模态。

作这一登记有两个缘由。这类情形确实发生，而一套对它们保持沉默的框架，将对其题材所含的一大类东西保持沉默。而它们正是传统在异化名目下所处理的诸关切所会落之处，从而框架无从触及它们，恰恰定位了一个空白，而非把它留作弥散的。就目前的诸资源而言，对这类情形所能说的，是发生了一次特性的改变，而这无需一个关于该空间本应包含什么的说明。所不能说的，是这一改变是否构成阻碍，而§9把它作为一个悬而未决的问题引入。

路径依赖。 这条性质在后文做最多的工作，而它由定义连同§3以一种值得逐步铺陈的方式随之而来。

定义 1 使阻碍成为生成能力的一次削减。§3确立了生成能力取决于一个系统的历史，因为一个系统接下来能生产什么，取决于它已经生产了什么以及生产的方式。把两者结合起来：一个给定的互动是否削减能力，取决于该互动所落入的那段历史。同一个互动，发生在一个系统历史的两个点上，在一个点削减能力而在另一个点扩大它，因为它所封闭与开启的诸延续，由该系统已经生产之物所决定。一个被施加在一个尚未发展出该约束所预设之工具的系统上的约束，封闭甚多；同一个约束在那些工具存在之后被施加，则封闭甚少，并可能开启一条此前不可用的路径。

由此得出两个后果，且两者都关乎阻碍如何能被评估。

第一个关乎瞬时评估。一个互动是否阻碍，抗拒从对该互动那一刻情境的描述读出，因为该评估汲取的是这样一个描述所略去的历史。在其瞬时描述上相同的两个情境，可能在一个给定互动是否阻碍上有别，其差别在于它们的历史。这是把当下这个原初对象同静态种类的诸原初对象区分开来的东西。权力分布、制度密度与身份配置，都是一个时刻的描述所供出的性质；能力是否已被削减，则立于这样一个描述所触及的东西之外。

第二个关乎对区间的评估。既然一刻的阻碍可能在此后扩大能力，而一刻的扩大可能在此后封闭，一个局限于单一时刻的评估便有误读其所落区间的风险。因而评估的单位是轨迹，而被评估的是阻碍跨该轨迹的进程。这里没有任何东西要求诸轨迹被同某个受偏爱的轨迹比较，而§7给出理由，对这样一个比较是否可用表示怀疑。可用的是一个系统能力跨该区间的进程，而那正是该评估所关乎的。

主张 2。 生成性阻碍是路径依赖的，而它的路径依赖由定义连同生成能力的历史条件化而随之而来，无需任何进一步的假设。一个互动是否削减能力，取决于它所落入的历史；因而评估的单位是轨迹而非时刻。

§1所作出的承诺被清偿到如此程度。路径依赖在那里被描述为该原初对象的一个后果，而非对它的一个附加，而上面的推导正是成问题的那个后果。该推导从§3所要求的，是那里所辩护的历史条件化的弱读法，即生产被生产所条件化；那个要求累积的强读法则无必要。

§5 冲突作为阻碍的显现，及该读法的界限

原初对象现已界定，§2所提议的化约可用§4的语汇来陈述。本节做三件事：它陈述阻碍与冲突之间的关系，它把诸机制不加特权地排列开来，且它供出一条判据，凭它一场给定的冲突被指派到框架的触及范围之内、或被置于其外。第三件是本节的主要工作。

5.1 阻碍与冲突之间的关系

冲突作为一个派生现象进入框架。这一关系可陈述如下。

定义 2。 两个系统处在相互生成性阻碍之中，当每一个都削减另一个的持续生成能力时。冲突，在这里所提议的读法上，是已在其各方之间变得具有关系意义的生成性阻碍的显现。

该定义中有两个从句需要评说，而第二个携带一份困难。

相互性是诸可能条件中的一个，其地位应予标记。阻碍容易单向而行：一方可能在其自身能力不受影响的情况下削减另一方的能力，而这种不对称的情形是常见的。不对称的情形所得出的，是支配而非冲突，且这一差别值得保留。一场冲突有两方相互作用；一场支配有一方作用、另一方承受。框架登记两者并以不同方式描述它们，这是它的一个可取之处，因为建立在权力分布之上的诸理论，在区分这两者上有一个已知的困难。

关系意义是第二个从句，而它所做的工作多于其简短所暗示的。并非每一次能力的相互削减都显现为冲突。两方可能各自略微缩减对方的诸延续，甚少察觉到这一缩减，并继续下去。

在框架的诸资源之内有一个候选阈值可用，且被临时提出。阻碍已变得具有关系意义，当它已进入各方彼此的工作模型之时：当每一方对另一方的模型把该阻碍作为一个词项包含在内，且每一方的行为被那个词项条件化之时。这使该阈值成为一个关于诸模型的事实，而非一个关于阻碍强度的事实，其优点是把定义 2 连接到§7所发展的装置，其缺点是它依赖于那些工作模型本身，而那一节发现它们地位有限。

该候选是不带信心地提出的。它可能被证明过于宽松，因为诸模型吸收了大量并不组织任何关系的东西，那么该阈值便会容纳一些根本不是冲突的冲突。可作为缓解而言说的是，这一含糊附着于显现而非附着于原初对象：诸延续是否已被封闭，容许比该封闭是否已变得有意义更为锐利的处理，而那个更锐利的问题正是框架所建立于其上的问题。供出一个可辩护的阈值被引入§10。

5.2 诸机制，不加特权

§2所概览的诸机制，如今作为通往阻碍的诸路径而各就其位。资源竞争在一方对一个资源的获取取走了另一方延续的一个条件时产生阻碍。权力不对称在能力落差被转化为封闭时产生它。制度配置在诸安排以牺牲另一方为代价来稳定一方发展时产生它，并在它们为双方稳定发展的诸条件时压制它。误解在一方依据对另一方的一种构想而行动、以致封闭它当作已然封闭之物时产生它。未获满足的需求在下一小节所界定的诸条件下产生它。象征性与规范性承诺在该承诺把对方的皈依或遏制当作作出承诺一方自身延续的条件时产生它。

关于这份名单的两点观察。诸机制会复合：数种可能同时运作，而它们共同产生的阻碍可能超过每一种单独产生的。且这份名单是敞开的。其他机制存在，而框架的主张关乎这些机制所产生的现象，把这一列举按设计留作不完备。一套关于机制的理论会要求这一列举；一套关于它们所产生之现象的理论则不会。

5.3 范围判据，在未获满足之需求上受检

对一套这种框架最迫切的反对，是有些冲突产生而其中并无阻碍，且这一反对是对的。未获满足之需求的案例把它尖锐地陈述出来，而通盘处理那个案例得出本小节所供出的判据。

考虑两方，其中一方想要另一方拒绝提供的某样东西。那份挫败是真实的，而随之而来的冲突是真实的。阻碍在哪里？有两个答案可用，而其中一个是陷阱。

那个陷阱是把阻碍定位在挫败本身之中：主张一个未获满足的欲望削减了想要的一方的生成能力，因而那场冲突毕竟是阻碍性的。这一动作在每一个案例中都可用，因为想要某样东西而缺它，正是想要任何东西的条件。若认真对待，它使每一个未获满足的欲望都成为一次阻碍，而一个适用于其领域之内一切的概念，在其领域之内什么也不辨别。这一动作还把一场显然在两方之间的冲突，移置进其中一方的内部，而这与一套其诸承诺是关系性的框架格格不入。这个答案被拒绝。

第二个答案照单全收地接受这个案例，并问要什么为真、阻碍才会在场。浮现出两个条件，且任何一个都足够。

第一个是满足该需求会封闭另一方的诸延续。当一个需求只能以另一方继续发展之能力的代价被满足时，该需求的满足构成对满足它的一方的关系阻碍，而这场冲突落在框架的触及范围之内。这一条件关乎该需求的内容以及满足它会有何代价。

第二个是该需求的受挫可度量地败坏想要的一方自身的能力。这是内部所在，而可度量性的要求正是使它不至坍缩进那个陷阱的东西。所要求的是能力以一种可独立于挫败本身而描述的方式被削减：曾经可用而已不再可用的诸延续、曾在使用而如今不再的诸工具、曾在运作而已然失效的诸过程。一份使想要的一方能力完好无损的挫败，无论被感受得多么尖锐，都不满足该条件。

当两个条件都不成立时，这场冲突落在框架之外。这类冲突确实发生。两方可能想要互不相容的东西，可能就那不相容争辩，可能发现这一分歧痛苦而持久，并可能自始至终使彼此的生成能力毫发无损。框架描述这样一场冲突不是什么，而把它是什么留给另一种说明。

主张 3。 一场冲突落在当下这套框架的触及范围之内，当满足一方的需求会封闭另一方的诸延续时，或当该需求的受挫以一种可独立于挫败而描述的方式败坏想要的一方的能力时。两个条件都不满足的冲突是真实的，且落在本框架所解释的东西之外。

关于该判据的三点评说。

该判据关乎范围，且不携带评价性内容。一场落在框架触及范围之内的冲突，仅凭这一点，并不比一场落在其外的冲突更糟或更迫切。该判据所决定的，是框架能就一个给定案例回答哪些问题。

该判据达不到自我适用。满足一个需求是否会封闭一方的诸延续，以及一份挫败是否已独立于其自身而败坏能力，都是关于该案例的、要求调查的问题。该判据陈述要调查什么，而把调查留作未做。

该判据承继了§4中留其悬而未决的参照问题。两个条件都取决于能力的比较，而能力的比较预设了对能力本应为何的一个界定。因而该判据在与定义 1 相同的方面是临时的，而这一临时性，就其所能，在§7中被清偿。

本小节开篇的那一观察值得给一个末语，因为读者可能把它当作一个让步。一套拒绝去解释一类冲突的框架，或许看似弱于一套解释它们全部的框架。反面才成立。一套吸收每一个案例的框架，是借一个诸竞争框架同样可用的规定来这样做，而在吸收了一切之后，它辨别甚少、禁止甚少。陈述一套框架在哪里不适用，正是使它的诸适用富有信息量的东西，而供出一条判据、凭它这一不适用能在一个特定案例中被辨认出来，正是使这一陈述不止于一份免责声明的东西。

5.4 价值作为生成方向的定向

需求的案例确立了一个模式：一个看似要求其自身原初对象的候选，一经考察，便化解为一种作用于能力的机制。价值容许同样的处理，而这一处理值得铺陈，因为价值是在冲突诸说明中最常被授予原初地位的候选。

自然的看法主张，承诺于互不相容价值的各方凭那不相容而冲突，而§2引入了第一条限定：规范性分歧是常见的且大多是惰性的，一旦与一个要求对方皈依或屈服的要求相接，便成为冲突。那一限定把冲突定位在那个要求之中。有一个第二的、更富信息量的说明可用，而它关乎价值对一个生成性过程做了什么。

价值使生成完好无损，而作用于它的方向。一个承诺于某价值的系统继续生成，其轨迹偏向其可能性空间的某些区域而非另一些。以 $\phi$ 记一个系统的生成性动态，以 $\phi_v$ 记那在一个价值承诺之下的动态，二者的关系是一种定向：

$$\phi ;\longmapsto; \phi_v .$$

该映射保存生成性并重分配它。系统能生产什么仍是它本能生产的东西；它移向那些生产中的哪一个则已被重新赋权。

这把价值置于一个已被定义的模态之内。§4把一个延续空间的收缩与它的变形区分开来，其中变形把可用的诸延续换成另一些范围相当而特性不同的延续。价值在变形模态中运作：一个价值承诺不彻底封闭任何延续，而在余下的那些中偏置该轨迹。这一对应是不勉强的，该模态本是出于别的缘由被引入的，而它供给价值一个在框架中无需任何新装置的位置。

该说明在耦合系统的案例中挣得其价值，而亲密关系提供了最清晰的实例。处在一个紧密耦合中的两方，常被描述为拥有不同的价值，而这一描述在所说之事的层面上是准确的。它在动态上所命名的，容许更锐利的陈述。一方的诸承诺把耦合的系统定向到一个以事业、探索与流动为特征的区域；另一方的诸承诺把它定向到一个以家庭、稳定与在地为特征的区域。每一种定向都是融贯的，每一种都使另一方的能力在形式上完好无损，且没有一种是一个错误。

困难在于耦合。当两个系统共同生成时，它们沿一条轨迹生成，而一条轨迹一次容许一种定向。因而各方是在协商一个共享生成性过程的方向，而非在协调它们各自分开的过程，而关于价值的分歧，是关于它们共同构成的那个系统之未来动态的分歧。

主张 4。 价值承诺作为定向的诸机制而运作：它们保存一个系统的生成能力，并把其轨迹偏向其可能性空间的特定区域。当两个系统共同生成时，它们的诸承诺关乎一条轨迹，因而价值分歧是关于一个共享生成性过程之方向的分歧。

三点评说，第一点关乎该主张所保留不说的东西。

这一定向自始至终是描述性的。主张 4 归于一个价值承诺的，是它确实施加的方向，把一个耦合本应遵循的任何方向搁在一边。这一区分要紧，因为一个归于耦合本身的受偏爱方向，会重新引入§7所发现不可用的那个界定，而这一重新引入会遭到同样的反对。

该主张预测一个对照情形，而这一预测要求一条关于共同性的判据，而框架只是局部地供出它。生成是共同的，在成问题的意义上，当各方的延续空间被耦合、以致一方所取的一个延续从另一方移除诸延续之时：它们的诸轨迹在诸定向所关乎的方面被约束到一条路径上。生成是平行的，当那耦合无论另一方取什么都使各方的诸延续保持可用之时。

该判据容许程度，且抗拒锐利的适用，因为耦合鲜有全然的、也鲜有全然缺席的。§7.1主张各方是在它们的诸耦合之内被构成的，由此得出，耦合各方之间的生成在每一个案例中都在某种程度上是共同的，因而这一对照相应地是程度而非种类上的。该主张所预测的，如此限定之后，是价值分歧关乎冲突的程度与共同性的程度成比例，而其延续保持相互可用的各方能持有相分歧的诸定向，冲突相应地更少。这说明了一个熟悉的观察，即在同事之间可容忍的价值差异，在其未来被共同生成的各方之间变得尖锐，而它弱于一个断然的对照会是的样子。

该主张连接到随后的东西。价值承诺在§6所界定的意义上是象征性的，且它们以那里所描述的方式闭合：一个承诺条件化着哪些互动被着手进行，而如此被条件化的诸互动供出用来评估该承诺的经验。因而价值冲突持续到产生它们的诸条件之外，这把它们同需求种类的诸冲突区分开来，并解释了它们特征性的持续时间。

§6 象征性闭合与阻碍的持续

象征性机制在§5中与其他机制并列出现，而那里所给的处理是正确的：一个对某规范或象征秩序的承诺在可界定的诸条件下产生阻碍，而它不加特权地在名单中就位。本节问是什么担保去扩展那一处理，并以单一的性质作答。该性质是狭窄的，而这狭窄正是要点，因为一个更宽的主张会把象征安装为第二个原初对象，并瓦解§2的架构。

6.1 成问题的性质

§4的模态性质包括持续时间，而那里引入了一则如今被清偿的评说。在产生它的互动已停止后仍持续的阻碍，要求解释，因为一个已结束的互动对随其而来的东西什么也解释不了。当阻碍长过它的原因时，某样东西已接管了产生它的工作。

象征性机制以其恰恰能做到这一点的能力而与众不同。

主张 5。 象征性机制是那些凭之阻碍变得自我再生产的机制。借固定一种诠释，它们规约本会修正它的后续互动，而如此被固定的阻碍持续到产生它的诸条件之外。

该主张关乎一种持续的模式，而不触及更远。它对以下这些什么也没说：象征构成主体、中介经验，或结构化社会场域，这一切可能都为真，而其中没有一样是本文所要求的。它所说的，是一个固定诸诠释的机制，就持续时间这条性质而言占据一个独特的位置，而这个位置正是使本节挣得存在的东西。

6.2 闭合的机制

固定凭之产生持续的那个结构，可分步铺陈。

一种诠释，一旦被持有，便条件化着一方留意什么以及它如何构想它所发现之物。这一点是不足为奇的，且一般地适用于诸诠释。就闭合而言要紧的，是这样一种情形：该诠释条件化着本会供出反对它之证据的那些互动本身。

考虑一方持有另一方怀有敌意的诠释。该诠释条件化着它的行为：它有所保留，它有所戒备，它拒绝它本会把握的那些开口。那一行为反过来条件化着对方的行为，而对方，遇到保留与戒备，也报以保留与戒备。第一方如今发现它的诠释被它自己的诠释帮助产生的行为所确证。本可能修正该诠释的那次互动，已被该诠释塑造成一次确证它的互动。

这一结构的两个特征值得标记。

该环路就证据而言是闭合的。诸互动继续发生并继续被诠释，而本会算作否证的东西已被该诠释自身的运作弄得不大可能。该方接收到它自己的行为已安排好的证据，而这不同于拒绝证据。这把闭合同单纯的固执区分开来，且这正是闭合抗拒适用于固执之诸补救的缘由。

该环路对它所固定之物的准确性无动于衷。一个形成时准确的诠释，以与一个错误的诠释相同的方式闭合，而错误的情形更具后果，因为该环路供出了该诠释所缺的确证。这连接到§2中就身份引入的观察：在对立中被构成的诸身份在不准确的同时仍完全运作，而闭合正是容许这一点的那个结构。

6.3 象征性机制在其他机制之间

本节与§5之间的关系要求精确，因为这里所授予的扩展容易被过度解读。

象征性机制仍是机制。它们以其他机制所用的方式产生阻碍，而主张 5 对它们的生产性角色什么也没添加。主张 5 所关乎的，是生产之后发生什么：它们所产生的阻碍与持续时间有一种特征性的关系，而这一特征性的关系正是本节所描述的。

这一扩展在诸条件下对其他机制可用。闭合是一条结构性的性质，而一个非象征性的机制能获得它，当它的运作同样条件化着本会修正其效果的那些互动之时。一个决定谁被准入到本可改变该安排之审议的制度安排，在相同的结构意义上展现闭合。使象征性机制与众不同的，是闭合对它们而言是特征性的而非偶然的，因为固定一种诠释正是它们所做的事。

该性质是一个程度问题。诸诠释被固定得或紧或松，诸环路被闭合得或紧或松，而修正本可凭之进入的诸开口或多或少。主张 5 关乎一种倾向，且自始至终容许那分级的读法。

6.4 这为何作准备

闭合在此被陈述为固定诠释之诸机制的一条一般性质。它在本文中的主要应用较为狭窄，且在§7中到来。

定义 1 所要求的诸参照模型，是一方对另一方生成性动态的诸诠释。它们因而以所描述的方式受制于闭合：$A$ 对 $B$ 动态的一个模型，条件化着 $A$ 对 $B$ 的行为，后者条件化着 $B$ 的行为，后者供给 $A$ 用来评估该模型的证据。一个如此闭合的模型已不再能被它所建模的那一方所修正，而§7的论证是，正是这一可修正性的丧失，而非该模型的内容，与阻碍相关。

主张 5 供出那一论证所需的机制。闭合解释一个参照模型如何变得不可修正，而§7陈述它一旦如此便有何随之而来。

§7 参照问题与工作模型的刚性

定义 1 被留其悬而未决，附着着一份困难，而§5双重地承继了它。该困难在于阻碍是比较的：能力的一次削减，是相对于能力本应为何的一次削减，而这一比较要求对那个别样方式的界定。本节接手这一困难，论证该界定无法以困难看似要求的形式被供出，并陈述随这一失败而来的东西。

7.1 一条享有特权之轨迹的不可用

自然的界定是一条系统在没有该互动的情况下本会遵循的轨迹：该方倘若阻碍未曾发生本会生成什么。有两个论证不利于它，且它们相互独立。

第一个是休谟的。谈论一个系统本会遵循的轨迹，当那条轨迹在做规范性工作时，就是从关于所是者的一个描述，转到关于本应是者的一个主张。这一转移要求一份担保，而描述性的诸事实本身供不出任何担保。一个系统的历史、它的构成与它的诸约束，决定它做了什么以及它能做什么；它们把它本应做什么留作未定。一个携带规范性分量的、关于那个别样方式的界定，会需要这样一份担保，而本框架所接纳的诸资源无从提供任何担保。

第二个论证内在于框架，且就当下的目的而言更具决定性。§3确立了生成能力是被关系条件化的：一个系统的可用延续取决于它所处的诸耦合。生成性关系存在把这一点推得更远，主张成问题的那些系统是在它们的诸耦合之内被构成的，而非以业已成形的姿态进入诸耦合。在这一承诺之下，自然界定所要求的那个反事实是畸形的。该方在没有其诸耦合的情况下本会遵循的轨迹无从界定，因为该反事实所要求的那个减除，移除了其轨迹本要被界定的那一方。该操作溶解了它本要孤立出来的那样东西。

两个论证收敛于同一结果，且值得保持有别，因为第一个是对推断的一条一般约束，而第二个是本框架自身诸承诺的一个后果。一套怀有个体主义承诺的框架，会单独面对第一个。

有一个后果应立即被登记。异化这一概念，在把它定义为对一个主体本会本真地成为之物的偏离的那种形式中，恰恰要求那个被证明不可用的界定。因而本文拒绝该概念，而§4已然拒绝了它，做法是把一个延续空间的变形当作一次特性的改变，而这无需一个关于该空间本应包含什么的说明。

7.2 工程学读法

一条享有特权之轨迹的不可用提示了两种回应，而其中一种是一个假二分。可能会得出结论说，轨迹根本无法被谈论，而定义 1 因缺其所要求的比较而失败。也可能会得出结论说，某条享有特权的轨迹毕竟必定存在，而上面的诸论证应被抵制。有一个第三立场可用，而它正是这里所取的。

把参照轨迹当作一个不携带任何本体论承诺的工程学构造物。

这一动作在数门成熟框架中有先例，而这些先例共享一个结构。在控制论中，一条参照轨迹是相对于一个控制目标而被选定的路径；理论中没有任何东西主张系统在超出该目标自身的任何意义上本应遵循它。在模型预测控制中，参照随视界的推进而在线被修正。在贝叶斯推断中，先验是为其在后续更新中的角色而被持有的工作性承诺。在预测编码中，大脑运作起来仿佛它拥有一个关于其环境的生成模型，而该模型是由它在预测与行动中的表现来评估的，而非由它同环境终极结构的对应来评估。

这些所共享的，是该模型是一个工具。它是为它所使能之物而被持有的，由它是否使能之而被评估，并在它未能如此时被修正。没有任何关于世界结构的主张被引入，且不需要任何这样的主张。

应用于此：以 $\hat{\gamma}$ 记一个关于某系统生成性动态的工作模型，由一方为协调之目的而持有。那个尖号标记该模型的地位：这是一个受制于修正的临时构造物，而非一个关于它所建模之系统的事实。该模型随互动的进行而变化，而它的变化是所研究之动态的一部分，而非对一个外在于它们的标准的偏离。

由此得出两个问题，而本文的结果落在第二个之中。

7.3 许可约束

第一个问题是什么许可这个构造物之为构造物。一个工具由它所服务的目标来辩护，而该目标必须被界定，因为一个由一个未被界定之目标来辩护的工具，是由无物来辩护的。

这里可用的界定是狭窄的。之所以要求一个参照模型，是因为对阻碍的诸归属要求一个参照：说 $A$ 削减了 $B$ 的能力，就是说对 $B$ 可用的某样东西不再可用，而这陈述了一个比较。缺某个关于什么曾可用的工作模型，该归属便没有内容。因而该模型是使该原初对象得以运作的极简装置。

这一辩护的狭窄，正是使该立场保持稳定的东西，而这一点值得强调，因为该立场容易被破坏稳定。倘若该辩护被扩展，以致该模型服务于该方的繁盛、或它的本真发展、或它的固有目的，那么§7.1所拒绝的本体论承诺便以另一个名目回来。该模型于是会对一个超出协调的标准负责，而那个标准恰恰会是休谟论证所禁止的东西。工程学读法只在它的目标保持可运作时成立。

这部分地清偿了§4中引入的债务。定义 1 的比较由一个工作模型来许可，而该工作模型由它在使该比较可运作中的角色来许可。该说明所不供出的，是任何特定模型都是那个正确模型的保证，而没有任何这样的保证可用。

有一条进一步的局限需要陈述，因为它关乎框架触及多远，且在迄今为止的阐述中被低估了。

上面所许可的参照模型，是一个由关系中的一方为在其内协调之目的而持有的模型。一个由关系之外的观察者作出的阻碍归属，要求一个由那个观察者持有的参照模型，而上面所给的辩护未能许可它。该观察者不与任何一方协调，因而缺一方的模型由之被辩护的那个目标。一个观察者的模型对该说明所界定的任何东西都不负责。

由此得出两个后果，而它们收窄框架的诸主张。

对阻碍的诸归属，就这里所集结的诸资源而言，从关系之内是良构的、从关系之外是病构的。一方可能说它的诸延续已被封闭，持有一个关于什么曾可用的工作模型；第二方可能说同样的话；一个代它们说这话的观察者，持有的是一个框架无法许可的模型。这是一个可观的限制，且它被接受。

这一限制关乎§8以及§10的经验纲领。一个由一位理论家作出的、跨两个条件对阻碍的比较，是一个观察者的归属。因而§8的构造所能确立的，比一个阻碍有别的证明更为狭窄：它能确立各方自己的诸归属会有别，而这正是§8所主张的。一个观察者的归属究竟能否被许可，被留其悬而未决，而§10把它作为框架最迫切的未决问题引入。

在关系之内所能供出的，是一个关于诸模型如何被持有的条件，而那是下一小节的主题。

7.4 刚性作为与阻碍相关的变量

第二个问题是什么把一个被持有得好的工作模型同一个被持有得坏的工作模型区分开来。这里与预测编码的类比要求一则限定，而该限定得出本文的结果。

在预测编码中，模型是单个系统对一个环境的模型。环境不持有一个关于该系统的模型，且它不对被建模作出回应。成问题的情形在两方面都有别。每一方都持有一个关于对方的模型；每一方都被对方所建模；且每一方都能借以模型未能预料到的方式行动来修正对方的模型。诸模型是相互指涉的，而被建模的各方是能动者。

这一差别有一个后果。在单系统情形中，一个失败的模型被环境所矫正，环境无差别地供出否证的证据。在相互情形中，一个模型是否被矫正，取决于建模的一方是否对被建模的一方之矫正保持开放，而这一开放是关系的一个变量。

§6供出开放凭之丧失的机制。$A$ 对 $B$ 动态的一个模型，条件化着 $A$ 对 $B$ 的行为；$A$ 的行为条件化着 $B$ 的行为；而 $B$ 被条件化的行为，是 $A$ 作为关乎该模型的证据而接收到的东西。当该环路闭合时，该模型不再对 $B$ 负责，而 $B$ 修正它的能力已被移除。

闭合是否败坏矫正通道，容许一个论证而非一个规定，而该论证如下推进。

设 $A$ 持有一个关于 $B$ 的模型，$A$ 把它当作准确的。$B$ 所发出的诸矫正被 $A$ 对着那个模型接收，而该模型决定它们如何被构想。当该模型被当作准确时，一个指示它不准确的矫正是反常的，而 $A$ 有两种构想可用：该模型错了，或 $B$ 的行为前后不一。把该模型当作准确，正是使第二种构想成为可用的东西；一个临时地持有其模型的一方只有第一种。

§3确立了生成是被历史条件化的，从而一个系统所生产之物进入它后续生产的诸条件。应用于 $B$ 的矫正性行为：以第二种构想接遇的诸矫正，条件化着 $B$ 后续的矫正性行为，而它们向下条件化它，因为一个被构想为前后不一的矫正，所成就的少于该矫正所意图的，而它使 $B$ 付出的多于它所回报的。因而矫正的速率下降。

这对 $A$ 的模型的后果是，关乎它的证据在本会扰动它的那个特定方面减少。$A$ 的模型因而被对着一个更薄的矫正通道持有，而容许第二种构想的诸条件被加强。这正是§6的闭合在一个关于另一方之模型被持有的特定情形中出现。

该论证要求该模型被当作准确而非临时地持有，而这正是§7.2的工程学读法被构造来阻止的东西。因而本节的两半是相连的：工程学读法是使第二种构想保持不可用的东西，而放弃它是容许闭合开始的东西。

主张 6。 一个关于另一方生成性动态的工作模型，在它保持可被它所建模的那一方修正的期间，作为一个工具是可接纳的。当这样一个模型变得被当作准确而非临时时，上面的论证表明矫正通道败坏，而该模型趋向不可修正。在这一说明上，与阻碍相关的，是一个模型被持有的刚性，而非该模型所断言的内容。

由此得出四个后果，而它们是本节的实质。

该变量是关系的一条性质。刚性关乎一个模型如何被持有以及它能被谁修正，而两者都是关于各方之间耦合的事实。因而它是§4意义上的一个关系所在现象，且它需要两方才能生起：一个不可被它所建模的那一方修正的模型，预设了一个它所建模的那一方。

偏离失去它的分量。一旦参照模型被理解为非本体论的，一方对另一方之模型所预测者的偏离，便是模型误差而无其他。该偏离不携带任何规范意义，而框架缺乏可供给它任何规范意义的资源。这正是异化之被拒绝在可运作的词项上所等于的东西。

该主张是可证伪的。它预测不可修正的诸模型伴随着上升的阻碍，且该关系在诸模型内容受控的情况下成立。一个诸模型被刚性地持有而阻碍未能上升的案例会不利于它，一个在可修正性得以维持的情况下阻碍上升的案例亦然。§8构造该预测所要求的比较。

该主张连接本文两个后来的关切。象征性闭合，在§6中作为固定诠释之诸机制的一条一般性质来处理，是刚性凭之生起的机制；刚性，在此处处理，是闭合在被闭合者是一个关于另一方之模型时所产生的东西。这两节在不同的一般性层次上描述同一个现象。

§8 一个构造案例 - 内容固定不变下的刚性

主张 6 断言刚性与阻碍相关，而内容无关。确立这一点要求一个内容固定不变而刚性变化的比较，而这一比较在此为那个目的而被构造。

8.1 该案例的定位

随后的案例是构造的而非观察的，而这一定位上的差别应在该案例被给出之前陈述。

它的功能是孤立一个变量。一个形如”在 $Z$ 受控的情况下 $X$ 与 $Y$ 相关”的理论主张，要求 $Z$ 被控制，而在观察材料中控制 $Z$ 是困难的，当该材料是由对理论家的诸目的漠不关心的诸过程所生成之时。一个构造的案例凭规定供出该控制。它在控制上所供出的，它在证据分量上放弃；且它对频率、量级或发生什么也确立不了。

因而该案例例示§7.4的论证，而非确立它。那一论证从把一个模型当作准确连同§3的历史条件化，推导出矫正通道的败坏；下面的构造在一个特定情境中展现该推导，把诸模型的内容固定不变，以使刚性的角色可见。该构造所添加的，是可读性。它所保留不给的，是证据，而该机制是否在该构造之外运作，是§10如实引入的一个经验问题。

有一条进一步的限制适用，引入于§7.3。对阻碍的诸归属从关系之内获许可、从关系之外未获许可。因而下面的比较是以各方自己的诸归属会是什么的词项来陈述的，而非一个代它们作出的关于阻碍的主张。

因而读者的检验是狭窄的。它是下面两个条件是否仅在一个方面有别。倘若它们在第二个方面有别，该案例便未能孤立它所声称孤立的东西，而这一反对是决定性的。

8.2 构造

两方，$A$ 与 $B$，处在一个耦合之中，在其内它们共同生成：一项共享的事业，其轨迹取决于两者。摆在它们面前的一个决定要求其时机被敲定。

每一方在§7的意义上持有一个关于对方生成性动态的工作模型。$A$ 对 $B$ 的模型：$B$ 正在确立一批工作，其早期阶段容许甚少的中断，而 $B$ 需要大约三年在位，一个境况的改变才可行。$B$ 对 $A$ 的模型：$A$ 的工作是可携带的，而 $A$ 能在短时通知下吸收一个境况的改变。

这些模型的三个特征应予标记，因为该构造依赖于它们。

它们在实质上是合乎情理的。每一个都是一方会依据可用证据形成的那种模型，且没有一个是草率的。这一案例并不取决于任何一方是一个糟糕的建模者。

它们关乎生成性动态而非偏好。$A$ 的模型陈述 $B$ 的工作需要什么，把 $B$ 想要什么搁在一边。这把这一案例同一场关于目标的分歧区分开来，并把它定位在主张 6 运作之处。

它们在某种程度上是不准确的，且两者都是。$B$ 的第三年其实是那个灵活的年头，而头两年不是；$A$ 的可携带性在设备方面是真实的，而在 $A$ 于其间工作的合作者方面是不真实的。这一不准确被规定，因为它是必需的：一个其中一个条件含有误差而另一个不含的比较，会让两样东西变化。

8.3 两个条件

条件 I：诸模型可修正。

$B$ 陈述第三年是那个灵活的。$A$ 接收到这一点并相应地调整该模型。$A$ 陈述可携带性对设备成立、对合作者不成立。$B$ 接收到这一点并反过来调整。每一个矫正都由现有模型未能预料到的行为或陈述所引起，且每一个都被纳入。

关于时机的分歧持续。它可能持续一段可观的时期，因为被调整的诸模型使各方的诸约束处在张力之中，而这一张力是真实的。各方自始至终所拥有的，是一个在其协调所要求的容差之内追踪对方动态的、关于对方的模型。没有一方在对着一个虚构作打算。阻碍保持低：对每一方可用的诸延续被该耦合所影响，一如任何耦合影响它们，而没有一方的能力被对方对它的模型所削减。

条件 II：一个模型不可修正。

诸模型在内容与准确性上一如前述。$A$ 对 $B$ 的模型已在§6的意义上闭合：它条件化着 $A$ 对 $B$ 的行为，而如此被条件化的行为供出 $A$ 用来评估该模型的证据。

序列如§7.4所指示地推进。$B$ 的诸约束发展，一如诸约束所为，而 $A$ 继续对着如最初形成的那个模型作打算。因而 $A$ 的诸提议相对于 $B$ 的实际位置是错时的。$B$ 矫正，一如在条件 I 中。$A$ 对着一个本不该需要该矫正的模型接收到该矫正，并把它构想为 $B$ 之中的前后不一，而非该模型中的误差。$B$，接遇这一构想，矫正得更少，因为矫正被以一种惩罚矫正的构想所接遇。$B$ 的动态因而对 $A$ 变得更不可读，而 $A$ 的模型，如今接遇更少会扰动它的东西，被以增长的信心持有。

已被削减的，是 $B$ 在这一耦合下发展的能力。$B$ 的诸延续被一个 $B$ 再也无法影响的模型所塑造，而这一塑造经由 $A$ 的行为进行，而该模型条件化着这一行为。这是§4意义上的关系阻碍，而它以§6所描述的方式累积，持续到产生它的诸条件之外。

8.4 该比较所显示的

两个条件在两个模型的内容上、在两个模型的准确性上，以及在分歧的实质上，是相同的。它们在一个模型是被临时地持有还是被当作准确地持有上有别，以及在随之而来的可修正性上有别。每一方会归属的阻碍在第一个条件中是低的，而在第二个条件中是累积的。

有三种相竞的读法要求被排除，因为每一种都会在不诉诸刚性的情况下解释这一差别。

该差别是误差。 可能会有人主张 $A$ 只是对 $B$ 弄错了，而弄错说明了随之而来的东西。这一读法在该构造上失败。$A$ 在条件 I 中以相同的程度、在相同的方面弄错，而那里没有阻碍累积。误差在两个条件中都发生，而在其中一个中被吸收。所不同的，是该误差是否可被它所关乎的那一方矫正，而这是刚性而非误差。

该差别是分歧。 可能会有人主张各方分歧，而分歧产生了阻碍。这一读法同样失败。关于时机的分歧在两个条件中都在场，且在条件 I 中持续而无阻碍。把刚性从条件 II 中移除，留下分歧照旧立着且可处理。

该差别是被规定的。 可能会有人主张该构造内建了产生其结果的那个反馈，从而该环路之被观察到，是因为它被假定。这一反对会对一个作为证据提供的构造是决定性的，而它被§8.1中引入的定位所回答。该反馈在§7.4中从把一个模型当作准确连同生成的历史条件化被推导出来，而该构造展现该推导，而非假定它。所保持被规定的，是那个情境，而那正是一个构造案例所供出的。

该差别是坏信念。 可能会有人主张条件 II 中的 $A$ 是在以坏信念行动，而该案例描述的是一个道德失败而非一个结构失败。该构造排除这一点。$A$ 把 $B$ 的诸矫正构想为前后不一，随 $A$ 所持的那个模型而来：在那个模型上，那些矫正本不该是必需的，而一个把它的模型当作准确的一方，有理由作那一构想。该机制在没有恶意的情况下运作，这与定义 1 处引入的对意图的沉默相一致，且是一个有利于该主张而非不利于它的可取之处。

8.5 尺度

该构造被以两方陈述，它们的耦合是紧密的，而一个问题生起，关于该论证要求那个情境的什么。

该论证要求两个共同生成的系统，每一个持有一个关于对方动态的工作模型，而诸模型关乎行为。它其中没有任何东西要求那些系统是个体。诸国家持有关于其他国家发展约束的诸模型并据之行动；诸制度持有关于它们所管辖之诸人口的诸模型；诸研究共同体持有关于它们所评估之诸纲领的诸模型。在每一个之中，模型条件化行为，行为条件化被建模的一方所做的事，而该环路或闭合或保持敞开。

这是一个实质的主张而非一个假定，而它由该案例的结构来辩护：该构造自始至终指涉建模关系的诸性质，而无处指涉个人所特有的诸性质。倘若该主张失败，其失败会在于建模关系的某条性质，它在个人之间成立而在集体之间失败，而辨认出这样一条性质会是框架需要吸收的一份贡献。§10引入这一问题。

尺度不变的主张携带一个值得陈述的后果。倘若它成立，§8的案例便关乎§2开篇的那个国际情境，而现实主义关于诸国家误读彼此约束的观察，成为此处所描述之机制的一个实例，而非一个分开的现象。确立这一点会要求把当下的框架同本文所推延的权力基础的诸说明相比较，而§10把它命名为后续工作的第一项。

§9 框架所拒绝的东西与该拒绝的代价

一套借减法建立起来的框架，欠一个关于它移除了什么的说明。本节陈述什么被拒绝、缘由为何、代价几何，因为一条引入而不带其代价的限制，是一条被低估的限制。

9.1 规范性层次

正义、公平、评价意义上的价值、伦理与幸福，自始至终缺席，而它们的缺席是刻意的。缘由是架构性的。每一个都会作为一个进一步的原初对象进入，而一套尝试极简原初基底的框架，无法在第一个仍未被清偿之时接纳进一步的原初对象。所提议的次序是，阻碍先被处理出来，而规范性问题在那一基础上、由任何发现该基础可靠的人来提出。

代价是可观的，且应被直白陈述。就目前之姿态，框架把阻碍同它的缺席区分开来，而拒绝给诸阻碍排序。它供不出任何理由去主张一方对另一方诸延续的封闭是可反对的。它供不出任何理由去偏爱更少的阻碍甚于更多。一个问该说明在实践上有何随之而来的读者，在此阶段所接收到的答案是，实践上什么也不随之而来，因为实践性的结论要求该说明所保留不给的诸前提。

本文自身论证的一个后果使这一点更加尖锐。§7确立了参照轨迹不携带任何本体论地位，由此得出，一方对另一方之模型的偏离是不携带任何规范分量的模型误差。轨迹这一概念曾看似提供的任何规范抓力，因而被撤回。作为规范性处理之候选而存留下来的，是不可修正性，它是一个关系的一条性质，且会要求它自己的论证。框架辨认出该候选并止步。

9.2 意图

定义 1 附着于一个互动对能力所做的事，而对互动的一方所意图的东西保持沉默。这一沉默使该原初对象保持机械，而§8显示了这一沉默挣得其位，因为该案例转在一个于没有恶意的情况下运作的机制上。

代价是，道德评估要求框架所保留不给的那个区分。明知而施加的伤害，与在合乎情理的行为进程中施加的伤害，在任何关于责任的说明都必须登记的诸方面有别，而当下的说明两者都不登记。一套希望支撑这类评估的框架，会需要把意图作为一个进一步的词项重新引入，而这一重新引入会对该原初对象携带本文未曾考察的诸后果。

9.3 异化

该概念已被拒绝两次，在§4与§7，且出于相同的缘由：在把它定义为对一个主体本会本真地成为之物之背离的那种形式中，它要求一个关于本真的界定，而§7的诸论证发现它不可用。

有一个第二的排除同第一个并行运作，且引入于§4。异化种类的诸情形涉及一个延续空间的变形，而变形立于定义 1 所依凭的那个偏序之外。因而这类情形被排除两次：一次因缺那个本真的基线，一次因缺该序所能供出的诸空间之间的任何关系。

被保留下来的，是描述。一个其诸延续已被换成另一些幅度相当之延续的系统，已经历了一次特性的改变，而这可陈述而无需指涉那些延续本应为何。这样一次改变是否构成阻碍，被框架留其悬而未决，而这一悬而未决是一个空白，而非一个立场。

被放弃的，是该概念曾携带的批判之力。说一个安排造成异化，就是谴责它；说它变形一个延续空间，就是描述它。描述对框架可用，而谴责不可用，而一个因该概念的批判之力而看重它的读者，会发现这个替代物单薄。可用的回应是，一份倚靠一个不可用之界定的批判之力，是对着一个无人供出之说明借来的力，而辨认出这份债务是有所值的。它是否值它所耗的，是一个公道的问题，且被留其悬而未决。

9.4 从一个关系之外的归属

§7.3发现该说明许可由关系中各方持有的诸参照模型，而未能许可由关系之外的观察者持有的诸模型。这对框架的限制比§4至§5的阐述所暗示的更为严厉，而这一限制在此被完整陈述。

对阻碍的诸归属，就所集结的诸资源而言，从关系之内是良构的。一方持有一个关于什么曾对它可用的工作模型，可以说诸延续被封闭了。一个代那一方这样说的观察者，持有的是一个该说明无法许可的模型，因为该观察者不与任何人协调，而§7.3的辩护是一个凭协调的辩护。

这所耗的，是一套社会科学框架通常被想要来做的大部分事。第三方评估、跨案例的比较，以及关于一个人口中阻碍的总体主张，全都是观察者的归属。就目前之姿态，框架对它们无一支撑，而一个其兴趣在于这类评估的读者，应把该说明视作尚未为它服务。

有两种缓解可用，而这里都不予发展。一个观察者的模型或许能被观察者自己的一个协调目标所许可，在那种情况下诸归属会相对于那个目标，且会要求陈述它。或者一个观察者或许能报告各方的诸归属，而非作出一个归属，这是可用的且相当弱得多。这两者中哪一个能被弄得起作用，被作为框架最迫切的悬而未决问题引入§10。

9.5 关系意义的含糊

定义 2 借关系意义把阻碍转化为冲突，而§5不带信心地提出了一个候选阈值。该候选依赖于各方的工作模型，而§7发现它们地位有限，且它可能被证明过于宽松。可作为缓解而言说的，已在那里说过：这一含糊附着于显现而使原初对象独善其身，从而框架更锐利的问题是它所建立于其上的那个问题。所不能说的，是这一含糊无害。一套其核心派生概念有一个不精确阈值的框架，会在分类最要紧的那些案例中产出有争议的分类。

9.6 诸机制的列举

§5所处理的诸机制数为七，而它们的选取由它们在既有文献中的地位、而非由任何框架内部的原则所支配。其他机制存在。荣誉、怨愤与诸情感有被纳入的理据，而已被略去，且没有任何论证被提出说这七个是共同穷尽的或各自不可化约的。

这是一个真实的空白，而其形状应被理解。框架的主张关乎这些机制所产生的现象，而这按设计使该列举不完备。这一不完备所耗的，是任何关于诸机制如何彼此相关的说明：某些是否化约为另一些，它们是否以规则的方式复合，这份名单是否容许一个有原则的闭合。这些是框架所提出而不回答的问题。

§10 后续工作的方向

下面的诸项按它们对框架地位的关涉排序，从框架最为暴露的那些开始。

10.1 观察者归属的许可

这是框架最迫切的未决问题，而它的解决制约着随后的大部分东西。

§7.3凭诸参照模型在协调中的角色来许可它们，这把对阻碍的良构归属限制在关系之内的各方。一个观察者的归属究竟能否被许可，是悬而未决的。有三条路径提议自身。一个观察者或许能被授予它自己的一个协调目标，相对于它其模型会被许可、其诸归属会被索引。一个观察者或许能被限制于报告各方的诸归属，这是即刻可用的，且得出一套相当弱得多的框架。或者§7.3的辩护或许能被一个不经由协调的辩护所替换，在那种情况下那一小节的论证会要求通篇重做。

在这些中的一个被发展出来之前，框架的诸主张应被读作关于关系中各方能就它说什么的诸主张。

10.2 序的扩展

定义 1 依凭一个偏序，而不可比较的诸情形立于其外。该序能否被扩展以够到变形的诸情形而不引入评价性承诺，是悬而未决的。所会要求的，是延续空间之间一个既非幅度的量度、亦非按价值的排序的、关于幅度相当而特性不同者的关系。这样一个关系是否存在，是一个关于延续空间结构的问题，而它先于下面所描述的形式化。

10.3 同权力基础诸说明的比较

框架相较于§2所概览之诸传统有所增益的主张，已被论证而仍未受检。会检验它的，是一个权力基础的说明解释得很好的案例，以两套语汇加以处理，把每套说明所看到之物的诸差别记录下来。该案例会需要是历史的而非构造的，因为成问题的是对两套框架都未安排之材料的解释抓力。

§8指示这样一个案例会在哪里被找到。倘若尺度不变的论证成立，一个国家对另一个国家发展约束的误读，是刚性机制的一个实例，而这类误读贯穿变化中的能力分布而持续的诸案例，会把两套说明分开。一个权力基础的说明把这一持续归于分布；当下的说明把它归于诸模型的闭合，而这是一个容许不同证据的、有别的主张。

10.4 刚性主张的经验检验

主张 6 是可证伪的且未受检。检验它要求§8的比较在观察材料中进行：各方彼此模型的内容相当而它们的可修正性有别的诸案例。预测是，阻碍随不可修正性累积，且在可修正性得以维持的情况下保持低。

有两个困难伴随该检验，且应予预料。可修正性是一条倾向性的性质，且经由它的施行被观察，从而一个从未被挑战的模型供不出关于它本会不会让步的任何证据。而阻碍要求诸能力的比较，§7显示这依凭观察者所持的工作模型不亚于依凭各方所持的。这两个困难都不是本框架所特有的，且都不因指出这一点而被溶解。

10.5 形式化

该装置自始至终被保持在散文之中，而这一克制是刻意的：一套在诸对象稳定之前被引入的形式体系，装点该论证而非推进它。诸对象如今被以足够的精确陈述，以致形式化是一个合乎情理的下一步，而三个问题会组织它。什么结构表示一个延续空间，鉴于它容许收缩、扩张与变形，且变形保存幅度。阻碍如何沿一条轨迹传播并累积，鉴于它的路径依赖。以及一个关系上什么样的诸条件对应于一个在其内被持有之模型的可修正性。

同动力系统的连接是要去追寻的那个自然的连接，其中生成性动态答对一个流，延续空间答对可达集，而闭合答对一个不变区域的形成。这些对应是否经受住与诸现象的接触，是这样的工作会去敲定的问题。

10.6 闭合的形成与消解

§6描述了闭合而把它的动态留作未考察。什么样的诸条件使一个环路闭合，以及什么样的诸条件打开一个已闭合的环路，是框架所提出而要求别的资源去回答的问题。第二个尤其令人关注，因为闭合在此被描述为自我再生产，而自我再生产的诸结构凭之瓦解的诸机制，不如它们凭之形成的那些显然。

10.7 尺度不变

§8主张刚性机制在两个系统一边共同生成一边彼此建模的任何地方成立，并从结构上为该主张辩护。该主张对以下这样一种反驳是敞开的：辨认出建模关系的一条性质，它在个人之间成立而在集体之间失败，而这样一条性质可能存在。集体经由程序来建模，而程序可能以个人不会的方式闭合，或以个人不能的方式抗拒闭合。

10.8 关系意义的阈值

定义 2 的阈值在§5中被临时提出，且可能被证明过于宽松。所想要的，是一个把组织一个关系的阻碍同各方所吸收的阻碍区分开来的条件，以框架所能支撑的词项陈述，并对着分类有争议的诸案例加以检验。

10.9 规范性层次

无论什么样的规范性理论建立在这一基础上，都会从§9所辨认的那个候选开始：不可修正性，作为诸关系的一条性质，是在诸轨迹被剥夺规范地位之后所存留下来的东西。为什么一个关系对其各方之修正的开放应携带规范分量，是一个本框架所提出而缺乏资源去回答的问题。这是一篇第二论文会去接手的问题。

§11 结语

论题。 冲突被提议为受约束的关系动态之内生成性阻碍的显现。当前的工作不去穷尽此种阻碍凭之生起的诸机制，旨在确立一个极简的基础，数学的、象征的与规范的诸理论可建立于其上。

本文对那一提议所添加的，是一个实质结果，以及对该提议耗费几何的一个更清晰的视野。结果：对阻碍的诸归属要求参照模型；参照模型不容许任何本体论地位；一个被当作准确的模型败坏本可用来矫正它的通道；而与阻碍相关的，因而是这类模型被持有的刚性，而非它们所断言的内容。代价：该原初对象由一个把变形诸情形留作未分类的偏序所定义，而诸归属从一个关系之内获许可，同时从一个关系之外保持未获许可。两者都被作为悬而未决引入§10，而第二个制约着其余部分应被读到多远。