【(Preliminary)Perspective】Towards Generative Relational Governance - Discussion II

ENGLISH

Towards Generative Relational Governance

Discussion II

Wanhong Huang (huangwanhong@serendip.ngo)

Abstract

Two demands are made on contemporary governance, and this discussion treats them as distinct: the government of complex systems under uncertainty, and the sustained generation of value together with its return to those who produce it. Two research lineages are read here as responses to the one and to the other, the first assembled around adaptive, polycentric, and resilience-oriented governance, the second around the commons, generative justice, and peer production. Each is criticized from within for a limit that the other’s competence would address.

A theoretical gap is not by itself a reason to act. The discussion therefore examines a class of contemporary objects that are complex and generative at once: peer-produced software infrastructure, the encyclopedic commons, and platform moderation. In each, both lineages return a verdict of health by their own criteria, output improves, and the system fails. The failures appear to share a structure. What is consumed is the capacity of a particular set of participants to bear the cost of sustaining the arrangement, a quantity that neither framework measures.

A relational reading of this quantity is proposed as an induction from the three cases. It is distinguished from labor, from network centrality, and from social capital, and the conditions of its refutation are stated. The discussion establishes no theory of generative relational governance and proposes no institutional design. Its purpose is prior to both, being to determine whether such a theory is warranted at all, by asking whether a problem exists that the available frameworks cannot see. Questions of scale are posed and left open, since a quantity must be operationalized before it can be tracked across scales, and that operationalization is the work this discussion identifies as next.

Keywords: public governance; complexity governance; generative governance; relational structure; scale.

A note on the standing of this discussion. This is a discussion paper. It surveys existing governance thought and practice, compares two contemporary lineages, and examines three cases in which both appear to fail together. Its contribution is a diagnosis. A theory of generative relational governance is named in the title and left unbuilt; the work here is the prior one of asking whether such a theory would be warranted, and the answer depends on whether the problem identified in Part III is real.

The order of exposition follows the order of the argument. The taxonomy, the comparison, and the cases come first, and the relational reading appears in Part IV as what they suggest. Should the limits documented in §10 fail to hold, or should the three cases prove to share only a surface resemblance, the later parts lose their ground.

The author would welcome objection, correction, and counter-evidence, and can be reached at huangwanhong@serendip.ngo. The purpose of setting the argument out at this length is to make it easy to disagree with.

Part I. A Taxonomy of Governance

§1 The Question and the Method

The question that organizes this part is the following. As the scale of cooperation has grown and the relations to be coordinated have multiplied, what mechanisms has the human record produced for governing public affairs, what did each solve, and under what conditions did each fail.

The method requires two prior decisions, and both are departures from the usual practice.

The first concerns the unit of comparison. What will be compared is the coordination mechanism. The institutional form will not be, and the distinction is not pedantic. The institutional forms that the record presents are almost never pure. A contemporary open-source foundation coordinates simultaneously through hierarchy, through community norms, and through network reciprocity; a socialist state coordinates through hierarchy and, in its actual operation, through informal networks and quasi-market exchanges that its official self-description does not acknowledge; a corporation is internally a hierarchy and externally a market participant. If the comparison is conducted at the level of institutional forms, these composites cannot be analyzed, and the analysis collapses into a comparison of ideologies. Conducted at the level of mechanisms, the composites become tractable: one asks which mechanisms a given arrangement combines, in what proportion, and at which of its interfaces.

Four coordination mechanisms may be distinguished. Hierarchy coordinates through command and rule, and its characteristic instrument is the directive. The price mechanism coordinates through exchange and competition, and its characteristic instrument is the market signal. Norm coordinates through shared expectation and mutual monitoring, and its characteristic instrument is sanction by the group. Network coordinates through negotiated interdependence among parties none of whom commands the others, and its characteristic instrument is the ongoing relation itself.

Coordination mechanism. The means by which action is aligned among participants who cannot align it by direct mutual visibility alone. Four are distinguished. Hierarchy coordinates through command and rule. Price coordinates through exchange and competition. Norm coordinates through shared expectation and mutual monitoring. Network coordinates through negotiated interdependence among parties none of whom commands the others. Institutional forms are composites of these; the mechanism, not the form, is the unit of comparison.

The second decision concerns the shape of the history. It will be argued in §6 that these mechanisms stand in a determinate relation to one another, in which the residue of each becomes the problem of the next. This argument must be insulated at the outset against a misreading to which it is exposed. That one mechanism historically succeeds another does not establish that the successor is superior, and it does not establish that the predecessor failed by its own internal logic. The record contains, with some frequency, cases in which a functioning arrangement was destroyed by external force and in which the destruction was justified by an appeal to progress that the subsequent evidence did not support. Two of the cases examined below are of exactly this kind. A history of governance that could not distinguish an arrangement superseded by its own contradiction from an arrangement abolished by a state that considered it archaic would not be a materialist history. It would be a philosophy of progress with a materialist vocabulary.

§2 Centralized Governance

The Mechanism. Centralized governance coordinates through hierarchy. A center receives information, decides, and transmits directives downward through an apparatus whose members are bound by rule rather than by personal loyalty. Its achievement is that it permits coordinated action at a scale, and over a duration, that no face-to-face arrangement can reach: it makes conduct calculable, it makes outcomes predictable, and it renders the apparatus independent of the persons who happen to staff it at any moment.

The Practices. The Weberian ideal type is not the only realization of the mechanism and it is not the earliest. The Chinese imperial examination system, whose systematic implementation began under the Tang, established the principle of selecting officials by merit rather than by birth; recent research places the institutional turning point from pedigree-based aristocracy to bureaucratic recruitment in the seventh and eighth centuries, and by the Song the examinations had become the primary method of recruitment for official posts. The precedence is not merely chronological, and the direction of influence is documented: the Northcote-Trevelyan reforms that established merit-based entry into the British civil service were recommended with explicit reference to the Chinese examinations, and the method was first tried in the East India Company.

The Chinese case does more than antedate the European one; it modifies the concept. Analysis of Qing provincial appointments finds that the selection mechanism corresponded neither to the ideal type of meritocracy nor to Weberian patrimonialism, but that rulers adapted to conflict dynamics, with selection becoming more merit-based when internal conflict raised the demand for capability. And the examination system operated as something other than a recruitment device. It has been argued that its principal function was the projection of the ideational power of the state: it did not merely select officials but shaped how they thought, producing a culturally coherent bureaucracy bound by shared texts and habits of thought, so that the system was an instrument of political reproduction and not only of staffing. Those recruited through it were masters of a common scholarship and shared values with the local gentry, whose support was necessary for the provision of public goods and the resolution of local disputes.

Achievement and Limit. The advance of centralized governance, assessed against what preceded it, is genuine and it is threefold: speed of decision, capacity for coordination at scale, and the establishment of uniform standards across a territory. Against the alternative of allocation by birth, the examination system in particular represents an enlargement of access whose significance should not be diminished by the observation that the enlargement was partial.

The limit is internal to the achievement. The rule secures calculability by abstracting from the particular, and the same abstraction renders the particular invisible. Information degrades in ascent, since each level summarizes for the level above and every summary is a decision about what may be omitted. Local knowledge is lost, because it is precisely what the rule abstracted from. Innovation is constrained, since the categories through which the apparatus perceives are fixed in advance of what it might encounter. And the knowledge of the governed cannot ascend, for there exists no channel by which it could, the construction of such a channel requiring the very categories whose insufficiency is at issue.

The Chinese case discloses a further limit, which the Weberian type obscures. An apparatus that reproduces a common formation among those it recruits achieves coherence at the cost of variance. The uniformity that made the bureaucracy governable is the uniformity that narrowed what it could conceive.

§3 Market Governance

The Mechanism. Market governance coordinates through price. No agent need possess the dispersed information that the system uses, since the price aggregates it and each participant acts on the signal alone. Incentive is aligned with contribution automatically, dispensing with the supervisory apparatus that hierarchy requires.

The Practices. The mechanism is realized in market exchange as theorized from Smith onward, in the Hayekian account of the price system as an information-processing device, and in the sequence of liberalizations conducted under that account in the late twentieth century. It should be noted that no actual economy is coordinated by price alone, and that the firms which populate markets are internally hierarchies, which is a fact about the composite character of institutional forms rather than an objection to the analysis of mechanisms.

Achievement and Limit. The advance is precisely located: the mechanism solves what hierarchy could not, namely the use of knowledge that no center can assemble. It supports innovation because it does not require permission, and it coordinates dispersed decision without any decision about dispersal having been made.

The limits are equally located. The price registers what is exchanged and therefore fails to register whatever is produced or destroyed outside the transaction. Goods from which none can be excluded are undersupplied, because the mechanism that funds provision is the mechanism that excludes. Inequality is generated in the ordinary operation of the mechanism rather than as a deviation from it.

The deepest limit concerns the destination of value. The mechanism permits value to move toward whoever occupies the position from which it may be appropriated, and the movement of value away from the relations that produced it is constitutive of the mechanism. This is the point at which the sixth category below takes its bearings, and the observation is a structural description.

§4 Commons Governance

The Mechanism. Commons governance coordinates through norm. The users of a resource make the rules that bind their use of it and monitor one another’s compliance. Neither directive nor price is required, because the mutual visibility of the participants performs the work that both would otherwise perform.

The Practices. The practices are numerous, ancient, and distributed across cultures, and this distribution is itself the finding. Irrigation communities, forest commons, and inshore fisheries have been documented across a large body of cases as long-enduring, self-organized, and self-governed. The European Alpine and Iberian cases belong to this record, and so do the Japanese iriai forests, whose institutions were established largely during the Edo period and which supplied firewood, forage, and grass under local rules that protected the resource from overexploitation.

The Balinese subak deserves separate treatment, because it is the case in which the mechanism is seen in its greatest development. Irrigation cooperatives that manage the terraces are, together with the terraces themselves, approximately a thousand years old, and the system of water temples that links them in a nested hierarchy is of comparable antiquity, with subak and water temples described in royal inscriptions of the eleventh century. Its structure is worth stating with precision, since it is the structure that the later parts of this discussion will need. Each subak performs ritual offerings at its own water temple, and the temple is also the venue at which farmers elect leaders and decide irrigation schedules. Subak sharing a common water source form a congregation at a regional temple, at which agreement is reached on cropping schedules at the scale of the watershed. Coordination is thereby achieved at successive scales: the village temple governs the water entering the nearby terraces, the regional temple governs the water flowing into the larger area. The arrangement has been characterized as one in which sustainability and social coordination are emergent properties of a complex network of cooperation. When the system was modeled computationally, water-temple management was found to provide the optimal balance between pest suppression and water sufficiency.

Achievement and Limit. The advance is the use of local knowledge that neither hierarchy nor price can incorporate, rules that adapt to the particular conditions of the resource, and a level of participation from which compliance follows without external enforcement. The subak case suggests, further, that the mechanism may not be confined to the single small group. Nested coordination across an entire watershed, sustained for a millennium, is at least one instance in which norm-based coordination reached a scale that the mechanism’s critics have supposed closed to it. How far the instance generalizes is not settled by the instance.

The limits are two, and they must be distinguished sharply, because the distinction is the methodological point of this part.

The first limit is internal. Coordination through norm requires mutual visibility, and mutual visibility is what growth in the number of participants destroys. The mechanism requires a boundary, since trust is maintained by knowing who belongs, and the boundary that makes trust possible constitutes those outside it as external to the community’s obligations. These are properties of the mechanism.

The second is not a limit of the mechanism at all, and to record it as one would be to falsify the history. The iriai were dismantled in the postwar period by a government program to modernize them, undertaken on the view that collective management was archaic and inefficient, which shifted management from collective to individualized form. The subak were disrupted by the Green Revolution of the 1970s, when farmers were required to plant as often as possible with new varieties, fertilizers, and pesticides, bypassing the cropping patterns that the water temples had regulated; local authorities assumed control of water management and traditional schedules were ignored. The consequence was not improvement. Pest outbreaks followed and harvests were lost on a large scale, year after year. In the 1980s the Balinese government abandoned the imposed regime and the authority of the water temple system was restored.

The subak did not fail. The subak was overridden by a state acting on a theory of its own backwardness, and the theory was refuted by the outcome. Whatever else the history of governance may be, it is not a sequence in which the superior mechanism displaces the inferior. Any account that reads the decline of commons arrangements as their supersession has mistaken the exercise of power for the movement of history.

An arrangement superseded by its own contradiction and an arrangement abolished by a power acting on a theory of its backwardness are two different terminations, and a history that cannot distinguish them is a philosophy of progress in materialist vocabulary. The subak and the iriai did not fail. They were overridden, and in the Balinese case the outcome told against the theory that had justified the overriding, the imposed regime being abandoned after the harvests failed.

§5 Collaborative and Network Governance

The Mechanism. Collaborative governance coordinates through network. Government, firms, non-governmental organizations, residents, and universities participate in the governing of a matter that none of them commands, and coordination proceeds through negotiated interdependence rather than through directive, price, or the norms of a bounded community.

The Practices. The practices are those of co-production, multi-stakeholder governance, and network governance as developed over the past several decades, and they are found in urban regeneration, community development, and environmental governance. The mechanism is the characteristic response to problems that cross the boundaries of existing jurisdictions and that no single actor possesses the authority or the knowledge to address alone.

Achievement and Limit. The advance is that the mechanism operates where the others cannot: across boundaries, among parties of unequal kind, on problems that belong to no one’s jurisdiction. It permits local knowledge to enter without requiring the bounded community that commons governance requires, and it permits coordination without the authority that hierarchy requires.

The limits are three. Coordination costs are borne as unremunerated participation, and they do not distribute evenly but concentrate on those who take responsibility. Networks are not flat, and those through whom coordination must pass acquire an influence that no rule granted and that is therefore not subject to the procedures by which granted authority is checked. And accountability becomes diffuse in proportion as authority becomes shared, so that the mechanism’s characteristic virtue and its characteristic pathology have the same source.

§6 The Sequence and Its Residues

The four mechanisms may now be placed in relation. The relation is one of residues: each mechanism, in solving its problem, leaves something unsolved that becomes the problem to which the next mechanism is the response.

Hierarchy secures coordination at scale and cannot recover the local knowledge from which its abstraction departed. Price recovers exactly that: dispersed knowledge acting locally, its effect propagating through the signal, with no ascent through any hierarchy required. The remedy is specified by the contradiction it answers. Price, in turn, secures the use of dispersed knowledge and permits value to be drawn away from the relations that produce it; and in dissolving the traditional community it renders visible, in the only way available to it, what that community had been supplying unpriced: trust, situated knowledge, and the disposition to cooperate absent enforcement. Norm-based governance takes up what price consumed. Norm secures cooperation through mutual visibility and cannot survive the loss of mutual visibility that scale imposes, nor extend its obligations past the boundary that constitutes it. Network takes up what norm could not: coordination across boundaries, among unlike parties, without a bounded community. And network leaves the coordinator’s position unaccountable and the coordinator’s burden unremunerated.

Two residues have not been taken up by any of the four. The first is the government of systems that do not hold still, which hierarchy could not manage and which network addressed only at the level of the individual problem. The second is the destination of value, which price made into a structural feature and which none of the subsequent mechanisms has recovered. This discussion reads the two contemporary lineages as descending from these two residues, and offers the descent as an account of why the two bodies of work have engaged one another so little.

Two residues survive the sequence unclaimed: the government of systems that do not hold still, and the destination of the value produced. The complexity lineage descends from the first and the generativity lineage from the second. Their descent from different residues would account for their having developed in isolation, and their having reached, at approximately the same moment, the limit that the other’s competence would address is the reason offered here for thinking them together.

The comparison may be summarized. The table records, for each category, the dominant coordination mechanism, whether local rule-making is admitted, whether the arrangement learns, and how it treats the destination of value. The last column is the one that the received comparisons omit, and its omission is the reason the sixth category had to be constituted separately.

Category	Mechanism	Local rules	Learning	Destination of value
Centralized	Hierarchy	Low	Weak	Upward to the center
Market	Price	Medium	Strong	To the appropriating position
Commons	Norm	High	Strong	Retained by the user group
Collaborative	Network	High	Strong	Unspecified; diffuse
Complexity	Any	High	Very strong	Not theorized
Generativity	Norm, network	Very high	Strong	Returned to generators

The horizontal rule in the table is not decorative. Above it are coordination mechanisms realized in institutional forms; below it are lineages that combine mechanisms and have not settled into a form. (The last two rows, Complexity and Generativity, are these lineages.) Reading the table as a single list of six options would obscure precisely the distinction on which Part II depends.

The digital condition should be recorded here, since it supplies the material possibility of both. By collapsing the cost of coordination, it makes possible a coordination neither hierarchical nor priced at scales beyond the reach of mutual visibility, which is the condition of contemporary peer production. And it reproduces the contradiction of price in a new medium, since the infrastructure through which coordination passes becomes a coordinating node, and data becomes the new object of extraction, drawn from the relations whose coordination the platform enables.

Part II. The Two Contemporary Lineages

§7 The Standing of These Two

The two categories examined in this part are not a fifth and sixth mechanism to be set beside the four. They are of a different kind, and stating the difference is necessary if the comparison in §10 is to be intelligible.

The four preceding categories are coordination mechanisms that have been realized in durable institutional forms. The two that follow are research and practice lineages, which is to say programs that are still in motion and that have not settled into an institutional form. Complexity governance is a governance perspective rather than an institution, and it can be pursued within a state apparatus, within a firm, or within a community. Generative practice is a family of experiments, and it can adopt the norm-based form of the commons or the network form of peer production.

That these two are the ones that a relational reading approaches most closely is therefore not a coincidence and not a preference. It follows from what they are. A mechanism that has settled into an institutional form has, by settling, fixed its answer to the question of what it coordinates and how. A lineage still in motion has not, and it is in the space of what has not yet been fixed that the questions of this discussion can be posed at all.

§8 The Complexity Lineage

The first lineage descends from the residue of coordination. Its problem is the government of systems that do not hold still, and its formative context is environmental and natural-resource management, where the inadequacy of centralized expert control became evident earliest.

Adaptive governance is understood in this literature as the range of interactions among actors, networks, organizations, and institutions that emerge in the pursuit of a desired state for social-ecological systems, invoked in response to the complexity and uncertainty of rapid environmental change; its emergence is traced explicitly to the failure of earlier management regimes to construct governance structures robust enough to secure sustainability under uncertainty. Its normative content has been summarized as the devolution of management authority to promote participation, the construction of collaborative networks across levels, the maintenance of arrangements flexible enough to permit response to surprise, and learning through those networks. Polycentricity supplies the structural correlate: an arrangement is polycentric when its constituent decision-making entities possess considerable de facto autonomy from one another, whether or not they are formally independent, and adaptation in such arrangements emerges among multiple centers of agency as a relatively self-organized process marked by innovation and social learning. Resilience supplies the criterion of success, characterizing the degree to which a system endures perturbation without collapsing into an alternative stable state.

The lineage’s governing question is thereby fixed. It asks how a system may persist.

§9 The Generativity Lineage

The second lineage descends from the residue of extraction. Its problem concerns the value that circulates within a system and whether that value returns to those who produced it. The persistence of the system is a separate question, and this lineage does not pose it.

Its diagnosis is that both received alternatives extract. In the formulation that has organized much of the recent work, communism extracts value for the state and capitalism extracts it for the corporation, so that both destroy communities and individuals; against both is proposed the circulation rather than the extraction of value, defined in terms of the universal right to generate unalienated value and participate directly in its benefits, the right of value generators to create their own conditions of production, and the right of communities of value generation to sustain paths for its circulation. The criterion follows: sustainable systems return value to its sources, and crises arise where value is expropriated with little return. It is worth recording that this account takes its bearings explicitly from indigenous practice, holding that a better model than the top-down solutions is found in the bottom-up approach of indigenous traditions, which practiced value circulation rather than value extraction.

The empirical field is populated by practices in which the return can be observed: peer production of the open-source and encyclopedic kind, which has challenged proprietary control over intellectual property through a gift exchange of labor value; agroecological practice, which establishes a return cycle of ecological value through composting and intercropping; worker-owned cooperatives and artisanal production, in which profits, skills, and responsibilities circulate back to those generating the original value. Community currencies and the maker movement belong to the same family.

The lineage’s governing question is thereby fixed, and it is not the question of the first lineage. It asks how value may continue to be generated and to return.

§10 The Blind Spot of Each Lineage

The claim of this section is that each lineage has a limit, that the limit of each is documented within its own critical literature, and that the two limits are complementary. This is the section on which the remainder depends.

The Limit of the Complexity Lineage. The lineage answers the persistence of systems. It has been criticized from within for the thinness of its treatment of power and of the sources of cooperation, and the criticism has become one of its recognized internal problems rather than an external objection.

The critique has been formulated as concerning the process, power, and meaning dimensions of environmental and natural-resource governance. Resilience research has been characterized as depoliticized, as failing to recognize social contingency, and as failing to recognize the variable capacities of different actors to shape governance systems. The judgment has been rendered that resilience thinking has failed to deal with issues of power and has attended too much to governance structures and functions and too little to political, historical, and cultural context, so that the literature has emphasized the management of policy problems at the expense of the politics of governing itself. The point has been put with particular sharpness in the observation that adaptive co-management, like all environmental management, is an inherently political and not simply a technical undertaking, and that management institutions are embedded in unequal relations of production such that their reform is unlikely by itself to produce equitable outcomes.

A second and structurally distinct limit concerns what happens when adaptive arrangements meet incumbent institutions. In the study of adaptive governance initiatives in urban regeneration it has been found that such initiatives can institutionalize and materialize, but that the resilience of incumbent institutions against them causes the emergent adaptive institutions to lose their core features over time, so that the transformational effect remains limited; the mechanism identified is that existing institutions run on a particular distribution of power maintained by those who benefit from it. The limit is therefore not that the lineage fails to describe complex systems. It is that the adaptive form, on entering a field structured by power, is reshaped by that field.

What the lineage does not supply is an account of why participants continue to cooperate, of where the value sustaining their cooperation originates, and of what becomes of that value once produced. Its criterion is persistence, and a system may persist while extracting comprehensively from those who sustain it.

The Limit of the Generativity Lineage. The lineage answers the return of value. Its limit concerns scale, and it appears with unusual clarity because its exemplary practices have now run long enough for their failure modes to have been documented.

Peer production exhibits the limit as the exhaustion of those who maintain. The reported condition of open-source maintenance is that a majority of maintainers work unpaid, that a majority have quit or considered quitting, and that a substantial proportion of those who leave cite exhaustion as the reason. The consequence is not confined to individual well-being: a widely deployed component of a major infrastructure project was retired not because it had become obsolete but because those maintaining it could no longer sustain the work. The structure of the failure is what matters here. Value circulates, in the sense the lineage requires, since the software returns to the community that produced it. What does not circulate is the labor of maintenance, which concentrates on a few and is drawn upon by an ecology of corporate users that treats the resource as free. A vicious circle has been described in which overworked maintainers lack the time to onboard assistance, which deepens the exhaustion and raises the probability of abandonment. This is a governance failure occurring inside a system whose value circulation is functioning as designed.

Commons arrangements exhibit the limit as the difficulty of extension. The received response is the nesting principle, according to which local organization is neither absorbed nor sidelined but included within a multi-level polycentric system operating from local to regional levels. The difficulty is that the up-scaling of community-based approaches has run well ahead of knowledge about how such up-scaling might work. It has been argued that the design principles scale for some classes of global public good and are not conducive to others, and it has been acknowledged that the principles proposed for global commons are not yet solidly established or strongly tested, resting on a limited empirical base at that scale.

A qualification is required here, and it is the qualification that Part I was constructed to make available. The scale limit of the commons is not established by the historical decline of commons arrangements, because that decline was, in the cases examined, not a failure of the arrangements. The subak coordinated a watershed for a millennium and were disrupted by state imposition, not by their own limits; the iriai were dismantled by a modernization program premised on their obsolescence. The evidence for the scale limit is the contemporary evidence: the documented difficulty of extending commons governance to global resource problems, and the documented concentration of maintenance burden in peer production. These are limits observed in functioning systems, not inferred from the fate of systems that were overridden.

The Complementarity. The two limits are the inverse of one another. The complexity lineage possesses an account of persistence at scale and lacks an account of value and of why cooperation is sustained. The generativity lineage possesses an account of value and of its return, and lacks an account of persistence at scale. Each is strongest where the other is weakest.

The complexity lineage possesses an account of persistence at scale and lacks an account of value and of why cooperation is sustained. The generativity lineage possesses an account of value and of its return and lacks an account of persistence at scale. Each is strongest where the other is weakest. This discussion reads the complementarity as the residue structure of the historical sequence, and not as a coincidence of two literatures.

This discussion reads the complementarity as the residue structure of Part I, and not as an artifact of the characterizations offered here. On that reading the complexity lineage inherits the problem of coordination, which hierarchy raised and which price and network transformed without resolving, and the generativity lineage inherits the problem of extraction, which price made structural and which the digital condition reproduced. These are the two residues the sequence leaves unclaimed, and the two lineages are the forms in which each is presently being worked. Their descent from different residues would account for the slightness of the engagement between them. The reason offered here for thinking them together is that each has reached, at approximately the same moment, the limit that the other’s competence would address.

The separation itself is asserted here as a reading and not as a documented fact about the two literatures. What the discussion documents is the limit of each and the complementarity of the two limits. Whether the two bodies of work have in fact proceeded in mutual ignorance is a bibliometric question, and it is one this discussion does not settle.

§11 The Question Left Open by the History

The question left standing may now be stated. Whether there is a mode of governance that answers both demands at once, sustaining complex systems while sustaining the generation and return of value within them.

The status of this question requires precision. It is not posed because a framework was available that could answer it. It has been arrived at by constructing a taxonomy, examining the practices under each category and the limits of each, and reading the critical literature internal to the two lineages that the sequence produced. Had that literature not documented these limits, the question would not arise and the remainder of this discussion would have no object.

Part III. The Convergence

§12 The Insufficiency of a Theoretical Gap

Part II established that each lineage has a limit and that the two limits are complementary. This establishes the existence of a theoretical gap, and a theoretical gap is not a reason to do anything. Gaps in the literature are numerous and most of them deserve to remain open. A research program is warranted not by the incompleteness of the theories that precede it but by the existence of objects that those theories fail to describe, and the burden of this part is to show that such objects exist.

The claim to be defended is narrow and it should not be inflated in the reading. It is not that society requires a new governance theory, which is a proposition that could not be established by any evidence and which the discussion does not assert. It is the following. There exists a class of contemporary governance objects that possess complexity and generativity simultaneously; when the two lineages are applied to such an object, each returns a verdict of health by its own criterion; and the object fails nonetheless. Where this occurs, the failure is invisible to both frameworks, and it is invisible not because either framework has been misapplied but because each is functioning as designed.

An object of this kind is a double blind spot rather than a gap. The distinction matters practically. A gap is a region no theory has yet described, and it is discovered by surveying the literature. A double blind spot is a region that the available theories describe as unproblematic, and it is discovered only when the object fails. The first can be filled at leisure. The second is found by casualties.

Gap and double blind spot. A gap is a region no available theory describes, and it is discovered by surveying the literature. A double blind spot is a region the available theories describe as unproblematic, and it is discovered only when the object fails. The first may be filled at leisure. The second is found by casualties.

§13 The Structure of the Convergent Object

The digital condition, as recorded at the end of Part I, supplies the material possibility of both lineages at once, and it is therefore unsurprising that it should have produced objects in which both properties are present. Such an object may be characterized by two conditions holding together.

It is complex in the sense the first lineage requires. Many actors, no center, non-linear propagation of disturbance, and an environment that does not hold still. It is governed, to the extent that it is governed at all, through adaptive and polycentric arrangements rather than through directive.

It is generative in the sense the second lineage requires. Its value is produced continuously by its participants, it circulates back to them, and the system persists only for as long as the production continues. No stock underlies it. There is no reservoir on which it can draw when generation stops.

The two conditions were, for most of the history reconstructed in Part I, satisfied by different objects. A watershed is complex and its value is largely given rather than continuously produced. A workshop is generative and it is not complex. What the digital condition has produced is a class of objects in which both hold at once, and the members of this class are neither exotic nor peripheral. They include the software infrastructure on which the technical civilization runs, the largest reference work ever assembled, and the moderation systems that make the major discussion platforms usable.

§14 Three Cases

The three cases that follow are presented in the same order in each instance: the verdict of the first lineage, the verdict of the second, and the failure that both verdicts miss. The point is the isomorphism. One case is an anecdote and three with the same structure are a pattern.

Peer-Produced Infrastructure. Consider the maintenance of the open-source components on which contemporary computing depends.

The complexity lineage returns a verdict of health. The ecosystem is polycentric in the strict sense, its decision-making entities possessing considerable de facto autonomy; it is adaptive, learning through distributed networks and revising its arrangements in response to disturbance; it is resilient in the sense of the definition, having absorbed decades of perturbation without shifting into an alternative stable state. By the criteria the lineage most prominently supplies, the system reads as a success, and it is in fact frequently offered as an exemplar of the very properties the lineage recommends. Whether some criterion internal to the lineage would register the failure is the first of the refutation conditions stated in §22.

The generativity lineage returns a verdict of health. Value is generated by the participants and returns to them: the software produced by the community is available to the community, unalienated in precisely the sense the lineage specifies. There is no expropriation of the artifact. Judged by the criterion that sustainable systems return value to their sources, the system is functioning as designed.

The system is nonetheless failing. A majority of maintainers work unpaid; a majority have quit or considered quitting; a substantial proportion of those departing cite exhaustion. A widely deployed component of a major infrastructure project was withdrawn from security maintenance not because it had been superseded but because those maintaining it could no longer sustain the work. A vicious circle has been documented in which overworked maintainers lack the time to onboard assistance, which deepens the exhaustion and raises the probability of abandonment.

The failure is located precisely where neither framework looks. The artifact circulates; the labor of maintaining it does not. The burden of sustaining the system concentrates upon a small number of participants and is drawn upon by an ecology of corporate users for whom the resource is free. This is neither a failure of value circulation nor a failure of adaptive capacity. It is a failure in the distribution of the cost of sustaining the relations through which value circulates and adaptation occurs, and neither lineage possesses a term for that quantity.

The Encyclopedic Commons. Consider the largest collaborative knowledge project yet undertaken.

The complexity lineage returns a verdict of health, and the generativity lineage returns a verdict of health, on the same grounds as before. The project is self-organized, polycentric, adaptive, and has absorbed a quarter-century of disturbance. Its value is generated by volunteers and returns to the public without appropriation.

The output metrics confirm both verdicts. Content quality is up, article count is up, and the number of featured articles is up. The encyclopedia has never been better. This is the finding of those most concerned about the project’s condition, and it is offered by them as a warning.

The system is nonetheless failing. Acquisition of new contributors has fallen by more than a third while total edits have risen, which is to say that more work is being extracted from a shrinking base of committed volunteers. The system is concentrating. The administrator corps of the English project has declined from nearly eighteen hundred to slightly more than eight hundred, not all of whom are active, so that a project of increasing importance is held together by a shrinking body of unpaid volunteers. Efficiency gains, for years the compensation for a contracting contributor base, are reported to have reached their limit, an experienced editor who departs now costing more to replace for want of a pipeline behind them. The quality metrics, on this reading, are not evidence of health. They are evidence that the people maintaining the system have become irreplaceable.

The isomorphism with the preceding case is exact and it is worth stating without ornament. Value circulation is functioning. Adaptive capacity is functioning. The output is improving. And the relational structure that bears the cost of producing the output is being consumed by the production. Both frameworks measure the output; neither measures the consumption.

Platform Moderation. Consider the volunteer moderation systems on which the major discussion platforms depend.

Here the two verdicts are again favorable. The system is polycentric to a degree that its own operators emphasize, with communities self-organizing and devising rules tailored to their particular circumstances; it is adaptive, with moderators developing sophisticated understanding of local context and adapting general policy to local norm. Value is generated by participants and returns to the community, which is the ordinary self-description of the moderator and is not false.

The system is nonetheless failing, and this case makes visible a feature that the first two cases contain but do not display so plainly. The labor performed by volunteer moderators on a single platform has been valued at no less than several million dollars annually, against a platform revenue in the hundreds of millions, and the platform employs no paid moderators while relying on more than a hundred thousand unpaid ones. Moderators face exhaustion from the volume of work, from being targeted for their decisions, and from the emotional labor the role requires; the two principal reasons given for quitting are insufficient available time and conflict with other moderators. The position has been characterized as one in which unpaid volunteers bear a substantial burden of emotional labor while serving platform operators who may remove them at will.

What this case adds is the extraction channel. In the first two cases the beneficiaries of the uncompensated maintenance are diffuse, an ecology of users and readers. Here the beneficiary is a single corporate entity whose valuation depends on the labor it does not pay for. The generativity lineage possesses the concept that would name this, since it is extraction in the strict sense; but it does not detect it, because the value whose circulation the lineage tracks, namely the content, does circulate. What is extracted is the labor of governing the content. The content itself is not extracted, and the cost of governing it is a relational cost, not a value flow.

§15 The Consequences of the Isomorphism

The three cases have the same structure and the structure may now be stated in general form.

In each, the object is simultaneously complex and generative. In each, the first lineage’s criterion returns health, because the system persists and adapts. In each, the second lineage’s criterion returns health, because value is generated and returns to its generators. In each, the output improves even as the failure advances, so that the metrics both lineages would consult are actively reassuring. Their silence would be the lesser difficulty. And in each, the failure occurs in the same place: in the relations that bear the cost of sustaining the system, whose burden grows with the system while the relations bearing it do not.

Three consequences follow, and they should be distinguished carefully.

Three documented cases exhibit the same structure. Each object is complex and generative at once. In each, both leading frameworks return a verdict of health by their own criteria, the output metrics improve as the failure advances, and the system fails nonetheless. In each, the failure falls in the same place: on the relations that bear the cost of sustaining the system, whose burden grows with the system while the relations bearing it do not. If the three are indeed isomorphic, and §22 states what would show they are not, then neither framework is wrong and both are incomplete in the same respect. Each answers correctly the question it asks. Neither asks the question on which these objects die. Two frameworks that are each silent on a quantity do not become articulate about it by being placed side by side.

The first is that the frameworks are not wrong. Nothing in the foregoing constitutes a refutation of adaptive governance or of generative justice. Each answers correctly the question it asks. The difficulty is that neither asks the question on which these objects fail, and their conjunction does not ask it either, since two frameworks that are each silent on a quantity do not become articulate about it by being placed side by side.

The second is that the failure mode is invisible in advance. Where both criteria return health, the frameworks in use emit no warning, and the withdrawal of a critical component from security maintenance was not, so far as the documented record shows, anticipated by either. The community around such a project may know its condition well; the claim here concerns the analytical frameworks and not the tacit knowledge of participants. This is what distinguishes a double blind spot from a gap, and it is the reason the matter is urgent.

The third concerns the trajectory. The class of convergent objects is not shrinking. The material condition that produces them, namely the collapse of coordination cost that permits generative production at complex scale, is not being withdrawn. Each addition to the class is an object for which the available frameworks will report health until it fails.

§16 The Consequence of Non-Address

The question may be answered without prophecy, because the answer has already been observed three times.

What happens is that systems whose output is improving, whose value circulation is functioning, and whose adaptive capacity is intact, are consumed from within by the concentration of the cost of their own maintenance, and that this consumption is not detected until the moment of failure because the instruments in use are measuring other things. A component on which a large part of the technical infrastructure depends ceases to receive security patches. A reference work of unmatched scale becomes dependent on a corps of maintainers who cannot be replaced. A platform’s governance runs on labor that its own governance model does not acknowledge as labor.

None of this is a forecast. It is a description of events that have occurred, and the forecast that follows from them is the modest one that they will recur, because the conditions that produced them persist and because nothing in the available frameworks would announce their recurrence in advance.

This is the practical motivation, and the discussion advances no more than this. The claim is not that a relational account of governance is correct, and it is not that society requires one. The claim is that a class of objects exists for which the two leading frameworks report health while the object dies, that the failures share a structure, and that the structure concerns a quantity neither framework measures. Whether that quantity is well described as relational, and whether a governance account organized around it would do better, remains to be shown. It is shown nowhere in this discussion. What has been shown is that the question is not idle.

Part IV. A Relational Reading

§17 The Unmeasured Quantity

The three cases converge on a quantity that neither framework measures, and the task of this part is to characterize it as precisely as the evidence permits and no more precisely than that.

The quantity is not value, for value circulated in all three cases and its circulation was correctly registered. It is not the persistence of the system, for the systems persisted and adapted, and this too was correctly registered. It is not output, which improved in all three. What was consumed, in each case, was the capacity of a particular set of participants to continue bearing the cost of sustaining the arrangement through which value circulates and adaptation occurs.

This capacity has three features that the cases display jointly. It is borne by relations. The burden fell on the maintainers, the administrators, and the moderators, and it did not fall on the arrangement at large. It is consumed by the operation it enables, since the very activity through which the system produced its output was the activity that exhausted those who bore it. And it does not appear in the output, which is why the output metrics were reassuring throughout and why in the encyclopedic case the improvement of quality was itself the evidence of the danger, the maintainers having become irreplaceable precisely by maintaining well.

A quantity with these features is a property of the relations among participants. It is a property neither of the system nor of the value. This is the induction, and it is offered as an induction. The term that would have made the three failures visible is the same term in each case, and the discussion proposes, provisionally, to call it the relation.

§18 The Standing of the Term

The modesty of the proposal must be preserved against two inflations to which it is exposed.

It does not follow that relations are the object of governance, and the discussion does not assert this. The objects of governance remain what they have always been: resources, organizations, communities, enterprises, knowledge, technologies, and, among these, relations themselves. To reduce the object of governance to the relation would reproduce, in a relational vocabulary, exactly the reductive move for which the critical literature has already convicted the others.

Nor does it follow that a relational account will succeed. What the three cases support is the weaker proposition that some quantity, unmeasured by the available frameworks, is implicated in a documented class of failures. That such a quantity exists as a well-defined magnitude, and that it is relational in character, is an induction from three cases and awaits the operational account of §23. That a governance account organized around it would detect the failures in advance is a conjecture, and it is a conjecture that the present discussion does nothing to establish.

The distinction that permits the term to be used without inflation may be put as follows. The relational is a perspective, a way of seeing what a governance arrangement is doing to those who sustain it. The generative is an orientation, an account of what the arrangement should be producing. Complexity is a condition, the circumstance under which contemporary governance operates. These occupy three different positions, and the collapse of the three into a single object is the failure mode against which the whole of this part is a guard.

§19 The Distinction from Labor

The most parsimonious objection to the foregoing is that a name already exists for what the three cases exhibit, and that the name is unpaid labor. On this reading the maintainers, the administrators, and the moderators are performing work; the work is uncompensated; the beneficiaries are firms and publics that do not pay for it; and the classical analysis of the appropriation of surplus labor accounts for the whole of the phenomenon without any new term. The objection is serious and it must be met before the term relation is permitted to stand, since a concept that adds nothing to an available concept is not underdeveloped but superfluous.

Three features of the cases are not captured by the labor analysis, and they are stated in ascending order of difficulty for it.

The first concerns the direction of the remedy. If the quantity consumed were labor, compensating it would restore the system. This is a testable proposition and the cases speak against it. The reasons volunteer moderators give for departure include, prominently, conflict with other moderators, which payment does not address; the exhaustion documented in peer production is attributed to the volume of demand, to being targeted for decisions taken, and to the absence of tools and support, of which only the first is a quantity that money straightforwardly converts. The vicious circle identified in peer production is that overworked maintainers lack the time to onboard successors, and payment relieves this only if the paid maintainer’s time is thereby freed for onboarding. Where the same demand returns at greater volume, the time is absorbed again. Funding is a necessary component of any remedy and it is not sufficient, and the insufficiency is not incidental.

The second concerns the encyclopedic case, where the labor analysis produces a false negative. What was documented there was not primarily that volunteers were unpaid, which had been true throughout the project’s history including its period of growth. It was that the acquisition of new contributors fell while total edits rose, that the administrator corps contracted from nearly eighteen hundred to slightly more than eight hundred, and that the departure of an experienced editor now costs more to replace because no pipeline stands behind them. The quantity in decline is the capacity to replace those who bear the burden. A labor analysis that attended only to compensation would have registered no change, since the compensation was zero before and zero after. What changed was the structure of succession, and succession is a property of the relations among participants rather than of the exchange between a worker and a beneficiary.

The third feature, and the one on which the distinction principally rests, is the locus. Labor is a relation between a worker and one who appropriates the product, and it is measured on the axis of that appropriation. What the cases display is a burden that falls on the relations by which participants sustain one another: the relations of onboarding, of mutual support, of conflict resolution, of the transmission of tacit knowledge to those who might succeed. The moderator who departs because of conflict with other moderators is not describing appropriation by a beneficiary. The maintainer who cannot onboard is not describing an unequal exchange. In the platform case the appropriation is present and is correctly named by the labor analysis, and it is precisely there that we can see the labor analysis running out, since it names the extraction and remains silent on why the moderators who are not extracted from also leave.

The upshot is not that the labor analysis is wrong. In the platform case it is right, and the generativity lineage’s concept of extraction is the correct concept for what a corporation with a billion in revenue and no paid moderators is doing. The upshot is that the labor analysis is incomplete in a specific way. It analyzes the axis running from the participant to the beneficiary and it does not analyze the axis running from participant to participant, and the three cases fail on the second axis. The term proposed here is a name for the second axis and it earns its place only if the second axis is real. Whether it is real is a question the cases can be made to answer, and §22 states how.

§20 The Distinction from Network and Social Capital

Two further concepts stand close enough to require discrimination.

Network analysis supplies a formal apparatus for representing relations, and the objection is that the quantity at issue is a network property already available for measurement, such as the centrality of the maintainer or the connectivity of the moderator team. The apparatus is indeed the natural one for any subsequent formalization and nothing here is offered against it. The difficulty is that a network is a representation of relations at an instant and the quantity at issue is a rate. What the cases exhibit is not that maintainers occupy central positions, which is true and unremarkable, but that the burden borne by those positions grows while the positions do not multiply. The network is a snapshot; the finding is a divergence between two trajectories. A network representation can express the divergence, but only after one has been told which two quantities to place in ratio, and that is the contribution claimed here rather than the apparatus.

Social capital names a stock of trust, obligation, and reciprocity that a group possesses and can draw upon, and the objection is that the three systems are simply depleting theirs. This is closer, and the distance is worth marking exactly. Social capital is conceived as a resource that inheres in a group and that facilitates coordinated action; its characteristic questions concern how much of it a community has and what it enables. The quantity here is a load-bearing capacity. It is not a stock, and the relevant question concerns whether what it must bear is growing faster than it is. How much of it exists is a different question. A community may possess abundant trust and fail nonetheless if the burden placed on the relations that carry the trust outruns them, which is what the encyclopedic case displays: the community’s cohesion is not in question and its administrators are disappearing. Social capital would predict failure when the stock is low. The cases fail while the stock, so far as any indicator shows, remains high.

§21 The Danger of the Universal Variable

There is a standing objection to relational vocabularies which this discussion cannot evade and does not wish to. Trust is a relation. Power is a relation. Institutions, communication, identity, and norms are relations. If everything is a relation, then relation explains everything and therefore predicts nothing, and the term has purchased its scope at the price of its content. The objection has been put with force from within the relational literature itself, in the observation that pure relationality cannot create objects, that relations require substances as substances require relations, and that the opposition between the two has been overdrawn by its partisans.

The objection is accepted. It is accepted, further, that a relational governance account is more exposed to it than most, since governance is a domain in which every term of art can be redescribed relationally without effort and without gain.

What follows from accepting it is a restriction, and the restriction is the substance of this part’s claim to have said anything. The term relation is not offered here as a general ontology, and the discussion asserts nothing about what the world is made of. It is offered as a name for one specific quantity, which the three cases isolate and which the available frameworks do not measure: the load borne by the relations through which participants sustain one another’s continued participation, taken in ratio to the capacity of those relations to bear it. This quantity is not trust, though trust may condition it. It is not power, though its distribution is a question of power. It is not the network, which is its representation. It is not labor, which runs on the other axis. It is not social capital, which is a stock where this is a load.

The maintenance burden. The load borne by the relations through which participants sustain one another’s continued participation, taken in ratio to the capacity of those relations to bear it. It is not trust, though trust may condition it. It is not power, though its distribution is a question of power. It is not the network, which is its representation. It is not labor, which runs on the axis from participant to beneficiary. This quantity runs on the axis from participant to participant. It is not social capital, which is a stock where this is a load.

A term so restricted is falsifiable and it is nearly useless for most purposes, and those two properties are the same property. The discussion prefers a term that cannot explain everything.

§22 The Conditions of Refutation

The claim of Part IV is that a quantity exists, that it is relational in the sense just restricted, and that the two lineages do not measure it. Each conjunct can fail, and the ways it can fail should be stated so that the discussion can be held to them.

The claim fails if either framework detects the three failures by its own criteria. If it can be shown that a resilience metric, properly constructed, would have declined in advance of the withdrawal of the infrastructure component, or that a measure of unalienated value circulation would have registered the contraction of the administrator corps, then no unmeasured quantity exists and Part IV is empty.

The claim fails if compensation is sufficient. If projects that have secured funding for maintenance are found to exhibit no residual failure of succession, of onboarding, or of conflict resolution, then the labor analysis of §19 is complete and the second axis is not real.

The claim fails if the three cases are not isomorphic. If the maintainer’s exhaustion, the administrator’s disappearance, and the moderator’s departure are found to have distinct causal structures that share only a superficial resemblance, then the induction has been drawn from a coincidence.

The claim fails, finally, if the restriction of §21 cannot be held. If the quantity proposed here proves impossible to distinguish in practice from trust, from power, or from social capital, then the term has become the universal variable it was constructed to avoid, and it should be discarded rather than defended.

None of these tests is performed here. They are stated because a diagnosis that cannot say what would refute it is not a diagnosis.

§23 The Priority of Operationalization

The third of the four conditions carries the weight of the others, since a failure of isomorphism would leave nothing for the remaining tests to be performed upon. The obvious response is to enlarge the sample. If three cases are a slender basis for an induction, then twenty cases across domains, cultures, and scales would be a firmer one, and the natural next step appears to be their collection.

This discussion holds that the natural next step is the wrong one, and the reason is worth stating plainly because it is a constraint on the author before it is a claim about the world.

The quantity proposed in §21 is a ratio between a load and a capacity, and neither term has been given an operational definition. It has been said what the quantity is not: not labor, not centrality, not a stock of social capital. It has not been said, in terms that would permit a judgment in a new case, what would count as an instance of the load or as a measure of the capacity. In the absence of such terms, the addition of a twentieth case cannot confirm the induction, because there exists no criterion by which the twentieth case could be found to exhibit the same structure. Resemblance to the three is all that could be established. Cases would accumulate and the charge of superficial resemblance would accumulate with them. An enlargement of the sample conducted before the concept is operationalized does not strengthen an induction. It multiplies anecdotes.

An enlargement of the sample conducted before the concept is operationalized does not strengthen an induction. It multiplies anecdotes. Operationalization is prior to the collection of further cases, since without an operational criterion no further case can be found to exhibit the structure, and resemblance is all that could be shown.

The order of work is therefore the reverse of the apparent one. Operationalization is prior to the enlargement of the sample, and the three cases are sufficient for the purpose of operationalization because they were sufficient to isolate the quantity. What they are not sufficient for is its confirmation, and the discussion does not claim confirmation.

The materials for the operational account are present in the cases and may be indicated, though not developed, here. In each case the load is borne by an identifiable subgroup, the maintainers, the administrators, the moderators, and it is not borne by the membership at large. In each, the work that constitutes the load is of three kinds, and the three recur: the onboarding of successors, the resolution of conflict internal to the group, and the transmission of knowledge that is not held in the artifact. None of the three is output, which is the structural reason for their invisibility to both lineages, since both measure what the system produces and this work produces nothing that the system’s beneficiaries consume. In each case the load grows with a quantity the system exports, namely the population of dependents, of readers, of users. And in each the capacity is indexed by the size and the replaceability of the subgroup, which is why the contraction of the administrator corps and the bus factor of a software project are measures of the same thing.

An operational account built from these materials would be an account of the maintenance burden, and it would have a property that the present discussion lacks: it would generate a criterion by which an unseen case could be classified. Comparison is all the present account permits. Whether the criterion, once constructed, identifies the same structure in the household, the enterprise, and the state, or identifies it nowhere outside the digital commons, is the question on which the generality of the whole account turns, and it is not a question that further philosophical argument can settle. It is prior to Part V, since a renormalization requires a quantity to renormalize, and the discussion notes that the parts are here presented in an order that their dependency does not follow.

§24 Governance Design Principles

If the relation is admitted as a dimension of analysis, certain questions become available that were not available before. They are stated as questions, since questions are what the present state of the argument supports.

Which relations bear the cost of sustaining the arrangement, and how is that cost distributed among them. Whether the burden borne by those relations grows with the system while the relations bearing it do not. Which feedback loops permit the cost of maintenance to be drawn from participants without return, in systems whose value circulation is otherwise sound. Which relations obstruct generation, and which sustain the disposition to continue that neither rule nor price supplies.

These are analytical questions and not institutional prescriptions. They do not specify what any governance arrangement should be, and they do not substitute for the design principles the existing literatures have established. Their claim is to supply a dimension along which existing arrangements may be interrogated, and their value is to be judged by a single test: whether the interrogation would have disclosed, in advance, the three failures that the existing dimensions reported as health.

Part V. Scale

§25 The Problem of Scale

The practices in which the generativity lineage has demonstrated its results are practices of limited scale: the irrigation community, the forest commons, the makerspace, the software project. This is not an accident of research design. Those are the scales at which the lineage’s mechanisms operate, and the documented failures at larger scale are the evidence that they do not simply extend.

The problem may be stated exactly. The failures documented in §10 are not failures of the mechanisms, which function as specified. Value circulates in the open-source project, and the software returns to the community that produced it. The commons arrangement allocates the resource and enforces its rules. What fails is the passage from one scale to another, and this makes scale a problem of the first order. It is not a subsequent question of application. A governance account that holds only where mutual visibility obtains is an account whose domain of validity has been misdescribed, and it does not await extension.

The sequence at issue runs from the family through the community, the enterprise, the city, the state, and the international organization. Whether these share a common relational structure, and if so which of its properties survive the passage between them, is the question. Three sub-questions must be separated, because they are ordinarily conflated and because they have different answers.

§26 Relational Scale

The first concerns what a relational scale is. The sequence above is ordinarily read as a sequence of magnitudes, and read that way it is a sequence in the number of persons. This reading is inadequate, and its inadequacy is visible in the material of Part II. A software project with a hundred thousand users and three maintainers is not, in any sense that governs its failure, a large system. The relational scale that determines its behavior is the scale of the maintaining relation, which is small, and the failure occurs precisely because a small relational structure is loaded by a large population.

What is required, then, is a notion of scale internal to the relational field. A scale external to it will not serve. Candidate constructions suggest themselves: the number of relations a participant can sustain with attention; the length of the path through which value must travel to return to its generator; the depth of mediation between the point of production and the point of decision. Each of these varies to some degree independently of population, and each is a plausible determinant of where the documented failures occur. Which of them, if any, is the right variable is not settled here, and settling it is prior to any claim about invariance, since one cannot ask what is preserved across a transformation before one has said what the transformation is.

§27 Coarse-Graining and Its Residue

The second concerns what happens in the passage upward. To govern at a larger scale is to treat some multiplicity as a unit: the household stands for its members, the delegate stands for the constituency, the state stands for its population. This is coarse-graining, and it is the operation by which large-scale governance is possible at all. It is not an incidental administrative convenience.

The question is what the operation destroys. Every coarse-graining discards the internal structure of what it aggregates, and the discarding is not neutral with respect to the concerns of this discussion. The internal asymmetries of the aggregated unit become invisible at the coarser description, and they become invisible in exactly the way the critical literature of the complexity lineage has identified: the coarse description registers the system’s persistence while the fine description would have registered who within it bears the cost of that persistence. The bureaucratic contradiction reconstructed in Part I is a statement about coarse-graining. The degradation of information in ascent through a hierarchy is the residue of the aggregation performed at each level.

This yields a formulation of the problem more precise than the one with which the discussion began. The question is not whether governance can be conducted at large scale. It is whether there exists a coarse-graining that preserves what matters. And the two lineages, read in this light, may be understood as having answered a prior question differently. The complexity lineage coarse-grains toward the system, and its invariant is persistence. The generativity lineage refuses the coarse-graining, remains at the fine scale, and preserves the visibility of value and its return, at the cost of an account that does not extend.

§28 Invariance

The third concerns what, if anything, is preserved. The interest of the renormalization analogy lies here, and its interest is precise. The technique to which it refers is a systematic procedure for passing between descriptions at different resolutions, and its central discovery is that certain quantities transform in a controlled way under such passage, so that the behavior at one scale can be derived from the behavior at another. Applied here, the corresponding question is whether any property of a governed relational field transforms in a controlled way as the relational scale changes, and could therefore serve as the stable object that a governance account might hold to in place of a fixed institutional form.

The material of Part III permits one preliminary observation, and it is offered as an observation. It is not a result. The three cases converge on a single quantity, namely the capacity of a particular set of participants to bear the cost of sustaining the arrangement, and they converge further on a single mode of its failure: the burden grows with the system while the relations bearing it do not. The maintainers did not multiply as the dependent ecosystem multiplied. The administrator corps contracted as the encyclopedia grew. The moderator population did not scale with the platform’s valuation.

This suggests that the quantity to be tracked across scales is the ratio between the burden of sustaining the relational field and the relational capacity available to bear it. The number of participants would be the wrong quantity. The suggestion has a consequence worth stating, since it converts the observation into something that could be wrong. If the ratio is the right quantity, then the documented failure of commons arrangements to extend beyond the local is not a separate difficulty belonging to a separate literature. It is the same divergence appearing at a different point, the passage upward in scale being precisely the operation that increases the burden without increasing the capacity. Whether the ratio can be given a determinate sense, and whether it transforms in a controlled way as the relational scale changes, is what a generative renormalization would have to establish.

§29 The Limits of the Analogy

The analogy must be stated together with the points at which it fails, since an analogy defended only where it succeeds has not been defended.

The technique in its original setting presupposes a system whose components are of a determinate kind and whose interactions are specified in advance. A relational field satisfies neither condition. Its participants are constituted by the relations they enter, and their properties are therefore not given prior to the interaction. There is, further, no external observer for whom the coarse-graining is performed. The coarse-graining in governance is itself an act of governance, undertaken by participants within the field, and it therefore alters what it describes. This has no counterpart in the original setting and it may be the point at which the analogy breaks. Whether a procedure that presupposes an exterior can be adapted to a field that admits of none is not known, and the difficulty should not be minimized by the observation that the analogy is illuminating elsewhere.

What can be claimed is narrower and it is claimed without embellishment. The language of scale and of transformation between scales appears better suited to the difficulty than the language of institutional design, since the difficulty is a difficulty of passage between scales and not a difficulty of arrangement at any one of them. The mathematics that such a study would require is not developed here and should not be. What is recorded is that the question has been posed in a form that makes an answer conceivable, and that this is the whole of what the present discussion is in a position to do with it.

§30 The Standing of This Discussion

Parts I and II are the substance of the discussion, and they are the parts that can be completed. Part I constructs a taxonomy at the level of coordination mechanisms rather than institutional forms, examines the practices realized under each category across the cultures in which they are found, assesses the achievement and the limit of each, and arranges them in a sequence of residues. It insists on a distinction that the usual histories elide, between an arrangement superseded by its own contradiction and an arrangement abolished by a power acting on a theory of its backwardness, and it grounds the distinction in the record: the subak and the iriai did not fail. Part II shows that the two contemporary lineages are of a different kind from the four mechanisms, being programs still in motion rather than settled forms; it documents, from the critical literature internal to each, a complementary pair of limits; and it shows that the complementarity is the residue structure of Part I.

Part III is the practical motivation, and it is the part that answers the question of why any of this should be undertaken. It does not argue that society requires a new governance theory, which could not be established by any evidence. It argues that a class of objects exists for which both available frameworks report health while the object fails, that three such failures share a structure, and that the structure concerns a quantity neither framework measures. If the three cases do not exhibit that structure, this part fails and the discussion fails with it.

Part IV is preliminary in its positive claim and firm in its negative one. Its positive claim, that the unmeasured quantity is a property of relations, is an induction from three cases and is conjectured. It is not shown. Its negative claim is that the available concepts do not already cover the quantity, and this claim is argued: labor is shown to run on a different axis, network centrality to be a representation and not a finding, social capital to be a stock where the quantity is a load. The part accepts the standing objection that relational vocabularies purchase scope at the price of content, and it responds by restricting the term to a single quantity narrow enough to be wrong. Four conditions are stated under which it would be wrong.

Part IV also states, in §23, a constraint on what may be done next. The account is not to be strengthened by the collection of further cases, because no criterion yet exists by which a further case could be judged to exhibit the structure. Resemblance is all that could be judged. Operationalization of the load and the capacity is prior to the enlargement of the sample, and the discussion is therefore not a foundation on which more of the same may be built. It is a diagnosis awaiting the instrument that would test it.

Part V is programmatic, and its status requires distinguishing. It has attempted to convert the question of scale from a gesture into a problem, by separating the three questions that the gesture ordinarily conflates: what a relational scale is, what coarse-graining destroys, and whether anything is preserved under the passage. It has offered one preliminary observation, that the quantity on which the three cases converge is the ratio between the burden of sustaining a relational field and the capacity available to bear it, and it has stated the point at which the renormalization analogy may break, namely that the coarse-graining in governance is performed from within the field it describes. None of this is settled. What is claimed is that the question has been given a form in which it could be worked on, and not that it has been worked on.

It should be added that Part V stands on a quantity that Part IV has not operationalized, so that the order of presentation is not the order of dependency. A renormalization requires something to renormalize. The scale questions have been posed here because posing them is what the present state of the argument permits, and they cannot be worked until the instrument of §23 exists.

The conditional character of the whole should be stated without qualification. The discussion rests on two claims that could be false. The first is that the limits documented in §10 are real, and it fails if the critical literature does not sustain them. The second is that the three cases of Part III exhibit the structure attributed to them, and it fails if either framework can be shown to detect those failures by its own criteria, or if the failures do not in fact share the structure claimed. Should either claim fall, Parts IV and V have no ground and should be abandoned. A theory whose motivating problem does not exist requires withdrawal, and the diagnosis has been made falsifiable so that this judgment can be reached.

The title names a destination the discussion does not reach. What has been attempted is the prior task of asking whether the journey is warranted, and the answer rests on the three cases of Part III. Readers who find the cases unconvincing, who know of a fourth that breaks the pattern, or who can show that either framework detects these failures by its own criteria, would be doing the argument the service it most needs. Correspondence is welcome at huangwanhong@serendip.ngo.

End of Discussion II. The problem has been given a history and a diagnosis. Whether it admits of a solution, and at what scales, is not established here.

---

中文

走向生成性关系治理

讨论二

黄万宏 (huangwanhong@serendip.ngo)

摘要

对当代治理提出了两项要求，而本讨论把它们当作不同的两项：在不确定之下对复杂系统的治理，以及价值的持续生成、连同它向那些生产它者的返回。此处把两条研究谱系读作分别对这一项与那一项的回应，第一条围绕适应性的、多中心的、以韧性为取向的治理而聚集，第二条围绕公共领域、生成性正义与同侪生产而聚集。每一条都从内部因一个另一条的能力所会处置之界限而被批评。

一个理论上的缺口本身不是一个行动的理由。因此本讨论审视一类同时是复杂的与生成的当代对象：同侪生产的软件基础设施、百科式的公共领域、以及平台的内容治理。在每一个中，两条谱系都以它们自己的判准返回一个健康的裁断，产出改善，而系统失败。这些失败看似共享一个结构。所被消耗者是一组特定参与者承担维系那一安排之成本的能力，一个两个框架都不测量的量。

对这一量的一个关系性读法作为从这三个案例的一次归纳被提出。它被与劳动、与网络中心性、与社会资本区分开来，而它被反驳的诸条件被陈述。本讨论确立不出一种生成性关系治理的理论，也提不出一个制度设计。它的目的先于两者，即通过追问“是否存在一个可得框架所无法看见的问题”，来确定这样一种理论究竟是否被担保。尺度的诸问题被提出并被留敞，因为一个量必须先被操作化然后才能被横跨诸尺度追踪，而那一操作化是本讨论辨认为接下来之工作者。

关键词： 公共治理；复杂性治理；生成性治理；关系性结构；尺度。

关于本讨论之地位的一则说明。 这是一篇讨论稿。它勘察既有的治理思想与实践、比较两条当代谱系、并审视三个两者看似一同失败的案例。它的贡献是一个诊断。一种生成性关系治理的理论在标题中被命名、而被留为未建；此处的工作是那个在先的工作，即追问这样一种理论究竟是否会被担保，而答案取决于第三部分所辨认的问题是否是真实的。

阐述的次序遵循论证的次序。分类法、比较与诸案例先来，而那关系性的读法在第四部分作为它们所提示者出现。若§10所记录的诸界限不成立，或若这三个案例被证明只共享一个表面的相似，那么随后的诸部分便失去它们的地基。

作者会欢迎反驳、纠正与反证，并可在 huangwanhong@serendip.ngo 被联系。以这样的篇幅陈述论证，其目的是使它易于被反对。

第一部分. 一种治理的分类法

§1 问题与方法

组织本部分的问题如下。随着合作的尺度已增长、而有待被协调的诸关系已成倍，人类的记录为治理公共事务产生了什么机制，每一个解决了什么，以及在什么条件之下每一个失败了。

这一方法要求两个在先的决定，而两者都是对通常做法的背离。

第一个关切比较的单位。所将被比较者是协调机制。制度形式将不被比较，而这一区分不是迂腐的。记录所呈现的诸制度形式几乎从不纯粹。一个当代的开源基金会同时通过层级、通过社群规范、通过网络互惠而协调；一个社会主义国家通过层级、并在它实际的运作中通过它官方的自我描述所不承认的非正式网络与准市场交换而协调；一家公司在内部是一个层级、在外部是一个市场参与者。若比较在制度形式的层面上被进行，这些复合体便无法被分析，而分析坍缩为一种意识形态的比较。在机制的层面上被进行，这些复合体便变得可处理：人追问一个给定的安排组合哪些机制、以何种比例、以及在它的哪些界面上。

四种协调机制可被区分。层级通过命令与规则而协调，而它典型的工具是指令。价格机制通过交换与竞争而协调，而它典型的工具是市场信号。规范通过共享的期望与相互的监督而协调，而它典型的工具是群体的制裁。网络通过无一方命令其余者之诸方之间被协商的相互依赖而协调，而它典型的工具是那持续的关系本身。

协调机制。 那些无法仅凭直接的相互可见性来对齐行动之参与者据以对齐行动的手段。四者被区分。层级通过命令与规则而协调。价格通过交换与竞争而协调。规范通过共享的期望与相互的监督而协调。网络通过无一方命令其余者之诸方之间被协商的相互依赖而协调。制度形式是这些的复合体；机制、而非形式，是比较的单位。

第二个决定关切这段历史的形状。将在§6中论证：这些机制彼此立于一种确定的关系之中，其中每一个的残余成为下一个的问题。这一论证必须从一开始便被隔离，以抵御一个它所暴露于其下的误读。一个机制历史地承继另一个，并不确立那承继者是优越的，而它并不确立那先行者以它自己的内部逻辑失败了。记录以某种频率含有这样的案例，其中一个运作着的安排被外部之力所摧毁、而那摧毁被一个随后的证据所不支持的进步之诉求所辩护。下文所审视的案例中有两个恰是这一类。一段无法把一个被它自己矛盾所取代的安排与一个被一个认为它古旧之国家所废除的安排区分开来的治理史，将不是一段唯物主义的历史。它会是一种以唯物主义词汇写就的进步哲学。

§2 中央化治理

机制。 中央化治理通过层级而协调。一个中心接收信息、决定、并把指令通过一个其成员被规则、而非被个人忠诚所约束的机构向下传送。它的成就是：它准许一种在一个尺度上、并在一段持续之内、任何面对面的安排都无法企及的被协调之行动：它使行为可被计算、使结果可被预测、并使那机构独立于在任一时刻恰好充任它的诸人。

诸实践。 韦伯式的理想类型不是这一机制的唯一实现、而它不是最早者。中国的科举制，它的系统性实施始于唐代，确立了以才、而非以出身选拔官员的原则；近期的研究把从以门第为基的贵族制到官僚录用的制度转折点置于七、八世纪，而到宋代科举已成为官职录用的首要方法。这一先行不仅是编年的，而影响的方向被记录：那确立了以才为基进入英国文官系统的诺斯科特-屈维廉改革，是带着对中国科举的明确参照被建议的，而这一方法首先在东印度公司被试行。

中国的案例所做者多于先于欧洲的案例；它修改了那概念。对清代地方任命的分析发现，那选拔机制既不对应于精英制的理想类型、也不对应于韦伯式的家产制，而是统治者适应于冲突的动态，当内部冲突提高了对能力之需求时选拔变得更以才为基。而科举制作为一个招募装置以外的某种东西运作。有人论证它的首要功能是国家意念性权力的投射：它不仅选拔官员、而塑造他们如何思考，产生一个被共享的文本与思维的习惯所约束的、在文化上融贯的官僚体，因而这一制度是政治再生产的工具、而不仅是充任的工具。那些通过它被录用者是一门共同学问的能手、并与地方士绅共享价值，而士绅的支持对公共物品的供给与地方争端的解决是必要的。

成就与界限。 中央化治理的进展，对照它之前者被评估，是真实的、而它是三重的：决策的速度、在尺度上协调的能力、以及在一个疆域之上统一标准的确立。对照以出身分配的替代，科举制尤其代表一种通往之门的扩大，其意义不应被“那扩大是部分的”这一观察所贬低。

界限内在于那成就。规则通过从特殊抽象而确保可计算性，而同一抽象使那特殊不可见。信息在上升中退化，因为每一层级为它上面的层级作概括、而每一次概括都是一个关于什么可被略去的决定。地方知识丧失，因为它恰恰是规则所从之抽象者。创新被约束，因为那机构据以感知的诸范畴在它可能遭遇者之先便被固定。而被治者的知识无法上升，因为不存在任何它据以能上升的渠道，这样一个渠道的构造要求那些其不足正是问题所在的范畴本身。

中国的案例揭示一个更进一步的界限，即韦伯式的类型所遮蔽者。一个在它所录用者之间再生产一种共同养成的机构，以方差为代价达成融贯。那使官僚体可被治理的一致，正是那窄化了它所能构想者的一致。

§3 市场治理

机制。 市场治理通过价格而协调。没有一个能动者需要占有那系统所使用的分散信息，因为价格聚合它、而每一个参与者仅凭那信号而行动。激励自动地与贡献对齐，免去了层级所要求的监督机构。

诸实践。 这一机制在从斯密以降被理论化的市场交换中、在哈耶克关于价格系统作为一个信息处理装置的说明中、以及在二十世纪晚期依那说明所进行的一系列自由化中被实现。应当指出，没有一个实际的经济仅由价格所协调，而那些充填市场的公司在内部是层级，这是一个关于制度形式之复合品格的事实、而非一个对机制之分析的反对。

成就与界限。 那进展被精确地定位：这一机制解决了层级所无法解决者，即对没有一个中心能集拢之知识的使用。它支持创新，因为它不要求许可，而它协调分散的决策而无需任何关于分散的决策已被作出。

诸界限同样被定位。价格登录所被交换者、因而未能登录任何在交易之外被生产或被摧毁者。任何人都无法被排除于其外的物品被供给不足，因为那资助供给的机制正是那排除的机制。不平等在这一机制的寻常运作之中、而非作为对它的一个偏离被生成。

最深的界限关切价值的去向。这一机制准许价值移向无论谁占据那可从之被占用的位置，而价值离开那生产它之诸关系的移动构成这一机制。这是下文第六个范畴据以定位之点，而这一观察是一个结构性的描述。

§4 公共领域治理

机制。 公共领域治理通过规范而协调。一项资源的使用者制定约束他们对它之使用的规则、并相互监督彼此的遵从。指令与价格都不被要求，因为参与者的相互可见性执行两者否则会执行的工作。

诸实践。 这些实践为数众多、古老、并横跨诸文化而分布，而这一分布本身即是发现。灌溉社群、森林公共地、与近岸渔场，已横跨一大批案例被记录为长久持存、自组织、并自我治理。欧洲的阿尔卑斯与伊比利亚案例归属这一记录，日本的入会林也是，其制度大体在江户时期被确立，而它在保护那资源免于过度开采的地方规则之下供给薪柴、饲草与草料。

巴厘的subak（水利会）值得单独处理，因为它是那机制在它最大发展中被看见的案例。管理梯田的灌溉合作社，连同梯田本身，约有一千年之久，而那把它们联结于一个嵌套层级中的水神庙系统是可比的古老，subak与水神庙在十一世纪的王室铭文中被描述。它的结构值得被精确地陈述，因为它是本讨论后诸部分将需要的结构。每一个subak在它自己的水神庙进行仪式供奉，而那庙也是农民选举首领并决定灌溉时刻表的场所。共享一个共同水源的subak在一个区域庙形成一个会众，在那里就流域尺度上的种植时刻表达成一致。协调由此在接续的诸尺度上被达成：村庙治理进入邻近梯田的水，区域庙治理流入较大区域的水。这一安排被刻画为一个其中可持续性与社会协调是一个复杂合作网络之涌现性质的安排。当这一系统被计算地建模时，水神庙的管理被发现提供虫害抑制与水量充足之间的最优平衡。

成就与界限。 那进展是对层级与价格都无法纳入之地方知识的使用、适应于那资源之特定条件的规则、以及一种从之遵从无需外部强制便随之而来的参与水平。subak的案例进一步提示，这一机制可能不被限于单个小群体。横跨一整个流域、被维持一千年的嵌套协调，至少是一个基于规范的协调达到了这一机制的批评者曾设想对它关闭之尺度的实例。这一实例推广到多远，不由这一实例所了结。

诸界限有两个，而它们必须被锐利地区分，因为这一区分是本部分的方法论要点。

第一个界限是内部的。通过规范的协调要求相互可见性，而相互可见性正是参与者数目的增长所摧毁者。这一机制要求一个边界，因为信任由知道谁属于其中而被维持，而那使信任成为可能的边界把那些在它之外者构成为外在于该社群之义务者。这些是这一机制的性质。

第二个根本不是这一机制的一个界限，而把它记录为一个界限便会伪造历史。入会在战后时期被一个使它们现代化的政府计划所拆解，那计划是以“集体管理是古旧而低效的”这一看法被承担的，它把管理从集体形式移向个体化形式。subak被1970年代的绿色革命所扰乱，当时农民被要求以新品种、化肥与农药尽可能频繁地种植，绕过了水神庙曾调节的种植模式；地方当局接管了水的管理而传统的时刻表被忽视。其后果不是改善。虫害爆发随之而来、而收成在一个大尺度上、年复一年地损失。1980年代巴厘政府放弃了那被强加的体制、而水神庙系统的权威被恢复。

subak没有失败。subak被一个以一种关于它自己落后之理论而行动的国家所压过，而那理论被结果所反驳。无论治理的历史还是什么，它都不是一个优越的机制取代低劣者的序列。任何一个把公共领域安排之衰落读作它们之被取代的说明，都已把权力的行使误当作历史的运动。

一个被它自己矛盾所取代的安排与一个被一个以关于它落后之理论而行动的权力所废除的安排，是两个不同的终结，而一段无法把它们区分开来的历史是一种以唯物主义词汇写就的进步哲学。subak与入会没有失败。它们被压过，而在巴厘的案例中结果告发了那曾为压过作辩护的理论，那被强加的体制在收成失败之后被放弃。

§5 协作与网络治理

机制。 协作治理通过网络而协调。政府、公司、非政府组织、居民与大学参与治理一个无一者命令之事，而协调通过被协商的相互依赖、而非通过指令、价格、或一个有界社群的规范而进行。

诸实践。 这些实践是那些在过去数十年被发展的共同生产、多利益相关方治理与网络治理的实践，而它们被见于城市更新、社区发展与环境治理。这一机制是对那些跨越既有辖区之边界、而无单一行为者拥有权威或知识独自处置之问题的典型回应。

成就与界限。 那进展是这一机制在其余者无法运作之处运作：横跨诸边界、在不同类的诸方之间、在属于无人辖区的问题上。它准许地方知识进入而无需公共领域治理所要求的有界社群，而它准许协调而无需层级所要求的权威。

诸界限有三个。协调的成本作为不被酬偿的参与被承担，而它们不均匀地分布、却集中于那些承担责任者。网络不是扁平的，而那些协调必须经之通过者获得一种无规则所授予、因而不受制于那些用以核查被授予之权威的程序的影响。而问责随权威变得被共享而变得弥散，因而这一机制典型的德性与它典型的病理有同一个源头。

§6 序列及它的诸残余

这四种机制如今可被置入关系。这关系是一种残余的关系：每一个机制，在解决它的问题时，留下某种未被解决之物，而它成为下一个机制作为其回应的那个问题。

层级确保在尺度上的协调、而无法收回它的抽象所从之出发的地方知识。价格恰恰收回那：分散的知识在地方上行动，它的效果通过信号传播，无需通过任何层级的上升。那救治被它所回答的矛盾所指定。价格，轮到自己，确保对分散知识的使用、并准许价值被从生产它的诸关系抽离；而在溶解传统社群时它以它唯一可得的方式使某物可见，即那社群曾未定价地供给者：信任、情境化的知识、以及在缺席强制时合作的意向。基于规范的治理接手价格所消耗者。规范通过相互可见性确保合作、而无法在尺度所强加的相互可见性之丧失中存活，也无法把它的义务延伸过那构成它的边界。网络接手规范所无法者：横跨诸边界、在不同的诸方之间、无一个有界社群的协调。而网络留下协调者的位置不受问责、协调者的负担不被酬偿。

有两个残余不曾被这四者中的任何一个接手。第一个是对那些不静止之系统的治理，即层级所无法管理、而网络仅在个别问题之层面上处置者。第二个是价值的去向，即价格所使成为一个结构性特征、而随后诸机制中无一者曾收回者。本讨论把两条当代谱系读作从这两个残余降下，并把这一降下作为“为何这两批工作彼此交涉如此之少”的一个说明来提供。

有两个残余未被认领地熬过这序列：对那些不静止之系统的治理，以及所生产之价值的去向。复杂性谱系从第一个降下、而生成性谱系从第二个降下。它们从不同残余的降下会说明它们曾在孤立中发展，而它们曾在大约同一时刻抵达那另一条的能力所会处置之界限，是此处所提供的、把它们一同思考的理由。

这一比较可被概括。下表为每一个范畴记录：主导的协调机制、地方规则制定是否被容许、那安排是否学习、以及它如何对待价值的去向。最后一列是那些既有比较所省略者，而它的省略是那第六个范畴不得不被单独构成的理由。

范畴	机制	地方规则	学习	价值的去向
中央化	层级	低	弱	向上到中心
市场	价格	中	强	到占用的位置
公共领域	规范	高	强	由使用者群体保留
协作	网络	高	强	未被指定；弥散
复杂性	任意	高	极强	未被理论化
生成性	规范、网络	极高	强	返回给生成者

表中的水平横线不是装饰性的。它之上是被实现于制度形式中的诸协调机制；它之下是组合诸机制、并不曾安顿入一个形式的诸谱系。（最后两行，复杂性与生成性，即是这些谱系。）把这张表读作一份六个选项的单一清单，会恰恰遮蔽第二部分所倚的那个区分。

数字条件应在此被记录，因为它供给两者的物质可能。通过坍缩协调的成本，它使一种既非层级、亦非被定价的协调在超出相互可见性之触及的诸尺度上成为可能，而这是当代同侪生产的条件。而它在一个新的介质中再生产价格的矛盾，因为协调经之通过的基础设施成为一个协调的节点，而数据成为榨取的新对象，从那些其协调为平台所使成为可能的诸关系中被汲取。

第二部分. 两条当代谱系

§7 这两者的地位

本部分所审视的两个范畴不是一个待被置于那四者之旁的第五与第六机制。它们是不同的一类，而陈述这一差异是必要的，若§10中的比较要成为可理解的。

前四个范畴是已被实现于持久制度形式中的协调机制。随后的两个是研究与实践谱系，也就是说，是仍在运动中、并不曾安顿入一个制度形式的纲领。复杂性治理是一种治理视角、而非一个制度，而它可以在一个国家机构之内、在一家公司之内、或在一个社群之内被推行。生成性实践是一族实验，而它可以采用公共领域基于规范的形式、或同侪生产的网络形式。

因此，这两者是一个关系性读法最切近地趋近者，这既不是一个巧合、也不是一个偏好。它从它们是什么中得出。一个已安顿入一个制度形式的机制，通过安顿，已固定它对“它协调什么以及如何协调”这一问题的回答。一个仍在运动中的谱系不曾固定，而正是在那尚未被固定者的空间中，本讨论的诸问题才根本能被提出。

§8 复杂性谱系

第一条谱系从协调的残余降下。它的问题是对那些不静止之系统的治理，而它的形成语境是环境与自然资源管理，在那里中央化专家控制的不足最早变得显然。

适应性治理在这一文献中被理解为：在追求社会-生态系统之一个所欲状态时涌现的、行为者、网络、组织与制度之间的一系列互动，被援引来回应快速环境变化的复杂性与不确定性；它的涌现被明确地追溯到早先管理体制未能构造出足够稳健的治理结构以在不确定之下确保可持续性。它的规范内容已被概括为：管理权威的下放以促进参与、横跨诸层级的协作网络的构造、足够灵活以准许对意外之回应的诸安排的维持、以及通过那些网络的学习。多中心性供给那结构性的对应物：一个安排是多中心的，当它构成性的决策实体彼此拥有相当的事实自主，无论它们是否在形式上独立，而在这样的安排中适应作为一个相对自组织的、以创新与社会学习为标志的过程在多个能动之中心之间涌现。韧性供给成功的判准，刻画一个系统忍受扰动而不坍缩入一个替代稳定状态的程度。

这一谱系的统领性问题由此被固定。它追问一个系统如何可能持存。

§9 生成性谱系

第二条谱系从榨取的残余降下。它的问题关切在一个系统之内流通的价值、以及那价值是否返回给那些生产它者。系统的持存是一个单独的问题，而这一谱系不提出它。

它的诊断是两个既有的替代都榨取。在那组织了近期大量工作的表述中，共产主义为国家榨取价值、而资本主义为公司榨取它，因而两者都摧毁社群与个体；针对两者被提出的是价值的流通、而非榨取，以“生成未被异化之价值并直接参与其惠益的普遍权利、价值生成者创造他们自己生产条件的权利、以及价值生成之诸社群维系其流通之路径的权利”来定义。判准随之而来：可持续的系统把价值返回给它的诸源头，而危机在价值被以很少的返回所征用之处升起。值得记录的是，这一说明明确地从原住民实践中取它的方位，主张一个比那些自上而下之解决更好的模型，被发现于原住民传统自下而上的进路中，那些传统实践价值流通、而非价值榨取。

经验的场域被那些返回可被观察的实践所充填：开源与百科式一类的同侪生产，它已通过一种劳动价值的礼物交换挑战对知识产权的专有控制；农业生态实践，它通过堆肥与间作确立一个生态价值的返回循环；工人所有的合作社与手工生产，其中利润、技能与责任循环回给那些生成原初价值者。社区货币与创客运动归属同一族。

这一谱系的统领性问题由此被固定，而它不是第一条谱系的问题。它追问价值如何可能持续被生成并返回。

§10 每一条谱系的盲点

本节的主张是：每一条谱系都有一个界限，每一条的界限都在它自己的批判文献之内被记录，而这两个界限是互补的。这是随后一切所倚的一节。

复杂性谱系的界限。 这一谱系回答系统的持存。它已从内部因它对权力、对合作之诸源头的处理之单薄而被批评，而这一批评已成为它被承认的内部问题之一、而非一个外部的反对。

这一批判已被表述为关切环境与自然资源治理的过程、权力与意义诸维度。韧性研究已被刻画为去政治化的、为未能认识社会的偶然性、并为未能认识不同行为者塑造治理系统的可变能力。已被作出的判断是：韧性思维未能处理权力的诸问题、并对治理结构与功能留意过多、而对政治、历史与文化语境留意过少，因而这一文献以政策问题的管理为代价强调了治理本身的政治。这一点已被以特别的锐利道出，在如下观察中：适应性共同管理，如一切环境管理，是一个内在政治的、而非仅仅技术的事业，而管理制度被嵌入不平等的生产关系之中，以致它们的改革本身不大可能产生公平的结果。

一个第二的、在结构上不同的界限关切当适应性的诸安排遇见在位的诸制度时发生什么。在对城市更新中适应性治理倡议的研究中已被发现：这样的倡议能够制度化并物质化，但在位诸制度针对它们的韧性使那些涌现的适应性制度随时间失去它们的核心特征，因而那转变性的效果保持有限；被辨认出的机制是既有制度靠一个被那些从中获益者所维持的特定权力分配而运行。因此，这一界限不是这一谱系未能描述复杂系统。它是那适应性的形式，在进入一个被权力所结构化的场域时，被那场域所重塑。

这一谱系所不供给者，是一个关于为何参与者继续合作、关于维系他们合作之价值起源于何处、以及关于那价值一经被生产便成为什么的说明。它的判准是持存，而一个系统可以在从那些维系它者身上全面榨取的同时持存。

生成性谱系的界限。 这一谱系回答价值的返回。它的界限关切尺度，而它以不寻常的清晰出现，因为它示范性的实践如今已运行得足够久以致它们的失败模式已被记录。

同侪生产把这一界限展现为那些维护者的耗竭。开源维护被报告的状况是：多数维护者无酬工作、多数已退出或考虑过退出、而离开者中相当大的一部分把耗竭作为理由援引。其后果不被限于个体的福祉：一个被广泛部署的、一个重大基础设施项目的组件被退役，不是因为它已变得过时、而是因为那些维护它者不再能维系那工作。这一失败的结构是此处所要紧者。价值流通，在这一谱系所要求的意义上，因为那软件返回给生产它的社群。所不流通者是维护的劳动，它集中于少数、并被一个把那资源当作免费的公司用户之生态所汲取。已被描述一个恶性循环，其中过劳的维护者缺乏时间去引入协助，这加深了耗竭、并提高了被放弃的概率。这是一个发生在一个其价值流通如设计般运作之系统内部的治理失败。

公共领域的诸安排把这一界限展现为延伸的困难。既有的回应是嵌套原则，据它地方组织既不被吸收、也不被边缘化，而被包括于一个从地方到区域诸层级运作的多层级多中心系统之内。困难是：基于社群之进路的向上扩展已远远跑在关于这样的向上扩展如何可能运作的知识之前。已被论证：这些设计原则对某些类的全球公共物品可扩展、而对另一些不利，而已被承认：那些为全球公共领域被提出的原则尚未被牢固地确立或被强有力地检验，在那一尺度上倚于一个有限的经验基础。

一个限定在此被要求，而它正是第一部分被构造来使之可得的限定。公共领域的尺度界限不由公共领域诸安排的历史衰落所确立，因为那衰落，在所审视的诸案例中，不是那些安排的一个失败。subak协调一个流域一千年、而被国家的强加所扰乱，不被它们自己的界限；入会被一个以它们过时为前提的现代化计划所拆解。尺度界限的证据是当代的证据：把公共领域治理延伸到全球资源问题之被记录的困难、以及同侪生产中维护负担之被记录的集中。这些是在运作着的系统中被观察的界限，而非从被压过之系统的命运中被推断的界限。

那互补。 这两个界限是彼此的逆。复杂性谱系拥有一个关于在尺度上持存的说明、而缺乏一个关于价值以及为何合作被维系的说明。生成性谱系拥有一个关于价值及它返回的说明、而缺乏一个关于在尺度上持存的说明。每一条在那另一条最弱之处最强。

复杂性谱系拥有一个关于在尺度上持存的说明、而缺乏一个关于价值以及为何合作被维系的说明。生成性谱系拥有一个关于价值及它返回的说明、而缺乏一个关于在尺度上持存的说明。每一条在那另一条最弱之处最强。本讨论把这一互补读作那历史序列的残余结构、而非两个文献的一个巧合。

本讨论把这一互补读作第一部分的残余结构、而非此处所提供之刻画的一个人为物。依那读法，复杂性谱系承继协调的问题，即层级所提起、而价格与网络所转变却不曾解决者，而生成性谱系承继榨取的问题，即价格所使成为结构性、而数字条件所再生产者。这些是那序列未被认领地留下的两个残余，而这两条谱系是每一个当前正被工作的形式。它们从不同残余的降下会说明它们之间交涉之单薄。此处所提供的、把它们一同思考的理由是：每一条已在大约同一时刻抵达那另一条的能力所会处置之界限。

那分离本身在此作为一个读法、而非作为一个关于这两个文献的被记录之事实被断言。本讨论所记录者是每一条的界限与这两个界限的互补。这两批工作是否事实上在相互无知中进行，是一个文献计量的问题，而它是本讨论不了结的一个问题。

§11 那历史所留敞的问题

那仍立着的问题如今可被陈述。是否有一种一次回答两项要求的治理模式，在维系复杂系统的同时维系价值在它们之内的生成与返回。

这一问题的地位要求精确。它不是因为一个能回答它的框架曾可得而被提出。它是通过构造一种分类法、审视每一范畴之下的诸实践与每一者的界限、并阅读那序列所产生之两条谱系内部的批判文献而被抵达的。若那文献不曾记录这些界限，这一问题便不会升起、而本讨论的其余部分便会没有对象。

第三部分. 那汇合

§12 一个理论缺口的不充分

第二部分确立了每一条谱系都有一个界限、而这两个界限是互补的。这确立了一个理论缺口的存在，而一个理论缺口不是一个做任何事的理由。文献中的诸缺口为数众多、而它们中的多数应保持敞开。一个研究纲领不被那些先于它之理论的不完全所担保、而被那些理论未能描述之对象的存在所担保，而本部分的负担是显示这样的对象存在。

有待被辩护的主张是狭窄的、而它在阅读中不应被膨胀。它不是社会要求一种新的治理理论，那是一个无法被任何证据所确立、而本讨论不断言的命题。它是如下者。存在一类同时拥有复杂性与生成性的当代治理对象；当这两条谱系被应用于这样一个对象时，每一条都以它自己的判准返回一个健康的裁断；而那对象仍失败。在这发生之处，那失败对两个框架都不可见，而它不可见，不是因为任一框架曾被误用、而是因为每一个都如设计般运作。

这一类的一个对象是一个双重盲点、而非一个缺口。这一区分实践地要紧。一个缺口是一个尚无理论描述过的区域，而它由勘察文献被发现。一个双重盲点是一个可得理论描述为不成问题的区域，而它仅当那对象失败时才被发现。第一个可以从容地被填补。第二个由伤亡被找到。

缺口与双重盲点。 一个缺口是一个没有可得理论描述的区域，而它由勘察文献被发现。一个双重盲点是一个可得理论描述为不成问题的区域，而它仅当那对象失败时才被发现。第一个可以从容地被填补。第二个由伤亡被找到。

§13 那汇合对象的结构

数字条件，如第一部分末所记录，一次供给两条谱系的物质可能，而它因此产生了两个性质都在场的对象，这并不令人意外。这样一个对象可被两个一同成立的条件所刻画。

它是复杂的，在第一条谱系所要求的意义上。许多行为者、无中心、扰动的非线性传播、以及一个不静止的环境。它被治理，就它究竟被治理的程度而言，是通过适应性的、多中心的诸安排、而非通过指令。

它是生成的，在第二条谱系所要求的意义上。它的价值被它的参与者持续地生产，它流通回给他们，而那系统仅在那生产持续的期间持存。没有存量在它之下。没有储库供它在生成停止时汲取。

对于第一部分所重构之历史的大部分，这两个条件被不同的对象所满足。一个流域是复杂的、而它的价值大体是被给定的、而非被持续地生产的。一个作坊是生成的、而它不是复杂的。数字条件所产生者是一类两者一次成立的对象，而这一类的成员既不奇异、也不边缘。它们包括那技术文明据以运行的软件基础设施、那有史以来被组装的最大参考著作、以及那使各主要讨论平台可用的内容治理系统。

§14 三个案例

随后的三个案例在每一个实例中都以同一次序被呈现：第一条谱系的裁断、第二条的裁断、以及两个裁断都错过的失败。要点是那同构。一个案例是一则轶事、而三个有同一结构者是一个模式。

同侪生产的基础设施。 考虑当代计算所依赖之开源组件的维护。

复杂性谱系返回一个健康的裁断。这一生态系统在严格的意义上是多中心的，它的决策实体拥有相当的事实自主；它是适应性的，通过分散的网络学习、并回应扰动而修订它的诸安排；它在那定义的意义上是有韧性的，已吸收数十年的扰动而不移入一个替代稳定状态。以这一谱系最显著地供给的判准衡量，这一系统读作一个成功，而它事实上频繁地被作为这一谱系所推荐之性质本身的一个范例而被提供。是否有某个这一谱系内部的判准会登录这一失败，是§22所陈述之反驳条件中的第一个。

生成性谱系返回一个健康的裁断。价值被参与者生成并返回给他们：那被社群生产的软件对社群可得，恰恰在这一谱系所指定的意义上未被异化。没有对那制品的征用。以“可持续的系统把价值返回给它们的源头”这一判准衡量，这一系统如设计般运作。

这一系统仍在失败。多数维护者无酬工作；多数已退出或考虑过退出；离开者中相当大的一部分援引耗竭。一个被广泛部署的、一个重大基础设施项目的组件被从安全维护中撤出，不是因为它已被取代、而是因为那些维护它者不再能维系那工作。已被记录一个恶性循环，其中过劳的维护者缺乏时间去引入协助，这加深了耗竭、并提高了被放弃的概率。

那失败被精确地定位于两个框架都不看之处。那制品流通；维护它的劳动不流通。维系那系统的负担集中于少数参与者、并被一个对其而言那资源是免费的公司用户之生态所汲取。这既不是一个价值流通的失败、也不是一个适应能力的失败。它是那维系“价值经之流通、适应经之发生”之诸关系的成本的分配中的一个失败，而两条谱系都不拥有一个那量的术语。

百科式的公共领域。 考虑迄今被承担的最大协作知识项目。

复杂性谱系返回一个健康的裁断，而生成性谱系返回一个健康的裁断，以与之前相同的根据。这一项目是自组织、多中心、适应性的，并已吸收四分之一世纪的扰动。它的价值被志愿者生成并返回给公众而无征用。

产出的度量确认两个裁断。内容质量升、条目数升、而特色条目的数目升。这部百科从未更好过。这是那些最关切这一项目状况者的发现，而它被他们作为一个警告来提供。

这一系统仍在失败。新贡献者的获取已下降超过三分之一而总编辑数已上升，也就是说，更多的工作正从一个收缩的、忠诚志愿者的基础中被榨取。这一系统正在集中。英语项目的管理员团已从近一千八百下降到略多于八百，其中并非全部活跃，因而一个日益重要的项目被一个收缩的、无酬志愿者之团体维系在一起。效率的增益，多年来是对一个收缩的贡献者基础的补偿，据报告已达到它的界限，一个离开的资深编者如今因身后没有一条流水线而需付出更多以被替换。质量的度量，依这一读法，不是健康的证据。它们是那些维护这一系统者已变得不可替换的证据。

与前一案例的同构是确切的，而它值得被无装饰地陈述。价值流通在运作。适应能力在运作。产出在改善。而那承担产出之生产成本的关系性结构正被那生产所消耗。两个框架都测量产出；两者都不测量那消耗。

平台的内容治理。 考虑各主要讨论平台所依赖之志愿者内容治理系统。

在此，两个裁断再次是有利的。这一系统是多中心的，到一个它自己的运营者所强调的程度，各社群自组织并设计出为它们特定境况裁量的规则；它是适应性的，各版主发展出对地方语境的精细理解、并把一般政策适应于地方规范。价值被参与者生成并返回给社群，这是版主寻常的自我描述、而它不是假的。

这一系统仍在失败，而这一案例使一个前两个案例含有、却不那么平白地展示的特征成为可见。在单一平台上由志愿者版主执行的劳动已被估价为每年不少于数百万美元，对照一个在数亿之谱的平台收入，而那平台雇用零个有酬版主、同时依赖十万以上无酬者。版主面对来自工作之量、来自因他们的决定而被针对、以及来自这一角色所要求之情感劳动的耗竭；被给出的两个主要退出理由是可用时间不足与和其他版主的冲突。这一位置已被刻画为一个其中无酬志愿者承担情感劳动之实质负担、同时服务于可随意移除他们之平台运营者的位置。

这一案例所增添者是那榨取渠道。在前两个案例中那不被酬偿之维护的受益者是弥散的，一个用户与读者的生态。此处那受益者是一个单一的公司实体，它的估值依赖于它不付费的劳动。生成性谱系拥有那会命名这的概念，因为它是严格意义上的榨取；但它不察觉它，因为那谱系所追踪其流通之价值，即内容，确实流通。所被榨取者是治理内容的劳动。内容本身不被榨取，而治理它的成本是一个关系性的成本、而非一个价值的流动。

§15 那同构的诸后果

这三个案例有同一结构，而这一结构如今可被以一般形式陈述。

在每一个中，那对象同时是复杂的与生成的。在每一个中，第一条谱系的判准返回健康，因为那系统持存并适应。在每一个中，第二条谱系的判准返回健康，因为价值被生成并返回给它的生成者。在每一个中，产出改善，甚至在那失败推进之际，因而两条谱系都会查阅的度量都在积极地令人安心。它们的沉默会是较小的困难。而在每一个中，那失败发生于同一个地方：在那些承担维系系统之成本的关系中，它们的负担随系统增长、而承担它的诸关系不增长。

三个后果随之而来，而它们应被审慎地区分。

三个被记录的案例展现同一结构。每一个对象一次是复杂的与生成的。在每一个中，两个领先框架都以它们自己的判准返回一个健康的裁断，产出的度量随失败推进而改善，而那系统仍失败。在每一个中，那失败落于同一个地方：在那些承担维系系统之成本的关系上，它们的负担随系统增长、而承担它的诸关系不增长。若这三个确实是同构的，而§22陈述什么会显示它们不是，那么两个框架都不错、而两者都在同一方面不完全。每一个都正确地回答它所问的问题。两者都不问这些对象据以死亡的问题。两个各自对一个量沉默的框架，不会通过被并排放置而对它变得能言。

第一个是这些框架不错。前面的任何东西都不构成对适应性治理或生成性正义的一次驳斥。每一个都正确地回答它所问的问题。困难是两者都不问这些对象据以失败的问题，而它们的联合也不问它，因为两个各自对一个量沉默的框架，不会通过被并排放置而对它变得能言。

第二个是这一失败模式在事前不可见。在两个判准都返回健康之处，在用的框架发不出任何警告，而一个关键组件从安全维护中被撤出，就被记录的记录所显示，不曾被任一方所预期。围绕这样一个项目的社群可能很清楚它的状况；此处的主张关切分析框架、而非参与者的默会知识。这是那把一个双重盲点与一个缺口区分开来者，而它是这件事紧迫的理由。

第三个关切那轨迹。这一类汇合对象不在缩减。那产生它们的物质条件，即那准许在复杂尺度上进行生成性生产之协调成本的坍缩，不在被撤回。对这一类的每一次增添都是一个可得框架将报告健康直到它失败的对象。

§16 不予处置的后果

这一问题可被无预言地回答，因为答案已被观察三次。

所发生者是：那些产出在改善、价值流通在运作、适应能力完好的系统，被它们自己维护之成本的集中从内部消耗，而这一消耗直到失败的时刻才被察觉，因为在用的仪器在测量别的东西。一个技术基础设施的一大部分所依赖之组件停止接收安全补丁。一部规模无匹的参考著作变得依赖于一个无法被替换的维护者团。一个平台的治理靠它自己的治理模型所不承认为劳动的劳动而运行。

这一切都不是一个预报。它是对已发生之诸事件的一个描述，而从它们随之而来的预报是那个谦逊的预报，即它们将复现，因为那产生它们的条件持续、并因为可得框架中没有任何东西会在事前宣告它们的复现。

这是实践的动机，而本讨论提出不多于此。这一主张不是一个关系性的治理说明是正确的，而它不是社会要求一个。这一主张是：存在一类对象，对其两个领先框架报告健康而那对象死亡，那些失败共享一个结构，而那结构关切一个两个框架都不测量的量。那量是否被良好地描述为关系性的、以及一个围绕它被组织的治理说明是否会做得更好，仍有待被显示。它在本讨论中无处被显示。所已被显示者是这一问题不是空闲的。

第四部分. 一个关系性读法

§17 那未被测量的量

这三个案例汇合于一个两个框架都不测量的量，而本部分的任务是尽证据所准许地、而不比那更精确地刻画它。

这一量不是价值，因为价值在三个案例中都流通、而它的流通被正确地登录。它不是系统的持存，因为那些系统持存并适应，而这也被正确地登录。它不是产出，它在三个中都改善。所被消耗者，在每一案例中，是一组特定参与者继续承担“维系那价值经之流通、适应经之发生之安排”之成本的能力。

这一能力有三个特征，是这些案例共同展示者。它被诸关系所承担。那负担落在维护者、管理员与版主身上，而它不落在那安排整体上。它被它所使成为可能的运作所消耗，因为那系统据以生产它产出的活动本身，正是那耗竭了承担它者的活动。而它不出现在产出中，这是为何产出的度量自始至终令人安心、以及为何在百科式的案例中质量的改善本身即是那危险的证据，那些维护者恰恰通过维护得好而变得不可替换。

一个有这些特征的量是参与者之间诸关系的一个性质。它是一个既非系统、亦非价值的性质。这是那归纳，而它作为一次归纳被提供。那本会使这三个失败成为可见的术语，在每一案例中都是同一个术语，而本讨论提议，暂定地，把它称作那关系。

§18 那术语的地位

这一提议的谦逊必须被守护，以抵御两个它所暴露于其下的膨胀。

并不因此可得关系是治理的对象，而本讨论不断言这。治理的诸对象保持它们一向所是者：资源、组织、社群、企业、知识、技术，而在这些之中，关系本身也是。把治理的对象化约为那关系，会以一种关系性的词汇再生产恰恰是那批判文献已就其余者定罪的那个化约之举。

也并不因此可得一个关系性的说明将成功。这三个案例所支持者是那个更弱的命题，即某个量，未被可得框架所测量，牵涉于一类被记录的失败之中。那样一个量作为一个良好定义的量存在、而它在品格上是关系性的，是从三个案例的一次归纳、并等待§23的操作性说明。一个围绕它被组织的治理说明会在事前察觉那些失败，是一个猜想，而它是一个本讨论毫不确立的猜想。

那准许这一术语被无膨胀地使用的区分可如下道出。那关系性者是一个视角，一种看一个治理安排正对那些维系它者做什么的方式。那生成性者是一个取向，一个关于那安排应生产什么的说明。复杂性是一个条件，即当代治理运作于其下的境况。这些占据三个不同的位置，而这三者坍缩为一个单一对象，是本部分整个作为其防护的那个失败模式。

§19 与劳动的区分

对前面之论最简约的反对是：一个名字已存在于这三个案例所展示者，而那名字是无酬劳动。依这一读法，维护者、管理员与版主在执行工作；那工作不被酬偿；受益者是不为它付费的公司与公众；而关于剩余劳动之占用的经典分析在没有任何新术语的情况下说明这一现象的全部。这一反对是严肃的，而它必须在术语“关系”被容许立住之前被应对，因为一个对一个可得概念毫无增添的概念不是欠发展的、而是多余的。

这些案例的三个特征不被劳动分析所捕捉，而它们以对它之难度的升序被陈述。

第一个关切救治的方向。若所被消耗之量是劳动，酬偿它便会恢复那系统。这是一个可检验的命题，而诸案例反对它。志愿者版主为离开给出的理由，显著地包括与其他版主的冲突，而付费不处置它；同侪生产中被记录的耗竭被归因于需求之量、被归因于因所作决定而被针对、被归因于工具与支持的缺席，其中只有第一个是金钱直截了当地转换的量。同侪生产中被辨认的恶性循环是过劳的维护者缺乏时间去引入继任者，而付费仅当那付费维护者的时间因此被释放去引入时才缓解这。在同一需求以更大之量返回之处，那时间被再次吸收。资金是任何救治的一个必要组成部分、而它不充分，而这一不充分不是附带的。

第二个关切百科式的案例，在那里劳动分析产生一个假阴性。在那里被记录者首要地不是志愿者无酬，那在这一项目的整个历史（包括它的增长时期）中都为真。它是新贡献者的获取下降而总编辑数上升、管理员团从近一千八百收缩到略多于八百、而一个资深编者的离开如今因身后没有流水线而需付出更多以被替换。在下降之量是替换那些承担负担者的能力。一个仅留意酬偿的劳动分析本会登录不到任何变化，因为那酬偿在此之前是零、在此之后是零。所变化者是继任的结构，而继任是参与者之间诸关系的一个性质、而非一个工作者与一个受益者之间交换的性质。

第三个特征，也是这一区分主要所倚者，是那所在。劳动是一个工作者与一个占用那产品者之间的关系，而它被在那占用的轴上测量。这些案例所展示者是一个落在“参与者据以相互维系彼此”之诸关系上的负担：引入的诸关系、相互支持的诸关系、冲突解决的诸关系、向那些可能继任者传递默会知识的诸关系。那因与其他版主的冲突而离开的版主不是在描述一个受益者的占用。那无法引入的维护者不是在描述一个不平等的交换。在平台的案例中那占用在场、并被劳动分析正确地命名，而恰恰在那里我们能看见劳动分析用尽，因为它命名那榨取、而对“为何那些不被从中榨取的版主也离开”保持沉默。

其结果不是劳动分析错了。在平台的案例中它是对的，而生成性谱系关于榨取的概念是那个对“一个有十亿收入而零个有酬版主之公司正在做什么”的正确概念。其结果是劳动分析以一种特定的方式不完全。它分析那从参与者行向受益者的轴、而它不分析那从参与者行向参与者的轴，而这三个案例在第二个轴上失败。此处所提议的术语是那第二个轴的一个名字，而它仅当那第二个轴是真实的时才挣得它的位置。它是否真实是一个诸案例能被使之回答的问题，而§22陈述如何。

§20 与网络及社会资本的区分

两个更进一步的概念立得足够近以要求辨别。

网络分析供给一个用以表象诸关系的形式装置，而那反对是：所论之量是一个已可供测量的网络性质，如那维护者的中心性或那版主团队的连通性。这一装置确实是任何随后形式化的自然之选，而此处对它毫无提出反对。困难是一个网络是诸关系在一个瞬间的一个表象、而所论之量是一个速率。诸案例所展示者不是维护者占据中心位置，那是真实而不足为奇的，而是那些位置所承担的负担增长而那些位置不成倍。网络是一个快照；那发现是两个轨迹之间的一个背离。一个网络表象能表达那背离，但仅在人已被告知把哪两个量置入比之后，而那是此处所主张的贡献、而非那装置。

社会资本命名一个群体所拥有、并能汲取的信任、义务与互惠之存量，而那反对是：这三个系统只是在耗尽它们的。这更近，而这段距离值得被确切地标记。社会资本被构想为一个内在于一个群体、并促进被协调之行动的资源；它典型的问题关切一个社群有多少它、以及它使什么成为可能。此处之量是一个承载的能力。它不是一个存量，而相关的问题关切它所必须承载者是否比它增长得更快。有多少它存在是一个不同的问题。一个社群可以拥有充裕的信任而仍失败，若被置于承载那信任之诸关系上的负担超过它们，而这正是百科式的案例所展示者：那社群的凝聚不成问题、而它的管理员正在消失。社会资本会在那存量低时预测失败。诸案例在那存量、就任何指标所显示，保持高之时失败。

§21 普遍变量的危险

有一个对关系性词汇的长存反对，本讨论无法回避、而不希望回避。信任是一个关系。权力是一个关系。制度、沟通、身份与规范是诸关系。若一切都是一个关系，那么关系解释一切、因而预测不出任何东西，而那术语已以它的内容为代价购得它的范围。这一反对已从关系性文献本身之内被有力地道出，在如下观察中：纯粹的关系性无法创造对象，关系要求实体一如实体要求关系，而两者之间的对立已被它的党徒所过度描绘。

这一反对被接受。它进一步被接受：一个关系性的治理说明比多数更暴露于它，因为治理是一个其中每一个术语都能被毫不费力、毫无收获地关系性地重新描述的领域。

从接受它随之而来者是一个限制，而这一限制是本部分“说过任何东西”这一主张的实质。术语“关系”在此不作为一个一般本体论被提供，而本讨论就世界由什么构成断言不出任何东西。它作为一个特定量的名字被提供，即这三个案例所隔离、而可得框架所不测量者：那些参与者据以相互维系彼此之继续参与之诸关系所承担的负载，取与那些关系承载它之能力的比。这一量不是信任，尽管信任可能条件它。它不是权力，尽管它的分配是一个权力的问题。它不是网络，网络是它的表象。它不是劳动，劳动在另一个轴上运行。它不是社会资本，社会资本是一个存量、而这是一个负载。

维护负担。 那些参与者据以相互维系彼此之继续参与之诸关系所承担的负载，取与那些关系承载它之能力的比。它不是信任，尽管信任可能条件它。它不是权力，尽管它的分配是一个权力的问题。它不是网络，网络是它的表象。它不是劳动，劳动在从参与者行向受益者的轴上运行。这一量在从参与者行向参与者的轴上运行。它不是社会资本，社会资本是一个存量、而这是一个负载。

一个如此被限制的术语是可反驳的、而它对多数目的近乎无用，而那两个性质是同一个性质。本讨论偏好一个不能解释一切的术语。

§22 反驳的诸条件

第四部分的主张是：一个量存在，它在方才被限制的意义上是关系性的，而这两条谱系不测量它。每一个合取项都能失败，而它能失败的诸方式应被陈述，以便本讨论可被对照它们而被要求。

若任一框架以它自己的判准察觉这三个失败，这一主张便失败。若能被显示：一个恰当地被构造的韧性度量本会在那基础设施组件之撤出之先下降，或一个未被异化之价值流通的量度本会登录那管理员团的收缩，那么便没有未被测量的量存在、而第四部分是空的。

若酬偿是充分的，这一主张便失败。若那些已为维护确保了资金的项目被发现展现不出继任、引入、或冲突解决的残余失败，那么§19的劳动分析是完全的、而那第二个轴不是真实的。

若这三个案例不是同构的，这一主张便失败。若维护者的耗竭、管理员的消失、与版主的离开被发现有只共享一个表面相似的不同因果结构，那么那归纳便是从一个巧合被抽出的。

这一主张最后失败，若§21的限制无法被守住。若此处所提议之量被证明在实践中无法与信任、与权力、或与社会资本相区分，那么那术语便已成为它被构造来回避的那个普遍变量，而它应被丢弃、而非被捍卫。

这些检验中没有一个在此被执行。它们被陈述，因为一个无法说出什么会反驳它的诊断不是一个诊断。

§23 操作化的优先

这四个条件中的第三个承担其余者的分量，因为一个同构的失败会使剩下的诸检验没有可被执行于其上者。那显然的回应是扩大样本。若三个案例是一次归纳的一个单薄基础，那么横跨诸领域、诸文化与诸尺度的二十个案例会是一个更坚实者，而那自然的下一步看似是它们的收集。

本讨论持有那自然的下一步是错的，而那理由值得被平白地陈述，因为它在是一个关于世界的主张之前，是一个对作者的约束。

§21中所提议之量是一个负载与一个能力之间的比，而两个项都不曾被给予一个操作性的定义。已被说出这一量不是什么：不是劳动、不是中心性、不是一个社会资本的存量。不曾被说出，以那些会准许在一个新案例中作出判断的术语，什么会算作那负载的一个实例、或那能力的一个量度。在这样的术语缺席时，一个第二十案例的增添无法确认那归纳，因为不存在任何一个第二十案例据以能被发现展现同一结构的判准。对这三个的相似是所能被确立的全部。诸案例会累积、而表面相似的指控会随它们累积。一个在概念被操作化之前被进行的样本之扩大不加强一次归纳。它使轶事成倍。

一个在概念被操作化之前被进行的样本之扩大不加强一次归纳。它使轶事成倍。操作化先于诸进一步案例的收集，因为没有一个操作性的判准，没有一个进一步的案例能被发现展现那结构，而相似是所能被显示的全部。

因此，工作的次序是那表面之序的反面。操作化先于样本的扩大，而这三个案例对操作化的目的是充分的，因为它们曾足以隔离那量。它们所不足以者是它的确认，而本讨论不主张确认。

那操作性说明的材料在诸案例中在场，而可在此被指出、尽管不被发展。在每一案例中那负载被一个可辨认的子群所承担，即维护者、管理员、版主，而它不被成员整体所承担。在每一个中，那构成负载的工作是三类，而这三类复现：继任者的引入、群体内部冲突的解决、以及不被持于制品之中之知识的传递。这三者中没有一个是产出，这是它们对两条谱系不可见的结构性理由，因为两者都测量那系统所生产者、而这一工作生产不出任何那系统的受益者所消费之物。在每一案例中那负载随一个那系统所输出之量增长，即依赖者、读者、用户的人口。而在每一个中那能力被那子群的大小与可替换性所索引，这是为何管理员团的收缩与一个软件项目的巴士系数是同一件事的量度。

一个从这些材料被建造的操作性说明会是一个关于维护负担的说明，而它会有一个当下讨论所缺乏的性质：它会生成一个据以一个未被看见的案例能被分类的判准。比较是当下说明所准许的全部。那判准，一经被构造，是否在家户、企业与国家中辨认同一结构、抑或在数字公共领域之外无处辨认它，是那整个说明之普遍性据以转动的问题，而它不是一个进一步的哲学论证能了结的问题。它先于第五部分，因为一次重整化要求一个可被重整化之量，而本讨论指出诸部分在此被以一个它们的依赖所不遵循的次序被呈现。

§24 治理设计原则

若那关系作为一个分析的维度被容许，某些在此之前不可得的问题变得可得。它们作为问题被陈述，因为问题是当下论证之状态所支持者。

哪些关系承担维系那安排的成本，以及那成本如何在它们之间被分配。那些关系所承担的负担是否随系统增长、而承担它的诸关系不。哪些反馈回路准许维护的成本被从参与者无返回地汲取，在那些价值流通在别处健全的系统中。哪些关系阻碍生成，以及哪些维系那规则与价格都不供给的、继续的意向。

这些是分析的问题、而非制度的处方。它们不指定任何治理安排应是什么，而它们不替代既有文献已确立的设计原则。它们的主张是供给一个既有安排可沿之被审问的维度，而它们的价值将由一个单一的检验被判断：那审问是否会在事前揭示那三个既有维度报告为健康的失败。

第五部分. 尺度

§25 尺度的问题

生成性谱系已在其中示范它成果的诸实践是有限尺度的实践：灌溉社群、森林公共地、创客空间、软件项目。这不是研究设计的一个偶然。那些是这一谱系的诸机制运作的尺度，而在较大尺度上被记录的失败是它们不简单地延伸的证据。

这一问题可被确切地陈述。§10中被记录的失败不是那些机制的失败，它们如所指定般运作。价值在开源项目中流通，而软件返回给生产它的社群。公共领域的安排分配那资源并执行它的规则。所失败者是从一个尺度到另一个的通道，而这使尺度成为一个第一序的问题。它不是一个随后的应用问题。一个仅在相互可见性成立之处成立的治理说明是一个其有效性之域已被误描的说明，而它不等待延伸。

所论的序列从家庭经由社群、企业、城市、国家、到国际组织。这些是否共享一个共同的关系性结构，以及若是则它的哪些性质熬过它们之间的通道，是那问题。三个子问题必须被分开，因为它们通常被混同、并因为它们有不同的答案。

§26 关系性尺度

第一个关切一个关系性尺度是什么。上面的序列通常被读作一个量级的序列，而如此被读它是一个人数的序列。这一读法是不足的，而它的不足在第二部分的材料中可见。一个有十万用户与三个维护者的软件项目，在任何统领它失败的意义上，都不是一个大系统。那决定它行为的关系性尺度是那维护关系的尺度，它是小的，而那失败恰恰发生，因为一个小的关系性结构被一个大的人口所加载。

那么，所要求者是一个内在于关系性场域的尺度概念。一个外在于它的尺度将不服务。若干候选的构造自行呈现：一个参与者能带着注意维系的关系之数目；价值必须经之行进以返回它生成者的路径之长度；生产之点与决策之点之间中介的深度。这些中每一个都在某种程度上独立于人口而变化，而每一个都是那些被记录之失败发生于何处的一个可信决定因素。它们中哪一个，若有，是正确的变量，在此不被了结，而了结它先于任何关于不变性的主张，因为人不能在已说出那变换是什么之前，追问什么横跨一个变换被保存。

§27 粗粒化及它的残余

第二个关切在向上的通道中发生什么。在一个较大的尺度上治理，便是把某个多样体当作一个单位：家户代表它的成员，代表代表选区，国家代表它的人口。这是粗粒化，而它是那使大尺度治理根本成为可能的运作。它不是一个附带的行政便利。

那问题是那运作摧毁什么。每一次粗粒化都丢弃它所聚合者的内部结构，而那丢弃就本讨论的关切而言不是中性的。那被聚合之单位的内部不对称在那较粗的描述上变得不可见，而它们变得不可见，恰恰以复杂性谱系的批判文献所辨认的那种方式：那粗的描述登录那系统的持存、而那细的描述本会登录它之内谁承担那持存的成本。第一部分所重构的官僚矛盾是一个关于粗粒化的陈述。信息在通过一个层级之上升中的退化是那在每一层级被执行之聚合的残余。

这产出一个比本讨论由之开始者更精确的问题表述。那问题不是治理能否在大尺度被进行。它是是否存在一个保存那要紧者的粗粒化。而这两条谱系，在这一光之下被读，可被理解为已不同地回答了一个在先的问题。复杂性谱系朝向那系统粗粒化，而它的不变量是持存。生成性谱系拒绝那粗粒化、停留于那细的尺度、并保存价值及它返回的可见性，以一个不延伸之说明为代价。

§28 不变性

第三个关切什么，若有任何东西，被保存。重整化类比的兴趣在此，而它的兴趣是精确的。它所指的技术是一个用以在不同分辨率之诸描述之间通过的系统性程序，而它的中心发现是某些量在这样的通道之下以一种被控制的方式变换，因而在一个尺度上的行为能从在另一个尺度上的行为被推得。应用于此，那对应的问题是一个被治理的关系性场域的任何性质是否随关系性尺度改变而以一种被控制的方式变换，而因此能充当一个治理说明可能持守以替代一个固定制度形式的那个稳定对象。

第三部分的材料准许一个初步的观察，而它作为一个观察被提供。它不是一个结果。这三个案例汇合于一个单一的量，即一组特定参与者承担维系那安排之成本的能力，而它们进一步汇合于它失败的一个单一模式：那负担随系统增长、而承担它的诸关系不。维护者不曾随那依赖的生态系统成倍而成倍。管理员团随那百科增长而收缩。版主人口不曾随那平台的估值而扩展。

这提示那要被横跨诸尺度追踪之量是“维系那关系性场域的负担”与“可供承担它之关系性能力”之间的比。参与者的数目会是错误的量。这一提示有一个值得被陈述的后果，因为它把那观察转换为某种可能是错的东西。若那比是正确的量，那么公共领域诸安排未能延伸过地方之被记录的失败，便不是一个属于一个单独文献的单独困难。它是同一个背离在一个不同之点出现，在尺度上向上的通道恰恰是那个不增加能力而增加负担的运作。那比能否被给予一个确定的意义、以及它是否随关系性尺度改变而以一种被控制的方式变换，是一次生成性的重整化将不得不确立者。

§29 那类比的诸界限

那类比必须连同它失败的诸点一同被陈述，因为一个仅在它成功之处被捍卫的类比不曾被捍卫。

那技术在它原初的设置中预设一个其组件是一个确定之类、而其相互作用在事前被指定的系统。一个关系性场域满足两个条件中的哪一个都不。它的参与者被他们所进入的诸关系所构成，而他们的性质因此不在那相互作用之先被给定。此外，没有一个粗粒化为之被执行的外部观察者。治理中的粗粒化本身是一个治理的行动，被那场域之内的参与者所承担，而它因此改变它所描述者。这在原初的设置中没有对应物，而它可能是那类比破裂之点。一个预设一个外部的程序能否被适应于一个不容许外部之场域，是不被知道的，而这一困难不应被“那类比在别处是有启发的”这一观察所最小化。

所能被主张者更窄、而它被无修饰地主张。尺度与诸尺度之间之变换的语言看似比制度设计的语言更适合于这一困难，因为那困难是一个诸尺度之间通道的困难、而非一个在它们中任一个上之安排的困难。这样一项研究会要求的数学在此不被发展、而不应被发展。所被记录者是这一问题已被以一个使一个答案可被构想的形式被提出，而这是当下讨论有条件对它做的全部。

§30 本讨论的地位

第一与第二部分是本讨论的实质，而它们是那些能被完成的部分。第一部分在协调机制、而非制度形式的层面上构造一个分类法，审视在每一范畴之下、横跨它们被发现于其中之诸文化被实现的诸实践，评估每一个的成就与界限，并把它们排列于一个残余的序列。它坚持一个通常的历史所回避的区分，即一个被它自己矛盾所取代的安排与一个被一个以关于它落后之理论而行动的权力所废除的安排之间的区分，而它把这一区分奠基于记录：subak与入会没有失败。第二部分显示这两条当代谱系与那四个机制是不同的一类，是仍在运动中的纲领、而非已安顿的形式；它从每一条内部的批判文献记录一对互补的界限；而它显示那互补是第一部分的残余结构。

第三部分是实践的动机，而它是那回答“为何应承担这一切”之问题的部分。它不论证社会要求一种新的治理理论，那无法被任何证据所确立。它论证存在一类对象，对其两个可得框架报告健康而那对象失败，三个这样的失败共享一个结构，而那结构关切一个两个框架都不测量的量。若这三个案例不展现那结构，这一部分失败、而本讨论随它失败。

第四部分在它的肯定主张上是初步的、而在它的否定主张上是坚定的。它的肯定主张，即那未被测量之量是诸关系的一个性质，是从三个案例的一次归纳、并被猜想。它不被显示。它的否定主张是可得概念不已然涵盖那量，而这一主张被论证：劳动被显示在一个不同的轴上运行、网络中心性被显示是一个表象而非一个发现、社会资本被显示是一个存量而那量是一个负载。这一部分接受那个“关系性词汇以内容为代价购得范围”的长存反对，而它通过把那术语限制到一个窄到足以是错的单一量来回应。四个它会是错的条件被陈述。

第四部分也在§23中陈述一个关于接下来可做什么的约束。这一说明不要被诸进一步案例的收集所加强，因为尚不存在一个据以一个进一步的案例能被判为展现那结构的判准。相似是所能被判断的全部。那负载与那能力的操作化先于样本的扩大，而本讨论因此不是一个更多同类可被建于其上的基础。它是一个等待那会检验它之仪器的诊断。

第五部分是纲领性的，而它的地位要求辨别。它已尝试把尺度的问题从一个姿态转换为一个问题，通过分开那姿态通常混同的三个问题：一个关系性尺度是什么、粗粒化摧毁什么、以及是否有任何东西在那通道之下被保存。它已提供一个初步的观察，即这三个案例汇合于其上之量是“维系一个关系性场域的负担”与“可供承担它之能力”之间的比，而它已陈述那重整化类比可能破裂之点，即治理中的粗粒化是从它所描述之场域之内被执行的。这一切都不被了结。所被主张者是这一问题已被给予一个它能被在其上工作之形式，而非它已被在其上工作。

应被补充的是，第五部分立于一个第四部分不曾操作化之量上，因而呈现的次序不是依赖的次序。一次重整化要求某种可被重整化之物。尺度的诸问题在此被提出，因为提出它们是当下论证之状态所准许者，而它们无法被工作，直到§23的仪器存在。

整体的条件性品格应被无保留地陈述。本讨论倚于两个可能是假的主张。第一个是§10中被记录的诸界限是真实的，而它失败，若那批判文献不支持它们。第二个是第三部分的三个案例展现被归于它们的结构，而它失败，若任一框架能被显示以它自己的判准察觉那些失败、或若那些失败事实上不共享被主张的结构。若任一主张倒下，第四与第五部分便没有地基、而应被放弃。一个其动机问题不存在的理论要求撤回，而这一诊断已被弄得可反驳，以便这一判断能被抵达。

标题命名一个本讨论不抵达的目的地。所已被尝试者是那个在先的任务，即追问那旅程究竟是否被担保，而答案倚于第三部分的三个案例。那些发现这些案例不令人信服、那些知道一个打破这一模式之第四者、或那些能显示任一框架以它自己的判准察觉这些失败者，会对这一论证做它最需要的服务。欢迎在 huangwanhong@serendip.ngo 通信。

讨论二终。这一问题已被给予一段历史与一个诊断。它是否容许一个解决、以及在什么尺度上，在此不被确立。