A Commons-Evolved Meta-Architecture for Governance - A Preliminary Discussion within Generative Relational Governance 【（Preliminary）Draft】
ENGLISH
A Commons-Evolved Meta-Architecture for Governance
A Preliminary Discussion within Generative Relational Governance
Wanhong Huang · huangwanhong@serendip.ngo
Abstract
Governance design confronts a persistent difficulty. The targets of governance differ across scale, culture, kind, and historical period, so a single governance model transfers poorly among them. Bespoke design for each target does not accumulate, since a lesson learned in one setting seldom reaches another. And an attempt to author a universal governance standard from above is hard to legitimate, since no single body holds the standing or the knowledge to design governance for all targets at once. This paper offers a preliminary intuition in response, developed within the framework of Generative Relational Governance. We propose a meta-architecture for governance: an abstract specification of the components of a governance architecture, the standards those components answer to, and the constraints among them, together with the tools by which an adopter can implement the architecture for its own target, document it, analyze it, and revise it. The meta-architecture is not authored once and then adopted. It is evolved in a commons. An initial architecture is supplied as a starting point in the manner of the initial population of a genetic algorithm, each adopter implements and revises it in the course of governing its own target, the revisions become visible across a commons, and a working standard precipitates from differential adoption in place of being fixed in advance. The commons is where the contribution rests: the standard becomes the revisable output of a shared evolutionary process in place of its input, and this reverses the causal direction of conventional standardization. We separate the adopter, which is an organization or an individual, from the object of governance, which varies widely and need not be an organization. We adopt a cautious version that fixes a minimal normative floor by design, and we name the open problems a later programme must address, of which the representation of governance architectures in a shared and comparable form is the first.
Keywords: governance meta-architecture; commons; evolutionary standardization; generative relational governance; metagovernance; institutional infrastructure.
A note on the standing of this paper. This is a preliminary working paper. It states an early intuition and a research direction, and it does not claim a finished theory. Its aim is to identify a problem, state a motivation, and describe an envisaged approach clearly enough to be criticized and improved. The central claim is provisional, the parent framework is under development, and several of the tasks the paper names as necessary have not yet been carried out. Should the existing literatures on metagovernance, institutional design, and computational governance be shown already to answer the question posed here, the later sections lose their ground. Objection, correction, and counter-evidence are welcome at huangwanhong@serendip.ngo.
§1 The Problem of Governance Design
Three constraints hold at once, and together they form the difficulty this paper addresses.
The first is heterogeneity. The targets of governance differ widely. A family, a community, a shared resource, a firm, a transnational association, and a state call for different governance, and what counts as sound coordination in one setting can be dysfunction in another. A single governance model that fits all of these at once is unlikely to exist, and the diversity is the substance of the problem and not noise to be averaged away.
The second is non-accumulation. When each target is governed by a design built from the beginning, the resulting knowledge stays local. Solutions are reinvented, and failures repeat. No substrate carries a lesson learned in one setting over to another, and so the stock of governance knowledge grows far more slowly than the number of governance experiments would allow.
The third is the legitimacy of universals. One response to heterogeneity and non-accumulation is to author a general standard and ask that it be adopted. This is the path of institutional standardization, and it has produced instruments of real value. Its limit, for the present purpose, is that the authority to author a cross-context governance standard is hard to establish. No body has surveyed all settings, and no setting is obliged to accept a standard written elsewhere on its behalf.
The difficulty is that these three constraints leave only a narrow space between them. On one side lie incommensurable one-off designs, which fit their settings and do not accumulate. On the other lie universal standards, which accumulate and fit poorly and legitimate with difficulty. The question this paper opens is whether accumulation can be had without paying either price.
§2 The Adopter and the Object of Governance
Before the approach is stated, two roles must be kept apart, since running them together is a frequent source of confusion.
The adopter is the party that takes up an architecture, implements it, and revises it. An adopter is an organization or an individual, that is, an agent able to decide and to act. The object of governance is what the governance is about, the target upon which it is exercised. An object can be a family, a shared resource, a community, a process, an institution, a relationship, or a territory, and it need not be an organization, nor even an agent. We use the word object precisely because the target of governance varies so widely that any narrower word would smuggle in a restriction the framework means to avoid.
The architecture is accordingly wielded by an adopter and directed at an object. One adopter may govern several objects of different kinds, and one kind of object may be governed by many adopters. The revision of an architecture is carried out by an adopter, since only an agent can revise; the appropriateness of a revision is judged against how well it governs the adopter’s object. This distinction does the work of §4, where the site of revision and the site of selection are separated.
§3 The Object of Design
We do not think the answer is a better universal standard. We think the object of design should change.
The motivation is partly practical and partly principled. The practical part is a limit we accept openly. To learn which governance architectures work across scales, cultures, kinds, and historical periods, one would want to experiment across all of them. We cannot do this. No research group can run controlled trials across civilizations and centuries, and the very diversity that makes the question worth asking is what places it beyond the reach of ordinary controlled study.
This limit is usually treated as a weakness. We treat it as a direction. Because we cannot move across all settings ourselves, we decline to author the standard that would span them. We propose to build the field within which the many adopters generate and revise governance for their own objects, and within which their results become visible to one another. The inability to experiment across settings is the reason the generative-field approach is the appropriate one, and we present it as such openly.
The principled part of the motivation follows from a prior commitment, set out in §5. That commitment holds that structure and value are generated through relation and are not fixed ahead of it. A standard authored in advance of any implementation fixes structure ahead of the relations it is meant to govern, and so runs against the commitment. A standard that precipitates from the relations among many implementations honors it. The reframe this paper proposes is what that commitment yields once it is brought to bear on governance.
§4 The Meta-Architecture and Its Generative Field
We propose an abstract meta-architecture together with the field in which it evolves.
The meta-architecture is abstract in that it specifies the components of a governance architecture, the standards those components answer to, and the constraints that hold among them, while leaving the content of any particular governance to the adopter. It defines the form a governance architecture takes and the interfaces among its parts, and it stops there. Around it sits the generative field, an infrastructure that supplies affordances in place of commands. The field offers tools by which an adopter designs a governance architecture for its object, documents it in a shared and comparable form, analyzes its behavior, and reconciles or merges norms drawn from other settings. The field makes certain operations possible and certain results visible, and it leaves the governance itself to those who inhabit each setting.
The central reframe concerns the status of a standard. In conventional standardization a standard is authored and then adopted. A body produces it, adopters implement it, and revision is periodic and held by the body. In the approach proposed here a standard is emergent and then revised in commons. An initial architecture is supplied, and this initial architecture is not the standard. It is an initial population, in the sense a genetic algorithm gives the term, a starting point for variation and selection. The working standard, at any moment, is whatever has stabilized across many implementations through variation and differential adoption. The standard is an output of the process, and the initial architecture is the seed the process begins from.
The evolution runs at two grains, and this paper describes them as an intuition and not as a mechanism. At the finer grain, a single adopter revises its own architecture in the course of governing its object, guided by local fit, that is, by whether the revised architecture governs the object better for that adopter. At the coarser grain, the revisions do not stay sealed within their adopters. They are visible across the commons, and other adopters, sometimes governing objects of quite different kinds, borrow and revise from what they see. The finer grain is where a variant is produced. The coarser grain is where a variant spreads or fails to spread. A revision is at once a local adaptation and a contribution to the population, and the working standard that precipitates from the coarser grain becomes, in turn, part of the seed that later adopters begin from. So the site of revision is the adopter, and the site of selection is the commons, and the ancestor the population starts from rolls forward as the standard settles and resettles.
Central claim. A governance standard can be the emergent and revisable output of a commons evolutionary process. An initial architecture serves as an initial population and not as the standard, each adopter implements and revises the architecture in the course of governing its own object, and the working standard at any moment is whatever has stabilized across many implementations through variation and differential adoption. The task before us is accordingly to provide the abstract meta-architecture and the generative field within which standards are generated, documented, analyzed, and revised, and not to author the standard itself.
We state the guardrail in the same breath, since emergence alone is not a source of legitimacy. What spreads is not thereby justified. A norm can propagate because it serves the strong, through network effects, or through accidents in the initial population, and none of these is a mark of governing well. An approach that only celebrates bottom-up emergence launders power as legitimacy, and this is the known failure of every account that reads spontaneous order as self-justifying. For this reason the generative field is deliberative as well as populational. The infrastructure should make selection pressures visible and open to contest, so that adoption can be reasoned about and not merely undergone. The guardrail is developed as a design constraint in §5.
§5 Grounding in Generative Relational Governance
The reframe of this paper did not begin from software and arrive at a framework afterward. It began within Generative Relational Governance, the governance framework of the present programme, which itself rests on the deeper commitment of Generative Relational Being: that structure and value are generated through relation and are not fixed prior to it. The approach is what that commitment yields once it is turned on the design of governance. Naming the framework here is deliberate, and it explains why the standard is asked to emerge at all.
From that commitment the shape of the proposal follows. A governance standard authored from above and specified before any implementation fixes structure ahead of the relations it is meant to govern, and in the framework’s terms it is the anti-generative object. A standard that precipitates from the relations among many implementations is the generative object. This is why the proposal asks the standard to emerge: the relations among adopters, and between each adopter and its object, are what generate the architecture and the standard, and the standard is the settling of those relations into a shared form.
The generativity runs both ways, and this reciprocity is the framework’s specific contribution here. The relations generate the standard, and the evolution of the standard in turn reshapes the relations, since a revised architecture changes how an adopter governs its object and how adopters stand toward one another across the commons. A standard that settles is not an endpoint that halts the relations. It is a passing form the relations take, open to the next revision, so that generating the standard and revising the relations are one continuing process seen from two sides.
The alignment also explains the guardrail and keeps it consistent with the framework. The cautious version of the proposal fixes a small set of normative constraints by design while letting the rest evolve. One might worry that fixing anything at all runs against a generative commitment. It does not, so long as the fixed constraints are the conditions that keep generation open and are not substantive verdicts about what good governance should contain. The minimal floor concerns the contestability of adopted norms, the visibility of selection pressures, the reversibility of adoption, and resistance to capture. These are the conditions under which the commons keeps generating, and without them it settles into whatever the strong propagate. The floor protects generativity, and it does not dictate governance content.
The cautious floor. The generative field fixes a minimal normative floor by design: contestability, visibility of selection pressures, reversibility of adoption, and resistance to capture. These constraints are the conditions under which a commons keeps generating in place of freezing into the interests of whoever propagates most forcefully. They are not judgments about the content of good governance, and fixing them is consistent with a generative commitment because they protect the openness of generation.
The platform that hosts this process is itself governed, and we take the recursion as a design requirement in place of a paradox. The governance of the generative field should obey the same principles the field is meant to host. This requirement is what keeps the word emergent from sliding into the phrase whatever the strong propagate, since a field whose own governance is contestable, visible, reversible, and resistant to capture has furnished emergence with the conditions that mark it off from mere spread.
§6 Relation to Prior Work
We do not claim the meta level as our contribution. Governance of governance is a developed field. Work on metagovernance has theorized how plural modes of governance are coordinated (Jessop 2011; Meuleman 2008). Work on polycentric governance has shown how order arises among many overlapping centers of decision without a single apex (Ostrom 2010). Constitutional political economy has long studied the design of the rules under which ordinary decisions are made (Buchanan & Tullock 1962). Each of these lines has occupied the meta level for a long time, and each has met the problems named in §1. Standards for organizational governance already exist and are maintained by established bodies; ISO 37000 offers guidance said to apply to all organizations regardless of type, size, and purpose, and it names value generation among its own principles (ISO 2021). To claim novelty at the altitude of the meta level, or at the fact of cross-context ambition, would be to ignore this work.
The nearest neighbor is more specific, and we engage it directly. Schneider et al. (2021) propose Modular Politics, a governance layer for online communities assembled from computational components that are modular, composable, portable across settings, and interoperable across platforms, and they propose an open standard for networked governance built on the Institutional Analysis and Development tradition. The proposal is bottom-up, component-based, and standard-oriented, and it shares much of the spirit of the present one. Three differences mark the direction taken here. The first concerns scope. Modular Politics addresses online communities on digital platforms, whereas the object here varies across kinds, scales, cultures, and historical periods, and includes objects that are neither digital nor organizational. The second concerns the status of the standard. Modular Politics specifies a target architecture and a set of design goals, and in this respect it authors the framework, whereas the move here is that the standard precipitates from differential adoption and is revised in commons in place of being authored in advance. The third concerns the source of the normative floor. The floor here is grounded in Generative Relational Governance, which supplies a principled reason for the small set of constraints the field fixes, in place of a list of design desiderata.
The analogy that fixes the difference is with the infrastructures of open collaboration in place of any theory of governance. A version-controlled commons for software is a substrate on which practices of structuring software vary, propagate, and are selected, and on which no committee authored the standard that in fact emerged. The proposal here asks whether governance architectures admit a substrate of the same kind. That question is the one this paper means to open.
The claimed contribution. The contribution is not the meta level, which is old, and not the idea of a governance standard, which is old, and not modular or bottom-up governance, which is current work. It is the proposal that a governance standard be treated as the emergent and revisable output of a commons evolutionary process in place of an authored input to it, that the object of design be accordingly the abstract meta-architecture and the generative field within which such evolution occurs, and that the whole be extended to governance objects of any kind. The claim is held lightly, as a direction to be tested by building.
§7 Open Problems and Further Work
This paper states a direction. The programme it opens has a first and load-bearing task, and several that follow.
The first task is the problem of representation. An evolutionary account requires something that varies and recombines. If governance architectures cannot be represented in a shared and comparable form, then variation and differential adoption have nothing to operate on, and the evolutionary language stays a metaphor. The programme must ask what the reusable components of a governance architecture are, and in what form two architectures drawn from different settings can be compared, merged, or made to exchange parts. Until this is answered, the generative field is an aspiration, and the distinction drawn in §6 between an evolutionary substrate and an authored framework rests on a promise not yet kept. We name representation as the central open problem in place of deferring it quietly, since the weight of the proposal rests upon it.
The second task is the design of the field itself. This includes tools for authoring and documenting governance architectures in the shared form, tools for analyzing their behavior, and tools for reconciling and merging norms across settings. The merging of norms is likely the hardest of these, since it is where heterogeneity resists comparison most strongly, and it is also where the cautious floor of §5 does its work, because a merge that broke contestability or enabled capture is one the field should refuse.
The third task is the specification of the minimal normative floor. The cautious version of the proposal depends on identifying constraints that protect generativity without dictating governance content, and on showing that this small set is at once sufficient to prevent capture and modest enough that it does not become a universal standard reintroduced by the back door.
The fourth task is the evolutionary environment itself. Where controlled experiment across settings is unavailable, the alternative is an environment in which many implementations coexist, adopt from one another, and revise. The programme must describe what this environment requires in order to function, and how emergence within it can be observed and studied while the designer observes from outside in place of steering the outcome from within.
We offer this as a beginning. The claims are provisional, the framework is under development, and the purpose of the paper is to make the direction clear enough to be criticized and improved.
A note on method. No formal apparatus has been used here, and the abstention is deliberate. The proposal is at the stage where its work is to fix a direction and a vocabulary, and a formalism introduced now would borrow a rigor the direction has not yet earned. Two honest developments lie ahead. One is the representation problem named in §7, which when solved would give the evolutionary language a genuine object to operate on. The other is a connection to the language of evolutionary and dynamical systems, in which an initial architecture would answer to an initial population, differential adoption to a selection pressure, and a stabilized standard to an attractor of the commons dynamics. Neither is attempted here, and what is claimed is stated in prose meant to be legible without the apparatus that might later formalize it.
References
Buchanan, James M., and Gordon Tullock (1962). The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor: University of Michigan Press.
International Organization for Standardization (2021). ISO 37000:2021, Governance of Organizations - Guidance. Geneva: ISO.
Jessop, Bob (2011). Metagovernance. In Mark Bevir (ed.), The SAGE Handbook of Governance, 106–123. London: SAGE.
Meuleman, Louis (2008). Public Management and the Metagovernance of Hierarchies, Networks and Markets. Heidelberg: Physica-Verlag.
Ostrom, Elinor (2010). Beyond Markets and States: Polycentric Governance of Complex Economic Systems. American Economic Review, 100(3), 641–672.
Schneider, Nathan, Primavera De Filippi, Seth Frey, Joshua Z. Tan, and Amy X. Zhang (2021). Modular Politics: Toward a Governance Layer for Online Communities. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW1), 1–26.
中文
一种由共域演化的治理元架构
生成性关系治理之内的一次初步讨论
黄万宏 · huangwanhong@serendip.ngo
摘要
治理设计面对一个持久的困难。治理的诸目标跨尺度、文化、种类与历史时期而不同，故单一的一个治理模型在它们之间迁移得很差。为每一个目标量身定制的设计并不累积，因为在一个情境中习得的一个教训鲜少抵达另一个。而从上方撰写一个普适治理标准的尝试难以取得正当性，因为没有单一的机构持有为所有目标一次性设计治理的地位或知识。本文对此提供一个初步的直觉，在生成性关系治理的框架之内展开。我们提出一个治理的元架构：一个关于一个治理架构之诸组件、那些组件所回应之诸标准，以及它们之间诸约束的抽象界定，连同诸工具，凭之一个采纳者能为它自己的目标实现该架构、记录它、分析它并修订它。该元架构不是被撰写一次然后被采纳。它在一个共域中被演化。一个初始架构以一个遗传算法之初始种群的方式作为一个起点被供出，每一个采纳者在治理它自己目标的过程中实现并修订它，那些修订跨一个共域变得可见，而一个可用的标准从差异化采纳中析出、而非被事先固定。共域正是这一贡献所在之处：那个标准成为一个共享演化过程的可修订输出、而非它的输入，而这翻转了常规标准化的因果方向。我们把采纳者，它是一个组织或一个个人，同治理的对象，它变化甚广且无需是一个组织，分开。我们采纳一个谨慎的版本，凭设计固定一个极小的规范底线，并命名一个后续纲领必须处理的诸开放问题，其中以一个共享且可比较的形式表示诸治理架构是第一个。
关键词： 治理元架构；共域；演化式标准化；生成性关系治理；元治理；制度基础设施。
关于本文定位的一则说明。 这是一篇初步的工作论文。它陈述一个早期的直觉与一个研究方向，且不声称一套完成的理论。它的目标是辨认一个问题、陈述一个动机，并把一个所设想的进路描述得足够清晰以被批评与被改进。那个中心主张是临时的，那个母框架正在发展之中，而本文所命名为必要的诸任务中有数个尚未被执行。倘若能证明关于元治理、制度设计与计算治理的现存诸文献已然回答此处所提的问题，后面的诸节便失去它们的立足之处。反对、更正与反证欢迎寄至 huangwanhong@serendip.ngo。
§1 治理设计的问题
三个约束同时成立，而它们合起来构成本文所处理的困难。
第一个是异质性。治理的诸目标差异甚广。一个家庭、一个社区、一份共享资源、一家公司、一个跨国联合体，以及一个国家，呼唤不同的治理，而在一个情境中算作健全协调之物，在另一个中可能是功能失调。一个同时适合所有这些的单一治理模型不大可能存在，而这一多样性是那个问题的实质、而非要被平均掉的噪声。
第二个是非累积。当每一个目标都由一个从头建起的设计来治理时，由此产生的知识停留在局部。诸解决方案被重新发明，而诸失败重演。没有任何基底把在一个情境中习得的一个教训带到另一个，故治理知识的存量增长得远慢于诸治理实验的数目所会容许的。
第三个是普适物的正当性。对异质性与非累积的一个回应是撰写一个一般标准并要求它被采纳。这是制度标准化的路径，而它已产生真正有价值的诸工具。就当下的目的而言，它的界限在于撰写一个跨情境治理标准的权威难以确立。没有任何机构勘察过所有情境，而没有任何情境有义务接受一个在别处代它写就的标准。
那个困难在于这三个约束在它们之间只留下一个狭窄的空间。一侧躺着不可通约的一次性设计，它们适合它们的情境而不累积。另一侧躺着普适标准，它们累积、适合得差，且艰难地取得正当性。本文所开启的问题是，累积能否在不付出任何一种代价的情况下被拥有。
§2 采纳者与治理的对象
在陈述那个进路之前，两个角色必须被分开，因为把它们混为一谈是一个频繁的混淆来源。
采纳者是取用一个架构、实现它并修订它的那一方。一个采纳者是一个组织或一个个人，也就是说，一个能决定并能行动的能动者。治理的对象是治理所关乎之物，即它被施于其上的那个目标。一个对象可以是一个家庭、一份共享资源、一个社区、一个过程、一个制度、一段关系，或一片领土，而它无需是一个组织，甚至无需是一个能动者。我们恰恰使用对象这个词，因为治理的目标变化如此之广，以致任何更狭窄的词都会偷运进一个该框架意在避免的限制。
因而该架构由一个采纳者所执握、并被指向一个对象。一个采纳者可能治理数个不同种类的对象，而一种对象可能被许多采纳者治理。一个架构的修订由一个采纳者来执行，因为只有一个能动者能修订；一次修订的适当性对照它治理该采纳者对象的好坏来判断。这一区分做§4的工作，在那里修订的场址与选择的场址被分开。
§3 设计的对象
我们不认为那个答案是一个更好的普适标准。我们认为设计的对象应当改变。
这一动机部分是实践的、部分是原则的。实践的部分是一个我们公开接受的界限。为了解哪些治理架构跨诸尺度、诸文化、诸种类与诸历史时期起作用，人会想要跨它们全部做实验。我们无法这样做。没有任何研究团队能跨诸文明与诸世纪运行受控试验，而恰恰是那使这个问题值得一问的多样性，把它置于寻常受控研究的可及范围之外。
这一界限通常被当作一个弱点。我们把它当作一个方向。因为我们自己无法跨所有情境移动，我们谢绝撰写那个会横跨它们的标准。我们提议建造那个场域，在其之内众多采纳者为它们自己的对象生成并修订治理，而在其之内它们的结果彼此变得可见。跨诸情境做实验的无能，正是生成性场域进路成为那个适当进路的缘由，而我们公开地如此呈现它。
这一动机的原则部分随一个在先的承诺而来，该承诺在§5中被铺陈。那个承诺主张结构与价值是通过关系被生成的、而非在它之前被固定。一个在任何实现之前被撰写的标准，把结构固定在它意在治理的诸关系之前，故与那个承诺相抵触。一个从众多实现之间的诸关系中析出的标准则尊重它。本文所提议的重新定框，正是那个承诺一旦被带到治理上便产出之物。
§4 元架构及其生成性场域
我们提议一个抽象的元架构，连同它在其中演化的那个场域。
该元架构是抽象的，在于它界定一个治理架构的诸组件、那些组件所回应的诸标准，以及在它们之间成立的诸约束，同时把任何特定治理的内容留给采纳者。它界定一个治理架构所采取的形式以及它诸部分之间的接口，而它止步于此。围绕它坐落着那个生成性场域，一个供出可供性、而非命令的基础设施。该场域提供诸工具，凭之一个采纳者为它的对象设计一个治理架构、以一个共享且可比较的形式记录它、分析它的行为，并调和或合并取自其他情境的诸规范。该场域使某些操作成为可能、使某些结果可见，而它把治理本身留给栖居于每一个情境的那些人。
那个中心重新定框关乎一个标准的地位。在常规标准化中，一个标准被撰写、然后被采纳。一个机构产生它，采纳者实现它，而修订是周期性的、并由那个机构持有。在此处所提议的进路中，一个标准是涌现的、然后在共域中被修订。一个初始架构被供出，而这个初始架构不是那个标准。它是一个初始种群，在一个遗传算法赋予该词的意义上，一个供变异与选择的起点。那个可用的标准，在任何时刻，都是通过变异与差异化采纳跨众多实现稳定下来的无论什么。那个标准是那个过程的一个输出，而那个初始架构是那个过程由之开始的那个种子。
那个演化在两个颗粒度上运行，而本文把它们描述为一个直觉、而非一个机制。在较细的颗粒度上，一个单一采纳者在治理它对象的过程中修订它自己的架构，由局部契合所引导，也就是说，由那个被修订的架构是否为那个采纳者更好地治理那个对象所引导。在较粗的颗粒度上，那些修订不封闭在它们的采纳者之内。它们跨那个共域可见，而其他采纳者，有时治理种类相当不同的对象，从它们所见中借用并修订。较细的颗粒度是一个变体被产生之处。较粗的颗粒度是一个变体扩散或未能扩散之处。一次修订同时是一次局部适应与一份对那个种群的贡献，而从较粗颗粒度中析出的那个可用标准，转而成为后来的采纳者由之开始的那个种子的一部分。故修订的场址是采纳者，而选择的场址是共域，而那个种群由之开始的那个祖先随那个标准安顿并再安顿而向前滚动。
中心主张。 一个治理标准能是一个共域演化过程的涌现且可修订的输出。一个初始架构充当一个初始种群、而非那个标准，每一个采纳者在治理它自己对象的过程中实现并修订该架构，而那个可用的标准在任何时刻都是通过变异与差异化采纳跨众多实现稳定下来的无论什么。因而摆在我们面前的任务是提供那个抽象元架构与那个生成性场域，在其之内诸标准被生成、被记录、被分析并被修订，而不是撰写那个标准本身。
我们在同一口气中陈述那道护栏，因为涌现本身不是正当性的一个来源。扩散之物不因此便被辩护。一个规范能因它服务于强者、通过网络效应，或通过初始种群中的诸偶然而传播，而这些中没有一个是治理得好的标志。一个只颂扬自下而上之涌现的进路，把权力洗成正当性，而这是每一个把自发秩序读作自我辩护之说明的已知失败。出于这一缘由，那个生成性场域既是审议的、也是种群性的。那个基础设施应使诸选择压力可见并对争议开放，从而采纳能被推理、而非仅仅被经受。那道护栏在§5中作为一个设计约束被展开。
§5 在生成性关系治理中的奠基
本文的重新定框不是从软件开始、其后才抵达一个框架。它在生成性关系治理之内开始，即当下这个纲领的治理框架，它本身依据生成性关系存在的那个更深的承诺：结构与价值是通过关系被生成的、而非在它之前被固定。这个进路正是那个承诺一旦被转到治理设计上便产出之物。在此处命名这个框架是刻意的，而它解释为什么那个标准竟被要求涌现。
从那个承诺，那个提议的形状随之而来。一个从上方被撰写、并在任何实现之前被界定的治理标准，把结构固定在它意在治理的诸关系之前，而在该框架的术语中它是那个反生成性的对象。一个从众多实现之间的诸关系中析出的标准是那个生成性的对象。这正是那个提议要求那个标准涌现的缘由：诸采纳者之间、以及每一个采纳者与它对象之间的诸关系，是生成那个架构与那个标准之物，而那个标准是那些关系安顿为一个共享形式。
那个生成性双向而行，而这一互惠是该框架在此处的特定贡献。诸关系生成那个标准，而那个标准的演化转而重塑诸关系，因为一个被修订的架构改变一个采纳者如何治理它的对象、以及诸采纳者如何跨那个共域彼此相立。一个安顿下来的标准不是一个使那些关系停止的终点。它是那些关系所采取的一个过渡形式，对下一次修订开放，从而生成那个标准与修订那些关系是从两侧看的同一个持续过程。
这一契合也解释那道护栏、并使它与该框架保持一致。那个提议的谨慎版本凭设计固定一小组规范约束、而让其余的演化。人可能担忧固定任何东西都与一个生成性承诺相抵触。它并不，只要那些被固定的约束是使生成保持开放的诸条件、而非关于好治理应含有什么的实质裁断。那个极小的底线关乎被采纳规范的可争议性、诸选择压力的可见性、采纳的可逆性，以及对捕获的抵抗。这些是那个共域持续生成所处的诸条件，而没有它们它便安顿进强者所传播的无论什么。那个底线保护生成性，而它不规定治理内容。
谨慎的底线。 那个生成性场域凭设计固定一个极小的规范底线：可争议性、诸选择压力的可见性、采纳的可逆性，以及对捕获的抵抗。这些约束是一个共域持续生成、而非冻结进无论谁最有力地传播之利益所处的诸条件。它们不是关于好治理之内容的判断，而固定它们与一个生成性承诺相一致，因为它们保护生成的开放性。
托管这一过程的那个平台本身被治理，而我们把这一递归当作一个设计要求、而非一个悖论。那个生成性场域的治理应服从该场域意在托管的那些同样的原则。这一要求正是使涌现这个词不滑入强者所传播的无论什么这一短语的东西，因为一个其自身治理是可争议的、可见的、可逆的且抵抗捕获的场域，已为涌现供给了把它同单纯扩散区分开来的那些条件。
§6 与先前工作的关系
我们不把那个元层级主张为我们的贡献。治理的治理是一个发达的领域。关于元治理的工作已理论化了多元的诸治理模式如何被协调（杰索普 2011；莫伊勒曼 2008）。关于多中心治理的工作已显示秩序如何在许多重叠的决策中心之间、在没有一个单一顶点的情况下生起（奥斯特罗姆 2010）。宪政政治经济学长久以来研究寻常决策据以被作出之诸规则的设计（布坎南与塔洛克 1962）。这些路线中的每一条都长久地占据那个元层级，而每一条都遇到§1中所命名的诸问题。关于组织治理的诸标准已然存在、并由既立机构维护；ISO 37000 提供据称适用于所有组织、无论其类型、规模与目的的指导，而它把价值生成命名在它自己的诸原则之中（ISO 2021）。在元层级的高度上、或在跨情境抱负这一事实上主张新颖性，将是忽略这一工作。
最近的邻居更为具体，而我们直接与它交锋。施耐德等人（2021）提出模块化政治，一个供在线社区的治理层，由模块化的、可复合的、跨情境可携带的、且跨平台可互操作的诸计算组件装配而成，而他们提议一个建立在制度分析与发展传统之上的、供网络化治理的开放标准。该提议是自下而上的、基于组件的，且面向标准的，而它分享当下这一提议的许多精神。三个差别标记此处所取的方向。第一个关乎范围。模块化政治处理数字平台上的在线社区，而此处的对象跨诸种类、诸尺度、诸文化与诸历史时期而变化，并包括既非数字也非组织的诸对象。第二个关乎那个标准的地位。模块化政治界定一个目标架构与一组设计目标，而在这一方面它撰写那个框架，而此处的动作是那个标准从差异化采纳中析出、并在共域中被修订、而非被事先撰写。第三个关乎那个规范底线的来源。此处的底线奠基于生成性关系治理，它为那个场域所固定的那一小组约束供出一个原则性的理由、而非一份设计愿望清单。
固定那个差别的类比是与开放协作的诸基础设施、而非与任何治理理论。一个供软件的版本控制共域是一个基底，在其上结构化软件的诸实践变异、传播并被选择，而在其上没有委员会撰写那个实际上涌现的标准。此处的提议问诸治理架构是否容许一个同种类的基底。那个问题正是本文意在开启的那个。
所主张的贡献。 这一贡献不是那个元层级，它是旧的；也不是一个治理标准这个观念，它是旧的；也不是模块化或自下而上的治理，它是当下的工作。它是这个提议：一个治理标准被当作一个共域演化过程的涌现且可修订的输出、而非一个对它撰写的输入；设计的对象相应地是那个抽象元架构与这样的演化发生于其内的那个生成性场域；而整体被扩展到任何种类的治理对象。这一主张被轻持，作为一个要通过建造来检验的方向。
§7 开放的问题与进一步的工作
本文陈述一个方向。它所开启的纲领有一个首要且承重的任务，以及随之而来的数个。
第一个任务是表示的问题。一个演化的说明要求某样变异并重组之物。倘若诸治理架构无法以一个共享且可比较的形式被表示，那么变异与差异化采纳便没有任何东西可作用于其上，而那套演化的语言停留为一个隐喻。那个纲领必须问一个治理架构的可复用组件是什么，以及取自不同情境的两个架构能以什么形式被比较、被合并，或被弄得交换部件。在这被回答之前，那个生成性场域是一个愿景，而§6中所划的、在一个演化基底与一个被撰写框架之间的区分，依据一个尚未被履行的承诺。我们把表示命名为那个中心开放问题、而非悄悄地把它推延，因为那个提议的分量依据它。
第二个任务是那个场域本身的设计。这包括用于以那个共享形式撰写并记录诸治理架构的诸工具、用于分析它们行为的诸工具，以及用于跨诸情境调和并合并诸规范的诸工具。诸规范的合并很可能是这些中最难的，因为它是异质性最强烈地抗拒比较之处，而它也是§5的那个谨慎底线做它工作之处，因为一次破坏了可争议性或使能了捕获的合并，是那个场域应当拒绝的一次合并。
第三个任务是那个极小规范底线的界定。那个提议的谨慎版本取决于辨认出保护生成性而不规定治理内容的诸约束，以及取决于显示这一小组既足以防止捕获、又谦逊到它不成为一个从后门被重新引入的普适标准。
第四个任务是那个演化环境本身。在跨诸情境的受控实验不可得之处，那个替代方案是一个环境，在其中许多实现共存、彼此采纳并修订。那个纲领必须描述这个环境为运作所要求的东西，以及在其之内的涌现如何能被观察并被研究，同时那个设计者从外部观察、而非从内部操纵那个结果。
我们把这作为一个开端提供。这些主张是临时的，那个框架正在发展之中，而本文的目的是把那个方向弄得足够清晰以被批评与被改进。
关于方法的一则说明。 此处没有使用任何形式装置，而这一克制是刻意的。这个提议正处在其工作是固定一个方向与一套语汇的阶段，而一个此刻被引入的形式体系会借来一份那个方向尚未挣得的严格。两个诚实的发展在前方。一个是§7中所命名的表示问题，它一旦被解决便会给那套演化的语言一个可作用于其上的真正对象。另一个是一个到演化与动力系统之语言的连接，在其中一个初始架构会答对一个初始种群、差异化采纳答对一个选择压力，而一个被稳定的标准答对那个共域动力学的一个吸引子。此处两者都不予尝试，而所主张的以意在无需那个日后或许会形式化它的装置便可读的散文被陈述。
参考文献
Buchanan, James M., and Gordon Tullock (1962). The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor: University of Michigan Press.
International Organization for Standardization (2021). ISO 37000:2021, Governance of Organizations - Guidance. Geneva: ISO.
Jessop, Bob (2011). Metagovernance. In Mark Bevir (ed.), The SAGE Handbook of Governance, 106–123. London: SAGE.
Meuleman, Louis (2008). Public Management and the Metagovernance of Hierarchies, Networks and Markets. Heidelberg: Physica-Verlag.
Ostrom, Elinor (2010). Beyond Markets and States: Polycentric Governance of Complex Economic Systems. American Economic Review, 100(3), 641–672.
Schneider, Nathan, Primavera De Filippi, Seth Frey, Joshua Z. Tan, and Amy X. Zhang (2021). Modular Politics: Toward a Governance Layer for Online Communities. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW1), 1–26.