Daemon Corporation

Global Leader in Excellence.

Research Report № 003

Coordination Without a Grand Unified Theory

Central finding

Coordination is not one phenomenon in the strong sense of a single problem with a common causal structure, mechanism set, or measure of success. Across organizations, markets, commons, governments, software systems, social movements, emergencies, open-source projects, deliberative forums, blockchains, and AI-agent systems, the term points to different objects and different achievements.

The most defensible commonality is thinner:

Coordination situations arise when two or more distinguishable loci of action are consequentially interdependent, so that their activities, decisions, or resource uses must somehow be made compatible under some criterion.

This identifies a recurring situation type, not a universal solution. The portable analytical move is to identify the interdependence and state the assumptions under which it must be managed. It does not follow that the same mechanisms, conditions, success criteria, or institutional forms will work across domains.

This distinction matters because operational fit is routinely confused with more demanding achievements. A distributed system can agree on a log without legitimately choosing its goals. An organization can function reliably without being safe. A commons institution can endure without being fair. A deliberative forum can reach agreement without producing free consent. A blockchain can be internally consistent while its community disputes which history should govern. People may also coordinate without a shared goal, a central coordinator, or a collective identity.

The evidence therefore supports a plural, layered account:

  • Strong unity is rejected at the level of mechanisms, undetermined at the functional level, locally supported for some bounded conditions, and weakly supported only at the level of situation type.
  • Family resemblance best describes the current conceptual landscape, but by itself explains or predicts little.
  • Narrow coordination is viable in two incompatible senses: a precise game-theoretic sense with limited coverage, and a broader interdependence-management sense with greater coverage but less discriminating power.
  • Overload is well supported both academically—through proliferating, weakly distinguished concepts—and rhetorically, where “coordination” can serve as aspiration, identity, advocacy, or ideology.

What travels best is not a technology or institutional blueprint but a discipline of analysis: name the dependency, actors, objective, authority structure, information conditions, fault or conflict model, power distribution, success criterion, and conditions of revision.

1. One word, several different problems

The ambiguity begins with the object being coordinated.

In coordination theory, the object is often a dependency among activities or resources. In transaction-cost economics it is adaptation around transactions. In collective-action theory it is contribution to a joint good. In commons governance it is appropriation, provision, and rule-making around a resource. In collaborative governance it is cross-boundary public decision-making. In distributed computing it is replica state, event order, or agreement under a fault model. In social movements it is mobilization and the formation of power. In deliberative democracy it is judgment and justification. In mission-oriented innovation it is political direction, portfolios, budgets, and cross-government authority.

These are not merely different applications of the same mechanism. They disagree about:

  • what the relevant actors are;
  • whether goals are given or formed within the process;
  • whether authority is legitimate, instrumental, or absent;
  • whether conflict is a fault, a constitutive feature, or the object of action;
  • whether incentives are central;
  • whether autonomy and exit must be preserved;
  • what counts as success;
  • and whether power belongs inside the explanation.

Comparative map of the traditions

Tradition Object and actors Characteristic failure and success Mechanisms and assumption profile
Coordination theory and organizational design Dependencies among activities, tasks, and resources; people are often represented as assignable to tasks Failure: unmanaged dependencies or a mistaken decomposition. Success: compatible interdependent work Rules, plans, roles, routines, schedules, shared representations, resource allocation. Goals are sometimes included as coordination processes, although later applications often bracket decision-making and sensemaking. Incentives, legitimacy, conflict, and power are acknowledged but weakly integrated.1
Institutional economics Transactions and adaptations among contracting parties, firms, markets, and hybrids Failure: maladaptation, hold-up, and governance structures poorly matched to transactions. Success: adaptive efficiency and economizing on transaction costs Prices support autonomous adaptation; hierarchy supports coordinated adaptation. Goals are largely given, incentives are central, authority is exercised through fiat and contract, and power appears principally through asset specificity.2
Collective-action theory Contributions by individuals to public or joint goods Failure: free riding or insufficient contribution. Success: provision and sustained participation Selective incentives, norms, thresholds, critical masses, and production functions. Goals and payoff structures are generally specified in advance; autonomy and exit are assumed; legitimacy is not normally the explanandum. The canonical zero-contribution prediction is empirically challenged, with no generally accepted unified successor.3
Commons and polycentric governance Appropriation, provision, monitoring, and rule-making around common-pool resources; actors include resource users and nested decision centers Failure: overuse, rule erosion, capture, or co-optation of collective choice. Success: institutional endurance and resource maintenance—not necessarily fairness Monitoring, graduated sanctions, conflict-resolution arrangements, nested institutions, and local rule-making. Local knowledge and rule autonomy matter. Authority is overlapping and qualified; conflict is managed institutionally. Power is consequential because durable arrangements can remain highly unequal.4
Collaborative and network governance Cross-boundary public decisions and joint action among agencies and non-state stakeholders; sometimes the whole network is the unit Failure: exclusion, domination, weak trust, power imbalance, or inability to act collectively. Success includes procedural inclusion, public purpose, deliberation, and network effectiveness Convening, face-to-face dialogue, trust building, shared understanding, negotiated rules, and alternative network-governance forms. Goals may be jointly formed. Legitimacy is definitional rather than incidental. Power is recognized but incompletely theorized.5
Distributed systems and computer science Machine states, logs, event ordering, decisions, or plans among processes under explicit timing and fault assumptions Failure: nontermination, inconsistent state, partition tradeoffs, or computational intractability. Success: safety, liveness, agreement, validity, and termination Leader election, quorums, log replication, membership rules, and fault-tolerant protocols. Objectives and validity conditions are inputs. Processes have no political autonomy; legitimacy and distributive power are not categories in the model.6
Open-source and standards communities Code changes, interfaces, releases, and specifications among volunteers, maintainers, firms, chairs, and committees Failure: burnout, unmaintained dependencies, capture, unresolved objections, and concentrated integration power. Success: shippable code or an implementable technical specification Maintainer jurisdiction, delegation, committees, appeals, implementation tests, and rough consensus. Participation may be voluntary, but authority remains institutionalized and partitioned. Public artifacts improve observability without eliminating hidden labor or concentrated control.7
Emergency management Under the Incident Command System, roles, tasks, and authority during volatile response; under the Dynes tradition, communication and coordination across established, expanding, extending, and emergent organizations Failure: either inadequate structure or a command-and-control model based on false assumptions of chaos. Success: effective response while preserving adaptability and continuity ICS uses formal roles, role switching, migrating authority, constrained improvisation, and system resetting. Dynes emphasizes continuity, coordination, cooperation, and shared governance. The disagreement concerns the environment and actor ontology—not simply formality versus flexibility.8
High-reliability organization theory Reliable operation in hazardous settings such as flight decks and control rooms Failure: operational unreliability, but also conceptual failure to distinguish reliability from safety or identify the population of “HROs.” Success criteria remain contested Attention to expertise and “heedful interrelating” are prominent, but the available evidence here is largely a critical review rather than the primary HRO literature. Reliability may increase while safety declines.9
Collective intelligence and human computation Group performance across tasks, or computational subtasks routed to human participants Failure: poor aggregation, weak validation, or mistaken causal attribution. Success: task accuracy or a measurable group-performance factor Collaboration processes, incentives, task decomposition, cross-checking, and aggregation. The designer typically supplies the task. Legitimacy, political representation, and collective responsibility are not measured. Evidence for a general collective-intelligence factor remains contested, and a published correction reduced a prominent reported average variance extracted from 44% to 19.6%.10
Deliberative and participatory governance Judgments, reasons, mutual understanding, knowledge integration, and the legitimation of public authority Failure: symbolic consultation, lobby subversion, structural inequality, hegemonic discourse, or agreement that is not free consent. Success: considered judgment, inclusion, depolarization, and legitimate authority Random selection, balanced information, facilitation, expert testimony, civility, and reason-giving. Goals and judgments are formed through the process. Legitimacy is the central success criterion; pre-existing inequalities can shape the forum before deliberation begins.11
Social movements and community organizing Constituencies, leadership, mobilization, shared purpose, and the conversion of resources into power Failure: inability to mobilize, develop leadership, or acquire power; conceptual stretching in movement theory. Success: a constituency able to act effectively on its purposes Narrative, relational organizing, strategic action, and leadership development. Goals are constituted through organizing; conflict and power are central rather than residual. Evidence for this tradition is comparatively thin in the present base.12
Mission-oriented innovation and cross-sector initiatives Political direction, portfolios, public funding, cross-ministerial authority, and coordination among governments, firms, and civil society Failure: fragmented policy mixes, unclear responsibility, cumbersome governance, or weak monitoring. Success: mission attainment and legitimacy of the initiative as a whole Political goal-setting, strong mission management, delegation, portfolio reallocation, and cross-sector collaboration. Partners retain some autonomy. Power and incumbent capture are risks. Structural frameworks are much better developed than comparative outcome evidence.13
Crypto, Web3, ReFi, network-state, and AI-agent communities Transaction order, Sybil resistance, token-weighted decisions, digital identity and exit, ecological asset claims, or agent roles and messages Failure: power concentration behind decentralized architecture, low participation, bad underlying assets, orchestration errors, and unresolved verification. Success criteria vary and are often unstated or promotional Proof of work, token voting, forks, smart contracts, agent topologies, and role assignment. Protocol goals are typically exogenous while community goals remain contested. Public ledgers do not establish legitimate authority or asset quality. “Autonomous” AI agents remain operationally delegated components unless a separate responsibility regime is supplied.14

This map reveals a recurring split. Technical and organizational accounts often ask whether interdependent components can work together. Governance, movement, and deliberative traditions additionally ask who may set the objective, whose interests count, how conflict is handled, and why an outcome should be accepted. The latter questions are not elaborations of synchronization; they are different problems with different evidence.

2. The thin common structure: consequential interdependence

The strongest candidate for a common core comes from coordination theory’s definition of coordination as managing dependencies among activities. Malone and Crowston identified shared resources, producer–consumer relations, simultaneity constraints, and task–subtask dependencies across computer systems, markets, and organizations.1

That framework is useful, but its portability has limits.

First, its generality was proposed rather than demonstrated. A later retrospective from the same research program acknowledged that applications had assumed rather than tested the generality of the mechanism taxonomy, that the theory had not been presented in a form conducive to testing, and that the design space remained far from characterized.15 Collective-action scholarship independently reported the absence of a mature typology of coordination mechanisms across its own heterogeneous models.16 The present evidence base found no matched study that operationalized the same dependency, mechanism, metric, and success criterion in two materially different traditions.

Second, “dependency” can become redescriptive rather than explanatory. Almost any causal relation can be called a dependency. To do analytical work, the account must specify:

  1. the relevant activities or decisions;
  2. the type and intensity of interdependence;
  3. what is controlled or adjustable;
  4. whose criterion defines compatibility;
  5. how failure will be observed;
  6. and which mechanisms are expected to change the outcome.

Third, the object of interdependence is itself contested. Some traditions locate it among tasks, others among actors, transactions, goals, adaptations, organizations, or decision centers. These choices are not interchangeable. Redesigning task interfaces is different from renegotiating authority, changing incentives, forming a political constituency, or establishing a legitimate goal.

Consequential interdependence is therefore best treated as the trigger for inquiry. It tells us when some coordination problem may exist, but not which problem it is or how to solve it.

3. Four hypotheses tested

3.1 Strong unity

Hypothesis: The different forms of coordination instantiate one general problem with a common causal structure, transferable mechanisms, or universal enabling conditions.

Evidence in favor

There are several genuine candidates for limited unity.

  • Okhuysen and Bechky propose that mechanisms such as plans, roles, routines, representations, and proximity create three integrating conditions: accountability, predictability, and common understanding. This has the right architecture for a general functional theory, although only the authors’ abstract—not the article body—was available.17
  • Ostrom’s design principles receive moderate empirical support across a review of 91 studies of community-based natural-resource management.18
  • Simon’s concept of near-decomposability explicitly spans physical, biological, economic, and social systems.19
  • Network governance repeatedly confronts tradeoffs between efficiency and inclusiveness, internal and external legitimacy, and flexibility and stability.20
  • Distinct formal models of collective action can converge on threshold and critical-mass behavior.16
  • Raft demonstrates real transfer of a decomposition—leader election, log replication, safety, and membership change—among systems sharing deterministic state machines, explicit membership, externally supplied goals, and defined fault assumptions.21

Evidence against

The negative evidence is stronger.

  • No comparative study in the evidence base tests the same mechanism against the same outcome in two substantially different traditions.
  • The only located application of the Okhuysen–Bechky framework outside organization studies—a small qualitative study of humanitarian coordination after the 2022 Cianjur earthquake—reported that accountability, predictability, and common understanding did not arise automatically from formal coordination mechanisms.22
  • Simon himself described near-decomposability as a strong property possessed by “vanishingly few” of all thinkable systems.19
  • Parnas showed that the effectiveness of modularization depends on the criteria used to decompose the system. Interfaces are not self-identifying; the relevant design decisions are domain-specific.23
  • Success criteria are incommensurable: safety and liveness, institutional longevity, technical implementability, legitimate authority, mobilized power, environmental additionality, and justice cannot be substituted for one another.
  • Formal solvability changes when assumptions change. FLP concerns possible nontermination in fully asynchronous consensus with one crash fault; CAP’s central result is scoped to an asynchronous model; finite-horizon decentralized planning can be NEXP-complete for two or more agents.6
  • Actor ontologies differ too sharply for a common mechanism to be presumed: processes in a protocol, resource users, randomly selected citizens, emergent disaster organizations, and social-movement constituencies are not equivalent units.

Verdict

Level Verdict
Common mechanisms Rejected with high confidence. No demonstrated cross-tradition mechanism set exists, and the main program proposing one acknowledged that generality had been assumed rather than tested.
Common functional architecture Undetermined. Accountability, predictability, and common understanding are a plausible candidate, but discriminating features and matched tests are absent.
Universal conditions Rejected. Ostrom-style principles have support within community-scale commons, not as a general blueprint.
Common situation type Weakly supported. Consequential interdependence recurs, but it is too thin to constitute a general causal theory.

Strong unity consequently fails where it would be most useful: at the level of mechanisms, conditions, and evidence. It survives only as a thin description of when a problem of fit may arise.

3.2 Family resemblance

Hypothesis: Coordination phenomena form a family whose members share overlapping features but no single defining causal structure.

This is the best description of the literature as it currently exists.

Different traditions recurrently display partial information, shared resources, mutual adjustment, contribution problems, incompatible local actions, authority allocation, feedback, thresholds, and adaptation. Yet no single one of those characteristics is present in every case. Several fields have reached similar conclusions about their own core concepts:

  • collaboration, cooperation, and coordination overlap so extensively that a systematic review found little basis for their conventional distinctions;24
  • polycentricity has multiple non-equivalent definitions and no accepted single representation;25
  • public administration contains a dense field of near-synonymous coordination concepts whose fragmentation has produced limited theoretical advance;26
  • cross-sector collaboration accumulated 95 theories, models, frameworks, and principle sets in one systematic review;27
  • commons scholarship has generated long lists of partly overlapping enabling conditions whose correlations are difficult to establish.4

Family resemblance accommodates this plurality. Its weakness is that it can accommodate almost anything. Goodwin and Jasper’s critique of concept stretching applies directly: as a concept broadens, it risks becoming a “sponge” that absorbs the entire environment and explains nothing.28

Verdict: Family resemblance is strongly supported as a description of the conceptual landscape, but it is not a theory of coordination. Without prior rules about which similarities count and which counterexamples would matter, it is difficult to falsify.

3.3 Narrow coordination

Hypothesis: Coordination is one bounded function or problem, while governance, collective action, mobilization, intelligence, and agency are distinct phenomena that may contain it.

The evidence supports two separate versions, which should not be merged.

Sense A: game-theoretic coordination

In the narrower game-theoretic sense, agents face multiple mutually advantageous equilibria and need to align expectations. The principal obstacle is informational rather than a conflict of interest. On this account, a Prisoner’s Dilemma is not a coordination game, and collective-action problems form a contrasting category.29

This sense has genuine discriminating power. It excludes many cases commonly called coordination: free-riding dilemmas, political conflict, mobilization, contested public decisions, and distributive struggles.

Its boundary is nevertheless disputed. Conventions need not always be coordination equilibria, and some collective-action situations may be better modeled as assurance games than as Prisoner’s Dilemmas. The relevant primary works by Lewis, Schelling, Ullmann-Margalit, and Runge were not available in the evidence base, so this hinge remains only secondarily sourced.

Verdict: A precise, useful technical sense, but one that covers only a small part of the territory.

Sense B: interdependence management

A broader operational definition is:

Coordination is the structuring and ongoing adjustment of interdependent activities, decisions, or resource uses so that specified joint constraints or performance criteria are sufficiently satisfied over time.

This requires:

  • identifiable activities, decisions, or resource uses;
  • consequential interdependence;
  • a specified compatibility or performance criterion;
  • and a mechanism that changes relationships or behavior.

Possible mechanisms include rules, roles, schedules, prices, queues, interfaces, communication, negotiation, hierarchy, voting, standards, incentives, shared representations, and feedback.

This definition travels farther than the game-theoretic one. It can include conflicting-interest cases, firms, markets, protocols, standards, disaster response, and commons. Its danger is low discriminating power: unless the dependency and criterion are specified, nearly any multi-actor process can be redescribed as coordination.

Verdict: Useful as a boundary-setting operational function, not as a universal explanatory theory.

3.4 The overloaded umbrella

Hypothesis: “Coordination” has been stretched across incompatible problems and often does rhetorical, aspirational, or ideological work rather than analytical work.

The evidence supports this at two distinct levels.

Academic overload

Management research, public administration, commons and polycentricity, and cross-sector collaboration have each documented the proliferation of overlapping concepts ahead of cumulative theory.24252627 This is not simply linguistic untidiness. It makes comparison harder because similar labels conceal different success criteria, while different labels may refer to the same empirical arrangement.

Promotional and ideological overload

In technology-adjacent communities, coordination can bundle incentive design, public goods, identity, sovereignty, ecological regeneration, decentralization, and moral aspiration.

  • Ethereum’s community-facing account described ReFi as an attempt to use Ethereum to solve global coordination crises. Yet the documented Toucan episode showed that ledger transparency did not establish the quality of underlying carbon credits; CarbonPlan reported that 99.9% of credits backing one pool were ineligible for the aviation offsetting program, after which Verra intervened with stronger institutional controls.30 This is evidence about that episode, not about ReFi as a whole.
  • The “network state” combines founder-led organization, national consciousness, a social contract, cryptocurrency, territory, an on-chain census, and diplomatic recognition. Its use of coordination is a program for political formation rather than a demonstrated mechanism.31
  • At the extreme, Gleichschaltung—also rendered as “synchronization”—was the official language for the Nazi regime’s coercive alignment of institutions in the name of national unity.32

Overload does not mean every use is vague. Technical communities also maintain precise internal meanings: RFC 7282 defines rough consensus as reasoned objection-handling rather than vote counting; Debian specifies jurisdiction and appeal; and Ethereum’s EIP-779 describes the DAO intervention as an exact “irregular state change.”733 Precision at one layer, however, does not validate broader political claims attached to it.

Verdict: The overload hypothesis is strongly supported, both as academic proliferation and as rhetorical or ideological expansion.

4. A map of distinct problem classes

The evidence supports at least twelve analytically distinct classes. They may coexist in one system, but no evidence demonstrates that solving them separately guarantees that the whole system will work.

Problem class Required input Success criterion Signature failure
1. Dependency and interface management A decomposition and a compatibility criterion Interdependent parts remain compatible Unmanaged dependency or incorrect decomposition
2. Contribution under free-riding incentives Payoffs and a production function Sustained provision Insufficient contribution
3. Agreement or state consistency under faults Membership, timing, fault, and validity models Safety, liveness, agreement, termination Nontermination, split state, or partition tradeoff
4. Judgment and information aggregation Signals, expertise, and an evaluable task Accuracy or considered judgment Aggregation artifacts or mistaken causal attribution
5. Resource and entitlement allocation Scarcity, claims, and entitlement rules Performance against an allocative objective Efficient execution of an unjust or substantively poor allocation
6. Direction setting and goal formation Preferences, mandates, and political contestation An authorized direction Fragmented objectives or unclear responsibility
7. Legitimate rule formation A constituency and a claim to rightful authority Legitimation and acceptable procedure Capture, symbolic cover, or consent distorted by power
8. Conflict management Standing objections or incompatible priorities Objections addressed through an accepted process Treating interest or value conflict as an information deficit
9. Commons governance A subtractable resource with difficult exclusion Institutional endurance and resource maintenance Overuse, rule erosion, or capture
10. Feedback-driven adaptation Monitoring, variance detection, and revision capacity Effective adjustment over time Inertia or incomplete feedback
11. Mobilization and power formation A constituency without sufficient power Capacity to pursue collective purposes Failure to build leadership, commitment, or leverage
12. Agency and accountability attribution A candidate collective actor and responsibility regime Coherent intention and answerability Inferring agency or responsibility from synchronization alone

These classes clarify several boundaries.

Coordination is not cooperation

Cooperation normally adds some form of aligned interest, voluntary contribution, or implementation of shared goals. Coordination in the broad operational sense does not require goodwill or common purpose. Markets can coordinate through prices among actors pursuing different ends; human-computation systems can route work to people who do not share the designer’s purpose; coercive regimes can align institutions.

Coordination is not collective action

Collective-action theory centrally concerns incentives to contribute to a joint good. Coordination may occur inside a collective-action system, but synchronizing contributions is not identical to motivating them. In the narrow game-theoretic sense, the two may be contrasting categories; in the broad interdependence sense, some collective-action problems can be redescribed as coordination. The boundary depends on which definition is used.

Coordination is not organization

Organization adds relatively durable roles, membership, decision rights, authority, and routines. Coordination can occur without an organization, as in price-mediated adaptation or emergent disaster groups. Organizations, conversely, may exist while coordinating badly.

Coordination is not governance or institutional design

Governance concerns who may make, interpret, enforce, and revise rules. Institutional design specifies rights, authority, sanctions, and procedures. Both often contain operational coordination, but they additionally confront legitimacy, power, conflict, and accountability.

Coordination is not collective intelligence

Collective intelligence is an empirical performance claim about groups completing varied tasks. Coordination may contribute to it, but measured performance does not establish representation, legitimate authority, wise judgment, or responsibility. Human computation makes the distinction especially clear: successful task routing can occur without collaboration or shared purpose.

Coordination is not mobilization

Mobilization forms constituencies and converts resources into power. It may require schedules, communication, and task allocation, but its explanandum is the creation of political capacity, not merely compatibility among activities.

Coordination is not collective agency

On influential shared-agency accounts, coordinated behavior is insufficient for acting together. Searle’s example of park visitors avoiding collisions while running for shelter illustrates coordination without a shared intention; dancers coordinating their movements provide the contrasting case.34 Other disciplines use “collective agency” more loosely, so this is a boundary within a particular philosophical family rather than a universal ranking.

5. What transfers—and where transfer fails

No direct comparative experiment supports the following as universal laws. The transfer map is therefore a hypothesis register grounded in observed contrasts, not a demonstrated general theory.

Dimension What appears to transfer Where transfer fails
Goal contestation The need to state the objective and who may change it Technical mechanisms generally take objectives and validity conditions as inputs. They do not authorize contested goals. This is the sharpest transfer boundary.
Actor autonomy Exit, refusal, and discretion can be represented as design variables Software processes do not possess the political autonomy of volunteers, citizens, firms, or emergent organizations. “Autonomous agent” usually means delegated operational discretion.
Dependency intensity Shared resources, prerequisites, simultaneity, and task decomposition recur The proposition that stronger dependency requires stronger coordination is often asserted without specifying the causal mechanism.
Scale Replication, standards, and interfaces can scale operations Participation, review, and authority often recentralize. Deliberative effects are easier to demonstrate in costly face-to-face settings, while commons principles are not established for large non-community systems.1118
Urgency Defaults, escalation paths, timeouts, and pre-delegated authority are reusable design ideas The right to set defaults or invoke emergency authority is institutional. A trauma-center study argued that formal information-processing typologies may be too rigid under urgency, novelty, surprise, and interpretive disagreement, although it did not establish causality.35
Trust Fault models are valuable prompts to state assumptions about failure and adversarial behavior Byzantine faults are not a sufficient model of political conflict, agenda control, ideology, or conflicting priorities.
Power asymmetry Concentration can be measured through voting weights, maintainer counts, centrality, decision rights, or ownership Network topology is not power. Resource ownership, agenda control, labor dependence, discursive authority, and control of defaults may diverge.
Voluntariness Opt-out, resignation, forking, and refusal are genuine institutional variables Coordination can also be coercive. Voluntary entry does not settle the legitimacy of effects on nonparticipants.
Resource control Auctions, prices, queues, budgets, and tokens can be compared as allocation mechanisms Each operates only after a prior entitlement rule specifies who may claim what. That rule is political and institutional.
Information distribution “Who knows what, and when?” is a robust diagnostic question Distributed information does not imply that decentralized solutions are computationally or institutionally easy.
Observability Logs, open code, public ledgers, and monitoring can improve auditability They do not reveal hidden labor, identities, preferences, asset quality, or all causal responsibility.
Legitimacy Very little transfers beyond the instruction to identify the relevant constituency Correctness, participation, transparency, or successful synchronization does not establish rightful authority.
Exit Forks and resignation can preserve dissent Whether exit is meaningful depends on network effects, assets, identity, infrastructure, and switching costs. The distribution of such costs in the DAO fork was not established in the evidence base.
Reversibility Versioning, rollback, system resetting, and forks are robust operational patterns Social harms, sunk investments, lost trust, and ecological damage may be irreversible even when the technical state can be rolled back.
Uncertainty Feedback, redundancy, escalation, and experimentation recur Under deep uncertainty, dependencies and success criteria may not be specifiable in advance.
Accountability Causal traces can support investigation Responsibility remains a legal, moral, and institutional allocation; it does not emerge automatically from observability.

The most portable pattern: assumptions before mechanisms

Distributed computing is instructive not because society can be governed as a computer but because formal computer science states its assumptions explicitly. FLP, CAP, and Raft ask about membership, timing, failure, validity, and required guarantees before proposing a solution.621

The corresponding institutional questions are:

  • Who counts as a participant or affected party?
  • Who supplies the objective?
  • Which failures are anticipated?
  • Which behavior is strategic, adversarial, negligent, or merely mistaken?
  • What must always remain true?
  • What may be sacrificed under stress?
  • Who may revise the arrangement?
  • Who bears the costs of failure?

This methodological discipline transfers more reliably than any protocol.

Interfaces transfer conditionally

Interfaces and modular decomposition are robust as analytical moves. They help locate changing decisions, ownership boundaries, and dependencies. As design solutions, however, they require near-decomposability. Simon’s own limit and Parnas’s criteria-dependence mean that no general theory can identify the correct modules without substantive domain knowledge.1923

Legitimacy transfers least

The evidence is especially consistent here:

  • proof of work establishes resource-weighted chain selection, not political equality;36
  • IETF rough consensus addresses technical objections, not the standing of a bounded citizenry;7
  • durable commons arrangements may remain unfair;4
  • deliberative agreement may be distorted by structural inequality;37
  • the DAO fork required social rule choice outside protocol execution;33
  • public ledgers did not establish carbon-credit quality;30
  • successful human computation does not create a legitimate collective agent.38

Operational coordination can be necessary for legitimate governance, but it is not sufficient for it.

6. Robust and superficial analogies

Robust, with conditions

  1. Dependency identification. Shared resources, prerequisites, simultaneity, and producer–consumer relations recur across organizations and technical systems. This is reliable as diagnosis, not yet as causal theory.

  2. Near-decomposability. Modular treatment can transfer where interactions within components are substantially stronger than interactions across components. The condition must be established rather than assumed.

  3. Assumption specification. Fault, timing, membership, information, and validity assumptions should be made explicit in both technical and institutional design.

  4. Threshold and critical-mass behavior. Distinct formal collective-action models can produce discontinuous contribution dynamics. This is a real convergence within that tradition, not evidence of a universal mechanism.

  5. Partitioned authority with exit. Debian’s jurisdictional technical committee resembles polycentric governance’s interdependent centers with qualified independence. This is a plausible structural analogy, not a tested equivalence.

  6. The operational–normative gap. Reliability may diverge from safety, durability from fairness, and agreement from free consent. This is one of the strongest recurring cross-tradition patterns—although it may reflect a general problem of social measurement rather than a distinctive property of coordination.

Superficial or failed

  1. Protocol consensus equals political consensus. Binary agreement among processes, IETF objection-handling, and democratic consent share a word but not an actor model, success criterion, or source of authority.

  2. Proof of work equals democratic voting. “One-CPU-one-vote” denotes resource-weighted chain selection under an honest-resource-majority assumption, not equal citizenship.36

  3. Decentralized architecture equals decentralized power. A study of Compound, Uniswap, and ENS found that 8, 11, and 18 delegates respectively could exceed half of voting power, with less than 10% of tokens participating in Compound and Uniswap votes during the observed period.39 This should not be generalized to all DAOs.

  4. Transparency equals truth. A ledger may accurately record tokens whose underlying claims are weak. The Toucan episode demonstrates the distinction without refuting ReFi as a category.30

  5. Byzantine faults equal political conflict. Arbitrary process behavior is not a model of ideology, distributive conflict, power, agenda control, or hegemonic discourse.

  6. Collective intelligence equals collective agency. A latent performance factor is not a theory of representation, intention, legitimacy, or answerability.

  7. Military command equals emergency management. Dynes argued that a military analogy mischaracterized disaster settings as social chaos to be rectified through command and control. His alternative emphasized continuity, coordination, and cooperation across established and emergent social units.8

  8. High-reliability organization as a universal template. The available critical review reports that the field lacks objective criteria for identifying its own cases and sometimes conflates reliability with safety.9

  9. AI-agent autonomy equals accountable actorhood. Contemporary multi-agent systems assign roles, messages, tools, and termination conditions through human-designed orchestration. MAST’s taxonomy identified recurrent system-design, inter-agent-alignment, and verification failures; its large corpus was LLM-annotated, while 21 traces received three-way expert annotation. A targeted intervention reportedly improved task success by 9.4%, showing that some orchestration failures may be tractable.40 None of this establishes moral or legal agency.

7. Precise and vague uses of “coordination”

Register Examples Appropriate interpretation
Precise technical Coordination equilibria; crash-fault consensus; CAP under an asynchronous model; finite-horizon DEC-POMDP complexity; Raft; IETF rough consensus; EIP-779’s state change Terms with explicit objects, assumptions, and success criteria
Precise but bounded Ostrom’s design principles for community commons; ICS structuring mechanisms; MAST’s failure taxonomy Useful only with population, scale, and evidence limits attached
Aspirational Better policy coordination, collective impact, mission coordination Names a desired state without identifying the causal mechanism
Identity-forming ReFi, DAO, network-state, and some HRO usage Helps constitute a community or project identity; analytically underdetermined
Metaphorical Human brains as processors, organizations as computers, swarms Potentially useful as a design heuristic, risky as an ontology
Ideological Coordination as a warrant for sovereignty, national unity, or presumed technological decentralization Performs political work while borrowing the appearance of technical neutrality

Two cautions follow.

First, criticism of hype should not erase precise practices inside the criticized communities. Debian’s constitution, RFC 7282, Bitcoin’s chain-selection rule, and EIP-779 are exact institutional or technical artifacts.

Second, precision inside a subsystem does not validate the claims made for the whole arrangement. A precise voting contract says nothing by itself about fair representation; a precise consensus protocol does not confer political legitimacy; and a transparent asset ledger does not verify the asset.

8. Which umbrella concept is least misleading?

No candidate covers the whole terrain without distortion.

Candidate What it captures What it loses
Coordination, game-theoretic sense Equilibrium selection and conventions Most conflict-of-interest, governance, mobilization, and institutional cases
Coordination, interdependence sense The operational layer across nearly every tradition Risks becoming a sponge; says little about goals, legitimacy, power, or agency
Coordinated action Restores actors to the analysis Leaves unresolved whether the relevant unit is an actor, task, organization, goal, or decision center
Collective action systems Contribution, incentives, public goods, and some commons Misses machine-only coordination and can import an overly narrow free-riding model
Collective agency Joint intention and answerability Describes a demanding achievement that many coordinated systems do not possess
Organized action Firms, agencies, projects, and durable roles Understates markets, emergent disaster organizations, machine systems, and spontaneous adjustment
Institutional design Rules, authority, sanctions, and revision Understates operational synchronization and machine substrates
Governance Authority, legitimacy, conflict, and rule-making Overloaded in its own right and too broad for technical coordination
Sociotechnical systems Human–machine hybrids Neutral but highly inclusive; the evidence base lacks the core CSCW and articulation-work literature needed to assess it fairly
Adaptive collective systems Feedback, learning, and revision Vague, and the cybernetics and systems-theory literature was absent from the evidence base
Layered composite Preserves distinctions among direction, operations, contribution, knowledge, allocation, governance, adaptation, and agency A checklist rather than a theory; composition remains untested

The least distorting solution is not a new master term. It is a two-level vocabulary:

  1. use coordination narrowly for the operational management of consequential interdependence; and
  2. identify explicitly which additional problem classes are present—goal formation, contribution, aggregation, allocation, governance, conflict, mobilization, adaptation, legitimacy, agency, and accountability.

This sacrifices terminological elegance for analytical clarity.

9. A working definition and its boundary

A useful working definition is:

Coordination is the structuring and ongoing adjustment of interdependent activities, decisions, or resource uses so that specified joint constraints or performance criteria are sufficiently satisfied over time.

This is a proposal, not a discovered essence.

It includes:

  • hierarchical and non-hierarchical adjustment;
  • cooperation and strategic interaction;
  • human, machine, and hybrid systems;
  • voluntary and involuntary arrangements;
  • systems with or without shared goals;
  • systems with or without a central coordinator.

It excludes:

  1. mere co-presence;
  2. independent parallel action;
  3. shared identity or a slogan without consequentially interdependent action;
  4. technical connectivity by itself;
  5. output aggregation where contributors do not interact consequentially;
  6. goal agreement considered apart from implementation dependencies.

Successful coordination does not entail:

  • legitimate authority;
  • justice or fairness;
  • collective intelligence;
  • collective agency;
  • decentralized power;
  • safety;
  • trustworthy underlying assets;
  • or accountability.

Nor is coordination excluded by coercion, illegitimacy, absence of shared goals, absence of a central coordinator, or absence of collective agency.

The unresolved goal-formation boundary

The most consequential definitional choice is whether goal formation belongs inside coordination.

There are three defensible positions:

  • Inside: Malone and Crowston include goal selection and decomposition among coordination processes and argue that some evaluative criterion is implied by the concept.1
  • Identical at one stage: Castañer and Oliveira propose calling the joint determination of common goals “coordination” and their implementation “cooperation.”24
  • Outside: Technical systems generally take objectives as inputs, while mission policy and organizing treat direction-setting as a distinct political process.

For analytical clarity, this report places goal formation outside operational coordination and treats it as a separate problem class. This is a deliberate boundary decision, not a finding established by the literature. It keeps the definition falsifiable and prevents successful dependency management from being mistaken for the authorization of the goal being served.

The decisive counterexample would be a case in which the same mechanisms that manage operational dependencies also demonstrably produce an authorized collective goal. No such case was established in the evidence base.

10. Implications for design and practice

10.1 Start with a problem profile, not a coordination solution

Before selecting hierarchy, markets, protocols, voting, deliberation, or networks, specify:

  • Object: What exactly is interdependent?
  • Actors: People, organizations, processes, tasks, or decision centers?
  • Objective: Given, negotiated, emergent, or imposed?
  • Authority: Who may decide and revise?
  • Incentives: Are contribution and compliance aligned?
  • Information: Who knows what, and when?
  • Conflict: Is disagreement informational, distributive, normative, or strategic?
  • Power: Who owns resources, sets agendas, controls defaults, or supplies indispensable labor?
  • Legitimacy: Which affected population must accept the arrangement?
  • Failure: What must never happen, and what degradation is tolerable?
  • Adaptation: Who monitors, learns, and changes the arrangement?
  • Exit: Who can leave, and at what cost?
  • Accountability: Who must explain and answer for outcomes?

Many failed analogies arise because a mechanism suitable for one profile is transferred to another without carrying its assumptions.

10.2 Do not ask one mechanism to solve a different class of problem

Examples include:

  • using token incentives to substitute for legitimate representation;
  • using transparency to substitute for verification;
  • using protocol consensus to substitute for political consent;
  • using communication to address distributive conflict;
  • using open participation to infer decentralized power;
  • using operational reliability to infer safety;
  • or using aggregation to infer collective agency.

A design should state which classes it addresses and which it leaves unresolved.

10.3 Treat mechanism existence and effectiveness separately

The existence of maintainers, committees, appeals, mission managers, sanctions, or deliberative procedures does not establish that they caused successful outcomes or outperform alternatives. This distinction is especially important in open-source governance, cross-sector collaboration, mission policy, and AI-agent systems, where design documentation is often richer than comparative outcome evidence.

10.4 Design for the operational–normative gap

Every arrangement should distinguish at least two scorecards:

  • Operational: Did the system synchronize, allocate, produce, recover, or decide?
  • Normative: Was the goal authorized, the process legitimate, the distribution fair, and responsibility assignable?

Success on the first scorecard should never be silently counted as success on the second.

10.5 Preserve revisability without romanticizing exit

Versioning, rollback, appeals, dissenting implementations, and forks are valuable. But reversibility is uneven. Technical state may be restored while losses, environmental harms, institutional distrust, or political exclusion remain. Similarly, formal exit may be available while switching costs make it impractical.

10.6 A useful experimental agenda

Several evidence-supported experiments could test the proposed structure:

  1. Matched mechanism transfer: Operationalize the same dependency, mechanism, and success criterion in two materially different domains—for example, a software project and a humanitarian network.
  2. Goal-contestation treatment: Hold the operational dependency constant while varying whether goals are given, negotiated, or imposed.
  3. Authority-concentration study: Compare when concentrated maintainership improves coherence and when it creates fragility or capture.
  4. Operational–normative dual measurement: Measure task performance and legitimacy separately in the same intervention.
  5. Interface validity test: Examine whether near-decomposability predicts when modular governance or technical interfaces succeed.
  6. Exit-distribution audit: Measure who can realistically fork, resign, or switch and how costs are distributed.
  7. Layer-composition test: In a mission-oriented initiative, test whether solving direction, contribution, allocation, governance, and operational coordination separately produces a coherent whole.

These are hypotheses for experimentation, not established prescriptions.

11. Important uncertainties and unresolved questions

The central conclusion is constrained by several major gaps.

A missing control condition

The evidence is rich in systems where goals are contested and authority is distributed. It contains no developed case of large-scale firm-internal coordination, military command and control, or state planning where goals are settled by ownership or command but coordination remains the binding problem.

The finding that operational coordination does not produce authorized goals, legitimacy, or collective agency may therefore be partly influenced by case selection. It should not be generalized to those omitted settings without further study.

Missing or thin literatures

  • Computer-supported cooperative work and articulation work—the principal sociotechnical rival to dependency-centered coordination theory—were absent.
  • Cybernetics and systems theory were absent, limiting assessment of “adaptive collective systems.”
  • Primary mobilization theory was absent; social movements and organizing were represented only thinly.
  • High-reliability organization theory was reached mainly through one critic rather than through its foundational texts.
  • The positive case for deliberative governance was available, but comparative evaluations of mini-publics were not.
  • ReFi evidence centered on one carbon-tokenization episode.
  • DAO evidence did not cover the full range of bicameral, reputation-based, identity-based, or conviction-voting designs.
  • Open-source evidence did not adequately cover moderation, documentation, release engineering, and community-care labor.
  • AI-agent evidence was benchmark- and framework-heavy, with limited evidence from consequential production deployments.

Questions that remain open

  1. Can any mechanism be shown to transfer across traditions under a matched operationalization?
  2. Is some collective action better modeled as assurance or coordination rather than as a Prisoner’s Dilemma?
  3. Is goal formation constitutive of coordination or analytically prior to it?
  4. Do accountability, predictability, and common understanding have discriminating causal roles across domains?
  5. Do separately solved problem classes compose into a functioning collective system?
  6. Is the operational–normative gap distinctive to coordination or a general property of social evaluation?
  7. When does concentrated authority create coherence, and when does it become capture or fragility?
  8. Who is the relevant constituency for legitimacy in standards, open-source projects, DAOs, and mission initiatives?
  9. When is a fork a meaningful exit institution rather than a formally available but unequal option?
  10. Is collective agency an emergent capacity, a legal attribution, a moral status, or a performance threshold?
  11. Can family resemblance be made falsifiable?
  12. How should power be represented when topology, ownership, agenda control, decision rights, labor dependence, and control of defaults diverge?
  13. Can AI agents become accountable actors, or only components within human responsibility structures?
  14. Would the central conclusions survive study of systems with settled goals and commanding authority?

Conclusion

Coordination is best understood neither as a grand unified phenomenon nor as a meaningless buzzword. It is a recurring problem situation: consequential interdependence among distinguishable loci of action. That situation invites a common analytical question—what must be made compatible, under whose criterion, and under what assumptions?—but not a common answer.

The traditions divide because they are solving different problems:

  • fitting tasks together;
  • eliciting contribution;
  • maintaining machine state;
  • allocating scarce resources;
  • forming goals;
  • authorizing rules;
  • managing conflict;
  • governing commons;
  • aggregating judgment;
  • adapting through feedback;
  • mobilizing power;
  • and attributing agency and responsibility.

These problems can coexist, and each may require coordination in the broad operational sense. But operational coordination does not absorb or settle the others.

The most portable structure is therefore modest but useful:

Name the interdependence. Name the actors. Name the assumptions. Name the criterion. Name who has authority to choose it. Then state which neighboring problems remain unsolved.

That approach preserves the real insights of coordination theory while preventing synchronization from being mistaken for cooperation, organization, intelligence, legitimacy, justice, governance, or collective agency.

Footnotes

  1. Thomas W. Malone and Kevin Crowston, “The Interdisciplinary Study of Coordination,” ACM Computing Surveys 26, no. 1 (1994): 87–119, doi:10.1145/174666.174668. ↩ ↩2 ↩3

  2. Oliver E. Williamson, “Transaction Cost Economics: The Natural Progression,” Nobel Prize Lecture, December 8, 2009, Nobel Prize; Friedrich A. Hayek, “The Use of Knowledge in Society,” American Economic Review 35, no. 4 (1945): 519–530, Library of Economics and Liberty. ↩

  3. Elinor Ostrom, “Collective Action and the Evolution of Social Norms,” Journal of Economic Perspectives 14, no. 3 (2000): 137–158, doi:10.1257/jep.14.3.137. The inspected representation was the author’s 1999 working-paper version rather than the journal version. ↩

  4. Elinor Ostrom, “Beyond Markets and States: Polycentric Governance of Complex Economic Systems,” American Economic Review 100, no. 3 (2010): 641–672, doi:10.1257/aer.100.3.641; Arun Agrawal, “Common Property Institutions and Sustainable Governance of Resources,” World Development 29, no. 10 (2001): 1649–1672, doi:10.1016/S0305-750X(01)00063-8. ↩ ↩2 ↩3

  5. Chris Ansell and Alison Gash, “Collaborative Governance in Theory and Practice,” Journal of Public Administration Research and Theory 18, no. 4 (2008): 543–571, doi:10.1093/jopart/mum032; Keith G. Provan and Patrick Kenis, “Modes of Network Governance: Structure, Management, and Effectiveness,” Journal of Public Administration Research and Theory 18, no. 2 (2008): 229–252, doi:10.1093/jopart/mum015. ↩

  6. Michael J. Fischer, Nancy A. Lynch, and Michael S. Paterson, “Impossibility of Distributed Consensus with One Faulty Process,” Journal of the ACM 32, no. 2 (1985): 374–382, doi:10.1145/3149.214121; Seth Gilbert and Nancy Lynch, “Brewer’s Conjecture and the Feasibility of Consistent, Available, Partition-Tolerant Web Services,” SIGACT News 33, no. 2 (2002): 51–59, doi:10.1145/564585.564601; Daniel S. Bernstein et al., “The Complexity of Decentralized Control of Markov Decision Processes,” Mathematics of Operations Research 27, no. 4 (2002): 819–840, doi:10.1287/moor.27.4.819.297. The CAP statement is scoped to the asynchronous model; the DEC-POMDP result cited here concerns finite-horizon problems. ↩ ↩2 ↩3

  7. Pete Resnick, “RFC 7282: On Consensus and Humming in the IETF,” June 2014, doi:10.17487/RFC7282; Debian Project, “Constitution for the Debian Project, Version 1.9,” March 26, 2022, https://www.debian.org/devel/constitution; Linux Kernel Documentation, “Feature and Driver Maintainers,” https://kernel.org/doc/html/next/maintainer/feature-and-driver-maintainers.html. ↩ ↩2 ↩3

  8. Russell R. Dynes, “Community Emergency Planning: False Assumptions and Inappropriate Analogies,” Disaster Research Center Preliminary Paper 145 (1990), later published in International Journal of Mass Emergencies and Disasters 12, no. 2 (1994): 141–158, doi:10.1177/028072709401200201; Gregory A. Bigley and Karlene H. Roberts, “The Incident Command System: High-Reliability Organizing for Complex and Volatile Task Environments,” Academy of Management Journal 44, no. 6 (2001): 1281–1299, doi:10.5465/3069401. Quotations from Dynes in the evidence base came from the 1990 preliminary paper, not the published version. ↩ ↩2

  9. Andrew Hopkins, The Problem of Defining High Reliability Organisations, Working Paper 51, National Research Centre for OHS Regulation, Australian National University, 2007, PDF. The foundational HRO sources were not directly inspected, so this conclusion rests principally on Hopkins’s critical review and his quotations from proponents. ↩ ↩2

  10. Anita Williams Woolley et al., “Evidence for a Collective Intelligence Factor in the Performance of Human Groups,” Science 330, no. 6004 (2010): 686–688, doi:10.1126/science.1193147; Christoph Riedl et al., “Quantifying Collective Intelligence in Human Groups,” PNAS 118, no. 21 (2021), doi:10.1073/pnas.2005737118; Riedl et al., correction, PNAS 119, no. 19 (2022): e2204380119, doi:10.1073/pnas.2204380119; Timothy C. Bates and Shivani Gupta, “Smart Groups of Smart People,” Intelligence 60 (2017): 46–56, doi:10.1016/j.intell.2016.11.004. The 2010 and 2021 studies share senior authors and should be treated as one research program rather than independent replication. ↩

  11. John S. Dryzek et al., “The Crisis of Democracy and the Science of Deliberation,” Science 363, no. 6432 (2019): 1144–1146, doi:10.1126/science.aaw2694. The inspected artifact was an accepted manuscript, not the version of record. ↩ ↩2

  12. Marshall Ganz, “People, Power and Change,” organizing framework, PDF; Jeff Goodwin and James M. Jasper, “Caught in a Winding, Snarling Vine: The Structural Bias of Political Process Theory,” Sociological Forum 14, no. 1 (1999): 27–54, doi:10.1023/A:1021684610881. ↩

  13. Philippe Larrue, The Design and Implementation of Mission-Oriented Innovation Policies, OECD Science, Technology and Industry Policy Papers no. 100 (2021), doi:10.1787/3f6c76a4-en; Larissa Calancie et al., “Consolidated Framework for Collaboration Research Derived from a Systematic Review,” PLOS ONE 16, no. 1 (2021): e0244501, doi:10.1371/journal.pone.0244501; European Commission, Interim Evaluation of the Horizon Europe Framework Programme for Research and Innovation (2021–2024), SWD(2025) 110 final. The Commission reported that missions were proceeding toward their goals despite cumbersome governance and incomplete monitoring; its account remains institutionally interested rather than an independent impact evaluation. ↩

  14. Satoshi Nakamoto, “Bitcoin: A Peer-to-Peer Electronic Cash System” (2008), https://bitcoin.org/bitcoin.pdf; Robin Fritsch, Marino Müller, and Roger Wattenhofer, “Analyzing Voting Power in Decentralized Governance: Who Controls DAOs?” (2022), doi:10.48550/arXiv.2204.01176; Qingyun Wu et al., “AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation” (2023), doi:10.48550/arXiv.2308.08155. ↩

  15. Kevin Crowston, Joseph Rubleske, and James Howison, “Coordination Theory: A Ten-Year Retrospective,” author draft, November 11, 2005, PDF. ↩

  16. Pamela E. Oliver, “Formal Models of Collective Action,” Annual Review of Sociology 19 (1993): 271–300, doi:10.1146/annurev.so.19.080193.001415. ↩ ↩2

  17. Gerardo A. Okhuysen and Beth A. Bechky, “Coordination in Organizations: An Integrative Perspective,” Academy of Management Annals 3, no. 1 (2009): 463–502, doi:10.5465/19416520903047533. Only the authors’ abstract reconstructed from OpenAlex metadata was inspected; the article body was not available. ↩

  18. Michael Cox, Gwen Arnold, and Sergio Villamayor-Tomás, “A Review of Design Principles for Community-Based Natural Resource Management,” Ecology and Society 15, no. 4 (2010): 38, doi:10.5751/ES-03704-150438. The authors recommend a probabilistic rather than deterministic interpretation and do not establish transfer across scales. ↩ ↩2

  19. Herbert A. Simon, “The Architecture of Complexity,” Proceedings of the American Philosophical Society 106, no. 6 (1962): 467–482. ↩ ↩2 ↩3

  20. Provan and Kenis, “Modes of Network Governance,” doi:10.1093/jopart/mum015. ↩

  21. Diego Ongaro and John Ousterhout, “In Search of an Understandable Consensus Algorithm,” USENIX Annual Technical Conference (2014), https://raft.github.io/raft.pdf. ↩ ↩2

  22. Leonita Agustina Setyawati, Herdis Herdiansyah, and Triarko Nurlambang, “The Dynamics of Coordination in Humanitarian Clusters in Indonesia,” ASEAN Natural Disaster Mitigation and Education Journal 4, no. 1 (2026): 48–68, doi:10.61511/andmej.v4i1.2026.3699. This was a single 26-interview case study in a low-authority venue and is used only as a qualified application. ↩

  23. D. L. Parnas, “On the Criteria To Be Used in Decomposing Systems into Modules,” Communications of the ACM 15, no. 12 (1972), doi:10.1145/361598.361623. ↩ ↩2

  24. Xavier Castañer and Nuno Oliveira, “Collaboration, Coordination, and Cooperation Among Organizations,” Journal of Management 46, no. 6 (2020): 965–1001, doi:10.1177/0149206320901565. Only the repository abstract and the authors’ public summary were inspected, not the article body. ↩ ↩2 ↩3

  25. Mark Stephan, Graham Marshall, and Michael McGinnis, “An Introduction to Polycentricity and Governance,” in Governing Complexity (Cambridge University Press, 2019), 21–44, author-hosted PDF. ↩ ↩2

  26. Philipp Trein et al., “Policy Coordination and Integration: A Research Agenda,” accepted preprint for Public Administration Review, University of Lausanne repository. ↩ ↩2

  27. Calancie et al., “Consolidated Framework for Collaboration Research,” doi:10.1371/journal.pone.0244501. ↩ ↩2

  28. Goodwin and Jasper, “Caught in a Winding, Snarling Vine,” doi:10.1023/A:1021684610881. ↩

  29. Michael Rescorla, “Convention,” Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/convention/; Garrett Cullity, “The Free Rider Problem,” Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/free-rider/. The relevant classic primary texts were not obtained, so the game-theoretic boundary is not primary-sourced here. ↩

  30. CarbonPlan, “Comment on the Energy and Climate Implications of Digital Assets,” May 9, 2022, https://files.carbonplan.org/OSTP-Digital-Assets-Comment-Letter-05-09-2022.pdf; Verra, “Verra Addresses Crypto Instruments and Tokens,” May 25, 2022, https://verra.org/verra-addresses-crypto-instruments-and-tokens/; Ethereum.org, “Regenerative Finance,” https://ethereum.org/refi/. These sources establish the Toucan/BCT episode, not a category-level conclusion about ReFi. ↩ ↩2 ↩3

  31. Balaji Srinivasan, The Network State: How to Start a New Country (2022), https://book.thenetworkstate.com/tns.pdf. This is a programmatic source and supports classification of the proposal, not evaluation of its efficacy. ↩

  32. United States Holocaust Memorial Museum, “Gleichschaltung: Coordinating the Nazi State,” Holocaust Encyclopedia, https://encyclopedia.ushmm.org/content/en/article/gleichschaltung-coordinating-the-nazi-state. ↩

  33. Casey Detrio, “EIP-779: Hardfork Meta: DAO Fork,” Ethereum Improvement Proposals, https://eips.ethereum.org/EIPS/eip-779. The EIP specifies that the fork changed no protocol rules and instead implemented an irregular state change at block 1,920,000. ↩ ↩2

  34. Abraham Sesshu Roth, “Shared Agency,” Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/shared-agency/. ↩

  35. Samer Faraj and Yan Xiao, “Coordination in Fast-Response Organizations,” Management Science 52, no. 8 (2006): 1155–1169, doi:10.1287/mnsc.1060.0526. The study was single-site and did not establish causality. ↩

  36. Nakamoto, “Bitcoin,” https://bitcoin.org/bitcoin.pdf. ↩ ↩2

  37. Iris Marion Young, “Activist Challenges to Deliberative Democracy,” Political Theory 29, no. 5 (2001): 670–690, doi:10.1177/0090591701029005004. ↩

  38. Luis von Ahn, Human Computation, Carnegie Mellon University doctoral thesis CMU-CS-05-193 (2005), https://csd.cs.cmu.edu/sites/default/files/phd-thesis/CMU-CS-05-193.pdf. ↩

  39. Fritsch, Müller, and Wattenhofer, “Analyzing Voting Power in Decentralized Governance,” doi:10.48550/arXiv.2204.01176. ↩

  40. Mert Cemri et al., “Why Do Multi-Agent LLM Systems Fail?” arXiv:2503.13657, version 3 (2025), doi:10.48550/arXiv.2503.13657; Kunlun Zhu et al., “MultiAgentBench: Evaluating the Collaboration and Competition of LLM Agents,” Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (2025): 8580–8622, doi:10.18653/v1/2025.acl-long.421. ↩