Atonement

An existing operator can carry a shared service while a new governing body is being organized; implemented constitutional powers can await the configuration needed to act. Selected Cosmos decisions make both relationships visible. Persistence Proposal 97 describes continuing relayer service and requests an extension during an anticipated operations-DAO transition. AtomOne’s public parameter responses return empty authority addresses; its v4.1.0 source and technical reference specify how those fields condition the named powers. Persistence supplies an operator’s account of provision, while the AtomOne readings identify a further configuration condition for authority.

Each decision also identifies who can fund a service, seek a response or revise an authority. Hub support delegates funding administration, a reporting proposal anticipates appeal to a supervisory authority, and AtomOne’s tested source excludes vetoes of decisions replacing the veto holder’s own address. These interventions reach different parts of a collective undertaking.

Allen, Berg and Davidson (2023) already connect sovereign decision rights to contracting, delegated powers and orderly exit. Knowledge-commons research and self-infrastructuring explain why technical cooperation includes continuing institutional work; Institutional Grammar distinguishes constituting an authority from regulating an actor’s conduct. Following who acts and what further decisions are required keeps collective authorization separate from service delivery.

IBC allows independently governed chains to verify cross-chain messages. The protocol needs someone to carry those messages; a commitment to a particular service level can additionally require maintained infrastructure and fault response. ICS-18 explicitly makes liveness depend on a correct, live relayer and leaves application-layer incentivization to surrounding arrangements (IBC 2023, “Desired Properties”). This division of work is part of the design. A protocol’s verification rules and a route’s continuing provision answer different questions.

The political question concerns the relationship among the actors doing that work. A Hub electorate can decide how to support Hub-side relaying. A relayer can undertake a service mandate. Neither decision automatically governs the counterparty chain or every organization supplying information and software. Cosmos’s commitment to independent chains makes the boundaries of those decisions consequential (Kwon and Buchman 2016). Authority over one component can coexist with dependence on actors supplying or controlling another. Reliance on relaying as a function does not establish dependence on one relayer when others can supply it.

The relevant relationship and any collective undertaking must therefore be specified. Hub voters, a route’s users and operators, an information-holding foundation, a shared-security relationship and a constitutional branch have different memberships and decision procedures. Sustaining cooperation need not mean keeping them under a single center. A branch can organize a different undertaking while retaining some operational relationships. What matters is which decisions the participants can make together, which remain separate and how the arrangements can be revised. A reporting demand addressed to a separate foundation establishes attempted influence, rather than demonstrating that the electorate and foundation share a reciprocal obligation.

The inquiry follows four questions: what is being decided; who can make that decision; what gives it practical or executable effect; and what related decisions remain elsewhere? These questions distinguish funding a service from constituting a body, and a demand addressed to another organization from authority to compel its response. They also allow an existing arrangement to count as cooperation, without requiring a visible repair or a new DAO.

The Governing Knowledge Commons approach already examines participants, resources, rules, histories, interactions and outcomes, including the continuation of a community (Murtazashvili et al. 2022, 110–113). Nabben’s self-infrastructuring likewise includes designing, operating, governing and maintaining technical and institutional arrangements (2023, 3–4). These accounts supply the starting point: cooperation includes maintaining the arrangements through which people act.

Institutional Grammar makes the comparison more precise. Frantz and Siddiki distinguish rules addressing actors' conduct from statements that constitute or modify entities, powers and institutional properties. Their grammar also specifies conditions under which a statement applies and constraints on its execution (2022, §2.1, Tables 2–3; §2.4). Establishing an office, supplying its decision procedure and directing an existing operator are consequently different institutional acts. The analysis uses these distinctions to compare the selected decisions.

Hirschman’s exit and voice direct attention to the organization a participant can leave and the authority to which a demand is addressed. Their interaction depends on institutional conditions: exit can support influence or obstruct collective adjustment, while continued membership alone cannot make voice effective (1970, 4–5, 18–19, 55). Selling a token, changing a service provider and creating a chain branch concern different objects of exit. Their costs and consequences cannot be inferred from the existence of a fork mechanism.

Frey and Schneider already distinguish speech with a binding procedural effect from expression intended to persuade, while treating the forms as complementary (2021, 3–4). The Cosmos comparison follows that distinction into the scope of the decision. A binding vote concerning chain state can leave another organization’s disclosure decision beyond its executable reach. A nonbinding demand can still exert political influence. Establishing that influence requires evidence of the organization’s response; the absence of a direct execution path does not make the demand useless.

Allen, Berg and Davidson place sovereignty within the economics of shared provision. Buying security can spare a chain the costs of maintaining its own validator set, while introducing contracting hazards between independently governed parties. Relationship-specific investments can make departure costly even when a chain retains the authority to leave (2023, 4–7, 11–13). Sovereignty therefore identifies a decision right; it does not establish the cost of exercising it. Their analysis directly addresses the problem of sustaining cooperation without absorbing consumer chains into the Hub.

Their proposed institutions also distinguish negotiating from authorizing and executing a change. Agreement-level committees would initially recommend changes for on-chain votes; later arrangements could delegate bounded powers while reserving other decisions and veto rights to token holders. Their onboarding template asks for service standards, implementation processes and termination consequences, alongside provisions for review, orderly exit and third-party arbitration (Allen et al. 2023, 20–27). Authority boundaries, institutional adaptation and implementation are already part of that account.

The documentary question here is how particular Cosmos decisions connect those elements. Relayer decisions authorize funding administration or propose an operator mandate; a demand for foundation reports attempts to influence a separate organization; a constitutional upgrade implements powers whose activation requires configured authorities. These records extend the objects of comparison beyond ICS agreements, while retaining the report’s distinction between independent governance and the arrangements for shared provision. They supply particular assignments and implementation conditions, rather than a new theory of contracting or evidence that the report’s recommendations succeeded or failed.

The selected episodes begin with a visible funding request, disclosure demand, security response, service migration or constitutional implementation. They are a documentary comparison, rather than a sample of routine operation across Cosmos. Each is read for the decision object, addressee, next required action and source-supported status. Their different functions and evidence prevent a test of whether formalization is necessary or sufficient for effective cooperation. Proposal texts establish requests and authorizations; source and documentation specify capabilities; public queries report returned state. The comparison reconstructs how the decisions connect without treating authorization as service delivery.

In December 2023, Cosmos Hub Proposal 862 requested 8,000 ATOM over three months to reimburse Hub-side relayer gas. It named relayer teams and a three-of-five multisignature administrator that would determine eligibility and supply fee grants. The retained record identifies a passed community-pool spending proposal to the named administrator (Cosmos Hub Governance 2023e). It authorizes funding for an existing activity. The administrator’s subsequent fee grants, actual disbursement and route coverage require separate records.

Persistence Proposal 97 makes the organizational distinction sharper. It requested 40,113.17 XPRT to extend CryptoCrew’s mandate for two months while an Infrastructure Operations DAO was being organized. The proposer represented its service as continuing proactively and specified ten counterparty chains, an uptime commitment and an alert-response target (Persistence Governance 2024, summary; details, opening paragraph, “Scope of service” and “SLAs offered”). This is an operator’s account and proposed service commitment, rather than independently observed uptime. Within that account, the existing operator carries the service while a different organizational arrangement is anticipated. An unfinished DAO therefore cannot be read as evidence that the underlying relayer function was unassigned.

Polkachu’s pinned gm-ibc README supplies an operator’s account of the work involved. It describes separate relayer infrastructure and the possibility that a broken channel remains unnoticed (Polkachu 2025). That concern explains why transaction fees and route monitoring need separate consideration. Paying a transaction does not establish that someone is watching the route, detecting faults and responding within a useful period.

ICS-29 fee middleware is a concrete alternative to external reimbursement. Its documented mechanism allowed fees to be escrowed and distributed to relayer addresses, including asynchronous payments by a module account or community pool. The pinned ibc-go changelog records the middleware’s removal in v10 (IBC-Go 2026a, 2026b). The design shows how part of financing can enter protocol-adjacent software. Route selection, adoption, service coverage and the source of the funds still require an account.

Where a deployed fee mechanism supplies financing, the remaining question is who chooses routes, monitors service and supplies the funds. Those functions should be followed separately. A route can combine automated fee distribution with an external monitoring agreement. The relevant question is how those assignments connect, including arrangements that work without a public dispute.

Two Hub votes make foundation reporting a governance object. Proposal 787 requested fuller Interchain Foundation financial transparency and passed in April 2023. Proposal 952 declared no confidence in its leadership, requested historical annual reports and passed in September 2024 (Cosmos Hub Governance 2023a; 2024). Their texts establish demands and signaling decisions. Allegations about conflicts, duties or financial management remain allegations.

The reporting question concerns the relation between a voting forum and a separate organization holding information. Proposal 952 also specifies a contingent route outside Hub voting: stakeholders would petition the foundation supervisory authority for an audit if reports were not delivered within 60 days. Its text connects a Hub demand to an intended appeal to a different authority (Cosmos Hub Governance 2024, §3.4). The proposal establishes neither that a petition was filed nor that it would meet the supervisory authority’s requirements. The proposals do not establish that the requested reports were produced, that a particular demand caused publication or that the votes themselves supplied an enforceable reporting obligation. Answering those questions would require organizational records and a supported account of the relevant authority.

Liquid-staking maintenance provides a supporting case about information entering a decision. Proposal 790 signaled support for the Liquid Staking Module in May 2023; Proposal 821 specified the v12 upgrade plan and passed in September (Cosmos Hub Governance 2023b, 2023c). In October 2024, All in Bits alleged serious contributor-provenance and disclosure problems. An Informal Systems maintainer subsequently described a provenance and maintenance concern, announced audits and proposed replacement architecture (AiB 2024; Poffenberger 2024). These are an interested party’s allegations and a maintainer’s response, rather than findings about what every organization knew when the earlier votes occurred.

The later repair decisions addressed the software through another set of actors. Gaia v22’s release notes attribute fixes to the OtterSec and Zellic audits and state that none of the identified issues was critical; Proposal 1000 authorized the v24 plan, including a refactored x/liquid module (Gaia 2025a; Cosmos Hub Governance 2025). The episode connects a governance authorization to information produced through maintenance and audit arrangements. It does not demonstrate that the earlier vote assessed complete information or that the later repair proves the original allegations. The detailed query and incident record is distinguished in the source appendix.

Across these episodes, governance depended on information produced by organizations around the vote: reports, contributor histories, audit findings and maintenance assessments. A vote can authorize a response to that information. Its existence does not show that the information was complete or that every demand was satisfied.

Interchain Security placed validator attachment and consumer-chain lifecycle obligations into a shared regime. Formal integration changed what the Hub and its validators undertook to provide. It also created responsibilities for operating additional infrastructure, coordinating changes and responding to incidents.

In February 2026, the Hub team’s deprecation notice attributed its proposed withdrawal to demand, revenue, validator costs and technical debt. These are the team’s stated reasons, not independently estimated effects. The notice proposed blocking new consumer creation before removing the provider module after offboarding. Gaia v27’s release notes documented the first step (Cosmos Hub Team 2026; Gaia 2026a).

The available v28 changelog states that v28.0.0 removed the provider module and left stubs for its query interfaces. Both queried Hub endpoints reported v28.3.0. The old launched-consumer query at the directory gateway returned HTTP 501 with Not Implemented (Gaia 2026b; Cosmos Hub Queries 2026). That failed query cannot supply a current count of consumer chains or establish how each chain completed offboarding. The July observation of two launched entries remains a dated observation in the earlier packet, rather than a current description.

Retirement makes the report’s distinction between a continuing relationship and its termination concrete. Allen, Berg and Davidson call for explicit exit conditions and orderly transitions, including the consequences for each party (2023, 24–27). The notice and release changes identify steps in retiring the shared regime. They do not establish the terms adopted by each consumer chain, the completion of each transition or whether the recommended governance bodies and reviews were used. The sequence cannot evaluate those recommendations without that history. It also supplies no comparison showing that sovereign operation is cheaper.

AtomOne provides a direct comparison between written authorization and a configured power. GovGen’s own account connects its initial distribution to NO and NO_WITH_VETO voters on Cosmos Hub Proposal 848, whose voting ended in November 2023. That chronology places the branch before the 2024 LSM disclosure and no-confidence vote. Those later controversies cannot be treated as its initiating cause (Cosmos Hub Governance 2023d; GovGen 2024).

AtomOne Proposal 2 ratified a constitution in January 2025. The pinned text specified Steering and Oversight DAOs while saying that neither existed at genesis. It also distinguished special-purpose DAOs and their initial arrangements (AtomOne Governance 2025; AtomOne 2024, Article 2 §4.b). The branch thus specified a new distribution of powers, rather than merely leaving an earlier collective. Its constitution, software implementation, configured authorities and subsequent practice are distinct parts of that redesign.

The October 2026 proposal response returns Proposal 21 as passed. The directory gateway and a separately listed Cogwheel endpoint both reported AtomOne v4.1.0. Both also returned the x/coredaos parameter endpoint with empty steering_dao_address and oversight_dao_address fields (AtomOne Queries 2026). The REST interface returned the module’s parameter object with those two authority fields empty.

AtomOne’s current technical reference explains the significance of that configuration. Its core-DAO module supplies endorsement, annotation, extension and veto hooks, but leaves treasury funds, membership and internal governance to separate arrangements. The reference states that empty addresses disable the associated powers until governance configures them. It also distinguishes a veto’s executable effect from the constitutional reason for using it: the protocol does not itself establish why a veto was cast (AtomOne 2026a, §§4.1–4.4).

The exact v4.1.0 source also limits the specified authority. It rejects an Oversight DAO veto of a proposal changing or disabling that DAO’s address, including proposals nested inside authorization wrappers. Selected tests of the exact v4.1.0 source passed for direct, wrapped and double-wrapped replacement. The restriction supplies a specific route for changing an authorized body; it does not judge the constitutional reasons for that body’s decisions. At queried height 10,863,655, both endpoints still returned empty authority addresses. The tested capability remains distinct from a configured authority and an observed institutional practice (AtomOne 2026b; AtomOne Queries 2026).

The technical reference and parameter readings support a precise inference: the two specified authorities were not configured at the returned states. Governance must configure the addresses before those hooks can act; membership and constitutional judgment require their own arrangements. Informal bodies, prospective appointments and future exercises of power require separate evidence. Later configuration would update the dated finding, with institutional effectiveness requiring evidence of how the powers were exercised.

The comparison identifies different objects of collective decision:

The named relayers in Proposal 862 rule out reading its funding request as evidence of an unassigned relayer role. Persistence likewise reports provision while a new DAO is anticipated, so an absent future administrator does not establish absent service. That operator account does not independently verify continued operation. AtomOne’s released module, by contrast, cannot establish active executable oversight while its required authority addresses are empty; the finding concerns those hooks, rather than all informal oversight. Organizational status takes its meaning from the function and conditions at issue.

The same distinction clarifies sovereignty and cooperation. Independent decision rights can support a service through negotiated mandates, shared modules, funding arrangements or a new constitution. Each arrangement places some decisions within a common procedure and leaves others with separate actors. Its adequacy depends on the service, the obligations undertaken and the ways those obligations can change. A public dispute identifies a point of contestation, while a causal explanation must establish why the arrangement produced the contested result.

Cooperation can involve maintaining a mandate, retiring a relationship or constituting different powers. Which outcome participants seek, and who bears its consequences, remain substantive questions. Keeping one center intact is not a general criterion of successful cooperation.

The behavioral boundary is observable exit and voice when those concepts are invoked; no psychological measurement is attempted. Funding proposals, public demands and branching identify available forms of action. They do not estimate participants' preferences, the relative attractiveness of departure or the causal effect of sovereignty on those choices.

An empirical test of the report’s recommendations would first establish which ICS relationships adopted agreement-level bodies, delegated powers, review triggers and exit terms. It would then follow their use, including decisions, ratification, implementation and transition costs. A later retirement cannot stand in for that adoption and operation history.

Across the other cases, a stronger comparison would reconstruct a service’s obligations, funding and authority before and after a change. Relayer analysis would follow actual fee distributions, route coverage and incident response. Reporting analysis would establish what was published, who could require it and how organizations responded. Constitutional analysis would distinguish configuration, appointments, decisions and contested exercises of power. Successful routine operation should enter that comparison alongside conspicuous disputes.

Attributing an intervention to a gap in responsibility requires showing the relevant gap and how actors addressed it. The explanation must be tested against alternatives such as a temporary implementation failure, ordinary specialization, changes in demand or an already effective institutional arrangement. A function located outside one module is not automatically ungoverned or inefficient.

This paper reconstructs selected interventions and compares their assignments. It does not establish a general theory of institutional completion or a comparative performance ranking. October query results describe particular public endpoints and returned heights; matching fields across two named providers do not certify every validator’s binary or the independence of their infrastructure. Proposal passage, available source code, deployed configuration and observed institutional exercise remain separate.

Sovereign chains can retain independent control while undertaking services whose operation crosses their decision boundaries. The selected Cosmos interventions identify funding, demands addressed to separate organizations, maintenance decisions, shared-security rules and constitutional redesign. Some organize a shared service; others seek to influence an actor outside the deciding collective. Each decision has an object and an addressee; its effect depends on how it connects to the other decisions the service requires.

That comparison avoids treating formalization as a uniform increase in governing capacity. A service can have an existing carrier before a new body is formed, while an implemented constitutional power can await the configuration needed to act. The records identify delegation, proposed escalation and limits on an authority’s control over its own replacement. Explaining their performance requires the next layer of evidence: actual provision, responses, implementation and contested exercise of power.

Historical proposal and specification findings use the retained May and July records. The October collection retains documents, query responses, the release archive, same-height responses and 63 selected native keeper/type test cases. The selected tests passed again during the final source review. A native test identifies an implemented rule; reported build identity and matching endpoint fields do not attest a validator binary or prove its state.

The ICF comparison relies on the proposal texts, whose passage establishes the Hub’s demand and proposed escalation route. It does not infer that the requested reports were published. Release notes support the stated characterization of audit fixes; inaccessible reports do not support exact finding counts. The retired provider query returned no current consumer population.

The two October Cosmos Hub endpoints reported Gaia v28.3.0 and returned matching x/liquid parameters (Cosmos Hub Queries 2026). The separate 2025 report on contributor cool-develope found no active vulnerability or malicious code attributable to that actor and described access and credential controls (Cosmos Incidents 2025). It concerns different contributions and does not resolve the earlier LSM allegations.

The complete November 2023 Allen–Berg–Davidson report, including its decision tree and onboarding template, supplies the closest institutional comparison. Two worked readings distinguish a spending authorization from constitutional authority. They are interpretive applications of Institutional Grammar, without a complete formal encoding or inter-coder reliability study. The comparison establishes documentary decision relationships; causal effects on cooperation, service performance and participants' exit costs remain unmeasured.