Summary

  • ICP-2 asked a new RIR to publish a budget tied to its activity plan and become financially independent through member support. That was a sensible entry test, but it did not test whether cash would remain legally accessible, whether a lawful body could approve spending, or whether core services could be financed through a governance emergency.
  • AFRINIC's own reports show the gap. It entered 2021 with substantial reserves and later described a healthy balance sheet, yet a court attachment froze all bank accounts for nearly three months and the organisation reported relying on stakeholder donations to meet commitments. Its later accounts reported strong finances alongside the absence of a quorate board, a chief executive and an approved annual budget.
  • A modern recognition standard should measure ring-fenced service runway, quorum and election continuity, litigation reserves, authority to spend during vacancies, tested transfer arrangements and access to protected records. Financial health is one layer of resilience, not a substitute for institutional design.

A healthy balance sheet can coexist with a disabled institution

Financial stability sounds like the least controversial RIR recognition criterion. A registry needs staff, connectivity, secure systems, offices, auditors and enough time to evaluate requests carefully. It should not depend on the political favor of a ministry or the recurring generosity of a sponsor. If its income is persistently below the cost of core registration services, every other promise becomes fragile. ICP-2 was right to require a credible budget and a path to member-funded independence.

The difficulty is that a balance sheet describes an entity at a date. Governance resilience describes what the entity can still do after authority, access or trust is disrupted. The two overlap, but they are not the same. Cash shown in an account may be restrained by a court. A reserve may be unrestricted in accounting terms yet unavailable because no valid board can authorize its use. A solvent association may have no lawful mechanism for filling enough vacancies to reach quorum.

A profitable registry may be unable to approve a budget, appoint a chief executive, defend several cases at once or share the records needed for temporary service support.

AFRINIC makes that distinction unusually visible. The organisation's published financial record before and during its governance crisis did not resemble the familiar story of a charity simply running out of money. Its 2020 annual report described a healthy financial position, a net surplus and growing reserves. Its 2021 annual report again described a very healthy balance sheet, record cash holdings and improved reserves. The same report said all AFRINIC bank accounts had been frozen from 23 July until 15 October 2021 under an attachment order, leaving the company unable to honor financial commitments and dependent on USD 504,000 in donations from stakeholders.

That is not a contradiction in the accounts. It is a contradiction in the old test. Accounting health measured the amount of value associated with AFRINIC. It did not measure whether the institution could reach that value under legal stress. The relevant failure was not necessarily insolvency. It was the conversion of liquid assets into operationally inaccessible assets at the moment they were needed.

The lesson is larger than one court order. Recognition standards should ask whether money and authority survive the same shock. A registry is not financially resilient merely because it owns enough assets. It is resilient when defined essential services can continue lawfully, independently and transparently across the plausible failure modes of the institution that owns them.

What ICP-2 actually required in 2001

The original ICP-2 criteria placed the funding model eighth in a list of ten essential conditions. The numbering was expressly said to be insignificant; all criteria mattered. The text expected a new RIR to be a not-for-profit association. It required a published budget related to a published activity plan and explicit support from the organisations backing the candidate. Initial grants and sponsorship were acceptable, but the registry had to be demonstrably independent and autonomous. Over time, it was expected to become financially supported entirely by its membership.

That formulation solved a specific entry problem. Africa and Latin America were moving toward their own registries, while incumbent RIRs still served networks in those regions. A candidate needed to show that it was more than an aspirational committee. It had to demonstrate a credible revenue base, a costed plan and enough community commitment to assume responsibility after incubation support ended. The test protected operators from being transferred into an institution that could not pay its staff or maintain the services promised in its application.

For that purpose, the criterion was proportionate. ICANN did not need to design the candidate's treasury rules or supervise its annual budget. It needed evidence that the new body could stand on its own. A published budget, member fees, start-up support and a transition plan could establish that threshold.

The 2001 text did not ask several questions that only become urgent after years of accumulated reliance. It did not distinguish unrestricted reserves from funds that remain usable during litigation. It did not require a separate operating entity or protected account for essential services. It did not specify who may approve emergency spending when a board falls below quorum. It did not price simultaneous court proceedings. It did not require funds for migration to an interim operator. It did not require an annual exercise proving that records, credentials and implementation knowledge could move without exposing confidential member data.

Those omissions are understandable in an entry document. They are no longer harmless if the same words are used to assess a mature RIR. A candidate's first budget answers, "Can this organisation start?" A resilience standard must answer, "Can the service continue when the organisation cannot act normally?" Financial independence remains necessary, but its entity changes. The money must protect the registry function, not merely the corporate shell.

AFRINIC's recognition file captured the promise, not the tail risk

The 2005 IANA evaluation of AFRINIC is a clear example of the original funding test in action. AFRINIC was a new institution completing a transition from incumbent RIR service. Its fee structure had membership categories linked to the quantity and type of resources held. Supporting organisations in several African countries agreed to carry parts of the first two years' costs. IANA recorded that AFRINIC was autonomous from those supporters and expected to assume full administrative and financial responsibility on its own membership revenue after incubation. The business plan was said to show the financial strength required for that handover.

Nothing in that assessment was obviously careless. The candidate had a revenue model, a cost plan, external support for the transition and a defined point at which support would end. It also had a distributed operating design: the recognition report described technical operations in South Africa, backup and disaster recovery in Egypt, training coordination in Ghana and incorporation in Mauritius. The other RIRs had participated in technical, legal, financial and administrative support. On the facts relevant to an entry decision, the plan looked sensible.

Yet an entry plan is an optimistic model by design. It models expected revenue, expected costs and a manageable transition. The rare event sits outside its center: a large attachment, a prolonged governance vacuum, a disputed election, an inability to constitute a board, a receiver appointed by a domestic court, or several of those conditions interacting. Tail risks appear implausible until an organisation with unique responsibilities experiences them.

The recognition file also illustrates why financial sufficiency cannot be detached from institutional authority. Membership fees arrive because a registry controls a unique service relationship within a region. That recurring income can make an RIR look safer than an ordinary association. But the same non-substitutability means resource holders cannot readily discipline poor governance by taking their business elsewhere. A predictable fee stream can continue even when members have weak practical leverage over the board, elections or remedies. Revenue persistence may therefore conceal institutional fragility instead of disproving it.

The right retrospective question is not whether IANA should have predicted every later dispute in 2005. No recognition review can forecast two decades of corporate history. The question is what the later experience teaches about the criterion. It teaches that a supported budget and future membership revenue are evidence of ordinary operating capacity. They are weak evidence about legal access to cash, continuity of lawful authority and the cost of transferring service under stress.

The 2020 figures looked like the success case

AFRINIC's financial position immediately before the bank restraint looked strong by conventional measures. The 2020 report stated that membership-fee revenue was about USD 5.53 million, down modestly from 2019, while operating costs fell to about USD 3.43 million. It reported a net surplus of about USD 2.14 million. Total reserves rose to about USD 8.04 million, strategic cash reserves to about USD 4.34 million and total cash holdings to about USD 9.25 million. Its reported liquidity ratio was 4.3 to 1.

The organisation's assessment of its 2016-2020 strategic plan was equally confident. A published strategic-plan assessment marked the financial-stability objective as met. It cited revenue growth, cost control, expanding cash holdings and a reserve target equal to two years of operating budget. On ordinary nonprofit finance measures, these are substantial cushions. A reviewer examining only year-end accounts could reasonably conclude that AFRINIC had room to absorb a revenue decline or an unexpected expense.

Those figures matter because they prevent an easy but misleading explanation of the later crisis. The institutional weakness was not proved by a thin cash balance. AFRINIC had accumulated the sort of reserve that governance codes often recommend. It had recurring membership income and positive operating results. The problem was that the reserve model assumed the organisation retained ordinary control of its banking, management and governing bodies.

A two-year reserve is useful against a gradual fall in income. It can fund a planned restructuring, a technology replacement, a recruitment delay or a temporary cost increase. It is less useful against an order that restrains all accounts. The accounting classification does not create a legal exception for payroll, registry security or reverse-DNS administration. Nor does a liquidity ratio answer who may direct funds if the governing body lacks quorum. The ratio divides one set of balance-sheet quantities by another. It does not test the validity of a signature.

This is why the phrase "rainy-day fund" is inadequate for an RIR. Rain arrives from the forecast; governance shocks can change who controls the umbrella. A resilience reserve must be designed around continuity events, not merely accumulated as excess cash. It needs legal structure, access rules, permitted uses, independent verification and an explicit connection to essential services. Otherwise, the reserve is a comfort signal until the institution encounters the kind of event that makes its accessibility uncertain.

The 2021 restraint converted liquidity into dependency

AFRINIC's 2021 report provides the sharpest financial stress test because it places the reassuring metric and the disabling event on the same page. The report said the year closed with a net surplus of about USD 1.92 million, total reserves near USD 10 million, strategic cash reserves of about USD 6.38 million and total cash holdings of about USD 11.91 million. It described the balance sheet as very healthy. It also recorded that the attachment order freezing all bank accounts made AFRINIC unable to honor its financial commitments for the relevant period.

The organisation's receipt of USD 504,000 in donations is the decisive detail. Donations are not evidence that AFRINIC lacked assets. They are evidence that assets and obligations had become separated by legal control. A body created to be financially independent found itself relying on outside support while reporting record cash. The old criterion's policy objective and its accounting proxy moved in opposite directions.

This does not mean every RIR should evade court orders by placing money beyond lawful reach. That would be the wrong lesson. Courts must be able to enforce valid claims, and counterparties must not bear the cost of institutional exceptionalism. The design problem is narrower: essential registry services should not become collateral damage of a dispute that can be secured through other assets, supervised payments, a dedicated continuity instrument or a court-approved operating allowance. Any protection must be transparent, limited to defined services and incapable of becoming a shield for directors or litigants.

Nor does the episode prove that more cash always solves the problem. AFRINIC could have held twice as much and still faced the same access constraint. Once all accounts are restrained, the marginal dollar does not improve operational runway. What matters is whether the institution has a legally recognized arrangement for paying essential staff and infrastructure, preserving records, maintaining security and complying with the order while the merits are litigated.

Financial independence must therefore include independence from emergency benefactors. If a registry can continue only because counterparties voluntarily extend credit or other organisations donate funds, the service is exposed to private discretion. The benefactor may be generous and well intentioned, but recognition should not rely on generosity. A protected continuity mechanism should exist before the crisis and should disclose who can activate it, who can spend from it, which costs qualify, how long it lasts and who reviews its use.

The bank restraint also reveals why litigation exposure belongs inside financial assessment. Legal disputes are not an incidental administrative expense for an RIR. A claim can reach the accounts through which service is funded. It can absorb management time, increase insurance costs, deter candidates from board service and affect the authority to contract. The test must model legal process as an operational dependency.

Strong 2024 accounts sharpened the paradox

AFRINIC's later accounts make the separation between money and governance even harder to ignore. The organisation's consolidated annual report for 2022-2024 reported another year of strong financial performance in 2024. It described a net surplus of about USD 3.53 million, total revenue of about USD 6.12 million, higher cash holdings, increased reserves and an improved liquidity ratio. Membership fees remained the core revenue base.

In the same financial discussion, the report said AFRINIC had no quorate board, no chief executive and no approved 2024 budget. It said the organisation nevertheless fulfilled contractual and operating obligations and controlled costs. Those facts should not be collapsed into a success or failure slogan. They demonstrate two different capacities. Staff preserved day-to-day services and finances remained strong. The institution's ordinary governing architecture, however, was incomplete.

That combination is precisely what a good stress test should detect. A narrow financial test would pass AFRINIC because cash, revenue and surplus were robust. A narrow service test might also pass because essential operations continued. A governance-continuity test would ask how long the organisation could rely on exceptional authority, who approved non-routine expenditure, how strategic decisions were deferred, how members could hold decision makers accountable and what would happen if a second shock arrived before ordinary governance returned.

Operational persistence can consume hidden capital. Experienced staff carry unwritten knowledge. Suppliers tolerate delayed decisions. Security projects can be postponed. Members continue paying because the regional service relationship has no easy substitute. A receiver or court can provide temporary authority. None of these supports appears as a liability equal to its institutional importance. The accounts can therefore improve while resilience is being borrowed from people and temporary arrangements.

This is not an argument for treating a governance vacancy as immediate financial failure. It is an argument for refusing to let positive accounts close the inquiry. A registry without a quorate board may be able to pay every invoice while remaining unable to make legitimate long-term commitments. It may avoid discretionary spending and produce a surplus partly because strategic work is delayed. Strong cash accumulation in such a period can reflect prudence, constrained authority or both.

The 2024 paradox gives the proposed test its answer. Budget health does not reliably predict governance resilience. At most, it shows one resource available to a resilient institution. The reviewer still needs evidence that authority, oversight, remedy and transfer capacity exist around that resource.

Four balances matter, not one

A modern assessment should separate four balances that conventional financial reporting tends to combine.

The first is the accounting balance. This includes assets, liabilities, revenue, expenses, reserves, cash flow, concentration of income and the assumptions behind going-concern reporting. Audited statements remain indispensable. They can expose persistent deficits, weak collection, excessive fixed costs, related-party transactions or depletion of reserves. Nothing in a broader resilience test should diminish their importance.

The second is the legally available balance. This asks what portion of cash can actually be used for essential service if accounts are restrained, management authority is disputed, signatories are unavailable or a court imposes conditions. It examines jurisdictions, account structures, insurance, secured claims, restricted funds, board delegations and pre-authorized emergency instruments. The answer must respect applicable law; the point is not asset concealment but lawful continuity.

The third is the operating balance. This prices the minimum service that must continue: secure custody of registration data, allocation and transfer administration where authorized, directory publication, reverse-DNS support, route-security functions, incident response, member authentication, essential billing and the staff and vendors behind them. Conference programs, grants, training and advocacy may be valuable, but they do not have the same continuity priority. A resilience calculation must identify the minimum service cost rather than treating the latest total budget as indivisible.

The fourth is the governance balance. It asks who has lawful power to approve spending, modify priorities, appoint executives, settle disputes, authorize record access and account to members under stress. Money without an authorized decision maker is not usable capacity. Authority without independent oversight can spend the reserve in ways that entrench the emergency. The two must be tested together.

These balances produce a different definition of runway. Traditional runway divides available cash by average monthly cost. RIR continuity runway should divide legally accessible, appropriately protected funds by the minimum monthly cost of essential services under the authority of a valid and reviewable decision maker. It should deduct realistic legal and transfer costs. It should disclose assumptions about staff retention, vendor credit and data access. The result may be far lower than the headline reserve figure.

The test should also consider correlation. A court dispute can simultaneously restrain cash, displace management, interrupt elections and complicate access to records. Treating each risk as independent understates the event. AFRINIC's experience shows that legal, financial and governance failure modes can arrive through one institutional channel.

Service isolation is financial architecture, not a slogan

The first improvement is to identify and isolate the service that recognition exists to protect. An RIR does more than keep servers online, but not every activity requires the same emergency guarantee. The institution should publish a minimum-service schedule: the systems, records, people, vendors and decisions required to preserve the accurate and secure administration of Internet number resources for a limited period.

Isolation can take several forms. A dedicated continuity instrument could cover a defined number of months of minimum service. Certain payments could be pre-authorized under bylaws or a court-recognized mandate when the board lacks quorum. Critical vendor contracts could include continuity clauses and rights for a temporary operator. Protected copies of essential records could be held under escrow and data-protection controls. None of these arrangements should transfer policy power or corporate ownership automatically. They should keep the function alive while legitimate institutions decide what comes next.

The legal design matters more than the label. A separate bank account at the same institution may be restrained with every other account. A reserve controlled by the same invalid signatories may remain unusable. An escrow that cannot release funds without the missing board is decorative. A broad emergency clause may be worse than no clause if it allows an executive to classify ordinary spending as continuity expenditure. The instrument must be tested against the events it claims to address.

Service isolation also requires a boundary around emergency spending. Eligible costs should be enumerated and independently reviewed. Payments should be logged. Related-party transactions should receive enhanced scrutiny. The authority should expire or require renewal. Members and the public should receive a report that protects confidential operational details while explaining how much was used, for what classes of service and under whose authorization.

The NRO's draft RIR Governance Document Version 2 moves in this direction by treating performance, continuity, governance and financial independence as separate ongoing requirements. It calls for continuity and redundancy procedures and for sharing records, systems and implementation material sufficient for an emergency operator, subject to escrow or data-protection controls. As of July 2026, the NRO's ICP-2 review page still identifies Version 2 as the current draft and describes further drafting and adoption work. The draft is therefore evidence of a needed architecture, not a claim that the problem has already been solved.

The useful principle is simple: the service should be able to survive a temporary failure of the corporation's ordinary controls without giving the temporary operator permanent institutional power. That is financial resilience with a constitutional boundary.

Quorum and election continuity belong in the funding test

Quorum rules can turn vacancies into a financial control failure. A board normally approves budgets, appoints executives, changes bank mandates, oversees audit findings, authorizes large contracts and manages litigation strategy. If too many seats become vacant and the constitution offers no narrow path to restore them, the organisation may have money but lack the body empowered to direct it.

A recognition assessment should therefore inspect the candidate's vacancy and election provisions alongside its budget. It should identify the minimum number of directors, the offices necessary for signatures, the time allowed to fill vacancies, the authority to call elections, the treatment of disputed results and the mechanism available if the very body that must call an election no longer exists. These are not abstract corporate-law questions. Each one can determine access to funds.

The test should distinguish continuity authority from ordinary authority. A temporary official may need power to pay essential bills, preserve data, renew insurance and organize a valid election. That official should not automatically receive power to alter regional policy, dispose of strategic assets, settle long-term questions of recognition or restructure the membership. The financial mandate should be broad enough to prevent collapse and narrow enough to prevent emergency rule from becoming a new normal.

Election funding also needs protection. If the only lawful route back to a board is an election, the costs of voter verification, independent administration, secure voting, dispute handling and possible rerun are continuity costs. They should not depend on the discretion of a faction whose power the election will determine. Nor should an election be rushed because the ordinary budget omitted the possibility of a challenge.

AFRINIC's years of governance difficulty show why an annual budget cannot assume that board formation will be routine. The NRO's 2023 receiver statement connected continued member services with the court-supervised task of restoring a board and appointing a chief executive. It also thanked staff for maintaining operations. The statement's own structure separates service persistence from functional governance. A resilient funding model must finance both the bridge and the lawful exit from the bridge.

This does not mean ICANN should prescribe the same election law to every RIR. Corporate statutes and regional traditions differ. It means recognition reviewers should require a demonstrable answer to the failure case: if ordinary quorum disappears, what valid, limited and financed mechanism restores representative authority?

Litigation reserves must model access as well as cost

Most organisations budget legal fees as an expense line informed by recent history. For an RIR, that is too narrow. Litigation can impose direct fees, adverse awards, account restraints, management distraction, disclosure obligations, reputational effects, insurance changes and constraints on governance. It can also raise a conflict between domestic judicial authority and global expectations for uninterrupted registry service.

A litigation reserve should therefore have at least three components. The defense component covers counsel, experts, evidence preservation and compliance with orders. The exposure component estimates adverse costs, security, settlement and insurance deductibles across plausible cases rather than assuming only one dispute. The continuity component funds essential service while ordinary accounts or decision powers are constrained.

The amounts cannot be universal. A small RIR and a large RIR have different budgets; legal systems differ; the value at issue can change quickly in an IPv4-scarcity environment. A sound rule would require a documented method, board and member visibility at an appropriate level, independent audit of the reserve's existence and a periodic scenario review. Sensitive litigation strategy need not be published, but the community should know whether the organisation has priced material legal risk and whether continuity funding is separate from the merits budget.

The reserve must not create perverse incentives. An enormous litigation fund under unchecked executive control could finance aggressive legal strategy rather than resilience. Release conditions should distinguish defense of the institution from protection of particular officeholders. Directors' and officers' insurance, corporate defense and continuity service are related but not interchangeable. Related-party counsel selection and unusual success fees warrant disclosure and conflict review.

The AFRINIC bank restraint supplies a concrete scenario: all ordinary accounts become inaccessible for several months. The test should ask whether salaries, hosting, security response and critical vendors can still be paid lawfully; whether the court receives a transparent minimum-service plan; whether the organisation can seek a supervised operating allowance; and whether another RIR or emergency operator can support narrowly defined services without assuming disputed corporate authority.

This approach respects courts rather than attempting to outrun them. It gives judges and parties a pre-existing, auditable way to distinguish essential service expenditure from unrestricted corporate spending. The registry does not receive immunity. Resource holders receive continuity that is less dependent on improvisation.

Migration readiness is a current liability

The final weakness in the old funding criterion is its treatment of transition as an entry event. AFRINIC's recognition plan priced migration into the new RIR. A mature standard must also price migration out of, or temporarily around, an impaired RIR. That does not assume derecognition. It recognizes that a transfer capability is valuable even when rehabilitation succeeds.

Migration has financial, technical, legal and human costs. Records must be complete enough for another operator to understand allocations, transfers, contacts, contractual status, disputes and confidentiality restrictions. Authentication and signing systems need controlled handoff paths. Reverse-DNS and route-security operations need continuity plans. Staff knowledge must be documented, and key personnel may need retention arrangements. Resource holders need notice and a way to correct errors. The receiving operator needs indemnity boundaries and funding. Courts and data-protection authorities may need to approve particular transfers.

A credible migration reserve should be tied to a rehearsed scenario. It should estimate the cost of preparing data, validating it, running parallel services, communicating with resource holders, resolving exceptions and returning control if the original RIR is rehabilitated. It should identify which party pays at each stage. If the remaining RIRs are expected to help, their commitment should be explicit rather than presumed after the crisis begins.

Data readiness is part of financial readiness because unavailable or poorly documented records make transfer slower and more expensive. Cheap storage is not the same as usable escrow. A protected copy must include the context needed to make lawful decisions, be refreshed at a known frequency, preserve an audit trail and be test-restored by people who do not depend on the impaired organisation's internal access.

The emergency operator must also be constrained. It should receive only the authority needed for defined services, for a defined period, under published review. It should not acquire the region's policy mandate merely because it can run the systems. Its costs should be auditable, and procurement should limit conflicts with incumbent RIRs that might benefit from a permanent transfer.

The draft RIR governance text recognizes readiness for transfer and emergency continuity as system responsibilities. The financial standard should make that commitment measurable. A promise that peers will cooperate is not a funded plan. Recognition should require the money, records, legal permissions and exercises that turn cooperation into a credible option.

A modern test needs scenarios, not one ratio

The redesigned criterion should begin with ordinary financial viability, then apply a small set of severe but plausible scenarios. The purpose is not to predict the exact next crisis. It is to discover common dependencies before they fail together.

One scenario should restrain all ordinary bank accounts for ninety days. Another should remove enough directors to break quorum while the chief executive position is vacant. A third should combine a disputed election with urgent security expenditure. A fourth should produce a material fall in fee collection. A fifth should require an emergency operator to perform minimum services with no access to undocumented institutional knowledge. A sixth should combine litigation expense, staff departures and a damaged registration record.

For each scenario, the RIR should identify the valid decision maker, accessible funds, minimum services, record location, essential staff, vendor rights, communications duty, member remedy, court interface and exit condition. It should state recovery-time and recovery-point objectives for each technical service, but it should also state an authority-recovery objective: how quickly can lawful representative governance be restored?

The test should publish a concise resilience statement rather than every security detail. Useful measures include months of legally accessible minimum-service runway; percentage of essential spending covered by a protected continuity instrument; maximum expected time without a quorate governing body; proportion of critical records successfully restored in the latest exercise; time required to activate a temporary operator; litigation exposure relative to unrestricted reserves; and concentration of cash, signatory authority and critical vendors.

No single measure should become a pass-fail talisman. A registry could game a twelve-month reserve target while leaving authority unresolved. It could distribute accounts across banks while a single court order reaches the entity. It could copy data without testing whether anyone else can use it. The assessment should examine how the controls interact.

Independent review is essential because management is likely to underestimate events that challenge management's own authority. Auditors can verify cash and controls, but the exercise also needs corporate-law, service-continuity, security and member-governance expertise. Results should go to the governing body and membership, with a public account of material gaps and remediation dates.

This is a demanding standard, but the burden is proportionate to the role. An RIR is a regional institution with global effects and limited substitutability. The cost of an annual exercise is small compared with improvising continuity after accounts, authority and records are already contested.

Financial resilience should not become financial supervision

Strengthening the criterion creates a risk of its own: an external reviewer could use resilience language to supervise ordinary budgets or punish policy disagreement. That would replace one weak test with an overbroad power. The standard needs boundaries for ICANN and the peer RIRs as well as obligations for the subject RIR.

The review should focus on auditable capabilities tied to essential services. It should not decide whether a registry spent too much on a meeting, chose the wrong office, funded an unpopular research program or set a fee at the level an external actor prefers, unless the decision creates a material continuity risk or violates a defined obligation. Regional members should retain control over ordinary priorities.

Triggers should be written and evidence based. A missed reserve target might require explanation and remediation, not an immediate recognition threat. A temporary deficit could be rational during planned investment. A court restraint might require activation of continuity controls without implying wrongdoing. The relevant question is whether the RIR can continue defined service, comply with law and restore ordinary governance within a credible period.

Confidentiality also matters. Bank details, security arrangements, personal data and litigation strategy cannot all be public. The reviewer should receive enough information under appropriate protections, while the public report states the method, scope, key findings, conflicts and required remediation. Secrecy about every detail would defeat accountability; publication of every detail could create new risk.

Remedies should escalate. A control gap can produce a dated correction plan. A failed exercise can produce independent verification and a repeat test. Material inability to fund essential service can activate limited continuity support. Derecognition should remain a last resort after a record of failure, opportunity to cure and a protected service path. Money should not become a shortcut around representation or due process.

The December 2024 implementation and assessment procedures published by ICANN recognize that a compliance review should turn on total circumstances and material impact, and that an ICANN-initiated review should be limited rather than a general supervisory role. Those limits are important. A modern financial standard should make them operational by specifying the evidence, thresholds and remedial sequence before crisis correspondence is sent.

What the AFRINIC stress test establishes

AFRINIC does not establish that reserves are unimportant. Its staff and services may have benefited greatly from the accumulated financial cushion. Nor does the case prove that service isolation would have eliminated litigation, election conflict or the need for a receiver. Resilience controls do not cure every cause of institutional failure.

What AFRINIC establishes is narrower and more consequential. A strong reported balance sheet did not prevent all accounts from becoming inaccessible. Record cash did not eliminate temporary reliance on donations. Persistent fee income did not ensure that a quorate board, chief executive and approved budget would remain in place. Continued service did not mean ordinary governance had returned. These are direct observations from the institution's own reports and official statements, not a speculative theory about hidden insolvency.

The case also demonstrates why the financial criterion cannot be assessed in isolation from ICP-2's other principles. Community support matters because members fund the system. Neutrality matters because a monopoly-like revenue stream should not finance unequal treatment. Record keeping matters because migration cost depends on usable records. Technical competence matters because money cannot substitute for operational skill. Bottom-up governance matters because someone must legitimately decide how reserves are used. The criteria are mutually reinforcing, not compensatory.

For a future recognition or periodic compliance assessment, the correct question is no longer, "Does the RIR have a balanced budget and enough reserves?" It is, "Can the RIR lawfully fund and govern minimum service through the loss of ordinary access, ordinary quorum and ordinary management, while preserving member rights and a tested path to restoration or transfer?"

That question changes what counts as evidence. Audited accounts remain the starting point. The rest of the file should include the minimum-service budget, legal-access analysis, continuity instrument, delegation matrix, election-recovery mechanism, litigation scenario, record-restoration result, migration exercise and public remediation report. A registry that cannot show those controls may be solvent yet fragile. A registry with modest reserves but strong legal and operating isolation may be more resilient than its headline ratio suggests.

The standard should reward candor. Publishing a failed exercise and a funded correction plan can show stronger governance than a flawless self-assessment. Tail risk cannot be eliminated, and a reviewer that demands certainty will encourage concealment. The objective is demonstrable capacity to absorb, explain and repair failure without using resource holders as leverage.

ICP-2's funding rule was written for the birth of a registry. AFRINIC's history supplies the missing adulthood test. Financial stability is not the size of the reserve at year end. It is the alignment of accessible money, limited authority, protected service, representative recovery and a credible exit when the institution is under its greatest strain.

Evidence and analytical limits

The ICANN ICP-2 text supports the historical account of the funding criterion: a published budget linked to the activity plan, community support, operational autonomy and eventual member financing. It does not contain the expanded continuity measures proposed here.

The IANA recognition report supports the description of AFRINIC's incubation support, membership-fee model, planned financial independence and transition arrangements. It is an entry assessment, not proof that later governance resilience was guaranteed.

AFRINIC's 2020 report, 2021 report and 2022-2024 consolidated report support the financial figures and the contrast between healthy accounts, the 2021 bank restraint and later gaps in ordinary governing capacity. The reports are AFRINIC's publications; they are used for facts they state and are not treated as independent adjudications of the disputes surrounding the organisation.

The NRO receiver statement and ICANN's March 2025 update support the limited claim that official actors distinguished continued services from the work of restoring functional governance. They do not establish that every action by a receiver, member, litigant, board candidate, court or external institution was correct.

The NRO Version 2 draft is cited as evidence of current reform direction, especially the separation of financial independence, governance, performance, continuity and emergency readiness. The NRO process page shows that final drafting and adoption remained pending in 2026. This analysis therefore proposes a test; it does not claim that the test is already binding.