When a court order meets a global routing table

Summary

A regional Internet registry is a legal person in a country and can be bound by competent courts. Global reliance on its services is not immunity from ordinary law, insolvency supervision, evidence preservation or an effective remedy between litigants.
Registry authority and routing operation are different. Registries maintain allocation and registration data and provide linked services; operators independently originate, accept, filter and propagate routes. A court order can alter inputs to those choices without directly commanding the global routing table.
Broad or ambiguous relief can still create spillovers. Changes to registration data, reverse DNS, RPKI materials, account control or service authority can be interpreted by thousands of independent parties, including parties with no notice and no practical opportunity to be heard.
Courts and registries need a standing continuity protocol: map every requested action to its technical effect, identify non-party dependencies, prefer the narrowest reversible measure, preserve records, separate disputed control from routine service, provide rapid technical review and publish a bounded implementation notice.
AFRINIC's court-supervised governance history demonstrates the institutional problem without proving a universal lesson about the merits of any one dispute. The public record supports careful claims about receivership, board reconstitution and continuity; it does not support saying that a Mauritius court decided how the world must route African address space.

The order arrives before the maintenance window

Imagine a regional Internet registry receiving an urgent order at 16:30 on a Friday. The order identifies the company, a claimant, specified assets and a disputed right. It requires preservation of a position, restrains a corporate act, appoints a receiver or directs the company to change a record. The legal team reads obligations, deadlines and exposure for contempt. The registry's engineers read a different set of questions. Which database row? Which account credential? Which signing system? Does the direction touch public registration data, reverse DNS, resource certification, a transfer, an allocation queue or only corporate control?

Which changes can be reversed? Which changes will be copied by others before the registry can explain them?

Neither team can safely answer the other's questions by analogy. Engineers cannot reduce judicial authority to an optional service ticket. Lawyers cannot assume that a sentence framed around possession or control maps cleanly onto a distributed technical system. A number registry does not operate a single switch marked "Internet access." Its actions change records and services that other organisations interpret through their own systems, contracts and risk rules. The legal effect can be direct while the network effect is contingent, delayed and dispersed.

That distinction makes the problem harder, not easier. If one central switch existed, a court could identify it and hear from the parties affected by turning it off. In the Internet numbers system, a registry action can become an input to thousands of independent decisions. Some networks may react at once, some after a validation cycle, some only when staff inspect a record, and some not at all. The court may have jurisdiction over the company carrying out the first action but no direct relationship with most organisations carrying the consequence forward.

The right response is not registry immunity. A company that supplies globally relied-upon coordination remains a company subject to law. Creditors need remedies. Members need enforceable constitutional rights. Contracting parties need courts. Employees, regulators and injured parties cannot be told that global dependence places corporate conduct beyond adjudication. The response is institutional translation: a repeatable method by which the court's lawful objective is converted into a technically bounded measure, with unnecessary spillover treated as a design failure rather than fate.

Since 2010, the value and operational use of Internet number records have become more visible. Address scarcity, transfers, resource certification, sanctions questions, fraud controls and registry governance disputes have increased the consequences attached to a record. The legal system and the registry system now meet more often at points where neither can pretend the other is merely background.

Two maps of authority occupy the same desk

The legal map starts with persons, jurisdiction and remedies. A regional registry is incorporated or organised under national law. It owns or controls assets, employs staff, enters agreements, holds records and exercises powers through directors or other corporate organs. A competent court can determine rights between parties, preserve evidence, restrain conduct, supervise insolvency, appoint an officer and compel compliance within the law governing the proceeding. The fact that the company's customers operate networks in many countries does not erase the legal map.

The technical map starts with functions and dependencies. The IANA Number Resources page describes global coordination of IP address space and Autonomous System Numbers. IANA generally allocates pools to five regional registries. RIRs register allocations and assignments for their service regions; local or national registries, providers and end users occupy further layers. Public records, reverse DNS and resource-certification services can provide information used by operators, security teams and counterparties.

These maps overlap at the registry company, but their boundaries are not identical. Corporate control over a registry does not amount to ownership of every route originated from addresses recorded by that registry. A member's contractual relationship does not make every downstream customer a party to the registry agreement. Recognition as an RIR does not remove directors from national company law. A route accepted in another country does not prevent the registry from correcting a record under an authorised process.

Confusion begins when a term migrates between maps without translation. "Control" may mean legal control of a company, administrative control of an account, authority to alter an authoritative record, possession of a cryptographic key, or practical ability to influence routing decisions. "Asset" may refer to company property, a contractual right, a registration entitlement or an address block described casually as property even when an agreement rejects that characterisation.

"Status quo" may mean preserving board authority, preserving a database state, preserving services to members or preserving the actual routes visible on the Internet. Each meaning leads to a different measure.

A sound order and a sound response therefore need a concordance. For every operative word, the registry should state the entity, system, authorised actor, technical operation, expected external signal, affected population, reversibility and restoration time. That statement does not tell the court what law to apply. It tells the court what the requested act would do. It also prevents a party from obtaining broad technical consequences through vague corporate language that neither the judge nor non-parties had a fair chance to examine.

Registration is consequential, but it is not routing

RFC 7020 is unusually useful because it states both the importance and the limit of registry functions. It describes a hierarchy rooted in the IANA role, with RIRs serving local registries and other customers. It identifies registration accuracy and uniqueness as core goals. It also states that whether addresses are announced, and how they are advertised into the routing system, are operational considerations outside the Internet Numbers Registry System.

This separation defeats two exaggerations. The first says that a registry record is merely clerical and cannot affect real connectivity. That is too weak. Operators and service providers use registration information for due diligence, incident response, address-policy decisions and commercial assurance. Reverse DNS can matter to services. Resource Public Key Infrastructure materials can affect route-origin validation signals. A transfer, revocation, account change or certificate action may alter how counterparties assess a route or organisation.

The second exaggeration says that an RIR controls the global routing table. That is too strong. A router accepts or rejects routes according to its operator's configuration, learned paths, filtering, commercial relationships and security policy. Even route-origin validation is implemented by relying parties; the registry publishes materials, while networks decide how to use validation states. A registration change may cause operational reactions, but the causal chain includes independent choices.

This architecture should shape evidence in court. A party seeking an urgent registry change should not be permitted to assert either "the Internet will certainly stop" or "nothing operational can happen" without showing the mechanism. It should identify the service, current state, proposed state, relying systems, update interval, likely reactions and available mitigation. The opposing party should meet the same standard. Technical consequence is an empirical question, not a metaphor.

The separation also shapes relief. If the legal objective is to prevent a disputed transfer, freezing transfer authority may be narrower than altering the registration holder. If the objective is to preserve evidence, taking a signed snapshot may be sufficient without changing public data. If account credentials are contested, dual control or escrow may preserve operations while preventing unilateral action. If corporate governance is disputed, keeping routine registry staff authorised under supervised limits may avoid making every member wait for the board dispute to end.

Precision preserves both judicial authority and network stability. An order that achieves its legal objective through a narrow control is easier to implement, explain and reverse. An order that treats registration, certification and routing as one indivisible entity risks doing more than the court intended.

What a court can bind, and what it should ask before doing so

A court's first question is jurisdictional: which parties and property are before it, and under what legal basis? For continuity, a second question is functional: which registry act would satisfy the remedy, and who beyond the parties is likely to rely on the result? The first determines authority. The second disciplines scope.

Several targets are relatively direct. The court can restrain directors from approving a transaction, order the company to preserve logs, place specified credentials under controlled custody, require a receiver to maintain ordinary services, direct delivery of corporate documents or prohibit destruction of records. These measures may be technically sensitive, but their primary target is the company and its governance.

Other requests need a fuller effects statement. An order to change a registrant name, revoke an entitlement, suppress a public record, alter reverse DNS, issue or withdraw certification material, approve a transfer or disclose confidential member information crosses from corporate control into registry service. The measure may still be lawful and necessary. It should, however, be treated as a consequential act with named dependencies rather than as an entry on a spreadsheet.

Before granting technically consequential relief, a court should receive answers to a compact set of questions. What exact record or service is affected? Is the act required by the legal right being protected, or merely convenient? Can the same objective be achieved with a temporary lock, annotation, escrow or dual authorisation? How many holders and downstream organisations are foreseeably affected? Is there critical infrastructure in the set? What notice can be given without defeating the remedy? What is the rollback plan if the order is varied? Which person will certify implementation?

What independent technical evidence supports the claimed effect?

Urgency does not eliminate these questions. It changes who answers them and when. A judge may grant a short preservation measure on limited evidence, then require a technical return hearing within days. A registry may be directed to freeze a disputed action while leaving routine service unchanged. A confidential annex can identify sensitive systems without publishing exploitable detail. The party seeking relief can be required to fund a neutral specialist where the measure creates unusual technical review costs.

The key is to distinguish preserving the court's ability to decide from granting the final technical consequence. In many cases, the status quo can be held with less disruption than either litigant claims. A temporary lock that prevents transfer or deletion may preserve rights while the court hears evidence. Once a record is changed and copied throughout external systems, later restoration can be slower and less complete than a legal order implies.

AFRINIC shows the company-continuity problem, not a judicial routing command

AFRINIC provides the clearest public illustration of why institutional continuity must be designed before litigation. It is a company incorporated in Mauritius and the recognised regional registry serving Africa. Its corporate constitution, service agreements and registry role intersect at one legal person. When its governance became subject to prolonged disputes and court supervision, the legal system had to address a company whose routine work carried consequences far beyond the immediate litigants.

The bounded public facts matter. AFRINIC's 16 October 2024 court update said that the Mauritius appellate ruling of 15 October restored an earlier judgment appointing the Official Receiver and providing a time-bound mandate to organise elections under AFRINIC's constitution. The hosted copy of the 2024 appellate decision records relief concerned with preserving the company and reconstituting governance through a receiver-supervised process.

Those materials justify saying that a domestic court exercised authority over the company and used receivership in connection with preservation and governance restoration. They do not justify saying that the court ordered global networks to accept or reject routes. They do not establish the merits of every claim in the wider history. They do not make every registry-service interruption the result of one judgment. A responsible account stays within the document it cites.

AFRINIC's Bylaws show why the distinction matters. Corporate entities, membership, board powers and meeting rules define how the organisation governs itself. The AFRINIC Registration Service Agreement addresses a different relationship: services between the registry and a resource member, member duties and consequences of termination. Neither document, read alone, describes every operational dependency. Together they show several legal surfaces within one institution.

A receiver charged with holding the ring is not merely guarding office furniture and bank accounts. Preserving the value of a registry company includes staff capability, system integrity, signing material, accurate records, member communication and confidence that ordinary decisions remain authorised. Yet a receiver should not be treated as possessing unlimited substantive number-policy power simply because the court has conferred corporate authority. The mandate, bylaws, agreements, community policies and court directions still need to be read as separate sources.

The institutional lesson is narrow and useful: court-supervised corporate continuity and number-registry service continuity should be expressly connected. If a receiver's order says preserve the company, a continuity schedule should state which services must continue, who may operate them, which exceptional decisions require leave and how members can challenge an error. That schedule protects the court's purpose from being undermined by technical ambiguity.

The service-effect ladder

Not every judicially compelled registry act carries the same risk. A continuity protocol should sort measures by service effect rather than by the emotional intensity of the dispute.

The first level is evidentiary preservation. Logs, correspondence, signed database snapshots, transaction history, access records and decision materials can be preserved without changing the public state. The main risks are confidentiality, integrity and chain of custody. Courts already know how to supervise evidence; the registry adds technical detail about formats, keys and retention.

The second level is control preservation. A disputed account can be locked, privileged actions can require two authorised persons, credentials can be rotated, or a transfer can be paused. Public records and routine services stay unchanged. These measures are often powerful because they prevent irreversible acts while imposing little external uncertainty.

The third level is annotated continuity. The registry preserves the operative record but adds a carefully designed status marker or sends a bounded notice to affected parties. Annotation is not harmless: counterparties may react to uncertainty. It should identify what is disputed and what remains valid, avoiding language that prejudges the merits.

The fourth level changes a service signal. Altering registration data, reverse DNS or certification material can affect operational and commercial decisions. The court should require a dependency assessment, a staged plan and a defined restoration route. Where feasible, the registry should notify directly affected holders before external publication, unless notice would defeat lawful relief.

The fifth level is irreversible or redistributive action. Reallocating disputed address space, approving a transfer to a new holder, destroying signing material or permanently closing a record can make later restoration difficult. Even when legally available, these measures call for the strongest hearing, evidence and review. A temporary freeze will often protect rights better than early redistribution.

This ladder is not a claim that lower levels are always sufficient. Fraud, security compromise, deliberate dissipation or an imminent unlawful transfer may demand rapid action. It is a requirement to explain why the selected level is necessary. The reasons should connect the legal harm to the technical act and reject less disruptive alternatives with specificity.

Sorting by effect also makes appellate review intelligible. A higher court can see whether the first judge preserved evidence, froze control or changed a service used by non-parties. Engineers can prepare rollback for the actual level ordered. Members can understand which functions remain ordinary. Precision turns continuity from an aspiration into an executable duty.

Non-parties are where domestic relief becomes global risk

The parties in registry litigation may be a holder and the registry, competing claimants, directors, creditors or a regulator. The people exposed to technical consequences are a larger group: downstream customers, transit providers, peers, hosting clients, security teams, certificate users, public services and people who have no idea that a domestic proceeding exists.

Not every exposed organisation needs formal standing. Courts could not function if every user of an address became a party. But the absence of standing does not make dependency irrelevant. Courts routinely consider effects on third parties when framing injunctions, preserving essential services and supervising insolvent businesses. A number registry needs a practical method to bring those effects into view without turning a focused case into a global assembly.

One method is a non-party dependency statement prepared by the registry and tested by an independent specialist. It should identify affected resource ranges, service types, estimated holder and customer classes, known critical functions, expected propagation times and mitigation options. It should avoid disclosing sensitive network data beyond what the court needs. Where a small number of directly affected entities can be identified, they can receive notice or an opportunity to submit technical evidence.

Another method is representative participation. An operator association, technical community body or court-appointed specialist can explain system effects without taking a position on who should win the underlying dispute. Its remit must be narrow: architecture, consequence, reversibility and continuity. It should not become a vehicle for political pressure or a substitute claimant.

Cross-border comity also matters. A domestic court need not ask foreign operators for permission to bind a domestic company. It should nevertheless avoid language implying authority it does not claim. The order can state that it directs the registry's act and does not decide the independent obligations of networks outside the proceeding. That sentence is more than diplomatic. It helps prevent parties from misrepresenting a company-level order as a universal direction to accept, reject or transfer routes.

The registry's public notice should use the same discipline. It should describe the action, legal basis, effective time, affected service and available contact. It should not announce that the Internet has been ordered to recognise a contested ownership claim. It should not turn a temporary measure into a reputational verdict. Accurate language reduces the risk that private counterparties amplify the relief beyond its scope.

Continuity must be built before the next case

An improvised response begins too late. By the time an urgent order arrives, parties are adversarial, deadlines are short and technical staff may fear contempt if they ask clarifying questions. Each RIR should maintain a court-readiness continuity plan approved through its governance and tested with external counsel and operators.

The plan should begin with a service map. For every legally addressable component, it should identify the corporate owner or controller, technical operator, authoritative data, credentials, dependencies, external consumers, backup arrangements and maximum tolerable interruption. It should distinguish the member registry, public directory, reverse DNS, resource certification, transfer controls, billing, elections and ordinary internal systems. A court cannot choose a narrow measure if the registry itself cannot describe its components.

Next comes an authority map. The plan should identify which decisions belong to the board, executives, ordinary staff, policy community, receiver, emergency key holders or external service providers. It should state which powers continue when a board lacks quorum, a chief executive is absent or bank access is restricted. The ICP-2 recognition criteria emphasise elements such as neutrality, technical capability, documented procedures, funding and continuity. Those expectations should be reflected in concrete succession arrangements, not left as recognition-era promises.

The plan also needs a legal intake rule. Every order should be logged with service classification, deadline, confidentiality status, exact operative language, responsible counsel and responsible engineer. Counsel should confirm authenticity and scope. Engineering should produce an effects statement. Senior authority should certify the chosen implementation. If words are ambiguous, the registry should seek clarification rather than silently choose the most expansive reading.

Finally, the plan needs exercises. A tabletop test can simulate a disputed transfer freeze, a receiver appointment, a demand for confidential member data, a signing-key compromise and an order affecting a critical service. The test should measure time to assemble authority, identify dependencies, obtain independent advice, communicate with members and restore from a reversed action. Findings should be reported to the board and membership in a form that protects security.

Continuity planning is not resistance to courts. It makes compliance reliable. A registry that can explain its systems quickly is more likely to receive precise relief. A court that receives credible alternatives is less likely to choose a broad measure merely because the parties present only extremes.

A standing technical friend of the court

Judges should not have to become routing engineers, and engineers should not decide legal rights. Complex cases often use neutral expertise to bridge specialised fact and judicial judgment. Number-registry disputes need a version designed for speed and independence.

A standing roster could include people with registry operations, routing, RPKI, security, insolvency and public-interest experience. Appointment in a case should include conflict checks covering employment, membership, address holdings, vendors, litigants and advocacy roles. The specialist's questions would be technical: What act is requested? What systems implement it? What external signals will change? How quickly? What alternatives preserve the legal position? How can the act be reversed? Which claims about outage or harmlessness are supported by measurement?

The specialist should not opine on who owns a contested entitlement unless the court asks for analysis of technical terminology. Nor should the specialist become a shadow registry executive. Operational responsibility stays with the authorised institution. The value of the role is independent translation at the moment when each party has an incentive to describe the technical consequence in absolute terms.

Funding can be proportionate. Large commercial litigants may bear the cost initially, subject to later allocation. A registry continuity reserve could cover cases involving governance or widespread non-party exposure. For smaller disputes, the court could use a brief written report rather than a full evidentiary exercise.

The report should enter the record where confidentiality permits. Over time, courts and registries would develop a body of bounded knowledge: which controls are reversible, how long external propagation takes, when notice creates risk, and which measures protect both adjudication and continuity. That knowledge is currently scattered among private counsel, registry staff and operators.

There is a useful precedent in the registry sector's own transparency practices. The RIPE NCC report on law-enforcement requests in 2023 classifies requests and explains institutional handling rather than presenting every legal demand as a self-executing technical instruction. Civil court matters differ, but the governing habit is transferable: authenticate, classify, assess legal basis, narrow disclosure or action, and report aggregate experience.

Neutral expertise will not eliminate disagreement. It will improve the entity of disagreement. The court can decide law and fact knowing whether it is freezing corporate control, altering a registry signal or creating an irreversible external state.

The order should contain an implementation schedule

Traditional orders identify parties, prohibited acts and deadlines. Technically consequential registry relief should add a short implementation schedule. The schedule is not software documentation. It is the legal instrument's statement of what compliance means.

First, it should identify scope positively and negatively. It may direct a lock on specified transfer authority while stating that ordinary registration, reverse DNS and certification services remain unchanged. Or it may require a defined record correction while stating that no finding is made about routes independently announced or accepted by operators. Negative scope prevents opportunistic expansion.

Second, it should name time. The schedule should include effective time, duration, review date and expiry. A temporary measure should not survive indefinitely because nobody diarised a return hearing. If external systems may cache a change, the order should account for propagation and restoration rather than assume instantaneous compliance.

Third, it should identify verification. A named registry officer can file a certificate stating which actions were taken, when and with what preservation. A neutral specialist can verify service state without receiving authority to alter it. Hashes or signed snapshots can preserve evidence of the before-and-after position.

Fourth, it should set communication boundaries. The registry should issue a factual notice to directly affected holders and, where material, the wider community. The notice should distinguish an interim direction from a final determination, avoid confidential detail and provide a channel for reporting unintended consequences. Parties should be restrained from describing the order more broadly than its terms.

Fifth, it should state rollback. If the order is stayed, varied or discharged, who acts, which prior state is restored, what new conflicts must be checked and how external consumers are notified? Restoration is not always the inverse of change. A transfer may have induced contracts; a certificate withdrawal may have changed validation; a public allegation may have affected trust. The schedule should preserve the evidence and authority needed for meaningful correction.

An implementation schedule also helps accountability. If harm occurs, reviewers can distinguish what the court required, what the registry chose and what operators independently did. Without that separation, every actor can attribute the result to the other two.

Registries should protect adjudication, not pick winners through architecture

A registry confronted with litigation can influence the practical result before the court decides the merits. It can lock an account, preserve one party's credentials, alter a public description, delay a transfer, change signing status or communicate with counterparties. Some interim choice is unavoidable. The legitimacy question is whether the choice preserves adjudication or decides it through technical leverage.

The proper default is reversible neutrality. Preserve records. Prevent dissipation. Maintain routine services. Freeze the disputed operation rather than all operations. Keep the last uncontested public state where lawful. Require dual approval for exceptional changes. Separate staff handling the contested matter from ordinary service teams. Record every privileged action.

Neutrality does not mean treating proven fraud and a credible holder identically forever. It means that interim architecture should not create a final outcome before evidence and authority are tested. Where immediate protective action is necessary, the affected party should receive prompt reasons and a rapid route back to the court or another independent reviewer.

Registries should also resist private demands dressed in judicial vocabulary. A letter threatening proceedings is not an order. An order against one party is not automatically an order against the registry. A foreign judgment may require recognition steps before domestic enforcement. A preservation direction does not necessarily authorise public accusation. Authenticity, service, jurisdiction and operative scope belong to counsel; technical staff should not infer them from urgency.

Conversely, a registry should not invoke community policy as a reason to disregard a binding order. Policies and agreements inform what the company may ordinarily do. Courts determine legal obligations within their competence. If compliance appears to conflict with other duties or create disproportionate third-party harm, the registry should seek clarification, variation, a stay or protective terms. Quiet non-compliance is neither continuity nor legitimacy.

The balance is institutional rather than heroic. Staff should not have to choose personally between contempt risk and network stability. A predefined escalation path, neutral evidence and an implementation schedule place that choice where it belongs: before authorised decision-makers with a record of reasons.

Operators remain responsible for their side of the boundary

Registry continuity cannot remove every risk because operators make independent routing decisions. Networks should not design critical connectivity on the assumption that every registry service and corporate structure will remain continuously uncontested.

Operators can inventory which decisions depend on public registration, RPKI, reverse DNS or registry account access. They can separate monitoring from enforcement, ensuring that a sudden data change creates review before a destructive automated response where risk permits. They can maintain contacts outside a disputed portal, preserve historical authorisations, test certificate rollover, and define exceptions for critical infrastructure. They can require vendors to expose why a route or customer state changed rather than presenting a single unexplained verdict.

This does not excuse registry error. It recognises distributed responsibility. RFC 7020's separation between registration and route announcement means resilience exists on both sides. A registry should publish accurate, stable signals and explain exceptional changes. An operator should decide how those signals interact with local risk, customer obligations and routing policy.

The same principle applies to counterparties. A registration dispute may be relevant to a contract, but it should not automatically be treated as proof of unlawful routing, fraud or lack of operational authority. Due diligence should examine the exact status, court terms, registry notice and technical evidence. Overreaction can magnify a narrow order into the global disruption everyone claims to fear.

Industry coordination can help. Operator groups and registries can agree on machine-readable but human-auditable status categories for court-affected records: frozen, disputed, preserved, changed by final order, or restored. The category should state its consequence and, equally, what it does not mean. No status should purport to direct routing. A structured notice can make independent decisions better without centralising them.

Operators should also participate in continuity exercises and provide evidence about realistic propagation. Courts often hear catastrophic predictions from parties but little measured detail from networks. Aggregated operator evidence about update cycles, filtering practices and restoration can help a judge choose a narrow measure. Responsibility includes making the architecture intelligible before a crisis.

Transparency without turning litigation into spectacle

Secrecy creates uncertainty, but unrestricted publicity can expose confidential data, prejudice proceedings and cause the very commercial reaction a preservation order seeks to avoid. A registry continuity regime needs layered transparency.

At case level, directly affected holders should receive the order or an accurate extract unless lawful secrecy prevents it. They should know the service action, effective time, duration, reason category and review route. Technical teams need enough information to mitigate effects; they do not need private allegations unrelated to the service.

At community level, the registry should publish a concise notice when an order materially changes common services or governance authority. The notice should identify the court, date, public case reference where available, bounded effect and next review point. It should distinguish the registry's statement from a party's characterisation. It should be corrected promptly when an order changes.

At aggregate level, every RIR should publish an annual legal-intervention report. Useful denominators include orders received, jurisdiction, case class, requested action, orders challenged, actions narrowed, emergency measures, service categories affected, notices issued, reversals and restoration time. Confidential cases can be counted without exposing parties. A zero should be published as a zero; silence prevents comparison.

At governance level, members should receive an after-action report for any event that tests continuity. The report should ask whether authority was clear, systems were mapped, independent advice arrived in time, non-parties were considered and restoration worked. Security details can remain restricted, but the institutional lesson should not disappear behind privilege.

Transparency should avoid score-settling. Registry notices must not sensationalise allegations or declare victory while an interim matter remains contested. Parties should not use a technically narrow order as marketing evidence of universal ownership or lawful routing. Courts can reinforce this discipline by requiring accurate descriptions of their relief.

Trust grows from bounded facts. In AFRINIC's case, it is possible to say that public materials document court-supervised receivership and board-reconstitution steps. It is not necessary to fill every evidentiary gap with a sweeping narrative. The same restraint should govern future cases: identify the company, court, order and service effect; state uncertainty; do not turn technical dependence into drama.

The strongest objections, answered

The first objection is sovereignty. Why should a domestic court modify relief because foreign networks rely on a domestic company? The answer is not that foreign reliance defeats jurisdiction. It is that courts routinely tailor equitable relief to necessity, proportionality and third-party effect. A technically precise order enforces sovereignty more effectively because compliance is clear and unintended consequences are reduced.

The second objection is delay. A continuity assessment could become a tactic for a wrongdoer to dissipate assets or preserve control. That risk is real. The answer is a two-stage process: immediate preservation of evidence and disputed authority, followed by a prompt technical hearing before service-changing or irreversible action. Emergency measures can be narrow and fast.

The third objection is cost. Small member-funded institutions cannot retain specialists for every dispute. Most demands will not need full review. A service-effect triage can reserve independent reports for measures that alter public records, certification, reverse DNS, resource status or critical governance. Standard forms and a standing roster reduce cost.

The fourth objection is that routing is independent, so no special protection is needed. Independence reduces direct control; it does not eliminate influence. Registry signals and corporate continuity matter precisely because operators use them independently. Distributed reaction can spread uncertainty beyond the original act.

The fifth objection is that RIR recognition and community governance should prevail over national courts. Recognition establishes a role within global coordination; it is not a licence to ignore company law. ICP-2 itself values stability, documented procedure and technical capacity. Court readiness is part of satisfying those expectations in the real jurisdiction where the RIR exists.

The final objection is that a continuity protocol will help powerful incumbents resist legitimate claims. It could, if written as a presumption against change. The better design protects adjudication rather than incumbency: freeze contested action, preserve evidence, maintain non-disputed service, provide reasons and set a rapid review date. A new claimant with a strong case benefits from the same clarity and from a final order that can be implemented without chaos.

A continuity covenant for courts, registries and operators

The institutional settlement can be stated in twelve commitments.

No immunity. RIRs accept that incorporation, contracts, employment, insolvency and records place them within domestic law and competent judicial authority.
No routing fiction. Courts, parties and registries state that registry acts influence but do not directly command independent route announcements and acceptance.
Exact entities. Every requested measure names the account, record, service, credential, corporate act or dataset affected.
Effect mapping. The registry supplies expected technical signals, external dependencies, propagation time, reversibility and critical-service exposure.
Narrow preservation first. Evidence preservation, transfer locks, dual control and time-limited freezes are considered before public or irreversible changes.
Independent translation. A conflict-checked specialist is available when parties dispute technical consequence.
Non-party visibility. Material dependencies are reported, and directly affected organisations receive notice where lawful and practical.
Reasoned implementation. The order or a filed schedule states why the selected technical level is necessary and why narrower measures are inadequate.
Service continuity. Routine, non-disputed registry functions continue under named authority during governance or insolvency proceedings.
Rollback. Every temporary service-changing action has preserved state, authorised restoration steps and a communication plan.
Bounded public language. Notices describe the order and service effect without converting an interim company-level measure into a universal routing claim.
Measured learning. Registries publish aggregate legal-intervention statistics and test continuity arrangements with courts and operators.

These commitments do not require a new global tribunal. They can be incorporated into registry bylaws, service agreements, board continuity policies, receiver terms, litigation protocols and judicial practice directions. They do not determine who wins a resource dispute. They determine whether the path to judgment protects the wider system from avoidable uncertainty.

The covenant also respects institutional competence. Courts decide rights and remedies. Registries explain and operate registry services. Operators decide routing. Recognition bodies assess institutional fitness. Members govern within the corporate constitution. Continuity comes from connecting these roles without collapsing them into one another.

The measure of legitimacy is controlled translation

Internet governance often speaks as though the choice were between national law and a borderless technical community. A regional registry proves that the choice is false. It is simultaneously a domestic legal person, a regional membership institution and a entity in a globally coordinated technical hierarchy. Each description is true, and none is sufficient alone.

A court order becomes dangerous to continuity not because domestic authority is illegitimate, but because legal language may travel through technical systems whose dependencies were not before the court. A registry response becomes illegitimate not because technical expertise is suspect, but because staff may expand or resist a remedy without transparent authority. An operator reaction becomes destabilising not because networks should ignore registry data, but because a narrow signal may be interpreted as a universal verdict.

The solution is controlled translation. Identify the legal objective. Name the exact registry function. Measure the external effect. Preserve the smallest state necessary. Hear independent technical evidence. Protect non-disputed service. Set a return date. Keep a rollback path. Describe the result without exaggeration.

AFRINIC's public court and governance materials make the need visible. A domestic court can supervise the company and seek restoration of valid governance. The wider registry community can insist, at the same time, that service continuity, documented authority and precise public communication remain central. Those propositions reinforce each other. A functioning registry is more capable of complying with law, and a legally accountable registry is more worthy of global reliance.

The global routing table will never fit inside one courtroom. The registry company can. Institutional legitimacy depends on knowing the difference and preparing, before the next order arrives, for everything that passes between them.