Summary
- Court orders can protect rights without turning a regional internet registry into a hostage to litigation, but only if injunctions, freezes, operating carve-outs and emergency powers are drafted for the ledger services that running networks depend on.
- The most important moment in a registry lawsuit may not be the day judgment is delivered.
The dangerous minute when a registry order is drafted
The most important moment in a registry lawsuit may not be the day judgment is delivered. It may be an earlier, quieter minute, when a lawyer converts grievance into a proposed order and a court is asked to decide which verbs are safe. Restrain. Preserve. Freeze. Maintain. Refrain. Disclose. Take down. Publish. Restore. Do not allocate. Do not transfer. Do not alter. Do not operate outside the ordinary course. In an ordinary commercial dispute those words are already consequential. In a regional internet registry dispute they can become infrastructure policy by accident.
AFRINIC makes the problem concrete because the institution is both small enough to be dragged through domestic corporate litigation and important enough that its records sit underneath large parts of African and Indian Ocean connectivity. It is registered in Mauritius. It serves as the regional registry for IP addresses and autonomous system numbers across its service region. It maintains registration records, supports public lookup services, handles reverse DNS, provides RPKI-related publication, processes resource requests and transfers, collects fees, and answers disputes about who may act for a resource holder. It is not a court, a state or a network operator. Yet a change in its recognised records can affect routing confidence, customer contracts, due diligence, abuse handling, security filtering, cloud onboarding and the market value of scarce IPv4 holdings.
That dual character is precisely why court orders matter. Courts exist to protect rights, restrain unlawful conduct, preserve disputed value and compel accountable conduct. A member, creditor, candidate, customer, supplier or claimant may have no other practical way to stop a registry from taking an action that would be hard to reverse. Injunctions, preservation orders, freezing orders and receivership directions are not hostile to continuity in themselves. Properly designed, they can be the tool that keeps a registry from becoming a private executioner while facts are tested.
But order design can also create a continuity hazard of its own. If the order freezes the wrong account, blocks all ordinary course operations, prohibits alterations without separating routine record maintenance from disputed changes, or leaves staff unsure whether a security update might amount to contempt, the court has not merely preserved the dispute. It has moved the dispute into the operating layer. The registry then faces an impossible choice: obey a broad legal restraint and let records, services or payments decay; or keep the ledger healthy and risk being accused of disobedience.
The economics of this problem are not dramatic in the cinematic sense. They are administrative and therefore more dangerous. A payment to a hosting provider may look mundane until the RPKI repository depends on it. A bank account freeze may look like value preservation until payroll, legal filing fees, data-centre invoices and insurance renewals all pass through the same account. An instruction not to issue resources may look prudent until it also captures returned space, fraud correction, reverse-DNS maintenance or a time-sensitive change needed by a hospital network, cable operator or national exchange point. The drafting line between preservation and paralysis is thin.
The lesson from AFRINIC is not that courts should keep out of registry disputes. The opposite is closer to the truth. When a private registry exercises high-consequence discretion over scarce number resources, judicial scrutiny may be essential. The lesson is that registry litigation needs orders written for infrastructure, not merely for corporate combat. Courts can protect rights while avoiding registry-continuity risk, but only if the order distinguishes the disputed action from the routine function, the claimant's remedy from third-party dependency, and the registry corporation from the ledger services that must remain boring.
AFRINIC turns ordinary remedies into infrastructure events
AFRINIC's crisis has been described in many languages: governance failure, resource dispute, corruption scandal, election breakdown, receivership, regional legitimacy struggle, institutional survival fight. For continuity purposes the label is less important than the channel through which conflict travels. The channel is corporate and contractual litigation in Mauritius. The effects, however, are not confined to a corporate balance sheet.
Reported accounts of the crisis show the pattern. The registry has faced years of disputes with Cloud Innovation, a Seychelles-based member associated with Lu Heng and Larus. AFRINIC accused Cloud Innovation of breaching its service agreement over use of IPv4 resources; Cloud Innovation contested AFRINIC's position and pursued litigation. The registry then became unable to appoint a stable board or chief executive for a prolonged period. Courts in Mauritius appointed a receiver to help restore governance and oversee elections. Election attempts in 2025 became contested, with allegations around powers of attorney and voting legitimacy. Subsequent reporting described a board finally being elected, criticism of the election arrangements, efforts to unfreeze accounts and resume work, and fresh litigation over dissolution, communications and the meaning of court orders.
This is not a tidy governance chronology. Nor should it be used as one side's morality play. For the economics of court orders, the key fact is that legal remedies aimed at one controversy can shape the registry's ability to perform unrelated functions. A court that restrains a resource withdrawal may protect a holder from irreparable harm. A court that freezes bank accounts may preserve money for a creditor or claimant. A court that appoints a receiver may protect an institution from management vacuum. A court that orders a correction to member status may prevent an election distortion. A court that restrains misleading public claims about judicial approval may protect the authority of the court. Each order may be understandable on its own record. Together, broad or poorly coordinated orders can produce a mesh of compliance risk around the very services that all sides say should continue.
The economic reason is that a registry is not a normal defendant. Its ordinary work is a set of low-margin trust functions. It is expected to keep uniqueness, accuracy and publication alive even while people contest who controls it, who pays it, who votes in it, which policies bind it and which rights attach to resources. That means the court's remedy can become part of the registry's operating environment. A bank freeze is not just a creditor remedy; it is a question about whether renewal invoices get paid. An injunction against allocation is not just a scarcity remedy; it is a question about whether the queue, returned space and fraud corrections are distinguishable. An order about communications is not just a reputational remedy; it is a question about whether staff can tell members what has changed without crossing an ambiguous line.
The problem is magnified by IPv4 scarcity. When addresses were administratively abundant, delay and ambiguity were irritating but less likely to alter large capital decisions. Scarcity changed that. AFRINIC's remaining IPv4 pool, its soft-landing rules, its transfer posture and its recognition of existing holdings all feed into operator planning, leasing economics, lender confidence and market pricing. A court order that delays a decision by three months can change bargaining power. An order that blocks transfer execution can affect liquidity. An order that clouds holder standing can alter credit risk. An order that leaves public records stale can make a perfectly functioning network look suspect to counterparties.
Court process therefore becomes part of the price of registry uncertainty. The parties may experience it as litigation strategy. Other networks experience it as a risk premium. They ask whether their resource request will be processed, whether their reverse DNS can be updated, whether their RPKI status will remain coherent, whether the bank account behind the registry will be available, whether the next board action will be challenged, whether a receiver has authority to approve routine changes, and whether an emergency instruction from a court will be operationally intelligible to engineers. Those questions are not abstract. They decide whether networks spend scarce cash on growth, lawyers, redundancy, brokers, cloud workarounds or delay.
This is why the judicial design question matters more than the institutional rhetoric. Official statements can insist on continuity. Litigants can insist they are protecting continuity. Outside coordination bodies can warn of systemic risk. Those positions are exhibits, not answers. The answer lies in the order's architecture: what exactly is restrained, what exactly is preserved, who exactly may act, how ordinary operations are carved out, which services must remain funded, and how quickly ambiguity can be resolved.
Preservation is not the same as stopping everything
The language of preservation is attractive because it sounds neutral. Preserve the status quo. Preserve assets. Preserve records. Preserve rights. Preserve the business. In a registry dispute, however, "status quo" can mean at least three different things. It can mean the last verified registration state. It can mean the corporate management position before a contested act. It can mean the practical continuity of services used by running networks. Those are not always aligned.
Suppose a resource holder challenges a threatened withdrawal. The last verified registration state may favour leaving the holder's entries unchanged while the dispute is heard. The corporate position may favour letting the registry enforce its interpretation of the service agreement. The network-continuity position may favour no immediate disturbance to routing, reverse DNS, RPKI or public lookup records, even if the registry is allowed to continue collecting fees and preserving evidence. A good order says which of these it is preserving. A weak order uses "status quo" as if the answer were self-evident.
The distinction matters because registries change records constantly. Some changes are contested and economically significant. Others are housekeeping: contact updates, abuse mailbox changes, reverse-delegation corrections, certificate repository maintenance, resource-request ticket handling, billing status entries, name changes after corporate reorganisation, security patches, and publication updates that keep public data consistent. If an order says "do not alter the registry" without a schedule of ordinary maintenance exceptions, staff may rationally stop too much. If it says "continue normal operations" without defining who may approve actions and which actions are normal, parties may later argue that every unfavourable update was an abuse of the carve-out.
The same ambiguity appears in account freezes. A freezing order can protect money from dissipation. But a registry's funds are not only distributable value. They are working capital for continuity. Salaries, data-centre payments, cloud services, legal compliance costs, audit fees, security tools, insurance, bank charges, election expenses and office obligations may all be necessary to keep the registry functional. A blanket freeze that lacks an operating-expense mechanism can turn preservation into slow insolvency. A carefully drafted freeze can preserve disputed value while allowing verified ordinary-course payments under reporting, caps, dual approval or receiver oversight.
AFRINIC's reported history gives this distinction special force. Outside observers have described the registry's accounts being frozen during earlier litigation and the institution struggling to perform core functions during prolonged dispute. That does not prove every freeze was wrong or every claimant was irresponsible. It proves that courts hearing registry disputes need evidence about operational cash flow, not only evidence about liability. The question is not merely, "Should funds be restrained?" It is, "Which funds, from which account, subject to which exceptions, approved by whom, reported when, and with what emergency mechanism if a service bill must be paid before the next hearing?"
Preservation also has a time dimension. A restraint that is tolerable for ten days may become destructive over six months. Courts commonly understand delay risk in commercial cases, but registry continuity has a particular shape. The damage is often not one spectacular outage. It is the accumulation of stale records, deferred approvals, staff attrition, missed security maintenance, members giving up on formal channels, counterparties demanding discounts, and engineers building private workarounds around the official system. The registry may still be online while its credibility drains away.
That is why registry orders need review dates and compliance windows that match operations. If the court restrains a category of action, it should know when the next billing cycle falls, when certificates or publication jobs need renewal, when staff contracts or service subscriptions require payment, how urgent member requests are triaged, and how emergency updates are logged. Preservation is an active discipline. It is not a legal pause button.
Injunction scope is an economic instrument
An injunction is usually described in legal terms: prohibitory or mandatory, interim or final, directed at a person or entity, enforceable by contempt. In a registry setting it is also an economic instrument. Its scope allocates risk among the claimant, the registry, other members and downstream users. A narrow order can stop a disputed act while leaving the registry's neutral services intact. A broad order can transfer delay costs to everyone who depends on the registry, including parties who have never heard of the lawsuit.
The first scope question is the verb. "Do not revoke" is different from "do not alter". "Do not transfer the disputed block" is different from "do not process transfers". "Do not make representations that the court has approved a leasing structure" is different from "do not discuss the commercial use of resources". "Do not allocate from this contested pool" is different from "do not issue IPv4 addresses". The difference may look technical to a judge pressed for time. To the registry and its members it is the difference between a bounded restraint and an operating ban.
The second scope question is the asset or record covered. In AFRINIC-related disputes, the same proceeding may refer to member status, shareholder classification, powers of attorney, resource registrations, account control, IPv4 blocks, policy implementation, communications, or corporate governance. Those categories should not be blurred. A member-status correction does not necessarily require a freeze on resource maintenance. A communications restraint does not necessarily require a ban on service notices. A challenge to election rules does not necessarily require an interruption to RDAP publication. The order should map each restraint to the exact risk it is meant to prevent.
The third question is who is bound. Orders often bind parties, affiliates, officers, representatives or intermediaries. In the registry context that language may be necessary to prevent evasion, but it can also create uncertainty for technical contractors, outsourced service providers, election vendors, hosting firms, banks or communications teams. A contractor maintaining a repository should not have to guess whether a routine publication job is a prohibited act by an intermediary in the litigation. The order should identify safe technical roles and reporting channels, especially where failure to act could harm users who are strangers to the dispute.
The fourth question is how the order handles future facts. Registry litigation changes quickly. A forged authority document may be discovered. A bank may refuse a payment. A security vulnerability may require a patch. A member may show that a reverse delegation is broken. A court may later clarify a prior order. A rigid injunction can become obsolete before the next hearing. A well-designed order includes a mechanism for urgent directions, limited emergency action, notice to parties after the event, and records sufficient for later review. That is not a loophole. It is how preservation survives contact with operations.
The economic point is simple: scope is price. Every additional noun and verb in an injunction creates a cost somewhere. Sometimes the cost is justified because the risk is grave. But registry continuity requires the cost to be visible. Courts should be given evidence not only on why the applicant fears harm, but also on what the requested wording would do to ordinary services, third parties, pending requests, security publication, bank access and staff authority. Litigants should be expected to draft with those effects in mind. If they seek a broad restraint, they should explain why narrower wording would fail.
That discipline protects courts as well as registries. A court that issues a precise order is less likely to be pulled into repeated contempt fights over routine acts. A registry that receives a precise order is less able to hide discretionary conduct behind continuity language. A claimant that obtains a precise order is more likely to preserve the remedy it actually needs. Precision is not procedural fussiness. It is the infrastructure-safe form of judicial power.
Bank freezes can protect claims while starving the ledger
Money is the easiest registry dependency to misunderstand. Courts and claimants see bank accounts as assets. Engineers see services. Staff see salaries. Members see the institutional capacity needed to answer tickets. All four views are real. A bank freeze that ignores any one of them can distort the remedy.
The purpose of a freeze is intelligible. If a claimant has a serious case and a risk that funds will be dissipated, the court may need to restrain movement of money. In a registry crisis, money can also be tied to disputed governance authority: who may sign, who may approve payments, whether a receiver controls the account, whether a board has been validly constituted, whether election or litigation expenses are proper, and whether member fees are being used for legitimate continuity or factional survival. Courts cannot simply assume that every payment request is innocent.
Yet a registry cannot be preserved by cutting off its oxygen. It needs a controlled operating allowance. That allowance should not be a vague promise that "ordinary expenses" may continue. It should identify the categories needed to keep the ledger alive: payroll for essential technical and member-services staff; hosting, connectivity and security costs; RDAP, Whois, reverse DNS and RPKI publication services; audit, insurance and banking fees; court compliance costs; member communications; and limited professional advice necessary to follow the order. Where mistrust is high, payments can be capped, reported and supported by invoices. A receiver, court officer or independent accountant can approve categories. But the existence of mistrust is not a reason to deny all operating spend. It is a reason to design controls.
The economic harm of a badly designed freeze arrives through substitution. If the registry cannot pay, members spend more. They hire lawyers to extract answers. They delay expansion. They pay brokers or intermediaries to work around uncertainty. They hold larger private reserves of addresses. They discount AFRINIC-region resources in transactions. They move customers to cloud providers with more stable public-address arrangements. They treat the registry as a risk source rather than a trust utility. In that sense, starving the registry does not only harm the institution. It taxes the ecosystem.
There is a second-order risk. When accounts are frozen and routine authority is unclear, the people with the most patience and legal budget gain relative power. Small operators, public-interest networks, universities, exchange points and rural providers may depend on timely registry service but lack the capacity to intervene. The freeze then redistributes delay costs toward the least represented users. A court may never hear from them because their harm is dispersed, operational and hard to plead. Good order design tries to see them in advance.
The remedy is not to immunise registry accounts. It is to divide funds by purpose. Disputed or extraordinary payments can be restrained. Routine continuity payments can be allowed under evidence. Litigation spend can be capped or separately approved. Election or governance expenses can be budgeted. Emergency payments can be made subject to immediate notification. The court can require a cash-flow statement, a service-dependency schedule and a list of payments due before the next hearing. That information changes the legal conversation from "freeze or do not freeze" to "preserve value without disabling value preservation".
AFRINIC's reported difficulties show why this matters. A registry can be kept legally alive while becoming operationally brittle. Conversely, a registry can keep services online while its governance legitimacy remains disputed. Bank-order design should not pretend those are the same state. The aim is not to reward the incumbent institution. It is to keep the public record, security publication and member-service channels functional while contested rights are decided.
Courts need operational evidence, not just party narratives
Judges decide on evidence. In a registry dispute, the evidence packet often arrives shaped by the parties' incentives. The applicant explains feared harm. The respondent explains why it must act. Outside coordination bodies may describe systemic stakes. Members may supply declarations about voting, fraud, delay or business impact. The missing document is often the operational map: what registry functions exist, which are affected by the proposed order, which can pause safely, which cannot, and who can perform them if ordinary governance is under challenge.
That map should be concrete. It should list the public services: RDAP, Whois, reverse DNS, RPKI repository publication, routing registry services where relevant, member portals, ticket systems, billing systems, zone or repository publication jobs, backup schedules and security monitoring. It should identify the staff or vendors responsible for each service, the bank payments needed to keep them running, the update frequency, the acceptable downtime, the next renewal or expiry risk, and the consequences of freezing changes. It should also identify which functions are discretionary, such as contested enforcement, policy changes, extraordinary transfers, large allocations or public advocacy.
This evidence helps courts separate two tasks. The first is adjudicating the dispute. The second is preserving the utility function while adjudication occurs. Parties tend to merge them. A registry may claim that any constraint on its chosen enforcement theory threatens internet stability. A claimant may claim that only a complete halt prevents irreparable harm. Both may be overstated. An operational map lets the court ask better questions. Can the disputed withdrawal be paused while contact updates continue? Can a resource transfer be held while reverse DNS remains editable? Can public statements be restrained while service notices continue? Can accounts be frozen except for approved operating expenses? Can policy implementation be paused without stopping ticket triage?
The AFRINIC record illustrates the cost of thin evidence. Public reporting has described allegations of corruption, disputed use of IPv4 resources, court orders, receivership, election irregularities, contested powers of attorney, claims about Cloud Innovation's member status, ICANN intervention, dissolution efforts and communications disputes over whether a court had approved a commercial continuity structure. Those facts matter. But none of them, by itself, tells a court which RDAP update should run, whether a reverse-DNS delegation can be corrected, or how an RPKI publication dependency should be funded. Without operational evidence, judges must infer infrastructure effects from institutional rhetoric.
Operational evidence also disciplines official claims. AFRINIC's own public material identifies its role in distributing and managing number resources and lists services such as reverse DNS, Whois, RDAP, Internet Routing Registry functions and RPKI. Those descriptions are useful as mechanics. They do not prove that every corporate power claimed by AFRINIC is necessary for continuity. Equally, critiques of registry overreach are useful in identifying why unchecked discretion can be harmful. They do not prove that every restraint sought against the registry is safe. Evidence must bridge the gap.
For courts, the practical question is whether the order can be audited after the fact. If staff make an emergency update, what record will show why it was necessary? If the registry refuses a request because of the order, what notice will explain the refusal? If a bank releases money for continuity, what invoice and approval trail will exist? If a party alleges contempt, what technical log will show whether the act was routine maintenance or a prohibited change? A registry order without audit design invites the next dispute.
The better approach is to require a continuity affidavit or schedule whenever a proposed order could touch registry operations. It need not be elaborate in every case. But it should be specific enough to let the court see which services are preserved, which actions are paused, which accounts are accessible, which third parties are affected, and how emergency issues return to the court. Litigation then becomes less likely to govern by surprise.
Carve-outs are where rights protection becomes continuity design
Carve-outs are sometimes treated as drafting afterthoughts: the boilerplate at the end of a freeze or injunction that says ordinary business may continue. In registry litigation, the carve-out is the heart of the order. It is the place where a court says that the claimant's rights are protected, but the public record will not be used as hostage.
An effective registry carve-out begins with routine record maintenance. The order should allow updates that do not change the contested entitlement: contact corrections, abuse-contact updates, name-server changes, reverse-DNS maintenance, public-record accuracy corrections, security patches, repository publication, and ticket handling that preserves existing service. If any of those categories could be abused to alter disputed control, the order can require notice, dual approval, escrowed logs or later challenge. But the default should be that the ledger remains accurate unless accuracy itself is the disputed act.
The second carve-out is for security continuity. RPKI, route-origin authorisation, repository freshness, certificates, manifests, revocation information and relying-party access do not fit comfortably into ordinary corporate-restraint language. They need explicit preservation. A court does not need to decide the technical merits of every publication act. It does need to know that ambiguity around the order will not cause staff to stop publishing necessary security material or to revoke status as a litigation tactic. The safest approach is to preserve the last verified operational state for contested resources while allowing routine publication and correction needed to keep that state coherent.
The third carve-out is for financial continuity. Approved expenses should be tied to service categories and reporting. A registry can be required to submit a rolling budget, identify critical vendors, keep payments within caps, and notify parties of extraordinary expenses. Where governance authority is disputed, a receiver or independent approver can authorise payments. The point is not convenience. It is to avoid the paradox of preserving the registry's value by preventing it from paying for the systems that create that value.
The fourth carve-out is for member communications. A registry under litigation must still tell members how to submit requests, whether elections are scheduled, whether services are degraded, how to preserve credentials, how to update records and how court orders affect their interactions. Communications restraints may be appropriate where a party alleges misleading claims about court approval or defamatory statements. But the order should preserve neutral service notices, court-directed notices and factual operational updates. Silence can be as harmful as speech when members must decide whether to act.
The fifth carve-out is for emergency action. Fraud, hijack attempts, credential compromise, forged authority documents, severe publication failures and security incidents cannot always wait for a scheduled hearing. A registry needs limited authority to prevent immediate harm, with strict logging and rapid notice. The standard should be high enough to prevent abuse, but not so high that staff stand aside while a known falsification enters the record. Emergency powers should preserve the last verified state or prevent a clearly unauthorised change; they should not become a route to decide the merits of a resource dispute.
These carve-outs are not gifts to AFRINIC or any registry. They are constraints on everyone. They tell the registry what it may not hide behind continuity. They tell the claimant what harm it may not impose on bystanders. They tell the court what evidence it will receive if the carve-out is used. They tell members that the litigation boundary has not swallowed the service boundary. In a scarce-address environment, that clarity has economic value.
Neutrality means the last verified state, not institutional victory
Registry neutrality is often invoked as a virtue but rarely defined under litigation pressure. It cannot mean that the registry always wins because it is the registry. It cannot mean that the claimant always wins because the registry is disputed. It cannot mean that nothing ever changes. The most workable definition is narrower: while contested rights are adjudicated, the registry should preserve the last verified operational state, block incompatible alterations, maintain public services and avoid taking merits positions through the ledger.
This principle separates dispute resolution from infrastructure paralysis. If a resource block is contested, the registry can mark that a dispute exists where appropriate, preserve existing operational data, prevent transfer to a third party, and continue publication services that keep networks stable. It need not revoke the holder, reissue the block, rewrite history or disable security status before the dispute is decided. If a signatory's authority is challenged, the registry can pause high-consequence changes while allowing low-risk maintenance through verified alternative contacts. If a court is deciding whether a board or receiver has authority, the registry can continue staff-level services under court-approved limits without treating every routine act as a governance endorsement.
AFRINIC's crisis shows how hard this is. The registry has been described by supporters as a fragile institution needing protection from paralyzing litigation. It has been described by critics as an overreaching gatekeeper over economically critical assets. Both descriptions may capture part of the risk. Neither should decide the ledger. Neutrality requires an order design that prevents the registry from converting its institutional position into a decisive operational act, while also preventing litigants from converting their grievances into a shutdown.
The last-verified-state principle is also economically legible. Operators and counterparties need to know what happens during litigation. If the default is "records stay coherent unless a court specifically orders otherwise", the market can price disputes more rationally. If the default is "any legal conflict may trigger revocation, publication uncertainty, transfer blockage or account paralysis", the market adds a larger discount to all resources tied to that registry. That discount falls not only on sophisticated litigants. It falls on ordinary networks that need financing, customers, upstreams and public-sector contracts.
Neutrality should also govern public communications. Courts may need to restrain parties from claiming judicial endorsement for a commercial model, a leasing programme, a governance structure or a transfer theory. At the same time, registry communications should not imply that a non-final interim order has decided questions it has not decided. Public statements around AFRINIC disputes have already shown how quickly parties argue about what a court order means. A continuity-safe order should identify who may summarise it, what must be said to members, and what claims about approval, legality or final status must be avoided. The aim is not to win the press release. It is to prevent the court's authority from becoming market propaganda.
The hardest neutrality cases involve fraud or corruption allegations. AFRINIC's past includes reported address-heist allegations involving a former executive and valuable IPv4 space, as well as later claims and counterclaims around proxy voting and member status. Fraud control cannot wait for philosophical purity. But the remedy still needs a neutral architecture. Stop the clearly unauthorised change. Preserve evidence. Notify affected parties. Allow cure where documents are defective rather than forged. Escalate contested questions to an independent forum. Keep the running network stable unless immediate technical harm requires action. That is how neutrality protects both rights and continuity.
Compliance windows prevent contempt risk from becoming ledger risk
Contempt risk is an underappreciated operational variable. Courts issue orders to be obeyed. If wording is broad and penalties are severe, rational staff and officers may over-comply. They may refuse routine updates, delay communications, avoid signing payments, or escalate every ticket to lawyers. From the outside that looks like registry dysfunction. From the inside it may be risk management.
Compliance windows reduce that risk. An order can distinguish immediate prohibitions from staged obligations. It can require disputed actions to stop at once, while allowing the registry a short period to identify affected services, notify the court of operational issues, propose a continuity budget, submit a list of pending member requests and ask for clarification. It can require parties to confer on carve-outs within a defined time. It can set a rapid return date for disputes about whether an action is routine or prohibited. Those mechanisms keep a temporary order from hardening into an accidental operating constitution.
The same discipline applies to mandatory orders. If a court requires a takedown, correction, publication or notice, the order should say what must happen, by when, through which channels, and what proof of compliance is sufficient. In 2026 reporting around AFRINIC, a dispute arose over claims that a court order supported a "shareholder-position continuity" structure tied to IPv4 leasing. AFRINIC's communications and responses from Cloud Innovation and Larus showed how contested the meaning and service of an interim order can be. Whatever one's view of that dispute, it illustrates the need for precision. A court order that affects public statements about registry resources should be specific enough that compliance does not depend on rival press interpretations.
Compliance windows matter for banks as well. If an account is frozen but operating expenses are allowed, banks need documents they can act on. A vague exception may not be enough for a risk-averse bank officer. The order should specify account numbers or categories where possible, authorised signatories, approval documents, payment limits and the status of recurring payments. Otherwise the court may believe it has preserved operations while the bank, fearing liability, blocks them.
Technical compliance also has lead times. Updating a public record is not the same as changing a sentence in a letter. RPKI publication can involve repository processes and relying-party propagation. Reverse-DNS changes can depend on parent-zone delegation, name-server tests and DNS caching. RDAP and Whois publication may involve database replication, validation rules and public-query systems. If an order demands immediate alteration of technical state, it should be told what "immediate" means in engineering terms and what rollback or verification is required. Speed is sometimes necessary. Unspecified speed can create errors.
A useful compliance order asks for logs. If the registry pauses a category of requests because of the order, it should maintain a list of affected requests, dates, reasons, risk category and proposed next step. If it takes emergency action, it should record the evidence, authority, person approving, affected records and notice given. If it spends money under a carve-out, it should keep invoices and approvals. These logs reduce later fights and help the court see whether continuity language is being used honestly.
The economic value of compliance windows is that they reduce fear. Staff can act within known boundaries. Members can understand delay. Courts can correct overreach. Parties can challenge misuse. Without those windows, the registry becomes a place where legal uncertainty is transmitted directly into operational hesitation.
Notice and cure separate fraud control from capital control
Registry orders often arise from fear that someone will do something irreversible: transfer resources, alter a record, cast a vote, drain an account, publish a misleading claim, revoke a block or destroy evidence. The fear may be justified. But in an address registry, speed and certainty have to be balanced against the risk of turning every defect into a weapon.
Notice and cure are the first safeguard. If a resource holder's paperwork is incomplete, the registry can notify the holder, identify the defect and allow a reasonable cure period before a high-consequence change. If a contact appears stale, the registry can use multiple channels. If a payment is late because accounts are under dispute, the order can distinguish inability to pay from refusal. If a proxy or authority document is challenged, the registry can pause the vote or change tied to that document while seeking verification, rather than annulling broader activity unless the defect is systemic.
Fraud is different. A forged power of attorney, account compromise, fabricated corporate authority chain or unauthorised transfer instruction may require immediate protective action. But even there, the action should be framed as preservation, not punishment. Freeze the specific change. Preserve logs. Notify the apparent holder through independent channels. Require proof from the person claiming authority. Escalate quickly to the court or a neutral decision-maker. Avoid revocation, redistribution or public accusation unless the evidence and authority support it. That approach controls fraud without letting the registry or a litigant convert fraud suspicion into capital control.
The AFRINIC election disputes show the wider point. Reporting in 2025 described allegations that powers of attorney had been used or lodged without consent, that the election was suspended and annulled, and that ICANN demanded greater explanation from the receiver. Those facts concern governance rather than resource records, but the mechanism is similar. A system can respond to suspected falsification by verifying the affected instruments, preserving the vote evidence, and explaining the threshold for annulment. Or it can leave participants uncertain about whether one disputed document, a cluster of documents or a systemic defect caused the result. Uncertainty then becomes its own legitimacy damage.
The same logic applies to resource administration. If a registry claims misuse, breach or false documentation, its process should separate ordinary verification from irreversible harm. Notice, cure, reasons, appeal and continuity protection are not bureaucratic luxuries. They are how a private registry avoids becoming judge and executioner over scarce inputs. A court order can reinforce that discipline by requiring that adverse registry action include reasons, cure periods where safe, emergency exceptions where necessary, and preservation of the last verified operational state during dispute.
For courts, notice and cure also reduce the need for broad emergency orders. If the registry already has credible bounded procedures for suspected fraud, a court may not need to freeze an entire class of operations. If the registry lacks such procedures, the court can impose them as interim safeguards. Either way, the order should avoid a binary choice between "let anything happen" and "stop everything".
The economic stakes are plain. Scarce IPv4 resources have market value, but their value depends on stable recognition. A holder facing a registry challenge may lose customers, financing or bargaining power long before final adjudication. A registry facing forged documents may lose credibility if it cannot act quickly. Notice and cure design makes those risks commensurable. It says: protect against falsification, but do not create value destruction as a default remedy.
RDAP, Whois, reverse DNS and RPKI need named protection
The most dangerous continuity failures are not always visible to non-specialists. A court may understand that the registry should keep "operating", but not know which operations need to be named. RDAP and Whois let others query registration data. Reverse DNS maps addresses back into names and supports reputation, email, security and operational diagnosis. RPKI lets networks publish route-origin authorisations and lets relying parties validate whether a route announcement is consistent with registry-backed assertions. Routing registry records, where used, inform filters and routing policy. These are not ornamental services. They are publication layers through which registry recognition becomes usable by the network.
An order that says "do not alter records" without preserving publication can be risky. If publication jobs stop, the last verified state may become inaccessible or stale. If RPKI maintenance is frozen, relying parties may encounter inconsistent or expired material. If reverse-DNS updates are blocked, an operator may be unable to maintain reputation or service continuity. If RDAP and Whois data are not corrected for harmless contact errors, due diligence and abuse handling suffer. Conversely, an order that allows all updates without restriction may let a party change control under the cover of maintenance. Named protection is the middle path.
For contested resources, the safest rule is usually preservation of the last verified operational state plus explicit permission for technical acts necessary to keep that state publicly coherent. That may include repository publication, renewal, correction of clerical errors, security patching, backup restoration, and changes that do not alter the contested entitlement. If an RPKI action would materially change routing authorisation for a disputed block, it may need notice or court approval. If a reverse-DNS change merely updates name servers for an uncontested holder, it should not be trapped by a broad litigation freeze. The order should make that distinction visible.
AFRINIC's own service descriptions provide the factual mechanics: number resource management, reverse DNS, Whois, RDAP, Internet Routing Registry functions and RPKI appear among the services it offers or supports. Those mechanics are enough to show why a registry order must speak in operational categories. They are not a reason to accept any official conclusion about institutional immunity. A court can protect RDAP, Whois, reverse DNS and RPKI continuity while still restraining disputed enforcement, communications, governance or transfer acts.
This matters because publication services create third-party reliance. A bank assessing an address-dependent borrower may query public registration data. A cloud provider may inspect address reputation and routing authorisation. A government customer may require stable reverse DNS or abuse contacts. An upstream provider may use route records or RPKI status in filtering. A security team may rely on public contacts to handle abuse. If litigation makes those signals unreliable, the harm spreads beyond the parties.
Named protection also helps staff. Engineers should not have to translate legal abstractions into service-risk decisions under threat of contempt. If the order says RPKI repository publication must continue for existing verified material, staff know what to do. If it says no new ROA affecting disputed resources may be issued without notice, staff know when to stop. If it says reverse-DNS delegations may be corrected for uncontested resources but not transferred for disputed resources, ticket handlers can triage. This is how legal certainty becomes operational certainty.
The broader lesson is that registry continuity is not one thing. It is a bundle of publication, security, record, payment, staffing and adjudication functions. Courts protect it by naming the bundle, not by invoking continuity in the abstract.
Courts can separate dispute resolution from infrastructure paralysis
A registry dispute should have somewhere to go. That place may be a court, arbitration forum, internal appeal mechanism, receiver-supervised process or some combination. What it should not do is use infrastructure fragility as leverage. The central design task is dispute isolation: contain the contested question without freezing the services that are not in dispute.
Dispute isolation begins with classification. Is the case about corporate governance, board validity, member status, a resource-holder agreement, alleged misuse, payment default, fraud, transfer execution, public communications, bank access or dissolution? Each category affects continuity differently. Governance disputes may require limits on extraordinary decisions but not on technical operations. Resource disputes may require preservation of specific records but not a general allocation halt. Bank disputes may require controlled operating budgets. Communications disputes may require wording restraints without cutting off member notices. Dissolution or derecognition questions may require transition planning rather than ordinary enforcement.
AFRINIC has faced several of these categories at once, which is why it is a demanding test case. The temptation in such a crisis is totalising language. One side says the institution must be saved. Another says the institution must be replaced. One side says litigation is paralysis. Another says litigation is accountability. The court's job is not to adopt either total claim as infrastructure policy. It is to define what can continue, what must pause, what evidence is needed, and what happens if the dispute outlasts the interim order.
Dispute isolation also reduces strategic behaviour. If litigants know that suing the registry can freeze all operations, litigation becomes a tool of paralysis. If the registry knows that invoking continuity can defeat all restraints, continuity becomes a shield for overreach. Both incentives are bad. A court that consistently preserves routine services while restraining specific disputed acts lowers the reward for both strategies. Claimants can obtain meaningful protection without threatening bystanders. Registries can maintain services without treating every lawsuit as an existential attack.
Transition planning is part of the same logic. Discussions in the wider number-resource community after AFRINIC's difficulties have raised the possibility of emergency support, successor operation or derecognition of a dysfunctional registry. Official coordination proposals may carry their own institutional biases, but they recognise an important fact: the registry function can be conceptually separated from the incumbent corporate shell. Courts do not need to decide global registry architecture in an interim order. They do need to avoid orders that make any future transition harder by letting records decay, services lapse or evidence disappear.
The court can therefore require escrowed backups, preserved logs, documented service dependencies, neutral access for a receiver or court officer, and continuity plans that do not decide the merits. It can require that any transition step protect uniqueness, registration accuracy, publication and running-network stability. It can prevent parties from using dissolution, receivership or derecognition rhetoric to justify unilateral control of records. It can ask who will maintain RDAP, Whois, reverse DNS and RPKI if a corporate question remains unresolved. Those are not policy adventures. They are prudent safeguards.
Infrastructure paralysis is often presented as an unavoidable byproduct of legal conflict. It is not. It is frequently a design failure: orders too broad, evidence too thin, carve-outs too vague, review dates too distant, payment channels too rigid, emergency powers too unclear and communications rules too politicised. Courts can do better, and litigants can be required to help them do better.
The measure of a good order is what remains boring
The best registry order is not the one that generates the strongest press release. It is the one that leaves the most important services boring. Members should still know how to submit routine requests. Public records should still answer queries. Reverse DNS should still behave. RPKI publication should still be coherent. Staff should still be paid. Bank officers should know which payments are allowed. Parties should know which acts are restrained. Courts should receive logs rather than rumours. Downstream users should not have to learn the names of every litigant to understand whether their network identifiers remain stable.
That standard is demanding because it refuses two easy stories. It refuses the story that registry continuity requires deference to every incumbent institutional claim. A registry that maintains critical records must be accountable precisely because the records matter. If it threatens withdrawal, transfer refusal, policy implementation or public statements beyond its legitimate scope, courts may need to restrain it. It also refuses the story that rights protection justifies broad paralysis. A claimant who freezes routine services may win leverage while harming networks that have no voice in the case.
AFRINIC's future will remain legally and politically contested. Its governance crisis, reported corruption history, dispute with Cloud Innovation, receivership experience, election controversies, board legitimacy questions, IPv4 scarcity and institutional continuity problems are too deep to be solved by one article or one order. But the court-order lesson is precise. The judiciary can protect rights and preserve infrastructure if it treats registry orders as operational instruments. The order must say what it restrains, what it preserves, what it funds, who may act, how emergencies are handled, how records are kept, how quickly ambiguity returns to court, and which services must continue regardless of who is winning the merits argument.
The economic payoff is lower uncertainty. Resource holders can distinguish litigation risk from service-collapse risk. Buyers, lenders and customers can evaluate disputed holdings without assuming the entire registry layer is unstable. Small networks can keep using formal channels rather than building costly workarounds. Courts can enforce their authority without being blamed for accidental outages. Registries can be held to account without being disabled as ledgers.
The principle travels beyond AFRINIC, but AFRINIC is the sharpest current illustration because its crisis has repeatedly forced domestic corporate law to touch transnational infrastructure. That contact is not illegitimate. Private registries are incorporated somewhere and should not float above ordinary law. The danger lies in pretending ordinary remedies can be copied into registry disputes without adaptation. They cannot. A regional registry is a private legal entity, but the continuity function it performs is a shared dependency.
Court orders can protect rights. They can stop abuse, preserve evidence, restrain false claims, maintain disputed value and force governance back into lawful channels. They can also create registry-continuity risk if they freeze the wrong action, omit routine-operation exceptions, starve the operating account, ignore publication services, or force staff to choose between contempt risk and ledger integrity. The difference is design.
In a registry case, justice is not measured only by who is restrained. It is measured by whether the record remains trustworthy while restraint is in force. The court should be able to decide the dispute without becoming the accidental administrator of every stale record, unpaid invoice and broken delegation. The registry should be able to continue its narrow function without using continuity as a claim to immunity. The parties should be able to fight over rights without conscripting RDAP, Whois, reverse DNS, RPKI and the customers of running networks into the fight.
The quiet triumph of a good order is that nothing dramatic happens. The disputed act stops. The evidence is preserved. The money needed for services is controlled but available. The court receives information. The registry does not expand its power. The claimant does not paralyse the ledger. Packets keep moving. Public records keep answering. The legal system does its work without making the infrastructure prove, once again, how fragile trust becomes when the wrong words are frozen into an order.