Who May Trigger an RIR Derecognition Review?

Summary

• The 2024 ICP-2 principles gave the NRO Executive Council exclusive origination power, while the August 2025 draft allowed any RIR, ICANN, or a coalition comprising twenty-five percent of the affected RIR's members or 2,000 members, whichever is lower, to submit a derecognition proposal.
• That model excludes affected resource holders and operators who lack voting membership, gives governments and wider community entities no direct route, and lets the same peer institutions that may face conflicts control whether a case reaches ICANN for a final decision.
• Standing should be graduated. Evidence reporting must be broadly accessible; an independent compliance review should have a lower but abuse-resistant threshold; a formal derecognition petition should require a substantial cross-regional constituency; and permanent service transfer should require separate resource-holder authorization and independent review.

Failure can be visible and still institutionally inaudible

Every accountability system has a threshold question before the merits: who may demand that the institution answer? Courts call a related concept standing. Corporate law speaks of member, director, creditor, regulator, or derivative rights. Administrative systems define complainants, petitioners, inspectors, and reviewing authorities. The RIR debate needs its own vocabulary, but the underlying choice is the same. If nobody with evidence can start a competent examination, the strongest substantive standards remain decorative.

This is why trigger eligibility deserves more attention than the final derecognition vote. The last decision is dramatic, but it is rare. The ordinary exercise of power happens earlier. A complaint is accepted or ignored. An audit is opened or delayed. A factual allegation is tested or left inside a mailing-list dispute. A member coalition is counted or rejected. An emergency is characterized as a temporary service problem or a governance crisis. Whoever controls those gates can make failure actionable or keep it unofficial.

The October 2023 NRO Executive Council request recognized that ICP-2 lacked procedures for validating continuing compliance, reviewing it, and specifying outcomes. The request was prompted by consideration of the AFRINIC situation, but the resulting rules were intended to apply across the registry system. That was the correct generalization: accountability cannot be invented for one crisis and then called neutral.

Yet a general standard can still be inaccessible. The question is not whether members, holders, governments, or operators can send an email, publish criticism, or lobby an RIR. They can. The question is whether the recipient must do anything: acknowledge the evidence, preserve relevant records, assess admissibility under published criteria, initiate an independent examination when a threshold is met, explain a refusal, and permit review of that refusal.

Standing, in this analysis, means that enforceable right to trigger a defined institutional response. It does not mean an automatic right to derecognize. Opening a file, ordering a compliance review, filing a formal removal petition, authorizing temporary continuity, and approving a permanent successor are different acts. They should not share one threshold.

In 2024, the door had one institutional handle

The October 2024 proposed principles were unambiguous about origination. Any proposal to recognize a candidate RIR or derecognize an existing one had to originate from the NRO Executive Council after a majority vote. ICANN would have final authority, following consultation with the RIRs.

That design had a certain logic. The NRO EC brings together the heads of the RIRs. It can assess system-wide implications and coordinate institutions that would have to implement a status change. Requiring a majority could block a unilateral campaign by one aggrieved peer. ICANN's final role could add distance from purely regional interests.

But the principle confused competence to coordinate with exclusive standing to initiate. A registry executive may have no incentive to advance allegations against a peer. The peers may fear reciprocal scrutiny, transition costs, or political controversy. They may sincerely disagree about whether facts warrant action. They may also lack direct visibility into the affected registry's member records, domestic litigation, staff constraints, or local operating conditions. Exclusive origination turns all of those judgments into a silent preliminary veto.

The composition problem is sharper during institutional failure. The affected RIR's chief executive might be absent, disputed, conflicted, or operating under a contested board. If that person participates in the NRO EC, how is recusal handled? If the seat is vacant, is the affected region heard? If the remaining executives vote, what regional evidence must each consider? The four-page principle did not say.

Nor did it give formal triggering power to the NRO Number Council, even though that council was conducting the revision and includes regional selections. The distinction matters. The NRO EC represents the registry institutions. The NRO NC, which also serves as the ASO Address Council, has a different composition and advisory role. Neither body is identical to the population of resource holders or to the full regional community.

A single institutional handle can make a door orderly. It can also keep it closed. The 2024 consultation correctly prompted questions about wider stakeholder input and gatekeeping, but the principle itself did not create a route around inaction.

The 2025 drafts opened the door, but only partway

The first RIR Governance Document draft, published in April 2025, replaced exclusive NRO EC origination with two paths. Any RIR or group of RIRs could submit a derecognition proposal. So could at least twenty-five percent of the members of the registry facing the proposal. The other RIRs, excluding the target, had to recommend removal unanimously before ICANN could make the final decision. ICANN itself could not initiate.

This was a major improvement in internal accountability. Members no longer had to persuade a peer executive to carry their case. The affected registry's own electorate could force the allegation into a formal channel if it assembled a large enough coalition. The draft also made peer responsibility explicit: each remaining RIR had to decide and publish reasons.

The model still had severe limits. Twenty-five percent can be an enormous organizing burden in a large membership body. It may be unreachable during a crisis if membership records are disputed, communications are controlled by the incumbent, or participating organizations fear service retaliation. Different RIRs define and count members differently. A percentage that is realistic in one region can be prohibitive in another. And the members entitled to vote for a governing body are not necessarily all the resource holders exposed to registry services.

The August 2025 Version 2 draft adjusted the rule. A petition could come from any RIR or group of RIRs, ICANN, or a group comprising at least twenty-five percent of the target RIR's total members or 2,000 members, whichever is lower. It also required the petition to identify the specific provisions allegedly breached, required publication, and gave the target a reasonable opportunity to respond before peer recommendations.

Adding ICANN created a route when peers or members would not act. The numerical cap reduced the disparity between differently sized registries. Requiring specified breaches made a petition more than a political demand. Publication and response improved fairness.

Still, the gate remained narrow. Members had a direct route; resource holders as such did not. Governments, operators, technical community entities, civil society organizations, staff, customers of members, and affected end users could provide information but could not compel the formal response unless they also qualified through one of the three listed doors.

“Member” and “resource holder” are different constituencies

The Version 2 draft defines a member as a legal or natural person entitled to participate in RIR governance by voting for the governing body. It defines a resource holder as a person or entity holding number resources registered with an RIR. The distinction is deliberate and important.

Membership is an institutional relationship. It may depend on an application, fees, corporate status, account standing, or other regional rules. Holding is an operational relationship reflected in registry records. Many organizations will be both members and holders. The categories need not be identical. A holder can depend on registration accuracy, reverse DNS, route-security services, transfer processing, and account authentication even if it does not possess an ordinary vote.

The standing rule privileges the governance constituency over the service constituency. That may be defensible for a complaint about a board election or member meeting. It is harder to defend where the alleged failure concerns records, service availability, discriminatory administration, security credentials, or a proposed transfer to an interim operator. The people directly exposed to those functions should not need a voting status unrelated to the harm in order to make the failure formally reviewable.

There is a second asymmetry. Recognition of a candidate requires broad support from resource holders in the proposed region. The draft therefore accepts that holders are a necessary constituency when a registry gains responsibility. Yet a coalition of holders is not listed as a direct petitioner when a registry may lose that responsibility. Entry asks whether they support the service institution. Exit asks whether enough voting members or external institutions will act for them.

Membership records can themselves become part of the dispute. If the allegation is that the RIR improperly suspended members, manipulated eligibility, failed to recognize valid representatives, or allowed affiliates to multiply votes, the target institution may control the denominator used to determine whether twenty-five percent has petitioned. A standing rule cannot depend entirely on a disputed party's unreviewed list.

The system needs a holder route. It should not treat every address assignee as interchangeable or allow one conglomerate to multiply signatures through affiliates. It should verify authority, aggregate controlled entities, protect confidential identity where necessary, and show regional coverage. But the difficulty of counting holders is not a reason to exclude them. It is a reason to design the count before a crisis.

Boards can speak for an RIR, unless the board is the problem

The drafts allow “any RIR” to submit a proposal, but an RIR is a legal organization. A human organ must decide to act. In most cases that will be a governing board, an authorized executive, or another body empowered under the registry's own rules. The global text says decisions should follow each entity's relevant procedures. That leaves board standing largely implicit.

Implicit board authority is ordinary and often sufficient. A board owes duties under its jurisdiction, supervises management, and can authorize legal or institutional action. If one RIR discovers credible evidence that a peer is endangering shared services, its board should be able to direct a petition. Requiring a separate global vote for every peer complaint would be cumbersome.

The difficulty appears when the affected RIR's own board seeks outside review. A functioning board might reasonably ask for an independent compliance examination, temporary assistance, or clarification of recognition status. But if board authority is contested, accepting its petition may decide the internal dispute indirectly. One faction can present itself as the institution and invite external bodies to validate its control. Rejecting the petition can be equally consequential if the lawful board is trying to protect records from an unauthorized faction.

The answer is not to deny boards a voice. It is to classify the request. A board with undisputed authority can speak for the RIR. A contested board can submit evidence as a claimant, but an independent screener should verify its legal authority before treating the request as the act of the corporation. Where domestic court orders exist, they should be read precisely. A temporary corporate order does not necessarily settle global recognition; a global status opinion does not settle corporate office.

Boards also should not be the only internal organ with access. An audit committee, ombudsperson, supervisory council, or valid member assembly may possess evidence that management withholds. The global standard can recognize submissions from those bodies without assuming they bind the RIR. Their evidence can trigger screening even when their legal authority to request final removal is absent.

This distinction between evidentiary access and institutional representation is fundamental. It permits facts to surface without letting every faction wield the corporate name.

ICANN can trigger, but should not investigate and judge its own case

Version 2 gave ICANN power to initiate a derecognition proposal. The drafters' summary of changes explains that the change responded to concern that RIRs might act in self-interest. ICANN could also initiate an ad hoc audit, while it still could not complete derecognition without unanimous recommendations from the other RIRs.

This provides an escape from peer silence. ICANN sits outside the five registry corporations and has an established public accountability structure. It oversees the IANA functions operator and already participates in global number-policy arrangements. If credible evidence shows that a regional failure threatens the registry system, it should not be powerless merely because no incumbent wishes to file.

But “ICANN may initiate” raises two role questions. First, who within ICANN decides? The Board, chief executive, staff, a designated compliance office, or a community mechanism could each produce a different accountability path. The 2024 comment record repeatedly asked for clarity about what “ICANN” meant. Version 2 names the corporation but does not, by itself, supply all internal procedures.

Second, ICANN is also the proposed final decision-maker after peer approval. An institution that formulates the allegations should not control the investigation and then assess its own petition without separation. The risk is not only bias toward removal. ICANN may narrow an investigation to defend its original framing, resist disclosing contrary evidence, or become institutionally invested in avoiding a finding that its intervention was unnecessary.

A credible design would place initial screening and fact finding with a qualified independent body under published terms. ICANN could refer information, request urgent preservation, and decide at the end, but the examiner would define the factual record and disclose exculpatory as well as inculpatory findings. The target, petitioners, holders, and peers would have controlled opportunities to respond. ICANN's final reasons would address that independent record.

Review must include refusals. If ICANN declines to act on a verified holder petition or an examiner's finding, the affected constituency should be able to challenge the refusal through a defined mechanism. Discretion without review can hide capture as easily as overreach.

Peer registries have expertise, conflicts, and an effective veto

Any RIR may trigger a proposal. That makes practical sense because peers see shared operational dependencies. They exchange data, coordinate policy, interact through the NRO, and understand what a loss of capability would mean. A peer may notice inconsistent records, failure to maintain an agreed service, or an inability to participate in system-wide coordination before the broader public does.

Expertise is not independence. A peer may compete for influence, disagree over regional structure, bear costs if it must help, or fear that a precedent will later apply to it. An incumbent may favor rehabilitation because removal would increase its burden. It may favor removal because a crisis damages the collective reputation. Either position can be honest and still conflicted.

Version 2 requires every remaining RIR to consider the proposal independently, publish a recommendation and reasons, and agree unanimously before the case reaches ICANN for a final decision. This prevents one peer from expelling another. It also gives each peer an unreviewed blocking position. A single registry can stop even a petition initiated by ICANN or thousands of affected members.

The draft created a recognition-review mechanism for candidate RIRs when incumbent objections may contain factual error or inadequate justification. It did not provide an equivalent independent override for a peer's rejection of derecognition. The asymmetry is striking. A candidate can challenge an incumbent's refusal under a defined process. A regional community seeking action against an incumbent can be stopped by one peer, even if that peer's reason is materially mistaken.

Peer unanimity may be appropriate for permanent removal if it is one safeguard among several. It should not bar an independent compliance finding, a public remedial order, or narrowly scoped continuity support. Nor should a peer veto be immune from review. A dissenting RIR should identify technical, legal, or proportionality grounds. An independent reviewer should be able to determine whether the refusal rests on a material factual error, undisclosed conflict, or reason outside the governance standard.

The peers should supply operational judgment, not final control over whether evidence can be heard.

Governments have public responsibilities, not an automatic removal right

RIR service regions cross national boundaries. Governments operate networks, regulate communications markets, protect public infrastructure, enforce corporate law, and represent citizens affected by connectivity. Their information can be essential during a registry crisis. The 2024 ICANN consultation included submissions from governments and Governmental Advisory Committee entities, some of whom sought formal consultation with governments in the affected region.

Yet the drafts do not give governments direct standing as governments. A ministry can petition if it is a qualifying member, persuade ICANN or a peer RIR, present evidence through consultation, or use lawful domestic powers. It cannot file a derecognition proposal solely by invoking governmental status.

That restraint is sensible. A unilateral government trigger could politicize number-resource administration, export sanctions disputes into registry status, or let one state speak over a multinational region. Derecognition must not become a tool for national control of a regional institution. Even a coalition of governments is not identical to the resource holders whose records and services would move.

Exclusion from the final-petition list should not mean institutional silence. Governments should have a recognized evidentiary channel, especially for court orders, corporate registration facts, criminal findings, sanctions that affect service capacity, infrastructure emergencies, and regulatory barriers. A regional group of governments could meet a threshold for independent screening without thereby proving non-compliance or choosing the remedy.

The safeguards should be explicit. Government submissions must identify legal authority and affected services. Political assertions should not substitute for verified operational evidence. The target and holders must be able to answer. Confidential material should be handled by an independent examiner. No government should gain access to non-public registration data merely by filing a complaint.

Government standing is therefore best understood as a right to evidence-based review, not a right to command removal. Public authority can alert, prove, and lawfully assist. Regional resource-holder authorization remains necessary for permanent institutional redesign.

Operators see service failure before governance bodies do

Network operators experience registry performance directly. They submit requests, maintain routing and contact data, rely on reverse DNS, publish route authorizations, transfer resources, and deal with account authentication. An operator can observe delay, inconsistency, discrimination, or technical degradation long before a board acknowledges systemic failure.

Some operators are RIR members and resource holders. Others receive resources through a Local Internet Registry, a parent organization, a national arrangement, or another service relationship. Their exposure does not always give them a vote. The Version 2 definition of the numbering community includes private-sector and technical entities, but the formal derecognition trigger does not extend to the numbering community as a whole.

Operators should not be able to launch a removal case merely because a ticket was mishandled. The distinction is scale and evidence. A public reporting channel can aggregate service indicators across the region: unavailable functions, integrity failures, inconsistent treatment, security events, unresolved authentication problems, missed service obligations, or inability to reach authorized personnel. Independent screening can separate isolated disputes from a pattern.

Operator evidence is particularly important for emergency continuity. The trigger for temporary service support should be an objective inability to provide specified services, not a political conclusion about the institution. A sufficiently severe, independently verified operational condition may justify temporary assistance even while governance allegations remain unresolved. Conversely, a heated board dispute does not justify moving services that continue to operate reliably.

Standing rules should therefore permit a defined number or distribution of affected operators to require a service-capability assessment. The threshold can include geographic spread, distinct corporate control, and evidence from more than one service category. That prevents a single commercial group from manufacturing a regional crisis while allowing dispersed failures to become visible.

Operators are witnesses to the control surface. Their role is not to decide the constitutional outcome alone, but no accountability system should force them to wait for an incumbent executive to translate service failure into an institutional complaint.

The twenty-five-percent threshold protects against raids and can protect failure

The high member threshold serves a legitimate purpose. Derecognition is an extreme remedy. A small faction should not be able to weaponize the process after losing an election, fee dispute, policy argument, or commercial case. A petition itself can damage confidence, consume resources, and distract staff. Repeated filings can become governance denial-of-service.

The May 2026 status report states that the NRO NC chose not to change the derecognition threshold. It described twenty-five percent or 2,000 members, whichever is lower, as deliberately high because removal is a final recourse. It also noted that a substantial petition falling short could still give an RIR or ICANN a credible basis to act. Concerns about affiliates or signatories concentrated in one country were acknowledged.

This reasoning shows both the threshold's value and its weakness. A near-threshold petition has no independent procedural effect. It depends on ICANN or a peer deciding that the evidence is credible. The insider gate returns at the exact point where the member route fails.

Twenty-five percent also confuses the threshold for starting a case with the threshold for proving a remedy. A petitioner should need substantial support to demand permanent removal. It should need much less support to obtain a neutral compliance assessment. Otherwise a captured institution can prevent the evidence needed to persuade a quarter of the electorate from ever becoming authoritative.

The solution is to separate stages. A low number of verified complainants can submit evidence. A higher, geographically and corporately diverse group can compel an independent review. A formal derecognition proposal can require the current high threshold or an equivalent holder threshold. Final removal can require findings, peer coordination, ICANN approval, cure analysis, and regional authorization. Abuse controls can impose waiting periods after meritless repeat complaints and costs for deliberate falsification without punishing good-faith reporting.

A high wall around the final remedy is prudent. A high wall around fact finding is impunity.

Time, confidentiality, and retaliation shape practical standing

A formal right is useless if using it exposes a petitioner to avoidable harm or if the receiving body can wait until the evidence loses value. Trigger design therefore needs clocks and protective rules, not only eligible categories.

Initial receipt should be acknowledged quickly. A screener should decide within a published period whether the allegation falls within the governance standard, whether urgent preservation is needed, and whether more evidence is required. Delays should be explained. A target RIR deserves freedom from an indefinite cloud, while petitioners and holders deserve protection against a case being quietly exhausted through time.

Confidentiality must be calibrated. Public allegations and public reasons are essential once a formal derecognition petition is accepted. Earlier stages may involve security details, personal data, legal advice, staff testimony, customer authentication records, or evidence of internal control. Automatic publication can endanger systems and discourage reporting. Total secrecy can enable factional intervention. An independent examiner should be able to receive protected material, test it, give the target a fair summary and response opportunity, and publish findings that omit only what is genuinely sensitive.

Retaliation is a service-integrity issue, not a private employment detail. A member or holder may hesitate to support review if it fears account suspension, delayed requests, exclusion from meetings, or public accusation. Staff and directors may fear dismissal or litigation. The standard should prohibit adverse treatment for good-faith participation, permit urgent complaints about retaliation, and require the examiner to distinguish legitimate enforcement from punishment for speaking.

Petitioners also owe duties. They should preserve evidence, disclose conflicts and affiliations, correct material errors, and avoid publishing credentials or personal data. Deliberate fabrication can lead to dismissal and proportionate consequences. Honest error should not be equated with abuse, particularly where the target controls the best records.

These protections make standing practical. Without them, the listed initiators are only those powerful enough to tolerate delay, disclosure, and retaliation. A bottom-up system must be usable by smaller entities before a crisis becomes so severe that only ICANN or another RIR can safely name it.

Affiliates, geography, and control determine whether a coalition is real

Counting signatures is not enough. A petition by 2,000 nominal members could be dominated by one corporate family, one country, one reseller campaign, or accounts created for governance influence. A petition by twenty-five percent could still omit most of a large multinational region. Conversely, a smaller coalition could represent severe failure across many countries and service classes.

The Version 2 draft defines affiliates and control in broad terms for governance purposes. That concept should be applied to standing. Entities under common control should generally count as one interest for threshold calculations, or at least be disclosed and capped. Authority to sign should be verified. Duplicate legal persons, subsidiaries, and representatives should not inflate the result.

Geographic distribution matters because an RIR serves a region, not merely a membership list. A petition concentrated in one market may identify a real regional problem, but it should not automatically authorize redesign for every other market. The threshold should report the number of countries or subregions represented, the share of holders affected, and the service evidence in each place.

No single metric is sufficient. Account equality protects small networks but can be manipulated through fragmentation. Resource-weighted voting favors large holders and can turn scarcity into political dominance. Fee weighting rewards wealth. Country weighting can give governments or small markets disproportionate control. A robust trigger uses a double or triple test: a minimum number of independent entities, a minimum geographic spread, and a meaningful share of the affected membership or holder population.

The count also needs a challenge process. The target RIR should be able to contest affiliation, authority, or eligibility, but it should not make the final determination. Petitioners should be able to challenge the denominator and claims that members are ineligible. A neutral verifier should preserve identities where publication could invite retaliation while releasing enough aggregate information to make the coalition credible.

Standing is not just a list of eligible names. It is a rule for deciding whether a claimed constituency is authentic.

Audit standing should be easier than derecognition standing

Version 2 applied the same listed initiators and member threshold to an ad hoc audit: a majority of the other RIRs, the same member coalition, or ICANN. The May 2026 status report says the NRO NC decided to separate audit-threshold issues from derecognition, rename the ad hoc audit as a compliance review, and continue drafting its scope.

That separation is necessary. An audit asks what happened and whether obligations are being met. Derecognition asks whether an institution should lose its status after other remedies prove inadequate. The evidentiary step should not require the political coalition needed for the sanction.

A compliance review can also protect the incumbent. Rumor is destabilizing. A credible independent examination can show that a service problem was temporary, an election allegation was unsupported, records remain intact, or a disputed action complied with local law and published policy. Making review accessible does not presume guilt; it replaces factional claims with a common factual record.

The trigger should use objective predicates. Examples include a documented service outage beyond a defined period, failure to publish required governance or financial records, a qualified audit concern, inability to identify a lawful governing body, verified refusal to follow a binding dispute outcome, material inconsistency in registry records, or evidence of undisclosed control. A petition that meets a predicate should compel screening. Several independent indicators or a cross-regional coalition should compel the full review.

To prevent harassment, the rules can limit repetitive claims arising from the same facts, require a signed evidentiary declaration, protect reviewers from lobbying, and publish reasons for dismissal. Waiting periods should apply only after a reasoned finding that a substantially identical complaint lacks merit. New evidence or urgent service risk must reopen the door.

The compliance review should end with more than a report. It should identify whether there is no breach, a technical remediation, a governance cure, an emergency service condition, or a potential basis for formal derecognition. Petitioners and the target should be able to challenge material factual errors. The review should not itself transfer service or status.

Emergency continuity needs an operational trigger, not a removal coalition

The August 2025 draft introduced emergency continuity when an RIR is unable to provide some or all of its services adequately. The other RIRs and ICANN could unanimously authorize a temporary operator after discussion with the affected registry and community where reasonably possible. The decision, rationale, and scope would be published; the initial period would be limited; the registry could resume service after capability was restored and verified; and a post-event review would follow.

This mechanism is distinct from derecognition and should remain so. Its trigger is present service incapacity, not a conclusion that the institution deserves removal. A flood, cyber incident, court-imposed limitation, staffing collapse, or systems failure can demand temporary assistance without proving governance unfitness. A captured board can also preside over services that still function, in which case emergency takeover may be unnecessary and disproportionate.

Standing for emergency assessment should therefore be broad and evidence-driven. The affected RIR should be able to request help. Operators and holders should be able to report verified inability to complete critical functions. Vendors and peer registries may have direct technical evidence. ICANN can coordinate a rapid independent test. Governments can report infrastructure conditions or legal restrictions.

Authorization to activate a temporary operator should be narrower. It should require a written description of the unavailable service, failed restoration attempts, data access required, security controls, maximum duration, and return test. The affected community should be heard when time permits. If immediate preservation is essential, participation follows as soon as possible and includes a right to challenge continuation.

Unanimity among peers and ICANN protects against unilateral intervention but can be too slow during an actual emergency. The safer design is not necessarily a lower political vote. It is pre-authorized, service-specific continuity arrangements adopted by each RIR and its community in advance. When an objective condition occurs, the agreed mechanism activates within its narrow scope. Any extension beyond the short initial period receives stronger approval and independent review.

Emergency standing should get a service restored. It should never be a shortcut to permanent institutional change.

Rejected cases need reasons and a route of appeal

The drafts devote more attention to the target RIR's ability to answer an accepted proposal than to a petitioner's ability to challenge rejection. Due process must run in both directions. A false or exaggerated petition can injure an institution. An unexplained refusal can leave an entire region inside a failing one.

At each gate, the decision-maker should publish a reasoned outcome. If a submission lacks verified identity, specified provisions, evidence, threshold support, or jurisdictional relevance, the rejection should say so and allow correction where possible. If a peer recommends against removal, it should identify whether it disputes the facts, materiality, persistence, cure record, proportionality, or successor plan. “Not in the system's interest” is not enough.

An independent reviewer should be able to correct material factual error, undisclosed conflict, threshold miscalculation, procedural unfairness, or reasons outside the governing text. Review should not let an examiner substitute its preferred policy for every institutional judgment. It should ensure that the judgment was made by the proper body, on evidence, for a permitted reason, and with due regard to holders and continuity.

Temporary relief needs separate rules. A rejected derecognition petition should not freeze ordinary service. A disputed emergency activation may require a stay on expansion of the temporary operator's scope while preserving the services already restored. Record destruction, credential misuse, or retaliation against petitioners may justify immediate protective orders.

The ICANN Bylaws provide reconsideration and independent review for covered ICANN conduct. A new RIR governance compact should connect its decisions to those mechanisms with precision and add independent review for peer recommendations and threshold verification where ICANN's mechanisms do not reach. It should identify who qualifies as materially affected, including holders and valid petitioner coalitions.

Accountability that reviews only intervention favors inertia. The power not to hear is still power.

A standing matrix can preserve openness without making removal easy

The final text should replace a single trigger rule with a standing matrix.

For evidence submission, any identifiable person or institution should be able to report facts: members, holders, operators, staff, directors, governments, civil society, technical bodies, other RIRs, and ICANN entities. Anonymous reports can be accepted when they contain verifiable evidence, though anonymity alone should not establish a contested fact.

For preliminary screening, a small independent office or panel should assess whether the allegation concerns a defined obligation and has credible support. It can request preservation and limited information. Dismissals receive short public reasons that protect confidential material.

For a compulsory compliance review, standing should extend to ICANN, a peer RIR, a defined lower member or holder threshold, a cross-regional operator coalition, a competent internal RIR organ, or a government coalition presenting evidence within public authority. Affiliate and geographic safeguards apply. Objective service or record-integrity events can also trigger review automatically.

For a formal derecognition petition, the threshold becomes higher. ICANN, one or more peers, and a substantial member or holder coalition can file after an independent finding or a documented failure to cure. A petition identifies breaches, evidence, affected services, attempted remedies, and the proposed consequence. A peer or ICANN petition should demonstrate meaningful support in the affected region rather than rely solely on institutional status.

For final derecognition, independent findings, peer coordination, an ICANN decision, a proportionality test, and regional authorization are required. The affected RIR receives a full response right. A single conflicted peer cannot secretly block the case, and no single claimant can complete it.

For emergency continuity, objective service incapacity activates only the pre-authorized minimum support needed. Extensions require escalating scrutiny. For permanent succession or regional redesign, a separate support process among affected holders and communities is mandatory.

This matrix makes the first gate accessible and the last gate demanding. It is harder to suppress evidence and harder to weaponize removal. That is the correct asymmetry.

Trigger power must belong to the governed as well as the governors

The RIR system does not need a universal right for every disappointed actor to commence a removal proceeding. It needs a credible path by which institutional failure can move from observation to evidence, from evidence to independent review, from review to cure, and only then from unremedied material failure to a status decision.

The evolving draft has improved since 2024. It moved beyond exclusive NRO EC origination, recognized a direct member petition, added a numerical cap, let ICANN act when peers remain silent, required specified allegations and a public response, and began separating compliance review from derecognition. The May 2026 decision to retain a high final threshold while reconsidering audit design points toward a staged model, even if the details remain unfinished.

The unresolved issue is whose dependence counts. Voting members are central, but they are not the whole service population. Resource holders may bear the direct consequences of inaccurate records, unavailable services, or a transfer to another operator. Network operators can see technical failure. Governments can supply lawful evidence and public-interest context. Staff and internal bodies can expose facts hidden from formal governance. Peers and ICANN carry system-wide responsibilities. Each needs a defined role; none should possess an unreviewable monopoly.

Standing also determines legitimacy after the crisis. A region is more likely to accept a difficult remedy if it knows that evidence could be submitted without permission from the accused, petitioners were verified, affiliates could not inflate support, dissent was recorded, facts were independently tested, refusal as well as action could be reviewed, and permanent service transfer required its own authorization.

The trigger rule should therefore state a simple principle: those materially exposed to an RIR's governance and services must have a practical route to make alleged failure answerable, while no claimant may convert that access into unilateral control. The institutions can coordinate the remedy. They cannot reserve to themselves the exclusive right to decide whether the problem is allowed to exist.