Summary
- The current second draft of the RIR Governance Document gives ICANN the final recognition or derecognition decision and says an affected RIR or candidate may use ICANN's applicable review procedures. That promise is uncertain because ICANN's current Bylaws expressly exclude disputes and claims relating to Internet numbering resources from both reconsideration and the Independent Review Process.
- A dedicated appeal should be independent of ICANN and the peer RIRs, admit the affected registry and directly affected resource holders, support confidential operational evidence, and offer rapid interim protection without freezing essential continuity measures.
- The reviewing panel should be able to correct factual, procedural, authority, proportionality, and remedy errors. Its orders should preserve accurate records and services, distinguish temporary operations from permanent institutional transfer, and produce an enforceable path back to lawful regional governance.
A final decision creates a new accountability problem
Recognition once looked like a gate at the entrance to a small institutional club. A proposed Regional Internet Registry showed that a region wanted it, that it could provide registration services, and that it could participate in global coordination. ICANN then accepted or rejected the application. The update to ICP-2 is attempting something more difficult. It would govern the entire life of a registry, including continuing obligations, emergency service arrangements, rehabilitation, and eventual derecognition.
At the end of the most serious case, ICANN would decide whether a regional institution should lose the responsibility it has exercised for years.
That decision is different in kind from an ordinary administrative refusal. A derecognized registry may lose its service region, its place in the Number Resource Organization, its authority to provide registry services, and its ability to direct the transition of records and operations. A successor or interim entity may acquire practical control over functions on which thousands of networks rely. The consequences can touch registration accuracy, reverse DNS, route-security credentials, transfers, contact changes, contractual expectations, membership rights, employment, pending litigation, and the local legal status of records and systems.
Calling such a decision final does not settle who may challenge it, who may pause it, or what happens if it is wrong. Finality is useful only after fair review. Before then, it magnifies error. A mistaken recognition can install a candidate that lacks regional consent or operational ability. A mistaken refusal can let an incumbent peer protect itself from competition. A mistaken derecognition can turn a remediable governance crisis into a service crisis. A mistaken refusal to derecognize can leave resource holders dependent on an institution that no longer meets basic duties. Every direction contains risk.
The NRO's current review timeline makes the institutional moment clear. The NRO Executive Council asked for an ICP-2 review in 2023; two consultation rounds followed; drafting of a final version remains in progress in 2026; and approval and adoption are scheduled for later in the year. The appeal question is therefore not a retrospective complaint about a settled instrument. It is a design choice that can still be made before an unprecedented power is exercised.
The central principle should be simple: no decision capable of reassigning regional registry responsibility should become practically irreversible before an independent body can test the decision and protect the people whose number resources depend on it. That does not mean every challenge must stop every action. It means the instrument must state who reviews, what the reviewer can examine, which services continue, and what remedy follows if the decision cannot stand.
The draft points toward a door that may be closed
The RIR Governance Document Version 2 contains several layers of decision. For recognition, each existing RIR independently reviews a candidate. Unanimous support normally sends the proposal to ICANN. If a candidate believes a negative peer recommendation contains material factual error or inadequate justification, it can request an independent third-party review. In a narrow final stage, an unjustified dissent from one RIR can be disregarded if the independent reviewer finds both error and satisfaction of all recognition requirements.
Derecognition is arranged differently. A proposal may come from an RIR, a qualifying group of the affected registry's members, or ICANN. The affected registry may respond. The other RIRs decide whether to recommend derecognition, and unanimity among them is required before the proposal reaches ICANN. ICANN then discusses the proposal with the RIRs, considers their input, and makes the final decision to approve or reject it. The draft requires publication of the decision and its reasons.
After that final decision, Section 2.3(c)(ii) offers one sentence on review: an affected RIR or candidate may petition under ICANN's then existing and applicable procedures. The sentence appears reassuring. It is also the most consequential ambiguity in the draft. It does not identify a forum, filing period, standard of review, right to interim relief, treatment of confidential records, participation by resource holders, available remedies, or the legal effect of a decision. Most importantly, it does not establish that any existing ICANN procedure is applicable to a numbering dispute.
Current ICANN rules make that uncertainty concrete. Article 4 of the ICANN Bylaws provides two major accountability mechanisms. Reconsideration permits a materially affected person or entity to challenge certain Board or staff action or inaction. The Independent Review Process permits a materially affected claimant to allege that an ICANN covered action violated the Articles of Incorporation or Bylaws. Yet the reconsideration section expressly excludes disputes relating to Internet numbering resources, and the IRP section expressly excludes claims relating to Internet numbering resources.
The exclusions matter more than the general availability of those mechanisms. An RIR derecognition decision would plainly concern the responsibility to provide services for IP addresses and Autonomous System Numbers. A candidate recognition decision would concern the same system. Unless the Bylaws are amended, a claimant could follow the draft's invitation only to be told that the obvious ICANN forums are outside scope. A review right that depends on a procedure from which the subject is excluded is not a review right. It is a future argument.
There may be interpretations under which a challenge is characterized as institutional governance rather than a dispute about particular resources. That possibility is not enough. The appeal must be predictable before the crisis, not reconstructed after counsel begins litigating jurisdiction. The instrument should not force a panel to decide whether a loss of regional registry status is sufficiently separate from the resources whose administration gives the status meaning. It should create a dedicated route or secure a clear amendment making the existing independent route applicable.
Reconsideration by the decision maker is not sufficient
ICANN reconsideration and independent review perform different tasks. Reconsideration asks the organization to revisit its own action under defined grounds: conflict with mission or policy, failure to consider material information, or reliance on inaccurate information. Independent review places a dispute before a panel outside the decision maker and asks whether ICANN complied with its constitutional commitments. In an RIR status case, both error correction and external judgment are needed, but the second is indispensable.
The final status decision would follow a chain in which the existing RIRs have already recommended an outcome and ICANN has already approved or rejected it. Sending the dispute back to an ICANN Board committee alone would preserve the same institutional center of gravity. The Board could correct an obvious omission, but it would still be judging a decision for which it is responsible.
If the challenge alleges that ICANN gave excessive weight to peer RIR interests, misread regional consent, ignored the effect on holders, or chose a remedy beyond its authority, internal reconsideration cannot supply the appearance or substance of full independence.
Peer review alone is equally inadequate. Existing RIRs possess operational knowledge and have a legitimate interest in global stability. They also have institutional interests. A recognition candidate can alter service-region boundaries and coordination costs. Derecognition of one peer can create transition burdens for the others while setting a precedent that may later apply to them. Their evidence and technical analysis should be part of the record, but their unanimity should not immunize the recommendation from external scrutiny.
An independent panel needs separation from both centers of decision. Panelists should not be current directors, officers, staff, paid advisers, or recent representatives of ICANN, the NRO, an RIR, the affected registry, a candidate, or a substantial petitioner. They should disclose professional, financial, and institutional ties. A party should be able to challenge a panelist for a material conflict. Independence must extend to the administrator and any technical expert, not only the person who signs the decision.
Expertise matters because this is not a generic corporate appeal. A panel may need to understand registry data, service dependencies, RPKI operation, allocation history, reverse DNS, membership structure, continuity arrangements, the relationship between global policy and regional policy, and the limits of domestic corporate law. The ICANN Bylaws already contemplate access to independent skilled technical experts in the IRP. A dedicated numbering appeal should do the same, with all substantive expert interaction recorded and disclosed subject to narrow confidentiality protections.
Standing must follow the harm, not only the institution
The draft gives the affected RIR or candidate a route to petition after ICANN's decision. Those entities plainly need standing. The candidate bears the cost of preparation and may be excluded despite meeting the criteria. The incumbent may lose recognition, service responsibility, NRO membership, and control of a transition. Neither should have to persuade the reviewing body that these direct effects are enough.
The harder question concerns resource holders. The draft defines them as legal or natural persons holding number resources registered with an RIR. They may experience the most immediate operational consequences of a status decision, yet Section 2.3(c)(ii) does not expressly give them a petition right. A holder may need a registration update during the transition, depend on a route authorization, face uncertainty about a transfer, or contest whether the designated successor can lawfully process its confidential information. A member may also lose voting rights or see its authorization attributed to a coalition it did not support.
It would be a serious design error to say that the institution adequately represents every holder. During a governance failure, the interests of the registry and its users can diverge. The incumbent may challenge derecognition to preserve itself while holders want only accurate records and a neutral service bridge. A candidate may challenge rejection while some holders believe its claimed mandate is false. A government or industry association may claim to represent a country, while individual operators in that country oppose the proposed transfer. The appeal should allow those differences to appear.
Standing should therefore have three levels. The directly affected RIR or candidate should have standing as of right. A resource holder or member should have standing to challenge a decision or remedy when it can show a concrete and particular effect on its service, legal rights, governance rights, data, or ability to use registered resources. Representative groups should be allowed to proceed when they disclose their membership, authority, financing, and conflicts and can show that they fairly represent the interests they claim.
Other entities may not need party status. Governments, technical associations, civil society organizations, emergency operators, potential successors, and other RIR communities can submit focused observations where they bring relevant public-interest, legal, or operational information. The panel should be able to reject repetitive or strategic submissions. Participation should improve the record, not turn the appeal into an unbounded conference.
This approach is consistent with a principle already present in ICANN's general IRP: a claimant must be materially affected through a directly and causally connected injury, and a claimant purporting to represent others must do so fairly and adequately. The numbering appeal can borrow that discipline without inheriting the numbering exclusion. It should also recognize a distinctive fact of registry governance: harm can be collective in service effect while still being particular to the entities whose records and authorizations are moved.
Notice has to reach the people who can lose something
Publication on an institutional website is necessary but not sufficient. Affected holders may not monitor ICANN resolutions, NRO pages, or specialist mailing lists every day. A final decision can become irreversible while the entities most exposed to it are still learning what occurred. Effective notice must be directed as well as public.
Before ICANN decides, the affected registry should be required to notify members and resource holders through verified service contacts, unless the registry cannot or will not do so. ICANN should then use available independent channels and require proof of delivery efforts. The notice should identify the proposal, the claimed failures or recognition basis, the recommended remedy, the proposed successor or interim operator if known, the services affected, the expected effective date, and the appeal deadline. It should explain how a holder can inspect public materials, submit evidence, request confidentiality, or challenge the use of its name.
The decision notice should do more than announce a result. It should separate findings of fact from interpretation and judgment. It should identify which requirements were met or breached, the evidence relied upon, the treatment of disputed evidence, the scope of any cure opportunity, the reasons lesser measures were rejected, and the expected effect on each service class. If confidential evidence was decisive, the public notice should describe its category and significance without exposing protected details, and the panel should have access to the unredacted material.
The appeal period should begin only when the decision and reasons are published together and direct notice has been attempted. A bare resolution followed weeks later by a rationale would impair review. The appellant cannot identify factual or proportionality errors until it sees the reasoning. The effective date should also be tied to adequate notice. An institution should not be removed, and a successor should not become permanent, through a timetable shorter than the time allowed to seek interim relief.
Language access is part of notice. English may be the official language of the registry system, but a service region can span many linguistic communities. The controlling legal filing can remain in English while plain explanations are distributed in major regional languages. Translation cannot be treated as proof of consent, but the absence of understandable notice can be proof that participation was not meaningful.
The record must exist before the appeal begins
An appeal cannot repair a decision whose factual basis was never assembled. The status proceeding needs a defined record: the initiating proposal, responses, audits, expert reports, peer recommendations, regional consultation results, conflicts, Board materials, legal advice to the extent disclosable, service-impact analysis, continuity plan, and comments from holders. Every material item should have a date, author, and disclosed institutional role.
The draft already requires independent RIR recommendations and published reasons. That is a useful foundation. But the final record must expose how the recommendation became ICANN's judgment. Did ICANN independently verify the facts or defer to unanimous peers? Did it test the member count behind a petition? Did it compare the incumbent's cure plan against the claimed harm? Did it evaluate the candidate's regional support? Did it identify who would hold copies of data, operate route-security services, answer registration requests, and bear transition cost? Those questions cannot first appear on appeal.
An administrative record also disciplines private communications. In a crisis, directors, governments, operators, courts, creditors, staff, and international bodies may all contact the decision maker. Not every logistical conversation belongs in public, but every substantive representation relied upon should enter the record. If a ministry asserts regional support, the assertion and its evidentiary basis should be disclosed. If a peer RIR warns of a technical risk, the warning should be available for response. If the affected registry offers a cure, the offer cannot disappear into a closed briefing.
The panel should ordinarily review the decision on that record. New evidence should be admitted when it could not reasonably have been produced earlier, when the decision maker withheld relevant material, or when developments after the decision bear directly on interim protection or remedy. This prevents strategic withholding while recognizing that service conditions can change quickly.
Record access must include a protected channel. Registry disputes can involve security architecture, authentication material, personal data, commercial information, and privileged legal analysis. A public-interest appeal does not justify disclosing material that could endanger networks or violate law. Protective orders, limited-access annexes, redacted public summaries, and logged expert access can preserve both accountability and safety. The rule should state that confidentiality may protect evidence but cannot conceal the existence of the decisive reason.
A stay should preserve services, not merely the incumbent
The word "stay" can mislead. In ordinary litigation it suggests freezing a decision. In registry governance, a complete freeze may be impossible or harmful. If a registry cannot issue essential credentials, maintain accurate data, or respond to urgent service requests, stopping all interim action would protect the institution at the expense of its users. Conversely, allowing a permanent handoff before review could destroy the value of the appeal.
The answer is a service-preserving stay. Once a timely appeal is filed, irreversible institutional effects should be paused for a short period while an emergency panelist considers tailored relief. The affected registry should not lose permanent status, dispose of critical assets, alter disputed records, or surrender exclusive control beyond what continuity requires. ICANN and the other RIRs should not install a permanent successor, extinguish membership rights, or represent the merits as finally settled.
At the same time, a neutral interim operator may perform specifically identified services if delay would threaten holders or registry accuracy.
The current ICANN IRP provisions offer a useful starting structure even though numbering claims are excluded from its scope. They allow an emergency panelist to consider interim relief, including a stay, where the claimant shows irreparable harm, a likelihood of success or serious merits questions, and a balance of hardships favoring relief. A numbering-specific rule should retain those disciplines but add a mandatory continuity factor: the panelist must identify the effect of relief on resource holders, registration integrity, route security, reverse DNS, and the global registry system.
Some safeguards should apply automatically during the short emergency review. No party should delete or materially alter historical registration data except for ordinary verified service changes. Audit logs, corporate records, election records, system images, and transition communications should be preserved. Existing route-security products should remain valid unless a specific security reason requires change. Neither side should solicit mass record transfers outside the published plan. Staff and contractors should receive clear authority instructions so that uncertainty at the top does not create conflicting operational commands.
The panelist should be able to split the decision into components. It might stay derecognition but permit emergency operation of one impaired service. It might let a candidate continue readiness testing while preventing final recognition. It might protect member voting rights while allowing an independent custodian to preserve data. It might order dual verification for sensitive record changes without creating two authoritative registries. Tailoring is not a compromise of principle. It is the only way to protect both due process and continuity.
Interim orders need short review intervals. An arrangement justified by a weekend outage may be excessive after a month. The panel should state the evidence required for renewal, restoration, expansion, or termination. The draft's 90-day limit on an emergency continuity period provides a useful outer rhythm, but an appeal stay and an emergency operation are separate instruments. One protects review; the other protects service. Their terms can overlap without becoming identical.
The reviewer needs a standard broader than factual error
The recognition review already described in Version 2 focuses on material factual error and inadequate justification in a peer RIR's negative recommendation. That may be suitable for the narrow question of whether one peer is blocking a qualified candidate. It is too limited for review of ICANN's final status decision.
A final appeal should test at least five classes of error. First is authority: whether the proposal and decision complied with the powers and limits of the governing instrument. ICANN should not recognize or derecognize without the required recommendation, modify a proposal beyond permitted bounds, or use emergency continuity as undeclared permanent replacement. Second is procedure: whether the required notice, response, consultation, conflict disclosure, audit, cure opportunity, and reasons were provided.
Third is fact. The panel should correct material findings unsupported by reliable evidence, including false member counts, misidentified resource holders, inaccurate service claims, undisclosed affiliate concentration, or incorrect statements about legal incapacity. Fourth is interpretation: whether ICANN applied the correct meaning to obligations such as impartiality, operational independence, regional support, performance, and ecosystem stability.
Fifth is proportionality and remedy: whether the decision reasonably compared the harm of continued noncompliance with the harm of intervention and selected the least disruptive measure capable of addressing the proven failure.
The level of deference should vary. A panel should not casually replace a technical judgment supported by independent evidence. Nor should it defer to an institution's interpretation of the boundaries of its own power. Questions of authority, basic fairness, and compliance with mandatory steps should receive independent determination. Technical and predictive findings can receive respectful review, but only if the record shows expertise, reasoned analysis, and consideration of contrary evidence.
The draft's presumption in favor of rehabilitation should have legal force on appeal. Derecognition is described as a last resort, justified where the harms of noncompliance outweigh the benefits of toleration. ICANN should therefore show what noncompliance was proved, what reasonable cure opportunity was given, what support was offered, why the response failed, and why a narrower order would not protect the community. A statement that trust has been lost is not enough. The panel must be able to test the chain from breach to remedy.
Review of a refusal to derecognize requires equal care. An incumbent should not be protected by a standard so deferential that no failure can be corrected. Petitioning members and affected holders should be able to show that ICANN ignored material evidence, applied the wrong threshold, permitted endless cure without milestones, or accepted peer protection as analysis. Independence serves both sides: it constrains intervention and institutional tolerance.
Affected holders need more than an audience
Consultation is often presented as participation, but the two are not synonymous. A meeting lets people speak. A legal participation right lets them see the case, submit relevant material, challenge false claims, and receive a reasoned response. In a status appeal, holders need the latter where their own authorization or service is at issue.
If a proposal claims that named operators support recognition or replacement, each named entity should be able to verify or contest that claim. If a petition count includes an entity as a voting member, the entity should be able to challenge eligibility, duplicate affiliation, proxy authority, or the date on which the count was fixed. If a remedy moves confidential records, affected holders should be able to address lawful transfer, security, data minimization, and service contact arrangements.
The appeal should permit intervention without requiring every holder to become a full litigant. A holder could file a concise statement identifying the affected resource relationship, the disputed fact or remedy, and the relief sought. The panel could group common issues, appoint liaison counsel or a representative committee, and create a secure evidence room for protected material. Small networks should not need the same legal budget as a regional institution to correct the misuse of their names.
Costs require deliberate design. The ICANN Bylaws say ICANN bears the administrative costs of maintaining the IRP, while parties ordinarily bear their own legal expenses. A numbering appeal should similarly prevent panel fees from becoming a bar. It should also provide limited support for credible community, nonprofit, and small-holder participation where the issue cannot otherwise be heard. Fee shifting should be reserved for abusive conduct, not used to frighten challengers with uncertain jurisdiction.
Representation should be tested throughout the case. A trade association may have authority from some members but not others. A government-supported group may speak for a public program without speaking for private operators. A registry board may be lawfully installed yet conflict with members on the remedy. The panel should state whose interests each entity represents and avoid treating the loudest submission as a regional mandate.
Participation also means receiving the outcome in usable form. The final decision should address the principal holder concerns, identify any protected service commitments, and explain how to seek correction if the transition does not follow the order. A judgment that resolves institutional status while leaving users uncertain about where to request service would be formally complete and operationally deficient.
Remedies should repair the decision without breaking the region
An appeal is only as credible as its remedies. If the panel can declare error but cannot prevent or reverse its consequences, review becomes an academic exercise. If it can simply substitute its preferred registry, it risks becoming another unaccountable decision maker. The remedy set must be strong, bounded, and continuity-aware.
The first remedy is remand with directions. The panel may require ICANN to consider omitted evidence, provide a cure period, disclose reasons, rerun a support assessment, resolve conflicts, or compare less disruptive measures. Remand is appropriate where the defect can be corrected without deciding the merits for the Board. It should carry a deadline and interim protections so that the same indecision does not continue indefinitely.
The second is vacatur or reversal of the decision where ICANN lacked authority, violated indispensable fairness, or could not reach the result on the lawful record. A recognition should not stand if the candidate lacked the required regional mandate. A derecognition should not stand if the proposal was not validly approved or if rehabilitation was never genuinely offered. A refusal should not stand where the decision ignored proved systemic harm and treated last resort as never resort.
The third is modification of remedy. A panel may uphold a noncompliance finding while narrowing an overbroad handoff, preserving member rights, limiting an interim operator's term, requiring independent data custody, or separating technical service from governance. It may uphold recognition but delay effectiveness until specified readiness conditions are verified. This is especially important because a single notice can contain both the status conclusion and the transition plan. One may be sound while the other is not.
The fourth is a protective order directed at continuity. The panel can require preservation of records, publication of service contacts, independent monitoring, audit access, periodic status reports, and defined return conditions. It can prohibit retaliation against members or staff who participated in review. It can order correction of public statements that misdescribe the legal status of the appeal. These measures do not decide who deserves institutional control; they keep the system trustworthy while that question is resolved.
The fifth is enforcement. A final panel decision should bind ICANN and the parties that accepted the governance instrument, subject to lawful court supervision. The current ICANN IRP provisions aim at final, binding decisions enforceable in a court with jurisdiction. A numbering appeal needs equally clear effect. Otherwise ICANN could receive an adverse decision, reconsider in public, and continue with the same outcome.
No global rule can erase domestic law. The affected registry is incorporated somewhere; records, employees, assets, creditors, and directors may be subject to local courts. The appeal should not pretend to transfer property that ICANN does not own or command a court. It should instead define the status consequences within the global registry system, the contractual commitments of participating institutions, and the conditions for technical coordination. Where local law blocks an ordered action, the parties should return promptly to the panel for an alternative that preserves the lawful objective.
Emergency continuity and appeal must not swallow each other
Version 2 creates emergency continuity for a registry unable to provide all or part of its services. Other RIRs and ICANN may authorize a temporary operator, with discussion involving the affected registry and community where reasonably possible, publication of rationale and scope, a community engagement period, a right to resume service after capability is restored and verified, and a post-event review. This is one of the draft's strongest ideas because it recognizes that service can be protected without immediately deciding permanent status.
But emergency authority can become a shortcut around appeal. Once another entity operates services, controls interfaces, employs transferred staff, and communicates with holders, temporary facts can harden into a permanent settlement. A later ruling for the incumbent may be technically possible but institutionally futile. That risk is why the emergency order must specify which services moved, which data was copied, which decisions remain prohibited, and how return will occur.
The reverse risk also exists. An incumbent can use appeal to obstruct urgent service support, withhold records, or extend dysfunction. The stay rule should deny relief that would expose holders to concrete, imminent harm. The incumbent's right to challenge status does not include a right to make continuity impossible. Cooperation duties, preservation orders, and neutral custody can be enforced while the merits remain open.
The appeal panel should review emergency measures when they materially affect the final decision, but it should not become the real-time operator. Its role is to enforce boundaries: necessity, scope, duration, neutrality, reversibility, and reporting. Technical implementation should remain with qualified operators subject to those limits. Every renewal should show what service remains impaired and why the same scope is still necessary.
The post-emergency report should be admissible on appeal but not treated as neutral by definition. The institutions that authorized and ran the arrangement may also be defending it. Holders should be able to challenge performance claims, identify service failures, and contest whether the temporary operator stayed within scope. Independent logs and monitoring are therefore more valuable than retrospective institutional assurances.
Time limits should be fast enough to matter
Registry status cannot remain uncertain for years. Candidates need to know whether to maintain staff and systems. Incumbents need to know whether they can govern, contract, and retain employees. Resource holders need one authoritative service path. Interim operators need a defined end. Speed, however, cannot be purchased by eliminating evidence or participation.
A credible schedule would contain several clocks. A party should have enough time after publication and direct notice to file a reasoned appeal, perhaps 30 days rather than the very short periods used for some internal mechanisms. An emergency panelist should address temporary protection within days. The record should be certified quickly. Intervention and confidential-evidence requests should have short, fixed windows. The merits panel should issue an initial timetable and aim for a decision within months, with extensions justified publicly.
The six-month aspiration in ICANN's IRP is a useful reference, not a guarantee. A numbering appeal may require faster interim decisions and a similarly disciplined final horizon. Cases involving only authority or notice can be decided on an expedited track. Cases involving contested technical capability, regional mandate, or domestic legal barriers may require experts and a hearing. The rule should allow both while refusing open-ended delay.
Deadlines should bind the decision maker too. If a panel remands for a new support assessment or cure decision, ICANN should not be able to leave the case unresolved. The remand order should state what must happen and when. Failure to act should permit return to the panel. A review right that ends in institutional silence is not effective.
Publication should be prompt. The parties may need a short period to identify confidential material or correct clerical errors, but the decision, reasoning, and operative orders should become public together. A short plain-language explanation should identify what changes immediately, what remains stayed, where holders obtain service, and what further step is expected.
Four scenarios expose the design choices
Consider a candidate rejected because one incumbent RIR says its operator support is inadequate. The candidate presents verified authorizations and alleges that the dissenting RIR counted affiliates inconsistently. Version 2 already offers independent recognition review before the proposal reaches ICANN. If ICANN later rejects despite a favorable peer process, the final appeal should test whether ICANN applied the criteria and considered the verified mandate. It should not restart the entire campaign or let the candidate rely on stale letters.
Now consider derecognition after an audit finds that a registry's governing body lacks effective control and services are deteriorating. ICANN approves removal and names an interim operator. The registry appeals, showing that a new board was lawfully elected and a cure can be verified within weeks. The emergency panelist might permit the interim operator to maintain one impaired service while staying permanent loss of status and ordering independent verification of the cure. The merits panel can then assess whether last resort was truly necessary.
In a third case, ICANN rejects derecognition after peers accept repeated promises of reform. A group of members and holders shows missed milestones, inaccurate registration service, and conflicts in the peer recommendations. Those directly affected should be able to appeal the refusal. The panel might require an independent audit and a time-bound rehabilitation order rather than immediately substituting its own derecognition decision.
Finally, imagine that local litigation blocks access to systems after ICANN approves derecognition. A successor exists on paper but cannot lawfully receive all data. The appeal cannot command the domestic court or pretend the obstacle is irrelevant. It can stay the permanent transition, authorize narrowly scoped continuity using lawfully available data, require the parties to seek local directions, and demand a revised remedy that does not depend on impossible transfer. This is where remedial flexibility protects legitimacy.
These scenarios show why a single sentence pointing to unspecified ICANN procedures cannot carry the load. Each case raises standing, evidence, interim protection, service scope, and remedy questions. Those questions should be answered in advance, not improvised under the pressure of an institutional failure.
The appeal clause should be part of the governing bargain
The NRO's May 2026 status report says work continues toward the final draft. The second-round community input report records the precise concern: commenters questioned reliance on changing ICANN review procedures, proposed a dedicated reconsideration and appeal route, sought an independent panel, and asked that any affected party be able to appeal. Those are not peripheral drafting preferences. They identify the mechanism that turns lifecycle authority into accountable authority.
The final instrument should therefore include the appeal's minimum guarantees, even if detailed rules follow in a separate published procedure. It should identify an independent forum; state that numbering status decisions fall within its jurisdiction; grant standing to the affected registry or candidate and concretely affected holders; require timely, reasoned notice; preserve a complete record; authorize emergency and final relief; establish continuity factors; and make the outcome binding within the registry system.
Implementation details can address filing format, page limits, panel selection mechanics, evidence exchange, and hearing logistics. They should not be allowed to withdraw jurisdiction, narrow standing below the instrument, eliminate stays, or reduce available remedies. The draft currently says later procedures cannot contradict the governing document. The appeal clause should use that hierarchy to protect the essentials.
Adoption also needs legal coordination. If ICANN intends its existing IRP to serve as the final appeal, the numbering exclusion must be addressed expressly and the scope must include decisions under the new instrument. If a separate tribunal is chosen, the RIRs and ICANN should execute commitments recognizing its authority, funding, confidentiality powers, and enforceability. The parties should not wait for the first derecognition to discover that the institutional document and ICANN's Bylaws point in opposite directions.
The clause should be tested before use. A tabletop exercise can simulate candidate rejection, emergency continuity, derecognition, holder intervention, confidential evidence, and domestic legal obstruction. The goal is not theater. It is to discover whether notices reach holders, a panel can be appointed, an interim order can be issued, technical records can be preserved, and service responsibility remains unambiguous. A right that has never been operationally tested is vulnerable at the moment it matters most.
Finality must be earned through review
The registry system needs decisive outcomes. Endless institutional conflict can itself damage trust and service. But decisive does not mean unreviewable, and final does not mean whatever the last institution in the chain says first. A sound final decision is one reached under known authority, on a disclosed record, after affected parties can be heard, with independent correction available before irreversible effects occur.
The updated ICP-2 effort has already recognized that mature registry failure requires more than the 2001 entry criteria. It has added audits, operating duties, rehabilitation, emergency continuity, derecognition, handoff, and readiness. Appeal belongs in that same architecture. Without it, the strongest new power rests on the weakest procedural sentence.
An independent appeal would not make derecognition impossible. It would make a defensible derecognition more durable. It would not give every holder a veto. It would let holders correct concrete errors and protect their services. It would not subordinate global coordination to every domestic dispute. It would force the global decision to acknowledge legal reality and design a remedy that can work within it.
The practical test is straightforward. On the morning after ICANN publishes a final decision, can the affected registry identify the forum and deadline? Can a resource holder whose records or rights will move explain its harm? Can an emergency panelist preserve necessary service without completing the permanent transfer? Can the merits panel inspect the same evidence ICANN used, correct a material error, and order an enforceable remedy? If any answer is uncertain, the appeal is unfinished.
Finality should arrive at the end of that process. Anything earlier is only power waiting to be challenged.