AFRINIC is a test case for how procedural judgment can move economic power without a formal vote. In a registry where IPv4 scarcity, transfer restrictions, disputed resource control, litigation and institutional recovery meet, a chair's sentence about scope, meeting time, rough consensus, material objection, last call, appeal or ratification record can change the practical value of scarce address resources. The power is not theatrical. It is quiet, usually written in the language of process, and often exercised before anyone says that an economic choice has been made.
The procedural sentence with balance-sheet consequences
The most powerful sentence in a registry policy meeting is often not a clause in the proposal. It is the chair's sentence: this objection is material, this point is outside scope, this issue belongs to implementation, there is rough consensus, there is not rough consensus, last call is closed, the proposal advances. It may appear in meeting minutes, in a mailing-list summary, in a report to the board or in a response to an appeal. It does not look like an allocation decision. In a scarce-address registry, it can still alter the commercial position of every holder whose resources will be governed by the resulting rule.
"Chair discretion" here means the discretion of the Policy Development Working Group chairs and meeting chairs who manage AFRINIC's number-resource policy process. It is not mainly about the discretion of AFRINIC's corporate board chair, though board ratification later magnifies the record that policy chairs create. The relevant acts are narrower: scope rulings, agenda and time management, rough-consensus calls, sorting material objections, closing last call, writing chair reports, handling appeals and creating the ratification record. These acts sound administrative because they are performed in the grammar of procedure. Their economic force comes from the asset they govern.
AFRINIC does not administer a ceremonial resource. It maintains the registry through which IPv4 address blocks, IPv6 resources and autonomous system numbers are recognised, recorded and serviced for Africa and parts of the Indian Ocean. IPv4 has been globally exhausted for years, yet it remains essential for mobile networks, broadband providers, hosting platforms, enterprise systems, cloud migration, security appliances, customer continuity and dual-stack transition. A block of IPv4 space may not be property in the ordinary land-title sense. But registry recognition of the holder, transfer status, registration contacts, reverse DNS, RPKI relationship, dispute state and contractual standing is an operational input with market value. Buyers, lenders, customers, counterparties and auditors all price continuity around that recognition.
AFRINIC's Consolidated Policy Manual describes a bottom-up Policy Development Process. Anyone may participate. Draft policy proposals are posted to the Resource Policy Discussion list and made available for review. A proposal must be published at least four weeks before a public policy meeting. The meeting agenda must be announced at least two weeks before the meeting. No change to a draft may be made within one week of the meeting, so participants can discuss a stable text. Chairs determine whether rough consensus has been achieved during the public policy meeting. A last-call period of at least two weeks follows. The chairs then evaluate feedback from both the meeting and last call, decide whether consensus exists and, if it does, recommend the proposal to the board for approval. Appeals exist, but they are procedural: the appellant must first discuss the matter with the chairs or working group, then may appeal with support from three working-group participants within two weeks of public knowledge of the decision. The Appeal Committee may annul a chair decision if the process was not followed.
Those steps are familiar in internet governance. They are designed to prevent policy from degenerating into a vote-counting contest and to let technically informed participants solve coordination problems without requiring unanimity. Yet the machinery also places considerable judgment in the chair. Rough consensus is not a number. Scope is not self-defining. An objection is not material merely because someone says it is. Last call is not a natural event; it is opened, extended, closed and interpreted. A report to the board is not a transcript; it is a constructed record of what the chairs think matters.
That discretion was easier to accept when RIR policy mainly governed future allocation from a pool that still had slack. If a rule determined how new applicants would receive addresses, a chair's decision looked like coordination among future users. Existing reliance was thinner. In the post-exhaustion world, the same decision can affect resources already held, contracts already signed and network plans already financed. A transfer rule can determine whether AFRINIC-issued resources can leave the region. A resource-classification rule can change liquidity. A verification rule can convert imperfect records into compliance risk. A last-call closure can end the period in which holders may object before the rule hardens into policy.
AFRINIC is a particularly revealing case because the institution has been under unusual strain. Public reporting and institutional statements describe allegations of historic address-record manipulation, a major dispute between AFRINIC and Cloud Innovation, court orders affecting resources and accounts, receivership under the Supreme Court of Mauritius, years of board discontinuity, an election attempt in June 2025 later annulled, subsequent board reconstitution, renewed policy activity and continuing litigation. These facts need not be turned into a single morality play. Their importance is institutional. They raise the price of a procedural mistake. In a trusted and stable registry, a close chair call may be treated as routine. In a contested registry governing scarce resources, it becomes a hinge between meeting procedure and economic power.
That is why official documents should be read carefully but not reverently. AFRINIC notices, NRO statements and ICANN correspondence are useful exhibits for dates, process steps, receivership status, policy classification and institutional claims. They do not, by themselves, settle the economics of chair discretion. Likewise, statements by Cloud Innovation, Larus or other interested parties may identify real costs or risks without becoming the frame through which the whole dispute must be seen. The analytical question is not which side owns the story. It is what power the chair's process creates, what objections it excludes or recognises, and whether the resulting record is strong enough for the economic consequences attached to it.
Scope rulings are the first allocation
The first act of chair discretion is not a call of consensus. It is a ruling about what the proposal is allowed to be about. Scope sounds technical until one notices what it removes from the argument. A chair can treat a concern as number-resource policy, business practice, staff implementation, contractual enforcement, legal risk, board authority, election politics or general governance. The ruling determines what evidence counts. It decides whether economic impact, existing-holder reliance, administrative burden and implementation risk are central to the policy record or merely background noise.
AFRINIC's manual needs a scope boundary. The Policy Development Process covers policies for handling internet number resources. General business practices and procedures are outside it. That distinction is sensible. A policy working group should not run payroll, instruct counsel in live litigation, manage every bylaw dispute or replace the board. The difficulty is that, in a post-exhaustion registry, number-resource policy and business consequence are no longer cleanly separable. A transfer rule may be resource policy and also alter asset liquidity, acquisition pricing, broker behaviour, member optionality and litigation exposure. A resource-review rule may look like registry hygiene and also become an inquiry into business model, customer geography or changed use. A contact-validation rule may look like directory maintenance and also affect transfer eligibility, service continuity and sanctions.
When the chair decides scope, the chair decides whether these consequences may be used to test consensus. If economic impact is out of scope, objections about liquidity, collateral, transfer pricing, contractual reliance and financing may be downgraded. If legal enforceability is out of scope, a policy may advance with the assumption that difficult authority questions can be solved later. If implementation burden is out of scope, staff and members may discover the cost only after political momentum has been created. If existing-holder reliance is out of scope, a rule can be described as future policy even when it changes the practical value of resources already issued.
The 2026 ratification of AFRINIC's Number Resources Transfer Policy shows why scope matters. The proposal had reached a consensus milestone in January 2022, was under appeal during part of 2022, and was ratified by the board on 4 February 2026 after years of governance interruption. The policy classifies resources for transfer purposes: AFRINIC-pool resources as "Regional", special-purpose resources as "Reserved", legacy resources as "Legacy" and resources transferred into the region from elsewhere as "Global". As AFRINIC's own overview explains, the result is that AFRINIC-issued IPv4 cannot be transferred out of the AFRINIC service region, while legacy resources and resources brought in from outside may be eligible for outbound movement under specified conditions.
That rule can be described in several plausible ways. Supporters may present it as regional stewardship, protection of African connectivity, a controlled transfer framework or a response to informal markets. Critics may describe it as an exit restriction, a liquidity reduction, a rule that leaves many holders more dependent on one contested institution, or a way of moving economic value between classes of holders. The chair's scope ruling determines whether the second group of descriptions counts as policy evidence or as ideological rhetoric. A process that treats transfer classification as a narrow technical matter will hear different objections from a process that treats it as a policy affecting already-held scarce resources.
The staff assessment for the transfer proposal recognised issues that were not merely semantic. It noted that allowing inter-RIR transfers was not strictly a legal question and involved a business decision for the PDWG and the board. It also discussed implementation changes, resource tagging, transfer logs, contractual revisions, hostmaster burden, due-diligence questions and consequences for unapproved transfers. These are economic facts even when the official vocabulary avoids the word "market". If a chair treats such consequences as peripheral, the discussion narrows before consensus is ever tested.
Scope discretion also affects who can speak effectively. A network engineer may explain that an outbound restriction will affect financing or acquisition terms. A broker may explain how it changes expected value. A small operator may say it reduces optionality during a merger or customer migration. A lawyer may warn about reliance or retroactivity. If the chair frames the proposal as a classification exercise, these interventions may sound off-topic. If the chair frames it as a policy with existing-holder consequences, the same interventions become material.
A good scope ruling should therefore classify the economic class of the proposal at the start: future allocation only, existing-holder impact, transferability or portability, registry-service continuity, resource status, compliance exposure, or governance-adjacent resource policy. The classification need not decide the result. It makes the terms of debate visible. It also prevents rough consensus from hiding an earlier, more consequential narrowing of the field. In scarce-resource governance, the first allocation is not the address block. It is the allocation of legitimacy to certain kinds of evidence.
Time management prices attention
The second chair function is agenda and time management. This is not the separate power to choose the community's political agenda. It is the narrower but still potent authority to order, pace, summarise and close discussion once a proposal is in the policy process. A proposal must reach people before it can bind them. It must be posted, placed on an agenda, presented, discussed, answered, returned to the list and eventually taken to last call. AFRINIC's rules require notice and a draft freeze before a public policy meeting. They do not eliminate chair discretion over order, time, remote participation, repetition, summaries and the moment at which the room is told that enough has been said.
Attention is scarce. Many AFRINIC members are not governance professionals. They are access providers, universities, mobile and fixed operators, hosting firms, data centres, enterprises, IXPs, public bodies and regional service providers. Their staff are dealing with outages, customer support, procurement, abuse tickets, routing incidents, billing and regulation. A policy meeting competes with operating a network. A mailing-list thread competes with customer escalation. A last-call announcement competes with everything else required to keep services running.
Time management converts unequal attention costs into influence. A proposal discussed late in a crowded session receives different scrutiny from one given prime time. A remote question read in summary has different force from a confident microphone intervention. A chair who asks authors to answer objections one by one creates one record; a chair who compresses objections into broad themes creates another. A chair who permits repeated support statements may create momentum. A chair who treats repeated objections as fatigue may create closure. None of this requires bad faith. It is simply how meetings produce records.
The economics matter because the record later becomes evidence of consensus. If a proposal restricts outbound movement of AFRINIC-pool IPv4, affected holders may include firms that did not attend because they did not realise the proposal would change exit value. If a proposal changes resource-review triggers, affected holders may include networks that do not read the list until a transaction or compliance request makes the policy visible. If a proposal touches contact validation, many holders may be affected but few may invest time in explaining how a directory rule becomes a service-continuity risk.
Language and geography sharpen the problem. AFRINIC's region spans many jurisdictions, operating environments and business cultures. English and French both appear in institutional life, but policy argument often rewards a narrow style: technical, legalistic, list-based and familiar to repeat participants. A small ISP manager who says, in imperfect policy language, that a verification step will fail in a jurisdiction with slow corporate records may be giving material evidence. If the chair hears only imprecision, the evidence loses weight. A chair's summary can either translate operational experience into the policy record or let it disappear.
This does not mean chairs should tolerate filibuster. A proposal cannot be held hostage by endless restatement. Authors need to know whether to revise or move forward. Staff need a manageable process. The problem is not that chairs manage time. The problem is that time management becomes economic selection when the subject is scarce resources. A high-consequence chair record should therefore say which affected classes were actually heard: small access providers, large holders, prospective recipients, transfer counterparties, downstream customers, abuse teams, registry staff and others. "Many comments were received" is weaker than an account of whose costs and evidence were represented.
The meeting is not a neutral container. It is a price mechanism for attention. Repeat participants can pay that price more easily. Less procedural but more affected holders may arrive late, speak awkwardly or remain absent. The chair's task is not to pretend these differences do not exist. It is to prevent the convenience of the meeting record from becoming a substitute for legitimacy.
Rough consensus is a burden of reasons
Rough consensus is a useful tool when a technical community needs to solve coordination problems without turning every issue into factional voting. It asks whether the group has broadly converged even though unanimity is absent. Properly used, it protects a correct minority from a simple majority and gives weight to technical and operational substance. Improperly used, it allows a small process class to declare convergence while affected parties are missing, confused or priced out of the discussion.
The phrase carries two different promises. "Consensus" suggests legitimacy. "Rough" gives chairs room to close despite disagreement. That combination is powerful because it authorises movement without a vote. AFRINIC's chairs must decide whether objections are serious enough, whether answers are sufficient, whether discussion has repeated itself, whether a late intervention introduces new evidence, whether a point is implementation rather than policy, and whether the remaining disagreement should stop the proposal. These are not mechanical judgments.
The transfer-policy history shows the economic stakes. A proposal declared to have reached consensus in January 2022 was eventually ratified in February 2026. The ratified rule affects which resources may leave the AFRINIC region. It need not revoke a block to change value. A holder whose addresses are less mobile has a different bargaining position. A buyer who cannot source from the region under the same conditions faces a different market. A lender or acquirer sees a different risk profile. A chair's earlier consensus call can therefore become, years later, part of the explanation for a rule that changes scarcity economics.
A proper consensus record should answer more than whether many people supported the draft. It should identify the objections, the evidence behind them, the affected classes they represent and the reason the chair judged them insufficient. A material objection in this setting is not only a claim that a proposal will break routing. It may be a claim that a rule reduces liquidity, changes reliance interests, pushes transfers into less transparent arrangements, imposes evidence burdens small networks cannot meet, grants staff or the board excessive later discretion, or makes AFRINIC governance risk more expensive for holders.
Economic objections can be self-interested. That is not a reason to ignore them. A large holder arguing for mobility may be protecting profit. A broker arguing for liquidity may be defending a business line. A registry insider arguing for regional retention may be protecting institutional relevance. A prospective entrant arguing for reservation may be protecting future access. None of these interests should be mistaken for public virtue. But interest does not make evidence false. In a scarce resource system, the people with economic exposure often see the cost of a rule before less-exposed observers do.
The better test is consequence. Does the objection identify a concrete operational, legal or economic effect? Does it connect the effect to proposal text? Does it explain why ordinary implementation cannot cure the problem? Does it show that the policy moves value between classes of holders? Does it expose a new discretion point that later staff or boards may expand? Does it describe an existing-holder impact that cannot be avoided? If so, the objection is material even if the objector has a commercial interest.
The opposite is also true. Not every objection should block policy. Some objections are rhetorical. Some repeat defeated points without new evidence. Some try to preserve private advantage at high public cost. Some are designed to exhaust the process. Chairs need authority to close against them. But closure should be reasoned. "This is commercial" is not a reason. "This objection identifies a transfer-value loss, but the proposal applies only to future allocations" would be a reason. "This objection identifies existing-holder mobility loss, but direct notice was sent, grandfathering is included and an independent review route exists" would be a reason. "This objection asserts market harm but gives no mechanism and no affected-holder evidence after targeted notice" would be a reason.
The most dangerous consensus call is the one that lets everyone avoid ownership. The chair says the community supported the proposal. The board later says the chairs followed the PDP. Staff say implementation follows policy. The market then asks who decided that the exit cost, documentation burden, implementation discretion or reliance loss was justified. If no one can answer, the decision may be procedurally valid but institutionally weak.
Rough consensus should therefore be understood as a burden of reasons. The chair need not count votes. The chair must show why the remaining objections do not defeat legitimacy given the resources and holders affected. If that explanation cannot be written in plain terms, the proposal is not ready for closure.
Material objections are where economics enters the record
Chair discretion is most visible when objections are sorted. A policy room rarely divides cleanly between support and opposition. Some objections say the text is unclear. Some say it violates the policy manual. Some say it exceeds the registry's role. Some say it is too hard to implement. Some say it harms a business model. Some say it is unfair to existing holders. Some say it threatens small networks. Some say it will improve security but create a new service risk. Some say it is simply bad philosophy. The chair turns this uneven material into a judgment about consensus.
That translation is economically loaded. If an objection is treated as drafting, an author can reword around it. If it is treated as implementation, staff can solve it later. If it is treated as legal uncertainty, the board may need advice. If it is treated as economic distribution, the proposal may require stronger notice and stronger reasons. If it is treated as personal or commercial grievance, it may be discounted. If it is treated as outside scope, it disappears from the core record. The same intervention can travel through any of these categories depending on chair judgment.
Consider an objection to an outbound transfer restriction: "This will reduce the value of my resources and make financing harder." One chair may hear narrow self-interest. Another may hear a market consequence. A third may ask whether the policy intentionally imposes that cost and whether existing holders were told clearly enough. A fourth may say price is not a registry concern because numbers are not commodities. The resulting record will differ radically even though the underlying objection is the same.
The official doctrine that number resources are not ordinary property can make objection sorting harder. It is true in an important sense. IP numbers are part of a coordinated addressing system, not parcels of land. A registry's role in uniqueness, registration and policy compliance cannot be erased by sale or lease documents. But denying property language does not erase economic reliance. Address blocks support revenue, customers, acquisitions, loan decisions, service contracts and lease income. Registry recognition is the condition that makes those arrangements usable. A chair who treats economic value as irrelevant because formal doctrine avoids property language is not being neutral. The chair is choosing an abstraction over market behaviour.
The Cloud Innovation dispute made this gap visible, even though the dispute should not be treated as a simple proxy for all policy questions. Independent analysis by the Internet Governance Project described AFRINIC's concerns about discrepancies between registered use and actual use, regional service obligations and requests for detailed information. It also described Cloud Innovation's objection that AFRINIC was trying to supervise changing network uses and customer geography. One need not accept Cloud Innovation's whole position to see the chair-discretion lesson. When policy or implementation language gives the registry broad room to decide whether changing use, leasing or out-of-region customers are acceptable, objections about discretion are not merely commercial complaints. They go to the boundary between registry coordination and business supervision.
Materiality should be defined around control points. Does the proposal create a new approval gate? Does it change the status or mobility of resources already held? Does it make a registry service conditional on evidence that some holders may not be able to produce? Does it let staff mark a resource as disputed, non-compliant or non-transferable without objective criteria? Does it rely on later board or staff judgment to decide the real economic effect? Does it treat a live dispute as a general rule? These are material questions because they determine how value moves through the registry layer.
Chairs also need to separate the objection from the objector. A large holder can raise a true distributional concern. A small operator can raise an implementation concern without knowing the right vocabulary. A staff member can identify a real operational problem while defending institutional convenience. An advocate can use public-interest language to cover a private preference. The chair's record should classify the claim, not the moral worth of the claimant.
That record matters in AFRINIC's current environment because trust is not abundant. A bare statement that objections were considered will not persuade holders who see transfer restrictions, resource categories or compliance escalation as value-moving acts. Nor should it. A registry emerging from receivership and litigation should make the chair's reasoning legible. Objection sorting is where procedural authority becomes economic authority. It decides whether stewardship language can override exit cost, whether regional protection can override liquidity, whether contactability can override continuity risk, and whether implementation can be trusted to narrow rather than expand discretion.
Last call is not just a clock
Last call is necessary. A proposal cannot remain open forever. Authors need closure, staff need implementation direction and the community needs a way to move from debate to decision. AFRINIC's process requires a last-call period of at least two weeks and gives chairs the task of evaluating feedback from the public policy meeting and the last-call period before deciding whether consensus has been achieved. In theory, last call is a final check. In practice, it can become a deadline for late discovery.
Late discovery is predictable. Many resource holders do not read every policy thread. Some may learn of a proposal only when a customer, broker, consultant, lawyer or peer warns them. Others may not understand the economic effect until a draft has been revised several times. A proposal title may sound harmless while the operative text carries a transfer restriction, compliance trigger or resource-classification consequence. A holder may have no reason to study the record until the rule looks likely to pass. By then, last call may be the only remaining window.
Process insiders experience last call differently. They have followed the thread, know the authors, understand the version history and can write a precise objection quickly. They know whether to challenge scope, wording, staff assessment, consensus or appeal posture. They know who else might support an appeal. A late-discovering operator first has to understand the proposal, read the manual, inspect old versions, identify the cost, draft a public objection and decide whether open disagreement is worth the risk. Two weeks may be generous for a policy regular and very short for an affected holder.
The appeal rules deepen this asymmetry. AFRINIC requires an appeal to be supported by three working-group participants and filed within two weeks of public knowledge of the decision, after discussion with the chairs or working group. That structure discourages frivolous appeals. It also means effective appeal depends on prior participation and social access. A holder that discovers exposure late may be directly affected and still poorly positioned to gather support from participants who were present. The appeal mechanism then protects procedural regularity more effectively than it protects late-discovering principals.
This matters because last call often arrives after the decisive framing has occurred. By then the proposal has a problem statement, revised text, meeting momentum, chair summaries and a record of support. Authors can say objections have already been answered. Supporters can describe new objections as delay. Chairs may be reluctant to reopen a settled room. A material objection may be true and still arrive when the institutional cost of admitting it has become high.
Last-call closure should therefore be scaled to consequence. For low-impact registry mechanics, an ordinary final review may be enough. For policies that affect existing-holder transferability, resource classification, portability, registry-service continuity, compliance exposure or approval rights, last call should not be the first clear notice that affected holders receive. Chairs should require direct, plain-language notice before last call begins. The notice should explain what may change, which resources are covered, whether existing holdings are affected, what objections have been raised and how a holder can respond without mastering the whole policy culture overnight.
Chairs should also publish an objection matrix at the opening of last call for high-consequence proposals. It need not be bureaucratic. It should identify the unresolved objections, the chair's preliminary view, the evidence still needed and whether each objection concerns scope, text, implementation, reliance, authority, market effect or operational burden. This would turn last call into a review of known issues rather than a scavenger hunt through archives. It would also make closure more defensible.
There is a real danger that more detailed last calls become instruments of delay. The answer is not endless extension. It is disciplined extension tied to missing material evidence from affected classes. If direct notice produces no new substantive evidence, chairs can close with stronger legitimacy. If it produces serious evidence that the proposal changes value or continuity in an unexamined way, closure should be delayed because the process has found information it should have gathered earlier.
Last call is not merely a clock. It is the final conversion point between debate and institutional authority. In AFRINIC, where a policy may be ratified years after its consensus milestone, a thin last-call record can become a durable asset for the winners and a durable liability for those later bound by the rule. Chairs should treat closure as a decision that will have to survive time, not as a meeting-management convenience.
Chair reports, appeals and the ratification record
The chair report is the audit trail of discretion. It tells the board, the community, staff, courts and markets what the chairs thought they were doing. A weak report says little more than that the proposal followed the PDP and reached rough consensus. A strong report states the scope classification, the participation record, the material objections, the chair's treatment of each objection, the implementation assumptions, the remaining risks and the reasons those risks do not defeat consensus. In high-consequence policy, the difference is not literary. It determines whether later ratification rests on evidence or on trust.
AFRINIC's governance history makes this point sharper. The NRO's September 2023 statement on AFRINIC's receivership described the official receiver's role as maintaining the status quo of assets, preserving business value, overseeing elections under AFRINIC's constitution, facilitating formation of a proper board and appointing a chief executive. That statement is useful as a factual exhibit: it identifies the receiver's reported role and the institutional problem receivership was meant to bridge. It does not answer whether a particular policy record was strong enough to support later economic consequences.
The 2026 transfer-policy ratification illustrates the problem of delayed authority. The consensus and last-call milestones occurred in late 2021 and early 2022. Ratification occurred on 4 February 2026, after governance interruptions. AFRINIC publicly described the policy as one that had progressed through the PDP and remained pending because of those interruptions, with restored governance structures allowing the board to consider proposals that had achieved documented community consensus. That sequence may be procedurally defensible. It also shows how chair discretion can become time-delayed economic power. A record made in one institutional period becomes the basis for a rule adopted in another.
Delay is not neutral. Between 2022 and 2026, AFRINIC experienced receivership, election controversy, board reconstitution, public dispute over member status and resource control, bylaw review and renewed litigation. IPv4 scarcity also continued to make transfer rules economically important. A board considering ratification after such a delay should not ask only whether the PDP calendar was once followed. It should ask whether the chair record still answers the questions that matter now: which existing holders are affected, whether they received meaningful notice, whether economic objections were treated as material, whether implementation assumptions changed, whether the policy depends on broad staff discretion and whether targeted review is needed before ratification.
Appeals are the other side of the report. A procedural appeal can annul a chair decision if the process was not followed. That is valuable, but the relevant process in high-consequence policy should include more than dates and meeting formalities. It should include whether the chairs correctly classified scope, identified affected holders, addressed material objections, gave adequate last-call notice and bounded implementation discretion. If appeals examine only whether a proposal moved through the required steps, they may miss the way discretion moves value inside those steps.
The requirement that appeals be supported by working-group participants has a rational purpose. It prevents an uninvolved person from using the appeal process casually. But where a policy affects existing resources, a directly affected holder may discover harm late precisely because the process did not reach it in plain terms. A legitimacy-preserving appeal path should allow such a holder to challenge whether meaningful notice and material-objection handling occurred, even if the holder was not a policy regular from the start. Otherwise, the appeal system can become a club good for participants rather than a safeguard for principals.
Board ratification can also let responsibility diffuse. Chairs say they found consensus. The board says it ratified documented community consensus. Staff say they implement the ratified policy. Each actor can be correct in a narrow sense while no one takes responsibility for the economic judgment. That is dangerous for AFRINIC because contested parties will not treat diffuse responsibility as legitimacy. They will treat it as another reason to litigate, delay transactions, demand indemnities or discount AFRINIC-linked resources.
A recovering board should therefore demand richer chair reports for high-consequence proposals. It need not replace the working group's policy judgment with board preference. It should insist that the record shows the economic consequence of the rule, the classes affected, the objections raised, the reasons for closure and the implementation controls. If the chair report is too thin, the board should send the matter back for targeted review rather than ratify into mistrust.
This is not board supremacy. It is ratification discipline. Bottom-up policy is strongest when the board can point to a record that would make sense to an ordinary affected operator. What changed? Who was heard? What objection was rejected? Why was it rejected? What discretion remains? If the report cannot answer those questions, ratification magnifies an evidentiary gap.
Implementation is where discretion can migrate
Policy chairs do not implement policy. Staff do. But chair discretion shapes implementation by deciding which ambiguities survive into the final text. If chairs declare consensus despite unresolved questions about how a rule will be applied, staff inherit discretion. That discretion may be necessary and narrow. It may also become a hidden second policy.
The transfer policy provides a concrete example. It uses categories, eligibility rules and approval language. It states that number resources are non-transferable unless AFRINIC has expressly approved the transfer request in writing. It tasks AFRINIC with making prudent decisions about whether to approve transfers. It requires the source holder to be the rightful holder of resources not subject to dispute, known or contemplated. It requires recipients to sign the Registration Service Agreement and, in certain cases, show plans, past usage and evidence of compliance. It also provides that resources transferred without prior approval may be deemed non-compliant and reclaimed. These safeguards may have legitimate purposes. They also leave practical questions: what counts as a contemplated dispute, what evidence is enough, how quickly decisions must be made, what "prudent" means, whether refusal reasons are reviewable, and whether one compliance concern can affect unrelated registry services.
If chairs treat those questions as implementation detail, economic power moves from the public policy process into staff decision-making. A holder may later discover that a broad phrase has become a document demand, a delay, a risk flag, a refusal or a compliance review. Staff may be sincere and within policy. The holder still experiences a discretion point that was not fully priced when consensus was declared.
Implementation discretion is not inherently suspect. Registry operations require judgment. Staff must deal with fraud checks, authority verification, duplicate claims, old records, sanctions, court orders, inaccurate contacts, abuse reports and technical consistency. A purely mechanical system would be brittle. The question is whether policy gives staff objective criteria and reviewable reasons, or whether it gives them broad permission to decide whose commercial arrangements feel acceptable.
AFRINIC's history makes the distinction important. Reporting in 2019 alleged historic manipulation of address records and showed why registry data around scarce IPv4 can carry large economic value. That supports strong provenance and verification controls. It does not automatically justify open-ended supervision of every holder's business model. The Cloud Innovation dispute likewise shows how resource review can shift from accuracy and authority into contested questions about changed use, customer geography, leasing and re-justification of old allocations. One can believe AFRINIC had legitimate concerns and still insist that policy language should define where registry coordination ends and commercial supervision begins.
Chair reports should therefore identify implementation control points before consensus is declared. If a policy requires staff approval, what criteria govern approval? If it creates resource categories, who applies them and how can a holder challenge the classification? If it treats a dispute as relevant, who records the dispute and on what evidence? If it allows reclaim or refusal, what notice, cure period and review route exist? If it affects RPKI, reverse DNS, RDAP, WHOIS or IRR services, what continuity safeguards apply? If it requires evidence of use, what level of evidence is proportionate and how will jurisdictions with different documentary systems be handled?
These are not minor operational details. They are the route through which policy becomes market risk. A proposal may be clear in purpose and still economically opaque in application. If affected parties cannot see how a rule will be applied, they cannot know whether to support, oppose or request safeguards. Consensus over unclear implementation is weaker than consensus over a visible decision path.
Implementation monitoring should also feed back into chair discretion. AFRINIC should publish aggregate data on transfer requests, refusals, processing times, refusal categories, cure periods, appeal outcomes, resource-classification disputes and service-continuity incidents. Future chairs should cite this evidence when assessing whether objections about burden are speculative or supported. If implementation has been more burdensome than promised, later proposals deserve higher scrutiny. If predicted harms did not occur, chairs should say so. Evidence should replace institutional folklore.
The chair's duty is not to write every form or bind staff to absurd rigidity. It is to refuse a false separation between policy and implementation where implementation will determine distributional effect. In AFRINIC, implementation opacity is not administrative housekeeping. It is where hidden economic governance can be created.
The contested case evidence
Cloud Innovation is often made to carry too much symbolic weight. To some observers it represents market freedom against registry overreach. To others it represents address arbitrage against regional stewardship. To others it is a litigation strategy, a governance crisis, a test of AFRINIC's survival or a proxy fight over the future of RIR authority. For the narrower question of chair discretion, its use is simpler. It shows why process must define the boundary between registry coordination and commercial control before a policy hardens into authority.
The Internet Governance Project's 2021 analysis described AFRINIC's concerns about discrepancies between registered usage and actual use, consistency between stated needs and later utilisation, and whether services originated in the region. It also described AFRINIC requests for detailed information and AFRINIC's asserted ability, in its own discretion, to terminate the registration relationship and reclaim resources. Cloud Innovation, according to that account, objected that AFRINIC was seeking intrusive control over changed network use and customer geography. Courts in Mauritius later became involved, including orders affecting resources and AFRINIC's bank accounts. Subsequent years brought receivership, election disputes, public argument about AFRINIC's capacity to operate and continuing litigation.
The point is not to decide the entire legal dispute. It is to identify the institutional boundary that chair processes must protect. If the registry asks whether addresses are uniquely registered, accurately recorded, controlled by the proper holder, contactable, not obtained by fraud and not subject to conflicting authority, it is acting like a registry. If it can review customer geography, changed business purpose, leasing arrangements and market strategy through broad discretionary language, it moves closer to supervising commercial conduct. A policy chair should notice when a proposal or implementation path enables that shift.
Cloud Innovation was a large and controversial holder. That makes it an imperfect symbol for small operators. It also makes the boundary visible. A rule that can trap, devalue or expose a large holder will be tested because the holder has enough at stake to fight. The same rule applied to a smaller operator may never reach court. The smaller operator may comply, avoid transfers, keep arrangements informal, accept a discount or abandon a transaction because litigation is too expensive. Chair discretion should not treat the visible dispute as an exceptional quarrel only between AFRINIC and one holder. It should ask what rule would apply to every holder if the discretion became ordinary.
The dispute also shows why chair reasoning should not adopt the framing of any interested party. AFRINIC's public position has presented number-resource administration as stewardship within the RIR system, and its 2026 communications have contested statements by Larus and Cloud Innovation about court orders, leasing and registry authority. Larus, Cloud Innovation and related advocates have presented AFRINIC's posture as overreach, lock-in and insufficiently accountable control over scarce resources. Both accounts are interested. Both may contain facts relevant to policy. Neither should be allowed to define consensus by itself.
This is especially important for the 2026 transfer rule. A regional restriction can be defended as preventing depletion of African resources. It can also reduce the outside option of holders and increase dependence on AFRINIC's process. If a chair treats the first description as public interest and the second as private profit, the record is biased. If the chair treats both as economic claims requiring evidence, the process becomes more credible. The same logic applies to resource review, contact validation and transfer approval. Each may have a legitimate coordination purpose. Each can also become a lever over scarce value.
The case also exposes the cost of weak internal review. If holders believe the policy process and appeals cannot constrain broad discretion, they will use courts. Courts are blunt, slow and expensive. They can freeze accounts, preserve resources, appoint receivers, hear winding-up arguments or invite external bodies to explain systemic consequences. Once disputes reach that level, every policy judgment becomes more politicised. Chair discretion is the place where boundary problems should be exposed earlier, while the rule is still a proposal rather than a litigation trigger.
The lesson is not that chairs should decide live disputes between AFRINIC and Cloud Innovation. They should not. The lesson is that live disputes must not quietly write general rules. A policy born in reaction to one controversial holder can bind ordinary holders for years. A scope ruling that treats holder-rights concerns as off-topic can convert a contested enforcement theory into a general policy norm. A consensus call that downplays market effect can later support board ratification of a rule that changes transfer value. Chair discretion is the point at which these translations should be made visible.
The small operator sees discretion only after it hardens
The cost of chair discretion often reaches small operators late. A small access provider may not know who chaired the meeting, how the last-call record was written or why an objection was rejected. It experiences the result as a form field, a transfer refusal, a documentation request, a contact-validation notice, a changed resource category, a limited exit option or a support conversation that stops moving until compliance is satisfied.
This delayed effect is why procedural discretion can be regressive even when the meeting is formally open. Large holders, brokers, consultants and institutional actors can follow policy early. They can submit written objections, appear at meetings, hire counsel, sponsor analysis and prepare appeals. Smaller operators usually meet the rule after it becomes routine. By then the question is no longer "should the rule exist?" but "can you comply?"
Imagine a small operator with a modest IPv4 block seeking to merge with a neighbouring provider. The commercial terms depend partly on whether the addresses will remain recognised after the transaction and whether unused space can be transferred, leased or retained for expansion. If policy categories restrict movement or approval criteria require months of evidence, the deal price changes. The operator may never have commented on the transfer proposal because it did not see itself as a policy actor. It pays later through delay, legal cost and lower optionality.
Or imagine a network in a jurisdiction where corporate records are slow, old registration details are imperfect and notarisation is inconsistent. A policy that sounds like record hygiene may become a recurring compliance burden. If chair reports treated documentation concerns as implementation detail, the operator now faces the burden without the record showing that its environment was considered. A chair's earlier decision not to treat operational friction as material becomes a later cost for the least procedurally equipped members.
Or imagine a network serving customers across borders. In a global routing system, customer geography can be fluid. If policy or implementation treats regional use as a thick compliance concept rather than a narrow allocation or membership condition, the operator may face questions about its business model, customer disclosure or changed use. A large holder can litigate. A smaller one may avoid transparent transfers, keep arrangements informal, decline efficient transactions or accept a discount.
This is how discretion changes behaviour without visible enforcement. Buyers ask for more warranties. Lenders discount. Customers ask for continuity guarantees. Operators delay updates because they fear scrutiny. Brokers become more valuable because they understand how staff interpret policy. Lawyers and consultants become more valuable because routine transactions now include registry risk. The policy process that was meant to coordinate scarce resources begins to create private rents around interpretation.
The small-operator lens also prevents a false choice between holder rights and public interest. A rule that reduces the value of a large holder's block may look like a blow against speculation. The same rule may reduce the options of small providers that need flexibility for mergers, financing, churn or local instability. A documentation duty may look like fraud control. The same duty can punish operators in countries where records are harder to produce. A regional retention rule may sound developmental. The same rule may discourage inbound resources or reduce transparent transactions that smaller networks could use.
Chairs should therefore make distribution visible early without turning every policy into a treatise. Who bears the cost: large holders, small operators, prospective entrants, downstream customers, staff, courts or the registry budget? Which costs are one-off and which recur? Which actors can pass them on and which cannot? If a proposal claims to protect the region, which actual networks gain and which lose? These questions keep chair discretion tied to the proposal's real effect. They do not require the chair to solve every economic problem. They require the chair not to pretend that procedure has no distributional consequence.
A discipline for chair discretion
The answer is not to abolish chair discretion. A policy process without discretion would invite filibuster, factional mobilisation, endless appeals and rules too rigid for technical reality. Chairs need authority to keep order, identify convergence, test objections and move proposals forward. The reform is to discipline discretion so it becomes accountable, reviewable and proportionate to economic consequence.
The first discipline is consequence classification at proposal intake. Chairs should classify each proposal as registry mechanics, future allocation, existing-holder impact, service-continuity impact or governance-adjacent resource policy. Registry mechanics covers narrow data formats and technical metadata. Future allocation covers resources not yet issued. Existing-holder impact covers transferability, portability, classification, revocation exposure, fees, leasing constraints, recognised control and documentation burdens on issued resources. Service-continuity impact covers RPKI, reverse DNS, WHOIS/RDAP, IRR, contactability and dispute-state changes. Governance-adjacent policy covers proposals whose effect depends heavily on board authority, bylaws, litigation posture or member categories. The chair should publish the classification and invite challenge before substantive debate.
The second discipline is a material-objection standard. AFRINIC does not need every objection to count equally. It does need a public standard for materiality. A material objection should identify a concrete operational, legal, economic or implementation consequence; connect that consequence to proposal text; and explain why ordinary implementation cannot solve it. Chairs should distinguish preference, ideology, drafting, implementation, existing-holder reliance, market effect, authority and adequacy of notice. That would reduce the temptation to dismiss economic objections as self-interest or to let rhetoric block necessary policy.
The third discipline is direct affected-holder notice for high-consequence proposals. Mailing-list publication is not enough where a rule affects existing-holder economics. AFRINIC should send plain-language notices explaining the proposal's practical effect, whether existing resources are covered, what changes in transferability or compliance, what objections have been raised and how to comment. If members ignore clear notice, the chair can close with stronger legitimacy. If members respond with new material evidence, the process has done its job.
The fourth discipline is a chair report that reads like a decision rather than a communique. For high-consequence proposals, the report should state classification, participation record, support themes, objection themes, chair reasoning, unresolved risks, implementation assumptions and safeguards before ratification. It should not merely assert rough consensus. It should explain why remaining objections do not defeat consensus. That would make board ratification meaningful and appeals focused.
The fifth discipline is implementation-condition consensus. If a proposal depends on staff discretion, consensus should be conditional on visible implementation criteria before or alongside ratification: approval criteria, evidence lists, deadlines, refusal reasons, cure periods, dispute flags, appeal routes and continuity guarantees. Chairs should not declare consensus over broad approval language while leaving the real decision path invisible. Implementation is part of the policy's economic effect.
The sixth discipline is a stronger appeal lens for high-consequence chair decisions. An appeal that asks only whether formal PDP steps were followed is too narrow. Review should also ask whether affected holders had meaningful notice, whether the chair addressed material objections, whether scope was classified correctly and whether implementation discretion was sufficiently bounded. Directly affected holders should have a route to challenge those issues even when they discovered the problem late.
Finally, chairs should not operate without institutional memory. AFRINIC should publish aggregate implementation data: transfer processing times, refusal categories, validation outcomes, classification disputes, appeals, cure periods, staff waivers and service-continuity incidents. Chairs considering later proposals should use this data to test claims about burden and benefit. Evidence should replace folklore.
These reforms protect chairs as well as holders. A chair in AFRINIC's contested environment can be accused of bias by whichever side loses. Clear classification, materiality standards and reasoned reports give the chair a defensible record. In scarce-resource governance, speed without legitimacy is not efficiency. It is risk transfer.
The legitimacy test
AFRINIC's chair-discretion problem is not that chairs have power. They must have power. The problem is that the power has changed character as IPv4 scarcity, transfer markets, litigation and institutional instability have changed the stakes of policy. A chair's declaration of rough consensus can help determine whether scarce address resources are portable or locked into a regional framework, whether objections about existing-holder reliance are material or dismissed, whether implementation discretion is bounded or deferred, and whether a board emerging from crisis has a record strong enough to ratify economically consequential rules.
The formal process remains necessary. Mailing lists, public meetings, last calls, chair reports, appeals and board ratification are better than closed rulemaking. But openness is not the whole theory of legitimacy. An open process can still be high-cost, insider-weighted, late for affected holders and vague about economic consequence. Rough consensus can still be a judgment by people whose procedural fluency exceeds their representativeness. Last call can still favour those who already know the record. Appeals can still protect regular participants better than directly affected but late-discovering members. Ratification can still magnify a stale consensus call.
AFRINIC's broader crisis makes this visible, but it is not unique to AFRINIC. Every RIR faces the post-exhaustion problem. Address resources remain technically coordinated but economically valuable. The old distinction between technical policy and economic governance is harder to maintain. A registry can insist that it does not own or sell property. It cannot plausibly deny that its records and rules affect value. Once that is true, procedural judgment must be upgraded. Chairs need to reason like institutional economists as well as meeting facilitators.
The upgrade is to treat every high-consequence chair call as an evidence act. Scope rulings should state what economic class of policy is being considered. Time decisions should recognise who has and has not been heard. Consensus calls should explain why material objections do not defeat legitimacy. Last-call closure should distinguish obstruction from late discovery. Implementation deferrals should specify what staff discretion remains. Appeals should test whether the process was meaningful for affected parties, not merely whether calendar steps were observed.
The payoff is lower risk. A holder that understands how objections were weighed will price AFRINIC discretion differently from one facing bare declarations. A buyer that sees objective transfer criteria will discount less. A small operator receiving plain-language notice before last call will treat policy as less alien. A court reading reasoned chair reports may be less tempted to manage registry policy through emergency orders. A board that ratifies with a strong record can claim legitimacy without asking the market to trust slogans.
The alternative is expensive. If chair discretion remains opaque, losing parties will treat policy as capture. Holders will use courts, delay transactions, avoid transparent updates, demand indemnities or price AFRINIC resources at a governance discount. Staff will inherit ambiguous rules. Boards will ratify into mistrust. The registry will say it is protecting community process while the market reads community process as another risk layer.
The final test is practical. Before a chair declares consensus on a policy affecting scarce resources, an ordinary affected operator should be able to answer five questions without hiring a policy insider: what will change, whether it applies to resources already held, what economic or operational costs were identified, why objections were rejected and how implementation discretion will be reviewed. If those answers are not visible, the process has not produced legitimacy strong enough for the consequence.
AFRINIC's future will turn partly on litigation, elections, finances and operational recovery. It will also turn on quieter sentences from policy chairs. This objection is material. This point is out of scope. There is rough consensus. Last call is closed. The proposal advances. In an abundant world, such sentences are meeting management. In AFRINIC's world, they are economic acts. The registry will be trusted when those acts are narrow, reasoned and reviewable. It will be discounted when they are broad, conclusory and insulated by the language of process.