The policy desk inside a small network
The policy file reaches the small operator at the worst possible time. A backbone router replacement is already scheduled. A municipal customer is disputing an invoice. The abuse mailbox is filling with complaints that mostly belong somewhere else. A manager wants to know whether the company can still obtain enough IPv4 space to support a wireless expansion that was sold before the address plan was finished. The engineer who normally reads AFRINIC notices is also the person who answers an upstream when a route leaks, signs off on customer renumbering plans, updates reverse-DNS delegations, and explains to sales why public IPv4 addresses now carry a market price.
The proposal on the screen does not look dramatic. It has the ordinary shape of internet-number policy: a problem statement, proposed text, references, revision history, mailing-list discussion, meeting discussion, last call, chair assessment, board ratification, and implementation. Its vocabulary is calm. It uses words such as stewardship, accuracy, fairness, scarcity, regional need, abuse handling, resource management, and operational stability. Yet the operator knows that one sentence in a registry manual can become a failed transfer, a lender's discount, a customer contract condition, a delayed acquisition, a support ticket that stops moving, or a lawyer's memo about whether an address block is still useful collateral.
To participate well, the operator has to do more than read the current draft. It must find previous versions. It must identify exactly which manual language is being inserted, deleted, or reclassified. It must compare the new wording with existing transfer provisions, soft-landing rules, abuse-contact duties, resource-review practice, membership agreements, and the recent history of court-tested registry authority. It must ask whether the proposal applies only to future allocations or also to resources already held. It must ask whether an implementation note could turn a harmless phrase into a service denial, whether a cure period exists, and whether a missed notice will later be treated as acquiescence.
That work can be done once. The harder problem is that the policy process asks for repetition. Policy does not arrive as a single constitutional moment. It arrives as drafts, comments, staff assessments, meeting slides, chair summaries, revised wording, board minutes, implementation notices, appeals, and later operational interpretations. A large holder, policy consultant, broker, lawyer, advocacy group, registry veteran, or other recurrent participant can treat the sequence as work. A small operator treats it as interruption. Its cost is not merely time. It is attention diverted from customers, operational risk accepted while reading unfamiliar legal language, money spent on outside advice, and the possibility that a missed procedural step will later be treated as proof that the operator had no meaningful objection.
That is the transaction-cost problem at the centre of AFRINIC policy. Formal openness is real but not enough. Anyone may submit a proposal, write to the list, attend a meeting, or object during final review. But the ability to convert those formal rights into effective influence is expensive. The more valuable IPv4 becomes, the more consequential the policy becomes. The more consequential the policy becomes, the more expensive it is to follow. And the more expensive it is to follow, the more the policy market rewards repeat participants who can spread the same knowledge across many fights.
AFRINIC's recent history makes this economics unusually visible. Public reporting and public institutional statements describe a registry that has faced allegations of address misappropriation, major litigation with Cloud Innovation, frozen accounts, receivership, years of board instability, an annulled election, a later board formation, continuing wind-up litigation, and disputed questions over transfers and member rights. Those facts should not be flattened into one official story or one litigant's preferred account. They do show why every procedural step now sits above scarce operational assets. In a stable registry, the cost of following a policy proposal may look like a governance chore. In AFRINIC, it can look like risk management around assets on which networks, customers, creditors, and counterparties rely.
Formal openness still has a price
AFRINIC's policy manual describes a recognisable bottom-up process. Policies for internet number resources are initiated and discussed by the community. Drafts are published, discussed on the Resource Policy Discussion list, and made available before a public policy meeting. Meeting agendas are announced. Text changes are frozen shortly before discussion so participants are not chasing a moving target in the room. Chairs assess rough consensus. A last-call period follows. If consensus is declared, the proposal moves to the board for ratification. Implementation is then announced through public channels.
Those features matter. They show that AFRINIC policy is not written in a closed legislative chamber. They also show why the cost of participation is easy to understate. An open door is not the same as a cheap seat at the table. To be effective, a participant must know what counts as a policy issue, which part of the manual is being amended, how the change interacts with other provisions, which objections have historically been treated as material, and how to keep an operational concern alive when the draft changes. The manual's steps are visible. The practical knowledge lives in the joints between them.
The first cost is discovery. A resource holder must know that the proposal exists before it can object. That sounds trivial until the operating reality of many AFRINIC members is taken seriously. They are not policy institutes. They are telecom operators, access providers, universities, hosting firms, data centres, public bodies, enterprises, research networks, and regional service providers whose first obligation is to keep service working. Their staff monitor outages, BGP sessions, customer tickets, hardware lead times, payment cycles, security reports, procurement, and local regulatory duties. A policy list competes with operational alarms that have immediate consequences.
The second cost is comprehension. A short proposal can have large implications. A rule about transfer status may affect liquidity. A rule about abuse contacts may affect support access or compliance exposure. A rule about resource review may affect lenders, buyers, and customers who depend on continuity. A rule about regional use may alter exit options even though packet routing remains global. A rule about proper use may give the registry a vocabulary for reviewing business models. A transition clause may matter as much as the substantive rule if it gives holders too little time to adjust.
The third cost is expression. It is not enough for a small operator to say, "this will hurt us." The process rewards objections that identify text, state an operational consequence, distinguish existing policy, propose alternative wording, and arrive at the right stage. That is not a defect by itself; serious rules require serious objections. But seriousness is not free. A lawyer can translate asset risk into legal language. A consultant can cite previous discussions. A repeat participant can speak in the dialect of rough consensus. An engineer running a network may understand the operational risk and still lack the spare hours to make that risk legible to the process.
The fourth cost is persistence. A proposal may be revised after comments. A risk that was obvious in draft one may be partly hidden in draft three. A staff assessment may introduce legal, implementation, or cost concerns in a form that shifts the debate. A meeting may resolve some points and leave others open. Final review may narrow the record. The operator who commented once must continue watching to know whether its concern survived revision. Repeat participants have the advantage of continuity; they remember which words disappeared, which objections were answered, and which unresolved points can still be pressed.
The result is not automatically bad faith. It is institutional selection. A process with high fixed costs selects for those able to pay them repeatedly. That includes public-minded experts and operational veterans. It can also include actors whose commercial, professional, or institutional interests are tied to policy outcomes. The small operator pays per proposal. The repeat participant spreads the same learning across transfers, abuse contacts, resource review, appeals, bylaws, elections, and implementation disputes. Over time, procedural fluency itself becomes power.
AFRINIC's legitimacy problem is therefore not solved by reciting openness. The relevant question is whether the process lowers or raises the cost of effective participation for the operator base it governs. If the cost remains high, policy does not represent the affected base merely because the door was open. It represents the subset that could afford to walk through the door often enough to matter.
The first mover writes the cost curve
The first repeat-player advantage appears before the public argument begins. A policy proposal is not only an opinion. It is a drafted instrument. AFRINIC's manual expects the proposal to name a problem, explain how the proposal addresses it, identify the sections to be modified, use clear references, include supporting material, and maintain a revision history. That template looks administrative. In a post-exhaustion registry, it is a production technology for turning a preferred interpretation into future authority.
Drafting costs are fixed costs. They must be paid whether the proposal affects ten members or a thousand. The person who pays them early controls the initial frame. Is the problem "regional leakage" or "transfer illiquidity"? Is the abuse-contact issue "network accountability" or "directory accuracy"? Is out-of-region customer use a policy breach, a normal feature of global routing, or something the registry should not police at all? Is a transfer rule meant to protect African networks, record changes of control, prevent fraud, preserve fee relationships, or manage capital movement? The first draft does not decide the answer, but it decides the terrain on which later arguments must fight.
The fixed cost includes legal imagination. A drafter must know whether a proposed change touches the policy manual, the Registration Service Agreement, bylaws, board authority, member rights, or implementation practice. AFRINIC's crisis makes that harder. Policy language now sits against a background of court orders, account freezes, receivership, contested elections, board reconstitution, and litigation over resource control. A sentence that might look benign in a stable registry can become risky in an institution whose authority has been tested in court. Drafting requires anticipating not only technical operation but legal leverage.
It also includes economic imagination. IPv4 scarcity changed what policy text does. Before scarcity, need-based allocation language could look like rationing a pool. After scarcity, similar language can restrict movement of resources that holders, customers, buyers, and creditors value. Before scarcity, discouraging stockpiling could look like conservation. After scarcity, broad anti-speculation language can reduce liquidity, impair collateral value, or push transfers into less visible channels. Before scarcity, abuse-contact publication could look like database hygiene. After scarcity, if tied to revocation, support denial, or transfer delay, it can become an enforcement hook over scarce assets.
Drafting advantage is why policy debates often favour those who know the old manual by muscle memory. They know that a change to one definition can alter several sections. They know that soft-landing language, transfer language, assignment language, and abuse-contact language interact. They know whether a proposal should be presented as a narrow correction or a structural reform. They know how to write a problem statement that makes opposition sound like resistance to fairness, stability, accountability, or regional development. That is not necessarily manipulation. It is how agenda formation works when text is the first asset.
The Cloud Innovation dispute illustrates the stakes without making one litigant's case the article's conclusion. Independent analysis of the dispute described contested claims about whether AFRINIC-issued resources had to be used within the region, whether customer geography mattered, whether changes in use required renewed justification, and whether revocation was proportionate. Later debates over transfers and resource classification became economically significant because they could harden or narrow exit options for holders. In that environment, drafting a transfer proposal is not clerical. It changes expectations about value.
Small operators are rarely absent because they have no interest. They are absent because drafting is expensive relative to their chance of success. A small ISP may know that a verification duty will be hard in a market with unreliable corporate paperwork, staff turnover, slow notarisation, and uneven access to specialised counsel. It may know that a transfer delay can kill a financing plan. But knowing the harm is not the same as drafting policy text that survives procedural scrutiny. Repeat participants have standard clauses, past examples, familiar arguments, and colleagues who will review language. Small operators have live networks and scarce time.
A healthier process would treat drafting asymmetry as a design problem. High-consequence proposals should arrive with plain-language impact statements, affected-holder notices, examples of operational effect, and alternative text routes. The drafter should not be the only participant with a usable theory of the proposal. If only insiders can understand a rule before it matures, the process has already tilted.
Archives make history a private asset
The second repeat-player advantage is archive capital. AFRINIC's process depends on public documentation: mailing-list archives, proposal versions, meeting minutes, staff assessments, last-call messages, board ratification records, implementation notices, and policy manual revision histories. Public archives are essential. They make the process less arbitrary. They also create a burden. To know what a proposal means, one often has to read what came before it.
Archive capital is cumulative knowledge that can be reused. A repeat participant remembers why a similar proposal failed, which objection was treated as decisive, which author accepted compromise wording, which chair summary was controversial, which staff assessment identified implementation burden, and which board action was delayed by governance interruption. A newcomer sees a wall of text. The repeat participant sees a map.
AFRINIC's own record makes the map difficult. The policy manual records changes to IPv6 assignment rules, ASN eligibility, soft-landing language, transfer rules, lame delegation, abuse-contact information, temporary resources, reverse delegation, IXP reservations, and anycast assignments. Public exhaustion material records soft-landing milestones: Phase 1 began in 2017, Phase 2 began in 2020, and Phase 2 uses smaller allocation and assignment limits. Public reporting in 2026 described AFRINIC as still holding a remaining pool of unallocated IPv4 addresses while trying to rebuild board and budget activity. Each detail can matter when a proposal concerns scarcity, eligibility, transferability, or implementation.
The archive is not merely AFRINIC's own material. A serious participant must also read public analysis and reporting. The Internet Governance Project's 2021 analysis framed the Cloud Innovation dispute around scarcity, transfer markets, regional-use claims, enforcement risk, and court escalation. KrebsOnSecurity's 2019 reporting described allegations of a large African IPv4 address heist linked to internal record manipulation and scarce address value. The Register followed boardlessness, receivership, election scheduling, nomination concerns, voting disputes, election annulment, a later board election, bylaw disputes, wind-up litigation, and ICANN's interventions. A 2023 public statement from the number-resource coordination community recorded the appointment of an official receiver and described the receiver's role as preserving assets, maintaining the status quo, overseeing elections under the constitution, helping form a board, and supporting the appointment of a chief executive. None of these sources should be treated as a complete explanation of AFRINIC's crisis. Together they show that AFRINIC policy operates inside a contested institutional environment, not a quiet technical workshop.
Archive capital matters because arguments in policy rooms often turn on history. A participant may say a concept was already debated. Another may say a proposal merely implements long-standing consensus. A third may say a previous draft answered the objection. A fourth may point to an old soft-landing clause, a past transfer proposal, or a staff interpretation. The participant without archive capital must either accept those claims or spend hours verifying them. The repeat participant already knows where to look.
Archive capital also makes vocabulary more valuable. Words such as "need," "utilisation," "reserved," "legacy," "regional," "global," "assignment," "allocation," "sub-allocation," "abuse contact," "resource review," and "authorised transfer" carry histories. An operator hearing them for the first time may understand the ordinary English meaning. A repeat participant understands the policy meaning, the litigation exposure, and the commercial implication. In a scarce-resource environment, that gap is expensive.
The cost falls unevenly. A large operator or broker can pay someone to read the archive. A small network may rely on informal summaries. But summaries are never neutral. A policy regular can compress the archive in a way that makes a proposal sound obvious. A critic can compress it in a way that makes the proposal sound dangerous. The absent operator receives rhetoric instead of a record. The more complex the archive, the easier it is for intermediaries to become unavoidable.
This is not an argument against archives. It is an argument for archive usability. High-consequence proposals should come with clean maps of prior decisions, rejected alternatives, unresolved objections, and implementation experience. The process should distinguish between a participant invoking history and a participant proving it. A link to a decade of messages is transparency in form. A structured account of what those messages established is transparency in substance.
When archive reading becomes a private cost, the policy market favours those who have already paid. Knowledge that should reduce collective uncertainty becomes a barrier to entry. AFRINIC's reform challenge is to turn archive capital from a private advantage into a public input.
Meetings convert presence into weight
Policy meetings are necessary because not everything can be decided by slow text. A meeting can surface misunderstandings, test operational objections, force authors to answer direct questions, and help chairs sense whether support is broad, shallow, technical, ideological, or merely repetitive. Yet meetings also convert travel, language, time zones, professional confidence, and social familiarity into influence. That conversion matters in a region as large and unequal as AFRINIC's service area.
Remote participation lowers cost, but it does not erase cost. A participant must know the meeting is happening, clear the calendar, follow specialised discussion in real time, speak in a public setting, and frame an objection with enough precision to affect the record. In-person attendance adds travel, visas, accommodation, lost work time, and the informal advantage of hallway discussion. Participants who meet regularly build trust, shorthand, and reputational capital. The small operator appearing once to object to a harmful clause enters a room that is already socially organised.
Language is a cost even when everyone uses English. AFRINIC's service region contains many legal and technical languages, but policy discussion tends to reward English institutional vocabulary. It also rewards those who can blend engineering, law, economics, and governance language. The operator who says "this will hurt customers" may be telling the truth. The repeat participant who says "the proposal creates an implementation externality, undermines reliance interests, and lacks a cure path" may be saying the same thing in a form more likely to survive the minutes.
Meeting cost is magnified by AFRINIC's crisis history. When a registry has been boardless, under receivership, affected by account freezes, exposed to litigation, and then pressed to rebuild legitimacy, a meeting discussion is not merely a technical exchange. Participants may wonder whether a board will ratify the proposal, whether litigation will challenge it, whether implementation will be delayed, whether court processes will affect authority, and whether policy is being used to change leverage in a broader dispute. The more uncertain the institution, the more background knowledge a participant needs to interpret what happens in the room.
Presence can therefore be mistaken for representation. In a low-participation institution, the people who can stay visible gain weight beyond their numerical share of the affected base. That does not make visible actors illegitimate. It does mean the institution should not treat attendance as a proxy for consent. The policy room may contain the most informed people, but it may still miss the ordinary holders that will pay for the rule.
Meetings also create fatigue. A long debate over a technical proposal may be productive for people paid to attend. For an operator managing outages, customers, and cash flow, every hour has an opportunity cost. Fatigue is not the same as consent; it is a cost of sustained presence. If the same participants can attend every session, correct every misstatement, answer every revision, and remain available through final review, their position gains weight partly because they can keep paying the attendance cost.
AFRINIC can reduce this cost without abolishing meetings. It can publish plain-language summaries before and after discussion, identify the economic class of affected holders, preserve remote questions in a form that chairs must address, separate technical objections from value-allocation objections, and avoid treating late or imperfect comments from operational members as noise merely because they lack meeting polish. It can also identify when the meeting record is thin relative to the proposal's economic stakes and require further targeted notice before the text advances.
The institutional-economics lesson is simple. A meeting is not a neutral vessel when participation costs differ. It is a market in attention. Those with lower attention costs can buy more influence. AFRINIC's policy legitimacy depends on whether it corrects for that asymmetry or merely documents it.
Revisions move value before consensus
Policy proposals often change through versions. That is healthy in principle. Drafting, criticism, and revision are how a rule improves. But version changes are also a transaction-cost machine. Each new draft requires comparison. What changed? Was a hard obligation softened? Was a narrow definition broadened? Was a cure period added? Was a grandfathering clause removed? Did a transfer restriction move from an explanatory note into operative text? Did staff assessment introduce an implementation practice that the policy text itself does not fully show? Did the title remain moderate while the operative effect grew larger?
Version tracking is a repeat-player advantage because the important change is often small. One verb can alter risk. "May" becomes "shall." "Record" becomes "approve." "Contactability" becomes "abuse handling." "Resources assigned after a date" becomes "resources administered by the registry." "Future allocations" becomes "all resources." "Reasonable review" becomes "whenever the registry sees fit." These shifts are not drafting style. In an IPv4 market they can affect transfer delay, revocation exposure, collateral value, customer continuity, and legal advice.
AFRINIC's policy environment contains several examples of why version discipline matters. Soft-landing rules define phases and allocation limits. Transfer rules define which resources may move, under what conditions, and whether incoming or legacy resources retain a status. Abuse-contact rules can begin as directory accuracy and become compliance machinery if implementation ties them to support denial or contractual breach. Resource-review ideas can begin as fraud control and become general supervision of customer geography or business purpose. A participant who tracks each version can identify the moment a proposal crosses from registry mechanics into economic governance. A participant who does not track versions may notice only after the text has acquired momentum.
Revision also changes the burden of objection. Authors can answer criticism by changing words without changing the control point. The objector then has to explain why the revised language still creates the same cost. That requires memory and time. Repeat participants can keep a table of objections and responses. Small operators usually cannot. A concern that remains valid may disappear from the record because the person who raised it does not return for the next draft.
This is not hypothetical in high-stakes registry policy. A proposal restricting outbound movement of AFRINIC-issued resources can be defended as regional development, conservation, or registry clarity. If critics object that it harms liquidity, a later draft may add wording about stability or speculation while leaving the exit restriction intact. The proposal has become rhetorically safer without becoming economically cheaper. A small operator, seeing familiar language about stability, may not detect the retained cost until implementation.
Version changes also create information asymmetry between authors and affected parties. The author knows why the change was made. The affected party sees only the text. A professional participant may infer the strategy. A new participant may not. Good process should reduce this asymmetry by requiring change logs that explain not only what changed but why, which objection it addresses, and which objections remain unresolved. The revision history should not be a decorative list of dates. It should be an accountability device.
The board and staff have a role here. Before ratification, a high-consequence proposal should be accompanied by a version-impact statement: what classes of holders are affected, whether existing resources are affected, whether transferability changes, whether revocation exposure changes, whether implementation requires new documentation, whether fees or support access are implicated, and whether the proposal is retroactive or prospective. Such a statement would not decide the policy. It would lower the cost of understanding it.
The deeper point is that policy value moves before the formal consensus call. By the time a proposal reaches final review or board ratification, much of the economics has already been embedded in language. Repeat participants understand this. They fight early over definitions, scope, and examples. Small operators often enter late, when the cost of changing the draft is highest. A process that wants real legitimacy must make early-stage language legible to those who will pay later-stage costs.
Late discovery is expensive by design
Final review is designed to prevent endless process. A proposal cannot remain open forever. There must be a final period, a point at which chairs decide whether remaining objections defeat consensus, and a path to ratification if the process has done its work. Without closure, repeat objectors could block useful changes indefinitely. The difficulty is that closure also prices late discovery. Those who learn of the proposal late must produce precise objections quickly, often after insiders have spent months building the record.
AFRINIC's manual sets a last-call period of at least two weeks. It also provides a conflict-resolution path for disagreement with chair actions. The participant is expected to first raise the matter with the chairs or the working group. If unresolved, an appeal may be filed with an Appeal Committee appointed by the board, with support from three people from the working group who participated in the discussions and within a short window after the decision becomes public. These safeguards have a rational purpose. They discourage frivolous appeals and require some connection to the process. They also favour people who were already present.
The late-discovering operator faces a stacked cost structure. It must understand the proposal, verify the history, identify the exact procedural act being challenged, write an objection or appeal in the correct form, find participating supporters if an appeal is needed, and do all of this before the deadline. A repeat participant can assemble that quickly. A small operator may first need to ask what the proposal actually does. By the time it knows, the window may have closed.
The narrower issue is not a broad theory of passive consent. It is that late discovery is foreseeable. AFRINIC's member base includes many organisations that will not track every policy thread. A high-consequence proposal should assume that many affected holders will discover the risk late unless the process creates direct notice and plain-language summaries. If late discovery is foreseeable, the process should not treat it as ordinary negligence by the holder.
Appeal costs are especially regressive. Filing an appeal can expose a small operator to reputational pressure, legal uncertainty, and the burden of being publicly associated with opposition. It may fear that objecting to an abuse-contact or resource-review proposal will make it look indifferent to abuse or fraud. It may fear that opposing a transfer restriction will make it look hostile to African development. Repeat participants are more comfortable in public conflict because they know the norms, have allies, and can survive rhetorical counterattack. Small operators may choose not to enter even when the economic harm is real.
The board's role in appeals is complicated by AFRINIC's governance crisis. Appeal committees and board ratification assume functional governance. Public reporting described years in which AFRINIC lacked ordinary board stability, followed by receivership, attempted elections, an annulled 2025 election, a later election producing directors, and continuing legal disputes. In such an environment, an appeal mechanism connected to board-appointed structures may not feel like a low-cost remedy to affected holders. Even if the remedy is formally available, trust in it may be weak.
The solution is not to remove closure. It is to scale closure safeguards to consequence. For registry-mechanics proposals - data format, publication clarity, technical contact fields - ordinary final review may be enough. For proposals affecting transferability, portability, leasing, revocation exposure, fees, security-service access, existing-holder recognition, or retroactive obligations, final review should not be the first moment affected holders receive clear notice. Such proposals need direct notice, impact statements, a longer objection window, and an appeal path that does not require a small operator to already be embedded in the discussion network.
Appeals should also ask the right question. The issue is not merely whether formal steps were followed. A process can be formally open and economically exclusionary. An appeal body reviewing a high-consequence policy should ask whether affected parties had meaningful notice, whether the proposal's economic effect was clearly described, whether operational-cost objections were answered on substance, whether retroactivity was justified, and whether implementation can preserve continuity during dispute.
Final review should be a last check on an understood rule, not a trapdoor for people who were absent when the rule became expensive.
Ratification risk enters the proposal itself
In the policy manual, board ratification appears after consensus. In a stable institution, that can look like a formal step: the community develops policy, chairs assess rough consensus, the board ratifies, and staff implement. In AFRINIC's recent history, ratification is more than a stamp. Board legitimacy, receivership, litigation, and election disputes shape whether the board can credibly convert a policy record into binding institutional action.
AFRINIC operated under severe governance strain for years. Reporting described the registry as unable to elect a board or appoint a chief executive from 2022 through much of 2025. The Supreme Court of Mauritius placed AFRINIC under receivership. A public 2023 statement from the number-resource coordination community recorded that the receiver's role included preserving assets, maintaining the status quo, overseeing elections under the constitution, helping form a board, and supporting appointment of a chief executive. Later reporting described efforts to run elections, questions over nomination arrangements, voting delays, allegations over member authority documents, annulment of a June 2025 election, and a later election that produced eight directors.
These facts matter for policy transaction costs because ratification risk becomes part of the proposal cost. A participant must ask not only whether the proposal has consensus but whether the ratifying board has uncontested authority, whether the policy will survive challenge, whether implementation will be paused by litigation, and whether a future board might reinterpret the result. The uncertainty is not abstract. Reporting in 2026 described continuing litigation over transfers, bylaw changes, member status, and attempts by Cloud Innovation to wind up the registry. ICANN sought permission to participate in the wind-up matter to explain AFRINIC's systemic role and the nature of number resources.
When ratification authority is contested, repeat participants again gain advantage. They can follow court filings, board announcements, communiques, member notices, and policy ratification records. They can ask lawyers whether a ratified policy is enforceable. They can structure transactions around uncertainty. They can delay, challenge, or accelerate depending on their interests. Small operators carry the uncertainty as risk. They may not know whether to plan under the old rule, the new rule, a challenged rule, or a likely future interpretation.
Ratification also creates a temptation for institutional self-protection. A board emerging from crisis may want to prove that the registry is functioning. It may prioritise visible policy action. It may ratify proposals that were delayed during governance interruptions. That can be sensible if the proposals are routine and well understood. It is dangerous if delayed proposals affect existing-holder economics and were formed under participation conditions that changed during the crisis. Ratifying a backlog is not the same as validating each proposal's cost distribution.
The policy-cost lens suggests a discipline. A restored board should separate routine service restoration from irreversible economic policy. It should ratify narrow registry mechanics when the record is strong. It should demand further affected-holder review for proposals that alter transferability, portability, leasing, revocation exposure, resource status, or security-service continuity. A board that wants legitimacy should not ask members to trust it because it survived crisis. It should make policy actions cheaper to understand and easier to challenge.
Ratification should include a reasoned statement for high-consequence policies. The statement should explain what invariant the policy protects: uniqueness, registry accuracy, fraud prevention, contactability, routing-adjacent coordination, security integrity, transfer recording, dispute-state recording, or operational continuity. If the policy's main effect is regional retention, anti-speculation, commercial-model discipline, or redistribution of scarcity value, the board should say why the registry has authority to impose that cost and why ordinary policy participation is enough to bind affected holders.
This would not make AFRINIC weak. It would make ratification credible. A board's value should lie in lowering uncertainty around policy adoption. When ratification adds uncertainty, holders price it as another risk layer. The most useful board in a scarce-resource registry is not the board that can move fastest from consensus to control. It is the board that can prove why a proposal deserves to move from discussion to obligation.
Implementation is the second draft
Policy does not end when the board ratifies. Implementation is often the hidden second draft. A manual clause may be broad or narrow, but staff must translate it into forms, portal fields, evidence requests, validation cycles, support rules, service-level deadlines, rejection reasons, publication changes, and escalation paths. That translation can preserve the policy's stated purpose or expand it. The cost of monitoring implementation is therefore part of the policy-proposal transaction cost.
AFRINIC's manual provides for implementation announcement and timing, including the possibility of delay where implementation requires more time. That is useful. It does not settle what implementation will demand from holders. A transfer policy can be implemented through objective evidence checks and clear deadlines. It can also be implemented as a discretionary approval regime. An abuse-contact rule can be implemented as a directory-reachability flag. It can also be implemented as a route to support denial or contractual breach. A resource-review rule can be implemented as fraud control. It can also become continuing supervision of customer geography or business purpose.
The operator who participated in the proposal debate must therefore continue watching after adoption. What forms are introduced? What documents are requested? Are requests standardised or case-by-case? Are rejection reasons published? Is there a cure period? Are reverse-DNS, RPKI, WHOIS, RDAP, and IRR services preserved during disputes? Are transfer applications processed on a predictable timeline? Are small errors corrected or escalated? Are old resources treated differently from new ones? Are incoming resources trapped by acceptance of services? Are implementation delays becoming informal policy?
This monitoring cost favours repeat participants because implementation is where comparison sets matter. A broker sees many transfer files and can detect patterns. A consultant sees many member requests and learns which evidence satisfies staff. A large holder can compare treatment across blocks. A lawyer can watch whether reasons align with policy text. A small operator sees only its own case. If implementation is harsh, the small operator may assume that is normal. If implementation is inconsistent, the small operator may not have the comparison needed to prove it.
Implementation monitoring intersects with record-integrity concerns without becoming the same issue. Public reporting about alleged address misappropriation showed that registry record changes can involve large economic stakes. That fact makes careful controls around implementation important. But the transaction-cost point is different. Even clean implementation can be too expensive if it requires small operators to produce excessive documentation, chase unclear staff requests, or pay advisers to understand why a policy-driven request is moving slowly.
AFRINIC's 2026 disputes over transfer policy, litigation, and bylaw review show why implementation is not a technical afterthought. Reporting described a newly adopted transfer policy that, in many circumstances, restricts transfers of AFRINIC-assigned IPv4 resources outside the region. It also described debate over whether resource members are registered members under Mauritian company law and whether bylaws should be changed to align governance rights with legal status. Those questions can affect who receives notices, who votes, who can challenge implementation, and whether policy obligations are seen as member consent or institutional imposition.
Implementation should therefore be designed to lower monitoring costs. AFRINIC should publish service levels for policy-driven actions, standard evidence lists, reasons for rejection, cure rights, appeal paths, redacted implementation statistics, and examples. It should distinguish between a policy violation that threatens uniqueness or fraud prevention and a paperwork issue that should be corrected without impairing service. It should preserve last verified operational state during disputes. It should not let one compliance question contaminate unrelated resources or services.
A useful implementation report would answer ordinary operator questions. How many transfers were requested? How many were approved, rejected, paused, or withdrawn? What were the rejection categories? How long did processing take? How many abuse-contact validations failed first attempt and passed cure? How many escalated beyond data-quality flags? How many resource reviews found fraud, documentation gaps, or no issue? How many appeals were filed and how were they resolved? These numbers would not reveal private files. They would reduce the informational monopoly of repeat participants.
Implementation is where the policy market either becomes cheaper or more expensive. A good registry uses implementation to make the rule predictable. A weak registry lets implementation keep discretion alive after the public debate has closed.
Small operators pay the highest marginal price
The repeated cost of policy participation falls hardest on small operators because their marginal hour is expensive. A large carrier may have regulatory staff, legal counsel, government-relations capacity, and enough address exposure to justify permanent attention. A broker or leasing business has direct commercial incentive to monitor every transfer rule. A policy consultant builds career capital from procedural fluency. A small operator has a narrower staff base and fewer spare hours. Yet it may be highly exposed to policy decisions because it depends on address continuity, customer growth, and predictable registry services.
This is the classic repeat-player problem from institutional economics and legal sociology. The repeat player has many cases, learns the forum, builds relationships, spreads costs, and can play for long-term rule change. The one-shotter or occasional participant appears only when a specific issue hurts enough to justify attention. The forum may be formally neutral, but experience changes outcomes. AFRINIC's policy process, because it is open, archived, procedural, and technically specialised, is especially susceptible to this dynamic.
The burden is not only the cost of writing comments. It is the cost of risk assessment. A small operator must ask whether a proposed transfer rule will reduce the resale value of unused space. It must ask whether a contact-validation rule will create service risk when staff change. It must ask whether a resource-review rule will require old customer documentation that no longer exists. It must ask whether board instability will delay a policy it needs. It must ask whether legal disputes around AFRINIC could affect allocation, reverse DNS, RPKI, or transfers. Each question may require external advice.
The burden is also strategic uncertainty. If policy is complex, small operators may avoid transactions that would otherwise be efficient. They may delay transfers, keep unused space idle, avoid transparent leasing, rely on informal arrangements, or buy from less visible channels because official paths feel risky. That harms the accuracy and liquidity the registry should support. A process that raises transaction costs in the name of control can push activity into less visible forms.
The burden is reputational. Policy debates around AFRINIC often use morally loaded language: stability, community, regional protection, speculation, abuse, capture, sovereignty, or disruption. Some of that language points to real concerns. But for a small operator, entering a public debate can mean being sorted into someone else's narrative. If it opposes a transfer restriction, it may be treated as indifferent to regional needs. If it opposes abuse-contact escalation, it may be treated as soft on abuse. If it supports portability, it may be treated as hostile to conservation. Repeat participants know how to survive these labels. Small operators often choose not to enter.
The burden is compounded by geography and market conditions. AFRINIC serves countries with different legal systems, languages, currencies, connectivity markets, and administrative capacities. Some operators work in low-margin environments where a few days of management attention matters. Some face unreliable corporate-record systems, slow notarisation, or limited access to specialised counsel. A policy that assumes every holder can produce perfect documentation, appear at meetings, monitor archives, and appeal on short deadlines is not neutral. It is calibrated to the most procedurally capable participants.
This is why policies marketed as protecting smaller or poorer networks can have the opposite effect. A regional transfer restriction may aim to preserve supply. If it reduces liquidity and discourages inbound resources, small networks face higher search costs. A strict contact rule may aim to improve accountability. If it creates service anxiety around ordinary email failures, small operators carry tail risk. A broad resource-review rule may aim to prevent fraud. If it requires old evidence that small firms cannot assemble cheaply, it raises compliance cost. A complex appeal path may aim to deter frivolous challenges. If it requires procedural supporters and fast filing, small firms lose practical remedy.
AFRINIC's crisis makes the small-operator burden more visible because institutional instability adds another layer. Operators must watch not only policy proposals but also elections, board authority, receivership orders, court cases, interventions by coordination bodies, bylaw debates, and communications from competing advocacy groups. The organisation's effort to rebuild in 2025 and 2026 is important, but recovery does not erase the cost imposed on members who had to operate through uncertainty.
The fair test is not whether a diligent operator could participate. The fair test is whether a normal operator, running a normal network with normal staffing constraints, can understand and respond to a high-consequence proposal before it becomes binding. If the answer is no, the policy process is taxing the wrong people.
High costs create policy rent
High participation costs do not merely exclude. They create policy rent. Rent is value captured not by producing a better service, but by controlling access, information, or procedure. In AFRINIC's policy environment, rent can accrue to consultants who interpret complexity, brokers who understand transfer paths, lawyers who manage appeals, organised groups that can mobilise comments, institutional actors who can stay present across years of debate, and policy veterans whose knowledge becomes hard for ordinary members to replace.
Some of these actors perform useful work. Consultants can help small networks. Lawyers can prevent arbitrary decisions. Brokers can reduce search costs. Advocacy groups can surface ignored risks. Policy veterans can remember past mistakes. The problem is not expertise itself. The problem is a structure in which expertise becomes valuable because ordinary process is too expensive for ordinary members. A healthy registry should need experts for genuinely hard cases, not for routine comprehension.
AFRINIC's history shows several rent channels. Scarcity created address-market value. Policy complexity created interpretive value. Litigation created legal-strategy value. Board instability created governance-intelligence value. Transfer restrictions created knowledge value around which resources can move, under what category, and with what evidence. Abuse-contact or review rules create compliance-navigation value. Each channel rewards those who operate repeatedly in the policy environment.
Policy rent can be mistaken for legitimacy. The same people appear at discussions. They answer questions. They know the process. They can draft. They can cite the archive. They can warn of implementation risks. Institutions come to rely on them because they are present. But presence may reflect lower participation cost, not broader representation. A small number of experts can become the process not because anyone designed capture, but because everyone else was too busy to pay the toll.
The rent is amplified when policy affects existing resources. If a proposal applies only to future free-pool allocations, applicants can decide whether to apply under the new terms. The cost is still real, but it is prospective. If a proposal affects existing holders - transferability, portability, leasing, fees, revocation exposure, security services, or recognised control - the policy changes the value of assets and dependencies already in place. The affected holders cannot simply avoid the rule. That makes the policy process a market in rights alteration. Interested repeat participants have stronger incentives to participate, while dispersed holders may still under-participate.
Transfer policy is the clearest example. A rule that restricts outbound movement of AFRINIC-issued resources changes liquidity. Holders with large positions, brokers, networks seeking supply, regional advocates, and registry institutions all have material interests. The cost of following the proposal is justified for them. For a small operator with a modest block, the expected individual effect may not justify constant monitoring, even though the cumulative effect across the membership is large. This is exactly the condition under which concentrated interests beat dispersed ones.
The remedy is not to ban interested participation. That would remove knowledge and create false neutrality. The remedy is to classify proposals by effect and raise legitimacy requirements where rent is likely. A proposal materially affecting existing holders should trigger direct notice, economic impact analysis, operational examples, anti-retroactivity review, and a meaningful challenge path. It should not rely solely on a room of participants, however open the room may be.
The process should also reduce rent by making routine knowledge public and structured. Clear transfer checklists, version maps, service-level data, appeal templates, implementation statistics, and plain-language summaries reduce the premium paid to insiders. They do not eliminate expertise. They shift expertise toward genuinely complex disputes rather than basic navigation.
The institutional conclusion is uncomfortable but necessary. A policy process with high transaction costs becomes a market. In that market, repeat participants can earn returns from procedural fluency. If AFRINIC wants policy to represent its operator base rather than its policy class, it must reduce the return to procedural fluency and increase the return to operational evidence.
Cost reduction should not become gatekeeping
AFRINIC's legitimacy requires lower policy-proposal transaction costs, but the cure must not be a new layer of discretionary gatekeeping. It would be easy to respond to low participation by letting the board, staff, receiver, or outside bodies decide which proposals are in the public interest before the community sees them. That might reduce noise and improve drafting quality. It would also shift power from repeat participants to institutional screeners. The problem would change form, not disappear.
The better reform is procedural cost reduction with narrow authority. The registry should make participation cheaper without deciding outcomes in advance. It should clarify, translate, notify, compare, and measure. It should not use quality control to suppress proposals or objections that challenge institutional preferences. A low-cost policy process is not managed consensus. It is a process in which affected parties can understand the stakes early enough to participate meaningfully.
The first reform is classification. Every proposal should be classified before discussion as registry mechanics, future allocation, or existing-holder impact. Registry mechanics covers data fields, publication formats, validation methods, routing-adjacent metadata, and security records that do not impair existing reliance. Future allocation covers resources not yet issued from any remaining pool. Existing-holder impact covers transferability, portability, leasing, fees, support access, revocation exposure, RPKI or reverse-DNS continuity, recognised control, or retroactive documentation burdens. Each class should have different notice and review requirements.
The second reform is a policy-cost statement. For high-consequence proposals, the author and staff should identify who pays the cost: current holders, new entrants, small operators, large holders, brokers, downstream customers, registry staff, courts, or counterparties. The statement should estimate documentation burden, monitoring burden, implementation cost, legal uncertainty, transfer delay, and continuity risk. It need not be perfect. The point is to force cost into the open rather than hiding it inside reassuring nouns.
The third reform is direct affected-holder notice. If a proposal can alter existing-holder economics, affected holders should receive a plain explanation outside the general policy list. The explanation should say what may change, when, whether existing resources are covered, what actions holders may need to take, how to object, and what happens if the policy is adopted. This is not paternalism. It is ordinary notice where a rule may affect existing reliance.
The fourth reform is version transparency. Each new draft should include a readable change log that states what objection or issue the change addresses and which material objections remain. Redline text should be available. Examples should show before-and-after treatment of ordinary scenarios: transfer to another region, merger, lease, abuse-contact failure, corporate name change, dormant record, disputed authority, and reverse-DNS continuity. Examples reduce the advantage of participants who can infer consequences from short clauses.
The fifth reform is implementation accountability. Before ratification, staff should publish a draft implementation plan for high-consequence policies: forms, evidence, timelines, cure periods, service effects, rejection reasons, appeal path, data publication, and transition. After implementation, AFRINIC should publish aggregate statistics. A policy whose implementation cannot be described cheaply should not be imposed cheaply.
The sixth reform is independent review for existing-holder impact. This review should not substitute its own policy preference. It should test whether the proposal is within registry function, whether it is retroactive, whether it protects a concrete invariant, whether less burdensome alternatives exist, and whether affected holders received meaningful notice. It should be independent enough that the losing side does not assume the registry judged its own power.
These reforms do not require AFRINIC to abandon bottom-up policy. They require it to distinguish discovery from authority. Mailing lists and meetings are good at discovering technical and operational knowledge. They are weaker at proving consent by absent principals. Boards are good at oversight and institutional responsibility. They are weaker when their own legitimacy is under dispute. Staff are good at implementation. They are weaker when implementation becomes policy expansion. Independent review is useful for scope and process. It should not become a new policy legislature.
The goal is a thinner, cheaper, more predictable policy environment. AFRINIC should make it easier to propose useful rules, easier to understand costly rules, easier to object with evidence, easier to challenge scope errors, and easier to monitor implementation. That is how a registry lowers transaction costs without becoming an editor of acceptable viewpoints or a discretionary gatekeeper over the policy agenda.
A cheaper process is the legitimacy test
AFRINIC's crisis is often told through larger dramas: a disputed resource holder, frozen accounts, alleged address misappropriation, receivership, an annulled election, ICANN intervention in litigation, a restored board, a dissolution fight, and arguments over whether IPv4 addresses are property, public resources, contractual rights, or something stranger. Those dramas matter. But the policy-proposal transaction-cost story is quieter and more durable. It asks who can afford to write, read, attend, revise, object, appeal, and monitor the rules that govern scarce number resources.
The answer, too often, is repeat participants. They may be sincere. They may be knowledgeable. They may be necessary. They may also be economically interested, institutionally invested, or professionally dependent on complexity. A registry policy market dominated by repeat participants does not need conspiracy to drift away from the affected operator base. It needs only high fixed costs, specialised vocabulary, long archives, meeting fatigue, rapid final-review windows, ratification uncertainty, and implementation opacity. AFRINIC has all of these conditions because its policy system sits on top of IPv4 scarcity and years of institutional stress.
The institutional-economics point is not that every operator must vote on every comma. Coordination would become impossible. The point is that rules with larger economic consequences require lower comprehension costs and stronger affected-party safeguards. A data-format rule can move through ordinary process. A rule changing transferability, portability, leasing, revocation exposure, or security-service continuity should not rely on the same participation model. The more a proposal resembles economic governance, the less ordinary openness can carry the legitimacy load by itself.
This is where AFRINIC becomes a test case for the RIR model. The older bargain treated registries as technical coordinators. The process could be light because the function was narrow. IPv4 scarcity, address markets, leasing, resource review, cross-border customer use, RPKI, reverse DNS, litigation, and board instability have made the function heavier. The temptation is to let the policy process grow heavier with it: more words, more compliance, more review, more discretion, more committees, more moral vocabulary. That path raises transaction costs and rewards those already inside.
The better path is the opposite. Narrow the common layer. Make the policy process cheaper. Use policy to protect invariants that running networks actually require: uniqueness, accurate records, fraud prevention, contactability, security metadata, transfer recording, dispute-state recording, and operational continuity. Where a proposal goes beyond those functions into commercial morality, regional capital retention, anti-speculation, customer geography, or existing-holder economics, require direct affected-holder legitimacy and clear authority rather than casual consensus.
AFRINIC's legitimacy will not be rebuilt by hiding weak rules behind public archives. It will not be rebuilt by treating every policy critic as a destabilising actor. It will not be rebuilt by letting a board emerging from crisis convert backlog into economic control without additional review. It will not be rebuilt by making small operators hire specialists to understand whether a sentence in the manual threatens their next transfer, loan, lease, customer contract, or support request.
Legitimacy will be rebuilt when an ordinary operator can see a proposal early, understand its economic consequence, compare versions, submit an objection without procedural translation, receive a reasoned answer, watch final review without surprise, challenge scope errors without needing insider sponsorship, and monitor implementation without hiring a guide to the institution. That is not a utopian demand. It is the minimum cost discipline for a private registry whose rules affect scarce operational inputs.
The final lesson is simple. When policy-proposal transaction costs are high, the policy market stops representing the affected operator base and starts rewarding repeat participants. The result may still be open. It may still be archived. It may still use the language of consensus. But it will not be cheap enough to be representative. AFRINIC's future legitimacy requires reducing those costs without turning policy into discretionary gatekeeping. A registry that lowers the cost of participation lowers the risk premium around its own authority. A registry that raises the cost of participation should not be surprised when operators, courts, markets, and customers begin treating its policy as another source of risk.