AFRINIC and the economics of reclamation and reuse

Summary

• Reclamation creates usable supply only when recovered IPv4 blocks pass through notice, cure, appeal, quarantine and condition evidence before another network relies on them.
• AFRINIC's scarcity makes recovered address space politically tempting, but reuse without proof discipline can turn registry correction into a discretionary capital-control discount.

Reclaimed addresses are used assets, not empty stock

A registry can make a scarce IPv4 block look simple by moving it from one column to another. The account is closed, the prefix is marked recovered, and a finite pool appears to have been replenished. In a region where new IPv4 supply has been rationed to small allocations, the attraction is obvious. A /20 that seemed dormant yesterday can look like relief for sixteen future /24 networks today. It can look like proof that stewardship is serious, that old waste is being corrected, and that late-arriving operators will not be left to buy every address from a costly secondary market.

The simplicity is deceptive. The recovered range is not a blank shelf in a warehouse. It is a used asset in a public coordination system. It may have been routed by one autonomous system and filtered by another. It may have RPKI material, IRR records, reverse-DNS delegations, abuse-history entries, mail-reputation memory, geolocation assumptions, customer contracts, escrow undertakings, lender files, procurement approvals, firewall allowlists and court filings attached to it. Some of those traces sit inside the registry. Many sit elsewhere. Some are accurate. Some are stale. Some record fraud. Some record legitimate reliance. A registry can amend its own database. It cannot make every other institution forget what the range once meant.

That is why reclamation and reuse are a distinct economic problem. Address-utilisation audits ask whether a holder can justify what it has. Waiting lists ask how the remaining pool should ration new demand. Reclamation asks what happens after the registry concludes that an old holder should no longer be recognised, and before another network is invited to rely on the same numbers. The question is not merely whether addresses can be taken back. It is how a scarce, history-bearing asset is made safe enough to put back into circulation without turning registry discretion into capital control.

AFRINIC is an acute test case. Its service region entered IPv4 exhaustion with a relatively small share of the global stock, fast-growing connectivity needs, and operators that often face hard-currency equipment costs, thin margins and uneven access to transfer markets. AFRINIC's public exhaustion material records a soft-landing regime in which allocations are bounded and verified need matters. Its policy tradition emphasises conservation, registration, aggregation and fairness. Those are real constraints. They also make recovered space politically tempting: if the ordinary cupboard is nearly empty, every old range begins to look like a policy reserve.

The temptation is understandable. It would be perverse to leave scarce public numbers trapped in dead companies, held under forged authority, diverted through stale records or warehoused behind claims that no longer survive scrutiny. Public reporting on alleged misappropriation of African address space, including claims that dormant or defunct-company records were manipulated, shows why inaction has a cost. But scarcity also changes the registry's incentives. Recovery can become a visible symbol of strength. It can also become a way to reopen old files, reinterpret old uses, impair market value and force holders to litigate before their networks are made safe.

The institutional answer is not to make reclamation impossible. It is to make it boring in the useful sense: evidenced, classified, noticed, appealable, quarantined, cleaned, documented and reissued by rule. A recovered block should not move from suspicion to reuse in a single administrative motion. It should pass through a procedure that closes the old reliance before creating new reliance. If AFRINIC can build that machinery, reclamation can strengthen the ledger. If it cannot, recovered addresses will carry a discount larger than any nominal supply gain.

Scarcity turns recovery into economic policy

In an abundant address economy, reclamation is mostly housekeeping. A terminated account is tidied, a typo is corrected, a dead contact is removed, and the opportunity cost is limited. Under exhaustion, the same act becomes distribution. A block recovered from one file can relieve another applicant, support a new access network, reduce dependence on leasing, improve a public-service operator's address plan or delay an expensive purchase. A block left unrecovered can be described as waste. The moral language of stewardship therefore arrives quickly.

AFRINIC's public materials help explain why. The soft-landing regime is designed to stretch a finite IPv4 stock, and late-phase allocations are constrained. Requests must be evaluated against documentation and conservation rules, not against an assumption of ample inventory. That official text is a factual exhibit, not a conclusion. It shows that the registry operates under scarcity. It does not by itself settle how far recovery power should reach, what process is required, or when a holder's old reliance should give way.

Scarcity changes the politics around every candidate block. Members waiting for addresses may see dormant ranges as hoarding. Incumbents may support enforcement against rivals while expecting tolerance for their own legacy mess. Brokers may see recovered space as future liquidity. Governments may ask why public numbers appear unused while connectivity projects struggle. A board under criticism may find recovery easier to explain than restraint. A registry staff member may feel that a conservative decision leaves value on the table.

The private costs are less visible. The current holder, its customers, lenders, counterparties and potential acquirers see a different risk: that the registry can revisit old paperwork and impair an asset without absorbing the loss. If that risk is unpredictable, it is priced into every AFRINIC-administered block. Buyers pay less or demand stronger warranties. Lessees insist on termination clauses. Lenders treat addresses as weaker collateral. Operators retain larger buffers. Customers ask for continuity commitments. Scarcity does not merely raise the value of recovered space; it raises the cost of procedural error.

That is the central economic tension. Recovery can be pro-competitive when it removes fraud, abandonment and false scarcity. It can be anti-investment when it creates a discretionary cloud over every old allocation. The difference lies in the criteria. A market can tolerate clear forfeiture for proven fraud, voluntary return, non-payment after notice, court-settled invalidity or genuine abandonment after reasonable inquiry. It cannot efficiently price an open-ended power to reinterpret historic use whenever the value of the addresses has risen.

Reclamation should therefore be understood as a supply policy with a legal and operational tail. It increases usable inventory only when the recovered block can be relied upon by the next user. It does not increase usable inventory merely because the registry has changed a status field. A nominally recovered /16 that remains disputed, reputation-damaged, unroutable or legally encumbered is not equivalent to a clean /16. The economics of reuse begin with that quality adjustment.

The registry ledger is powerful but narrow

The registry record is powerful because the internet needs uniqueness. If two firms can credibly claim the same address block in public records, routing, abuse response, contracting and security all become more expensive. AFRINIC's registration function exists to reduce that ambiguity. It tells network operators, customers and public users which party is recognised for a particular range and which contacts should be used for operational matters. Without that ledger, coordination costs rise quickly.

Yet the ledger's legal and economic effects travel far beyond the registry. A block may be named in a data-centre contract, a bank's security exception, a government tender, a merger schedule, an escrow agreement, an equipment configuration, a peering filter, a route-server file, a cloud onboarding check or a mail-provider reputation system. The registry did not create all of those dependencies. But its record helped others build them. When the record changes, those dependencies do not disappear at the same speed.

This is why a recovered block differs from never-issued inventory. A new allocation from untouched stock has prospective risk. A recovered allocation has historical risk. The old holder may have left customer dependencies, reputation damage, stale routing authorisations, reverse-DNS delegations, unresolved invoices, public allegations or private claims. Some of those facts may be irrelevant. Some may be decisive. The new holder will discover them only when banks, carriers, exchanges, security firms or customers react.

The registry can determine whom it recognises after due process. It should not pretend that recognition alone cleanses all consequences. That distinction is especially important in the AFRINIC setting, where public controversy around address use, alleged misappropriation, contractual enforcement and litigation has shown that registry actions can affect large commercial positions and downstream users who never attended a policy meeting. A termination or recovery notice is not merely clerical when the addresses support revenue, customers and market value.

None of this gives every reliance interest a veto. A fraudulent holder cannot defeat correction by signing customers. A hijacker does not gain legitimacy by routing stolen space. A spam operation should not preserve a block because it built a business on abuse. But a narrow ledger must distinguish between reliance that deserves protection, reliance that deserves transition, and reliance that deserves no weight. A successor entity with late paperwork is not the same as a forged holder. A quiet disaster-recovery reserve is not the same as abandonment. A contractual lease is not the same as theft. A dispute between commercial parties is not the same as an exhausted appeal.

The wider the reliance, the more carefully the registry must move. That does not weaken its authority. It locates the authority where it belongs: in maintaining accurate recognition, not in pretending that every economic consequence downstream is irrelevant. The registry should be able to correct the record. It should also be able to show that it understood what the correction would disturb.

Dormancy is a clue, not abandonment

The hardest reclamation cases often begin with silence. A prefix is absent from the global routing table. A contact address bounces. A company registry page is out of date. Reverse DNS appears lame. No one has opened a recent support ticket. To an exhausted registry, the range looks idle. To a waiting applicant, it looks wasted. To an opportunist, it may look vulnerable.

Dormancy is evidence. It is not a verdict. A prefix may be absent because its holder is gone, because the block is being prepared for sale, because a customer migration is under way, because an upstream filter broke, because the range is reserved for disaster recovery, because it is used behind a covering aggregate, because a merger froze changes, because a liquidator is collecting assets, because sanctions or political instability disrupted contacts, or because the block has been hijacked and the attacker is waiting for attention to fade. The same public symptom can point to opposite facts.

A policy that treats non-use as self-proving abandonment creates perverse incentives. Operators may announce address space solely to avoid appearing idle. They may maintain token services or unnecessary IRR records to prove life. Enterprises may avoid efficient consolidation if quiet reserves become vulnerable. Networks may route low-value traffic through a range merely to create evidence. The result is not conservation. It is defensive use, where addresses are made visible because the registry rewards visibility rather than need.

AFRINIC's policy tradition already implies more nuance. It values actual need, immediate use, registration accuracy and avoidance of stockpiling. That does not amount to a simple "announce or lose" rule. The question after an audit should be why the address block appears dormant and whether that condition defeats the basis on which the range is held. A serious inquiry looks at public routing, RDAP or Whois records, member correspondence, payment history, corporate succession, insolvency filings where available, reverse-DNS status, RPKI material, IRR records, abuse-contact responsiveness, contracts and court records. No single signal should carry the whole decision.

The economic purpose of this inquiry is to reduce false positives. A false negative leaves genuinely abandoned supply idle. That is costly under scarcity. A false positive can damage customers, contaminate future reuse, trigger litigation and reduce confidence in every other registration. Under IPv4 scarcity, both errors matter. But the second error is systemic: it teaches holders and buyers that the registry's recognition is conditional on later administrative judgment they may not be able to predict.

The standard should be practical, not impossible. The registry need not achieve perfect knowledge. It should build a rebuttable file strong enough to justify notice, cure, decision and eventual reuse. It should be able to say: we checked the relevant channels, contacted plausible authority holders, allowed a defined response period, evaluated the answer, classified the case and preserved evidence. That file is the difference between recovery and inference.

Procedure is the asset that makes reclamation credible

The most dangerous recovery decisions are not always the most aggressive. They are the least legible. If the holder cannot tell what triggered review, what evidence is required, who decides, how long cure lasts, what appeal exists, and what happens to dependent services while the dispute is open, the registry has not created a reclamation process. It has created a discretionary hazard.

A credible process separates stages. First comes triage: the registry identifies a candidate block and records the trigger, whether non-payment, failed contact, suspected fraud, invalid registration, court order, apparent abandonment, policy breach or voluntary return. Second comes notice: a precise statement goes to the registered contacts and, where the facts warrant, plausible successors, insolvency office-holders or administrative counterparts. Third comes cure: the holder can update contacts, prove current authority, explain reserve use, pay arrears, correct authority records, disclose protected reliance in a controlled way or propose migration. Fourth comes decision: the registry issues reasons and a proportionate remedy. Fifth comes review: a genuine appeal or external path exists before irreversible reuse, except in narrow emergencies.

This may sound slow. It is usually faster than litigation. The Cloud Innovation dispute is a warning here, though it should be treated cautiously because important allegations and contractual arguments remain contested in public accounts. AFRINIC's attempt to terminate or restrict recognition for a large holder became tied to court action in Mauritius, operational pressure on the registry and wider arguments over regional-use policy, registry authority and commercial reliance. Whatever view one takes of the parties, the episode shows that existential remedies invite existential resistance.

Procedure protects the registry as much as the holder. When recovered inventory has value, every decision will attract pressure. Applicants will ask why a range was not recovered sooner. Holders will allege favouritism. Brokers may speculate about timing. Governments may treat unused ranges as development failures. A clear process gives the registry a defensible answer: the evidence is not yet sufficient; the cure period is still open; the case is in appeal; the block is in quarantine; or the range is ready for rule-bound reissue.

Proportionality is part of procedure. Not every defect justifies full recovery. Some defects justify a record update. Some justify a transfer freeze while authority is clarified. Some justify temporary suspension of an ancillary service. Some justify partial recovery of unassigned space inside a larger holding. Some require a court order or independent decision. Only the strongest cases justify cancellation and eventual reuse. If every problem is treated with the same remedy, the registry will either under-enforce serious abuse or over-enforce clerical failure.

For AFRINIC, the process must be designed for the hard case, not the easy one. Voluntary returns and clear non-payment are not the true test. The test is a large, commercially embedded range with stale records, active customers, allegations of misuse, contested policy meaning and possible litigation. If the process can handle that case without collapsing into arbitrary power or paralysis, it can handle ordinary recovery.

Classification should come before remedy

Reclamation debates often jump too quickly to the remedy. Should the block be taken back? Should it be returned to the pool? Should it be held? Those questions cannot be answered until the case is classified. The same outward condition - an old record, a quiet prefix, a failed contact - may require very different treatment depending on why it exists.

Voluntary return is the cleanest category. A holder acknowledges that it no longer needs the space, confirms authority, and works with the registry to close records. The main tasks are record preservation, reverse-DNS transition, RPKI and IRR record cleanup, abuse-contact closure and quarantine proportionate to recent use. Non-payment after verified notice is also relatively clear, though large ranges and dependent customers may still require continuity planning before final reuse.

Fraud is different. If the initial allocation was obtained through forged documents or if authority was later manipulated, the registry must correct the record and preserve evidence. But even fraud cases need classification inside the category. Was the current user the wrongdoer, a successor, a customer, a purchaser or an intermediary? Are there innocent downstream networks? Is law enforcement involved? Is there a court order? The answer affects timing and continuity, not necessarily the final conclusion that the registry record must be corrected.

Stale corporate succession is different again. Many African and Indian Ocean businesses have changed names, merged, entered liquidation, been nationalised, moved assets, outsourced network operations or folded systems into parent groups. Public corporate registries can lag reality. A dissolved-looking company may have a successor. A bank, university, state agency or industrial group may have retained address rights even after the original network team vanished. Reclaiming too quickly can destroy value that belongs to creditors, public institutions or customers. Moving too slowly can allow false claimants to exploit a gap.

Policy breach is another category, and it requires special care. If a holder's use differs from historic representations or from a regional-use rule whose meaning is contested, full recovery before review is a severe tool. The registry may have legitimate concerns. The appropriate first remedies may be disclosure, prospective compliance, segmentation, transfer restriction or a freeze on new changes. Confiscatory remedies should be reserved for cases where the breach, the authority to act and the remedy are all strong enough to survive scrutiny.

Court-linked disputes belong in their own class. A registry may be ordered to maintain the status quo, freeze changes, recognise a receiver, preserve evidence or refrain from reissue. In those cases, administrative impatience is dangerous. The registry's task is to comply with the legal constraint while keeping public status clear enough that third parties are not misled.

Classification is not bureaucracy for its own sake. It prevents the registry from treating all uncertainty as abandonment and all non-compliance as fraud. It also prevents holders from laundering every defect into a demand for delay. Once the class is known, the remedy can be narrower, faster and more defensible.

Voluntary return needs a safe harbour

The least contentious recovered address is the one a holder gives back. A firm shuts a service, completes a merger, moves customers, finishes a migration, discovers an unused historical range or decides that the carrying cost and audit exposure are no longer worth the option value. In theory, voluntary return is the cleanest form of conservation. In practice, it will happen less often if the return process feels like an invitation to punishment.

Holders respond to incentives. If returning a range leads to hostile questioning about every past use, public suspicion, slow closure, uncertain treatment of related holdings or loss of standing in other registry matters, many firms will keep quiet. They may route defensively, leave old records untouched, or wait for a transfer opportunity rather than volunteer unused stock. That is bad economics. The registry wants information from the party best placed to know the range's history. It should not make candour irrational.

A safe harbour does not mean amnesty for fraud. It means that a holder that approaches the registry voluntarily, discloses the relevant history, cooperates with cleanup and is not concealing known adverse claims receives a predictable path to closure. The registry can still preserve evidence, correct records and refer serious wrongdoing where necessary. But it should distinguish a good-faith return from a forced recovery after evasion. That distinction increases supply by lowering the cost of doing the right thing.

The safe harbour should be procedural rather than informal. It can specify what the returning holder must provide: authority to act, the affected prefixes, recent routing status, known downstream reliance, reverse-DNS and RPKI status, relevant IRR records, known reputation issues, pending disputes and proposed dates for public status changes. It can also specify what the registry will do: acknowledge receipt, publish a neutral status, set quarantine, avoid public blame where none is established, and issue a closure record once the range is ready for future use.

This matters in the AFRINIC region because many old holdings sit inside institutions whose network history is messy but not necessarily abusive. A bank may have consolidated. A university may have shifted services to a national research network. A state agency may have reorganised. A commercial provider may have renumbered after an acquisition. If every approach is treated as a suspect file, the registry will learn less and recover less. If good-faith return is administratively safe, the registry turns private knowledge into public inventory.

Voluntary return also helps separate reclamation from confiscation. The more addresses enter quarantine through cooperative closure, the less every recovery discussion looks like a fight between the registry and a holder. The registry still needs hard enforcement for fraud, abandonment and non-payment. But a mature programme should make the cooperative path cheaper than the adversarial path. Conservation works best when the holder can leave cleanly.

Quarantine is where recovered supply becomes credible

The key institution between reclamation and reuse is quarantine. Without it, recovery and reissue collapse into a single event. The old holder loses recognition on Monday; a new holder receives the block on Tuesday; the market then discovers stale records, reputation damage, unresolved reverse DNS, old routing authorisations, confused geolocation, adverse claims or surprised customers. That is not conservation. It is risk transfer.

Quarantine creates a controlled interval in which the block is not ordinary inventory. The registry has recognised that the old claim has ended or is no longer sufficient, but it has not yet invited a new party to build reliance. During that interval, the registry confirms that notice and appeal windows have closed, authority records are reconciled, RPKI status is clear, reverse-DNS delegation is addressed, abuse contacts are reset or labelled, public disputes are reflected, and operational warnings are recorded. The length of quarantine should vary with the facts: block size, recent use, dispute history, customer reliance, reputation concerns and the complexity of associated records.

The economics are straightforward. A quarantined block has option value. It may become usable supply. It is not yet equivalent to clean stock. The interval reduces information asymmetry between the registry and future recipients. It also reduces the incentive to lobby for immediate recovery. If every recovered block first enters a documented cooling-off and cleanup process, recovery looks less like a prize transfer and more like an institutional act.

Quarantine should not become a hidden waiting list. That belongs to a separate rationing question. The purpose here is condition, not queue position. The registry should be able to explain why a block is in quarantine, what must be resolved before reuse, and whether there is any expected date for release. Applicants should not be encouraged to treat quarantined ranges as promised inventory. Holders should not be allowed to use quarantine as a way to preserve hopeless claims indefinitely.

The process also helps AFRINIC manage legitimacy. Because its remaining IPv4 pool is small, any large recovered block would attract attention. A /16 or /15 returned to the system is too valuable to reissue casually. If it is released without a public explanation, members may suspect favouritism or hidden bargaining. If it is held indefinitely without explanation, members may suspect paralysis. Quarantine with published status is the middle path: not immediate distribution, not permanent fog.

The registry should measure quarantine separately from recovery. Addresses under quarantine should not be counted as usable yield. They are recovered candidates, not yet deployable supply. That accounting discipline matters because headline recovery numbers can otherwise reward the wrong behaviour: taking addresses back quickly while leaving the harder work to the future user.

Cleansing is operational, not cosmetic

The word "cleansing" can sound like public relations. In IPv4 reuse it is concrete operations. A block is clean enough for reissue only when the main authority signals that affect its use have been reconciled or at least labelled. The registry's own RDAP or Whois data must be coherent. Maintainer and authority records must not point to the wrong party. IRR records should not authorise a stale origin without warning. RPKI material and ROAs must not make the new announcement invalid. Reverse-DNS delegation should not point to abandoned or adverse infrastructure. Abuse contacts, geolocation feeds and major reputation systems may need transition notices or remediation plans.

These are consequence channels, not the central thesis. IRR governance, ROA revocation, DNS delegation and address reputation each have their own economics. For reclamation, their importance lies in timing and burden. A registry should not reissue space while registry-controlled authority signals still contradict the reissue. Nor should it tell a new holder to discover every defect alone after allocation. The new holder can accept risk only if the risk is disclosed.

Some defects are within the registry's control. It can update public registration data, remove or mark outdated authority references, adjust reverse-DNS delegation after proper notice, and ensure that RPKI status does not conflict with recognised control. Other defects sit outside the registry. Mail blocklists, security vendors, geolocation services, carrier filters and cloud providers may update slowly or not at all. The registry cannot force them to forget. It can, however, publish enough transition evidence to make remediation cheaper.

The cost of cleansing should shape reissue policy. A small block with no recent routing, no dispute, no adverse public history and a clear voluntary return may need light quarantine. A large block recovered after contested use, with old IRR records, previous ROAs, customer dependencies and public allegations, may need a longer hold and a detailed condition report. Some ranges may be so impaired that immediate reissue is inefficient. They may be better segmented, held until records settle, or offered only to recipients with the capacity to remediate.

AFRINIC should avoid two extremes. It should not imply that a recovered block is pristine merely because the registry file is updated. That shifts costs to the next user and damages confidence. It should also avoid making perfect external cleanup a precondition for any reuse. Some internet memory is slow and imperfect by nature. The practical aim is not zero residue. It is known residue, bounded residue and registry-controlled residue fixed before the range is placed back into ordinary allocation.

This is where market evidence matters. The recipient should receive not only a prefix, but a history of what was checked and what remains. That history does not stigmatise the block. It makes the block more usable because it lets the recipient plan.

Alleged misappropriation shows why inaction is costly

A cautious account of reclamation must not become an argument for passivity. AFRINIC has operated in the shadow of public reporting that alleged serious address misappropriation. KrebsOnSecurity reported in 2019 on claims by researcher Ron Guilmette and others that African IPv4 ranges associated with defunct or acquired organisations had been moved or sold through questionable authority chains, with allegations involving an AFRINIC insider. AFRINIC said publicly that it was investigating. Those reports should not be converted into adjudicated findings against every named party or every range discussed in public. They do show why stale records are not harmless.

The alleged pattern is economically important. Dormant companies create opportunity. A valuable address range remains in public records, but the original holder is gone, merged, disorganised or unaware. The registry contact is stale. The successor is unclear. A broker, insider, claimant or attacker may be able to present enough paperwork to move control before anyone with a better claim notices. The more valuable IPv4 becomes, the stronger the incentive to exploit that ambiguity.

In such a world, a registry that refuses to reopen old files rewards the party most willing to manipulate silence. The cost is borne by legitimate operators, successors, customers and the wider community. Scarcity makes this worse because every misdirected block is both a private gain and a public loss. Recovery power is therefore necessary. It is part of the registry's anti-fraud duty.

But the same facts also warn against careless recovery. A defunct-looking company may have a successor. A public body may have reorganised. A university department may have changed name. A bank may have moved systems into a parent group. A liquidator may control the asset. A firm may have outsourced network operations without giving up the address range. If the registry treats stale corporate form as abandonment, it can repeat the injury it is trying to prevent, this time under administrative colour.

The remedy is chain-of-authority discipline. Before a stale range is reissued, the registry should document the historic holder, known changes, plausible successors, notices sent, responses received, evidence of fraud or abandonment, and the reason no better claim remains. In some cases, public notice may be appropriate. In others, a court, liquidator or government office may need to be contacted. Where evidence is weak, a hold is better than an irreversible transfer.

This discipline is also a corruption control. Valuable dormant ranges create temptations for insiders and outsiders alike. A recovery programme without chain-of-authority files becomes a hunting ground. A recovery programme with standard evidence, dual review, logs, appeal rights and public condition reports makes manipulation harder. It does not eliminate fraud. It raises the cost of successful fraud and lowers the cost of detecting it.

The Cloud Innovation dispute warns against total remedies

The Cloud Innovation episode should be handled with care. Public accounts describe AFRINIC's concerns about use of large address holdings, Cloud Innovation's objections, attempted termination or restriction, litigation in Mauritius, court restraints, bank-freeze pressure on AFRINIC and a broader governance crisis. The facts and legal arguments are contested in important respects. The policy lesson does not require treating every claim by either side as settled. It requires recognising what happens when a registry remedy threatens the whole economic base of a holder.

When a registry moves toward full termination or wholesale recovery, the holder experiences the act as existential. If the addresses support leasing, hosting, customers, collateral value or future sale, the expected loss can dwarf registry fees. A rational holder will litigate, seek injunctions, mobilise customers, lobby allies and frame the registry as confiscatory. The registry may see itself as enforcing policy. The holder may see itself as defending property-like reliance. The dispute then becomes larger than the original file.

Total remedies also make proportionality harder to see. If the alleged defect is inaccurate contact data, the remedy may be correction. If the issue is non-payment, the remedy may follow notice and cure. If the issue is a contested interpretation of regional-use rules, immediate full recovery before review is a high-risk tool. If the issue is proven fraud, stronger remedies may be justified. Lumping these situations together under one cancellation power makes the registry look arbitrary even when some enforcement is warranted.

The episode also shows how reclamation can become a proxy war over IPv4 scarcity, leasing markets, regional development, registry authority, governance legitimacy and commercial control. Once a case carries all those meanings, a narrow administrative file cannot contain it. The registry needs a process that prevents one dispute from becoming a referendum on all registry power.

That process should make three promises. First, AFRINIC will not ignore fraud, abandonment or serious policy breaches merely because enforcement is hard. Second, it will not use broad discretion to destroy going-concern value where narrower remedies can protect the ledger. Third, it will separate continuity of dependent networks from the final entitlement dispute as far as technically and legally possible.

This is not softness. It is institutional risk management. A registry that has faced governance stress and court pressure cannot rely on abstract trust. It must show bounded power. Even the party that loses should be able to see why the decision was made, what evidence mattered, what alternatives were considered and why the remedy matched the defect. That is the difference between stewardship and a confiscation drama.

Governance stress makes the file more important

Reclamation power is hardest to trust when the institution exercising it is under strain. AFRINIC has faced public controversy over board legitimacy, court proceedings, receivership arrangements and the continuity of ordinary corporate authority. Those facts should be handled as context, not as a verdict on every decision taken by every official. They nevertheless matter because recovery decisions depend on confidence that the person signing the decision is authorised, that the evidence file is complete, and that the remedy is not being used to win a governance battle by other means.

In a stable institution, members may accept a terse decision because they trust the machinery around it. In a stressed institution, the machinery must be visible. Who opened the review? Under what delegated authority? Who checked the evidence? Was there a conflict? Were legal restraints considered? Did a receiver, court-appointed officer, board, chief executive or staff committee hold the relevant power at the time? Was the same standard applied to comparable cases? Can the file be audited later without relying on personal memory?

These questions are not procedural decoration. They affect market value. A buyer or lender looking at a recently recovered block will ask whether the recovery can survive challenge. If the decision was made during a governance crisis, the condition report must do more work. It must show not only that the prior holder lost recognition, but that the registry itself had a valid path to act. Otherwise the risk follows the block into the next transaction.

Governance stress also increases the danger of mixed motives. A recovery decision may be technically justified and still be suspected if it appears to target a political opponent, a litigant, a critic, a brokered transaction or a class of holders unpopular in the policy room. The answer is not to stop enforcing during institutional difficulty. That would reward bad actors who wait for weakness. The answer is to make enforcement files more standard, more reviewable and less dependent on personalities.

Receivership or court supervision can even improve discipline if used correctly. A neutral officer may insist on records, authority chains, payment controls and reasoned decisions. But it can also slow action or create uncertainty about who may bind the registry. Public status labels, internal delegation records and court-order references become part of the economic infrastructure of recovered space. They tell the market whether a decision is provisional, final, restrained or appealable.

For AFRINIC, this means reclamation files should be built as if they will be read years later by a new board, a court, a recipient, a lender and a sceptical member. The file should not require trust in the mood of the day. It should stand on dates, notices, evidence, authority, reasons and review rights. That is how a registry under pressure prevents necessary recovery from being discounted as institutional improvisation.

Customer continuity should be separated from final entitlement

Reclamation decisions often speak to the registered holder, but the costs travel through customers. A hosting client may not know that its provider's address range is in dispute. A bank may have approved a source address years earlier. A hospital vendor may depend on a firewall exception. A school platform may use a static address for filtering. A public agency may have procurement documents tied to an address plan. If a registry action suddenly changes recognised control, those downstream users may experience the event as a service failure rather than a governance correction.

Customer reliance does not decide final entitlement. It should shape the path. A forged holder cannot keep addresses forever because innocent users exist. But the registry can still ask what transition is necessary to avoid unnecessary harm. There is a difference between stopping new unauthorised changes, freezing transfer, preserving current routeability during appeal, withdrawing authority records, changing reverse DNS, and final reissue. Those levers should not be collapsed.

This is the continuity firewall. It separates correction from destruction. If there is active hijack or immediate security harm, emergency action may be justified. If the defect is documentary, the registry can require cure while maintaining stability. If the holder loses after review, a migration window may be appropriate before final derecognition, especially where downstream users are identifiable and lawful. If a court requires status quo, the registry can preserve evidence and label the dispute rather than pretending the range is ordinary inventory.

The firewall protects the registry's legitimacy. Operators are more likely to cooperate with reviews if a documentation gap does not instantly become network death. Customers are safer if transition windows are explicit. Buyers and lenders can price risk if adverse claims have a defined path. The registry is less likely to face urgent injunctions if it has not made the remedy unnecessarily destructive.

The firewall also protects against capital-control fears. IPv4 addresses have market value. A registry that can freeze, revoke, refuse transfer, alter public authority signals and disclaim losses can move value without paying for it. That is precisely why reclamation must be bound to continuity-preserving rules. The answer is not to abandon enforcement; it is to ensure that enforcement affects only what the evidence supports at each stage.

Continuity is not permanence. It is a staging rule. Once the dispute is resolved, the registry can act. But until it is resolved, running networks should not be used as hostages in a record dispute. The public interest lies in accurate records and stable service, not in choosing one at the needless expense of the other.

Reissue design decides who captures the gain

Once a block is genuinely recovered and cleaned, the next question is distribution. Who receives the value created by reclamation? The answer affects incentives long before any particular range is released.

The simplest option is to return recovered space to AFRINIC's ordinary pool and issue it under existing soft-landing rules. This treats recovery as restoration of public stock and avoids a special market. It also has limits. Large recovered ranges may not fit neatly into small late-phase allocation rules. Some may be dirty or disputed in ways ordinary applicants did not bargain for. Some may be too valuable to release without explanation. Administrative simplicity can hide economic differences among blocks.

Waiting-list rationing is another possibility, but it should not be confused with cleanup. A queue may be a legitimate way to order claims on scarce supply. It does not solve the condition problem. If the first applicant in line receives a range with reputation damage, old authority conflicts or legal residue, the queue has allocated a burden as well as an asset. Refusal rights and condition disclosure matter.

Priced allocation or auction-like mechanisms recognise scarcity more explicitly. They can deter frivolous demand and reveal value. They also create a legitimacy hazard. If holders believe the registry can recover and monetise their ranges, every enforcement action becomes suspect. In a region where affordability and development are central concerns, auctions may favour capital-rich firms over smaller access networks. They may be efficient in price terms and corrosive in institutional terms.

Targeted reissue for public-interest uses has its own appeal: exchange points, rural connectivity, public services, critical infrastructure or new entrants. But it invites lobbying and requires criteria the registry may not be well placed to administer. The registry's comparative advantage is accurate recognition and policy implementation, not industrial planning across the whole connectivity economy.

The better answer may be plural but rule-bound. Small clean returns can enter ordinary distribution. Large recovered blocks can be segmented after technical and legal review. Impaired blocks can carry explicit condition labels and refusal rights. Special-purpose reserves should exist only where policy clearly supports them. Any priced mechanism, if used, should be separated from enforcement staff so that recovery decisions are not influenced by revenue expectations.

Most importantly, reissue should not erase the chain of events. The new recipient should know whether the block was voluntarily returned, recovered after non-payment, corrected after fraud, released after appeal, held in quarantine, subject to public dispute, or known to have operational residue. That information is not gossip. It is due-diligence infrastructure. In a scarce market, documentation is part of the asset.

A condition report should travel with recovered space

Old assets need condition reports. Real-estate buyers ask about surveys, liens, environmental risks and planning restrictions because the asset carries history. Reused IPv4 space needs a lighter version of the same discipline. The registry need not publish customer lists, confidential legal advice or sensitive security detail. It should publish enough for the next user and the market to understand the status of the block.

A condition report for recovered AFRINIC-administered space could include the prefix, prior public holder, recovery category, date review opened, notice date, cure deadline, decision date, appeal status, quarantine start and end, public dispute status, RPKI status, reverse-DNS status, known IRR-record transition, abuse-contact reset, voluntary-return flag, and material public warnings. For larger ranges, it could note recent routing visibility and known public reputation concerns. Sensitive details can be summarised, but the existence of material risk should not be hidden.

The report serves several functions. It protects the recipient from hidden defects. It protects the registry by showing that recovery was not arbitrary. It protects prior holders by creating a record of notice and review rather than relying on rumour. It helps brokers, lenders, auditors and procurement teams evaluate the range. It helps public observers distinguish between recovery from fraud, voluntary return, abandonment and unresolved dispute.

Some will argue that condition reports stigmatise recovered space. Silence is more likely to stigmatise it. If no report exists, buyers and operators assume the worst or spend heavily to reconstruct the history themselves. A standard report lets risk be priced. A block voluntarily returned by a dissolved company, quarantined for a defined interval and cleared of registry-controlled conflicts should become more attractive, not less. A block with known reputation residue can still be useful if the recipient understands the remediation burden.

Condition reports also discipline internal behaviour. Knowing that each recovered range will require a public account encourages better files, clearer classifications and more careful remedies. It makes easy cases only slightly slower if the registry is already keeping records. It makes hard cases much cheaper later because future recipients and counterparties have a reference point.

The report should travel with the allocation, not sit in an obscure meeting archive. Recipients should receive it as part of the issuance package. Public users should be able to find current status through ordinary registry interfaces or linked notices. Historical status should remain available for a reasonable period because external disputes and reputation questions often surface after reuse. This is how a reclaimed prefix becomes a bankable prefix.

Usable yield matters more than nominal recovery

Reclamation programmes are tempting to measure by address count. How many /24s were recovered? How much inventory was restored? How many dormant ranges were closed? Those numbers are easy to report and politically attractive. They are also incomplete. A recovered address is not a usable address until the next holder can rely on it at reasonable cost.

AFRINIC should therefore think in terms of quality-adjusted yield. A block under quarantine is not yet usable yield. A block subject to appeal is not ordinary inventory. A block that major counterparties reject because authority signals conflict is impaired. A block with severe reputation history may be usable only after costly remediation. A block whose recovery provokes broad litigation may have negative systemic yield even if it appears on a recovery dashboard.

The same principle applies to recipients. A large cloud platform or carrier may be able to clean a difficult range through staff, vendor relationships and operational leverage. A small ISP, community network or public-sector operator may not. If the registry gives impaired space to the applicant least able to remediate it, formal equality becomes economic inequality. The address is allocated, but the hidden cost lands on the weaker party.

Quality-adjusted yield also affects which cases are worth pursuing. Recovering a small range after years of dispute may consume more staff time, legal exposure and market confidence than the usable value recovered. Recovering a large range obtained through proven fraud may be necessary despite high cost. A mature programme should track both sides: nominal addresses recovered, usable addresses released, average quarantine time, remediation burden, disputes triggered, reversals, recipient refusal, and evidence of downstream harm.

This accounting would also improve public debate. Supporters of aggressive recovery could show real inventory gains rather than headline figures. Critics could point to specific impairment costs rather than treating all enforcement as dangerous. Members could see whether the programme actually benefits late entrants or merely produces status changes. The registry could learn which defects are common and which procedures reduce delay.

Pricing defects does not require AFRINIC to sell addresses at market prices. It requires acknowledging that defects have costs. Those costs appear as staff time, legal review, routing delay, support tickets, customer remediation, reputation cleanup, procurement hesitation and market discount. Ignoring them does not make reissue fairer. It merely hides who pays.

The aim should be a recovery balance sheet that subtracts what must be spent to make recovered ranges useful. Conservation accounting should be honest enough to distinguish a prefix recovered in name from a prefix ready for service.

Conservation without expropriation is the bargain

The strongest case for reclamation is both moral and economic: scarce public numbers should not be wasted, stolen, trapped in dead companies or held under false records while legitimate networks struggle. The strongest case against aggressive reclamation is also moral and economic: a registry should not turn a coordination function into a power to confiscate value, destroy customer continuity or rewrite commercial expectations without due process. The hard task is to hold both truths at once.

AFRINIC's history makes the task unavoidable. It serves a region with real connectivity needs, limited remaining IPv4 stock and a policy tradition built around conservation and fairness. It has also faced reported address-misappropriation allegations, a large and contested Cloud Innovation dispute, court pressure, governance stress and questions about institutional continuity. In that setting, recovery cannot be performative. It must be credible to those who want fraud removed and to those who fear arbitrary control.

The registry's proper role is narrower than some reformers may want and stronger than some market participants may prefer. It is not to guarantee every holder's business model, every lease or every speculative reserve. It is not to ignore fraud because customers exist. It is also not to become the owner of the economic destiny of address space. Its role is to maintain uniqueness, recognise legitimate control, correct invalid records and preserve service continuity where possible while doing so.

The practical sequence should be clear. Identify the candidate block. Classify the reason for concern. Build the evidence file. Notify the right parties. Allow cure. Decide with reasons. Provide review. Protect running networks where the evidence allows. Recover only what the evidence supports. Quarantine the range. Clean registry-controlled authority signals. Document external residue. Publish a condition report. Reissue by rule. Preserve history. Measure usable yield. Review mistakes.

If AFRINIC follows that sequence, reclamation can increase trust. It will show that the registry can remove fraud and abandonment without turning scarcity into discretionary power. It will give new entrants a chance at recovered addresses while giving existing holders confidence that their registrations are not revocable by mood. It will help markets price AFRINIC-administered space more accurately because buyers and lenders can distinguish bounded recovery from capital control.

If AFRINIC skips those steps, reclamation will have the opposite effect. Holders will route defensively, litigate earlier, hide customer data, resist voluntary returns, discount transfers, demand indemnities and treat every review as a threat. Recovered inventory may increase on paper while confidence falls in practice. The ledger will appear stronger because it has taken addresses back, yet the market will treat it as weaker because the conditions of taking are unclear.

IPv4 reuse is attractive because the cupboard is nearly bare. But the cupboard contains old files, not blank goods. A registry that understands memory can recover value. A registry that ignores memory merely moves risk from one account to another. The economics of reclamation and reuse begin with scarcity, but they end with legitimacy: a new reliance can safely begin only when the old reliance has been fairly closed.