AFRINIC's transfer politics are usually argued in the language of stewardship, regional fairness and institutional survival. That language is not empty. Africa's registry was created late, holds a small share of the global IPv4 stock, and serves networks that still need IPv4 capacity while the world continues its long, uneven migration to IPv6. But the language is incomplete. Once IPv4 became scarce, priced, leased, financed and litigated, a rule about whether addresses may move between regional registries stopped being an administrative detail. It became a rule about capital mobility.
That is the conflict at the centre of inter-RIR transfer politics. Regional Internet Registries were built to preserve uniqueness and maintain records in a world where addresses were treated as low-price technical inputs. They now sit above scarce operating assets whose value is determined in a global market. The registry still has a regional service territory; the asset does not acquire a regional economic nature merely because the database is regional. Packets do not carry passports. Customers, cloud workloads, hosting demand, reputation systems, financing arrangements, abuse risk and network operators all move through a global internet.
A transfer rule that blocks exit from one registry does not create more African connectivity. It changes who may monetise, buy, finance, lease, import or escape the address space recorded there. In institutional-economics terms, it changes the holder's exit right and the registry's hold-up power. It may be justified as conservation. It operates as a price rule.
Lu Heng's public notes make the sharpest market-side version of that argument. They should be read with attribution, because Lu is not a detached observer: he is the CEO of Cloud Innovation and Larus, and his companies are parties in the long AFRINIC dispute. Yet his notes ask the right questions for this case: what does a registry actually need to control; who benefits from transfer friction; who bears the cost when a registry is wrong; and why should a mailing-list consensus or board decision reduce the mobility of assets already embedded in live networks? Official RIR, NRO, AFRINIC and ICANN materials are useful factual exhibits for dates, policy text and institutional posture. They are not enough to settle the economics of exit, liquidity and registry self-interest.
Scarcity made geography a price rule
The economics begins with a simple fact. IPv4 did not become scarce because AFRINIC, ARIN, RIPE NCC, APNIC or LACNIC invented scarcity. IPv4 has a finite address space. The global allocation history then distributed that finite stock unevenly. The Internet Governance Project's 2021 analysis of the AFRINIC crisis noted that AFRINIC came late to the registry system and held only a small share of the global IPv4 pool, while much larger portions had already been distributed before the modern regional-registry structure existed. That history matters because a regional conservation argument in Africa starts from a real asymmetry. African networks entered the scarcity era with less legacy stock than many networks elsewhere.
AFRINIC's own exhaustion materials establish the administrative timeline. The registry says it has managed internet number resources since 2005; that the African community supported a Soft Landing policy in 2011; that Phase 1 of exhaustion began on March 31st 2017; and that Phase 2 began on January 13th 2020. In Phase 2, the minimum allocation or assignment size is /24 and the maximum is /22. Those are rationing numbers. They show that the old allocation model cannot satisfy ordinary expansion demand, let alone the address appetite of hosting, mobile access, cloud migration, enterprise continuity and transitional IPv4 services.
Scarcity then has a second life. Once a registry cannot issue enough new supply, the existing stock becomes a market. IGP reported that IPv4 transfer-market prices rose from roughly $8 per address in 2017 to about $30 by 2021, using a /16 block to illustrate the multi-million-dollar stakes. The precise price varies with timing, reputation, block size, routing history and closing risk. The institutional point does not vary. A registry fee schedule that charges service fees by category is not the price of the resource. The market price sits outside the registry. The approval gate remains inside it.
This gap creates the political economy of inter-RIR transfers. If an address can move to the buyer that values it most, regional scarcity becomes one factor among many: local networks can buy, sellers can sell, imported supply can enter, idle stock can surface, and price signals can do some allocation work once the free pool is exhausted. If an address cannot leave a registry region, the resource acquires a policy discount. A buyer in another region values it less. A financier lends less against it. A holder has a narrower exit. A potential inbound seller worries that the asset may be trapped after import. A registry may say it is protecting the region, but the immediate effect is to reduce the value and liquidity of the assets already held by networks in that region.
That is why geography becomes a price rule. The question is not mainly where packets are routed on a given afternoon. It is whether a block recorded in one registry can be sold, transferred or recognised in another tomorrow. A rule that says "regional" for registry purposes can be technically invisible and economically decisive at the same time. If it controls exit, it controls value.
AFRINIC's transfer politics are therefore not a morality play about hoarding versus development. They are a design problem in a scarcity institution. A region with too little IPv4 can try to ration what remains. It can also try to make itself a trustworthy place into which global IPv4 capital can flow. The first strategy relies on discretion. The second relies on liquidity and confidence. Inter-RIR transfer rules reveal which strategy a registry actually prefers.
Inter-RIR transfer is an exit right
The phrase "inter-RIR transfer" sounds technical, almost clerical. It is not. It is the right to move a recognised resource from one regional registry environment to another. That makes it a form of exit. In ordinary markets, exit disciplines institutions. A customer who can leave a bad bank, broker, depository or registrar has bargaining power. The institution must maintain enough trust to keep business. A customer who cannot leave is captive, and the institution's incentives change.
Lu's note on number-resource portability and the ICP-2 revision frames this as a structural weakness of the registry system. In his view, a network should have an unconditional right to move IP addresses or ASNs from one RIR to another, particularly when a registry suffers governance breakdown or operational failure. The analogy he draws to domain-name transfers is imperfect; domain registries and number registries differ in history, contract and technical function. The economic intuition remains strong. A registry that can lose members has to compete on neutrality, competence and predictability. A registry that cannot lose members can describe dependence as community membership.
AFRINIC's recent history makes the exit question concrete. The NRO's 2023 statement on the appointment of an official receiver said the Supreme Court of Mauritius had appointed a receiver to preserve AFRINIC's status, oversee elections, facilitate a board and appoint a chief executive. The Register later reported that AFRINIC had been unable to elect a board or perform some functions for years; that a June 2025 election was suspended and annulled after allegations around voter authorisation; and that a later 2025 election produced a board amid continuing legal and political controversy. ICANN intervened in 2026 in a winding-up matter to argue that numbering resources administered through AFRINIC were not corporate assets available for distribution.
None of those facts proves that every AFRINIC transfer decision is invalid. They prove something narrower and more important: registry continuity cannot depend on blind faith in one corporate shell. If the registry's governance fails, the records must continue. If the board is contested, live networks must continue. If litigation constrains management or freezes accounts, reverse DNS, RPKI, WHOIS, RDAP and transfer records should not become hostages. The more a registry resists exit, the more its own weakness becomes a systemic risk for the holders recorded in it.
This is where inter-RIR transfer politics differ from ordinary allocation policy. Allocation policy asks who should receive resources from a remaining pool. Exit policy asks whether a holder can escape an institutional environment without losing recognition. In a low-value world, that distinction was easy to blur. In a high-value world, it is decisive. A holder whose block is worth millions does not merely want an accurate database entry. It wants assurance that the registry cannot strand the asset inside a troubled jurisdiction, a contested board, a hostile policy environment or a discretionary review process.
AFRINIC's defenders can reasonably say that a region must protect its remaining resources from being drained by global demand. The harder reply is that exit restrictions also tell global supply not to enter. A one-way market is not a market. If inbound resources are welcomed but outbound freedom is denied, counterparties will treat AFRINIC as an asset trap. That is not a slogan. It is how capital prices jurisdictional risk.
The approval gate carries allocation logic into the market era
AFRINIC's Consolidated Policy Manual gives the older formal architecture. Section 5.7 covers IPv4 resource transfers within the AFRINIC region. It says the policy applies to an organisation with justified need for IPv4 resources that cannot be satisfied by AFRINIC. The source must be an existing AFRINIC member account or a legacy resource holder in the AFRINIC service region. The source must be recognised by AFRINIC and not involved in a dispute over the status of the resources. It cannot receive more IPv4 resources from AFRINIC for twelve months after approval, and it must not have received a transfer, allocation or assignment in the previous twelve months except in merger or acquisition cases.
The recipient side is even more revealing. AFRINIC must approve the recipient's need. The recipient must justify initial or additional usage according to policies in force, must be an AFRINIC member, must be subject to current policies and must sign the Registration Services Agreement for the resources received. Transferred legacy IPv4 resources no longer retain legacy status once transferred. This is not a mere recording system. It is a rationing system adapted to a secondary market.
Some controls are plainly defensible. A registry must verify the source. It must avoid duplicate claims. It must prevent a disputed block from moving into a new account before the dispute is resolved. It must ensure records, contacts and agreements are accurate. It must protect the database from fraud. A market without provenance is not liquid; it is dangerous.
The problem is that AFRINIC's model imports the old allocation question into the transfer transaction. The buyer must show need to the registry even after price, risk and commercial commitment have revealed need in the market. That creates the central asymmetry of modern IPv4 governance: the buyer bears the capital cost, the seller bears the opportunity cost, customers bear the continuity risk, but the registry retains a discretionary approval function while carrying limited commercial downside.
Need review had a plausible role when addresses were allocated from an unpriced pool. It prevented waste by applicants who could obtain scarce resources at administrative cost. After exhaustion, its purpose changes. It no longer decides how to distribute free-pool supply. It decides whether a privately negotiated movement of an already held resource will be recognised. The same word, "need", moves from conservation to permission.
That migration matters because the registry does not know the buyer's future as well as the buyer does. A data-centre operator may need capacity for customers not yet signed. A cloud platform may need reputation-segmented space. A mobile operator may need IPv4 for translation infrastructure while deploying IPv6. A buyer can be wrong, but a registry is not automatically better at forecasting demand. When the registry rejects or delays, the market loses the transaction even if the buyer was willing to bear the risk.
This is why Lu's "Policy Mirror" note draws a hard line between free-pool allocation and market transfer. It argues that need-based criteria may belong to unallocated pool issuance, but should not govern transfers, leases, financing, merger-related movement or changes of operational control involving already allocated resources. A more cautious version would allow objective recipient checks only where they protect registry invariants: uniqueness, accurate identity, fraud prevention, contactability, security continuity and dispute status. It would not ask whether the buyer's business model satisfies a regional allocator's moral view of proper use.
The distinction is not academic. If recipient approval remains broad, inter-RIR transfers become a political act. If approval becomes objective recording, inter-RIR transfers become ordinary market plumbing. AFRINIC's dispute sits exactly on that line.
Liquidity is not a luxury for the rich
Opponents of open transferability often treat liquidity as a rich-holder concern. That is a mistake. Liquidity is the mechanism by which scarce assets move from lower-valued to higher-valued uses and by which holders can borrow, sell, lease, insure, restructure or exit. In an address market, liquidity also improves discovery of dormant or underused resources. The less liquid the market, the more valuable it is to people with procedural knowledge, legal stamina and informal access to counterparties. The more liquid the market, the more useful it becomes to ordinary networks that need a transparent way to acquire capacity.
The IGP analysis of the 2021 dispute captured part of this point when it described the gap between global IPv4 market value and the lower administrative fees at which AFRINIC members had historically received resources. That gap created arbitrage opportunities, but the existence of arbitrage is not an argument against all market movement. It is evidence that rationing and price were disconnected. If a region has scarce stock distributed under old rules, the policy task is to make movement clean and accountable, not to pretend movement is illegitimate.
Lu's public notes push the argument further. In his essay on enforcement creep, he argues that audits, frozen transfers, retroactive review and opaque compliance standards suppress IPv4 liquidity by replacing price discovery with permission-seeking. The language is combative, but the mechanism is familiar. When approvals are discretionary and sanctions are severe, rational parties trade less, hold defensively, demand discounts, use lawyers instead of simple registry forms, or transact outside the cleanest record path. The registry then sees opacity and asks for more control, which deepens the cycle.
AFRINIC's policy environment is especially exposed because its public materials still contain allocation-era concepts. The policy manual describes conservation, need, non-portability of some PA space, assignments tied to documented purposes, and the idea that changed purpose can affect validity. Those ideas emerged from an engineering and conservation culture. In a market era, each one can become a liquidity tax.
The regional-development argument also has to confront a harder fact: illiquid assets are worse collateral. An African ISP that holds addresses but cannot sell them globally, pledge them clearly, lease them openly, or move them away from a troubled registry has less financing capacity than a comparable holder whose block is globally transferable. That means the local holder bears the cost of the regional restriction. A rule advertised as keeping resources in Africa can reduce the capital value of African resources.
This does not mean that every address should be flipped freely, or that fraud should be tolerated. Markets need integrity. But integrity and illiquidity are not the same. A clean registry should make it safer to trade, not harder. It should publish status, verify control, record disputes, preserve history and process transfers with predictable service levels. If it does those things well, addresses with AFRINIC provenance become more valuable, not less.
Liquidity also affects buyers. A buyer in the AFRINIC region benefits if global sellers know that an imported block can later leave. That reassurance lowers the price of entry. If the buyer later outgrows, restructures or sells the business, the resource remains useful capital rather than a stranded local entitlement. Conversely, if inbound resources lose legacy status, become subject to new restrictions, or face uncertain outbound rules, rational sellers price the trap. The region pays for that trap even when the policy was designed to protect it.
The market lesson is blunt. You cannot make a region richer by making its assets less mobile. You can make a registry more powerful that way. Those are different objectives.
Legacy status is the hinge between old and new markets
Legacy resources are politically sensitive because they predate or sit uneasily beside current registry contracts. They often carry expectations formed before today's scarcity regime. In AFRINIC's older in-region transfer policy, transferred IPv4 legacy resources no longer remain legacy after transfer. That clause looks technical, but it is economically large. It says that mobility converts status.
Status conversion has a market price. If a legacy holder can sell only by causing the buyer to accept a more restrictive policy environment, the buyer discounts the block. If the holder keeps the resource, it preserves status but loses liquidity. If a resource is imported into AFRINIC and then loses exit freedom, global sellers may avoid the region. A rule that seems to harmonise records can therefore reduce supply.
Lu's "Policy Mirror" note treats this as an import trap. It argues that inbound resources should retain their pre-transfer status unless the holder expressly opts into a different regime, and that acceptance of registry-publication services should not by itself surrender transferability or reliance interests. That is the maximal market position. The narrower policy point is that a region seeking IPv4 inflow should not make inflow look like a one-way conversion into lower liquidity.
AFRINIC and similar institutions have a reply. A registry cannot maintain multiple status regimes without administrative cost. Legacy claims can be unclear. Some dormant blocks have been abused or hijacked. KrebsOnSecurity's 2019 reporting on alleged AFRINIC-related address misappropriation showed why historical provenance matters. Ron Guilmette's investigation, as reported by Krebs and South African media, alleged that dormant or defunct-company address records had been manipulated and that the market value of affected IPs exceeded $50m. AFRINIC's then chief executive said the organisation was investigating; the individuals named in the report denied or did not respond in the ways reported. Those are allegations and reported responses, not a verdict on all legacy holders.
The answer to bad provenance is better provenance, not blanket status destruction. A registry can maintain historical status, record transfer chains, flag disputes, require proof of control, publish audit trails and refer competing claims to courts or independent review. It can distinguish a clean legacy transfer from a suspected hijack. It can reject fraudulent documents. What it should not do is use the existence of fraud risk to justify converting every legacy movement into a surrender of market freedom.
ICANN's 2026 intervention in AFRINIC's winding-up dispute made a narrow and necessary point, as reported by The Register: numbering resources administered through AFRINIC should not be treated as AFRINIC assets available for corporate distribution. That protects the ledger from liquidation logic. It does not follow that AFRINIC owns the economic destiny of the resources or that holders have no reliance interests. The middle ground is asset-neutral. The registry is not the owner. The holder is not a sovereign. The record should preserve recognised control, legal claims, transfer history and operational continuity while courts and contracts handle disputes beyond the registry's competence.
Legacy status is therefore the hinge between the old administrative world and the new market world. If designed well, old resources can move into productive use with cleaner records. If designed badly, legacy becomes either a loophole for abuse or a trap that discourages movement.
Transfer politics distributes bargaining power
Transfer politics are not neutral between buyers and sellers. A seller wants certainty that it can dispose of a recognised block. A buyer wants certainty that the acquired block will be recorded and usable. The registry wants to preserve its policy authority, member base, fee model and public legitimacy. These interests can coexist under objective rules. They collide under broad discretion.
AFRINIC's older in-region transfer policy imposes constraints on both sides. The source cannot be in dispute, must not have recently received resources and must wait before receiving more. The recipient must justify need and sign into the current policy environment. Those rules are meant to prevent churn and speculation, but they create different burdens. The seller's burden is timing and clean status. The buyer's burden is need, membership, documentation, future policy exposure and acceptance of AFRINIC's rulebook. In an inter-RIR setting, the buyer may also face uncertainty over whether the resource can cross regional borders at all.
This asymmetry changes bargaining. A seller with a globally transferable block can negotiate with buyers everywhere. A seller with a block restricted to a smaller buyer pool must accept a lower price or wait longer. A buyer facing registry need review must condition the deal on approval, increasing legal complexity. A broker must price closing risk. A lender must ask whether the transfer can be unwound or delayed. The registry may not charge a large transfer fee, but the market pays a fee in uncertainty.
The political incentives are equally asymmetrical. Existing holders may support rules that limit new supply to competitors. Buyers inside the region may support lock-in if they hope it lowers local prices, even though it may reduce inbound supply. Registry insiders may prefer rules that keep members and resources within the institution. Policy regulars may present this as stewardship. Large holders may present open transferability as freedom. Neither side is disinterested.
The Register's March 2026 article showed the adversarial claims in public form. AFRINIC accused Cloud Innovation, Larus and associated campaigns of litigation and procedural roadblocks that were obstructing recovery and community initiatives. Lu replied that the structural issue was a registry model concentrating high-consequence power over economically critical number resources while disconnecting that power from matching liability. NRS public messaging framed the same conflict as money, records and votes. These are claims by interested parties. They should not be treated as adjudicated facts. They do, however, expose the incentives. AFRINIC experiences litigation as institutional paralysis. Resource holders experience discretionary control as existential risk. Both reactions are predictable.
A well-designed transfer regime would reduce the reason to fight over the institution. If the registry's transfer role is narrow, predictable and reviewable, there is less payoff from controlling the board, the mailing list or the interpretation of regional policy. If the registry's role is broad, every election, bylaw review and committee appointment becomes a battle over capital mobility. In that sense, inter-RIR transfer politics are not only a symptom of governance crisis. They are also a cause of it. The broader the gate, the higher the value of gatekeeping.
This is why "buyer need" is not a harmless phrase. It locates economic discretion inside the registry. It lets the registry decide whether capital already committed by a buyer counts as real demand. It allows opponents to dress market exclusion as conservation. It allows supporters to dress institutional self-interest as community protection. A neutral system would ask narrower questions: is the seller recognised; is the buyer real; is the resource unique; is there a dispute; can records and security services transition; and is there a defined appeal if the answer is no? Once those questions are satisfied, the remaining issue is commercial. The registry may dislike the price, the motive, the lease, the geography or the buyer's business plan. Dislike is not a registry invariant.
Mailing-list consensus is not an economic mandate
The inter-RIR transfer fight does not happen only in court or in boardrooms. It begins in policy forums. AFRINIC's policy manual describes a bottom-up Policy Development Process built on openness, transparency and fairness. Anyone may participate in the working group. Proposals are posted to the Resource Policy Discussion mailing list, discussed, brought to a public policy meeting, judged for rough consensus, sent to last call and then recommended for board ratification if consensus is found. This is the familiar internet-governance production line.
The problem is not that mailing lists exist. They are useful: they make text public, preserve objections, expose technical problems and create memory. The problem is that low-cost participation can produce high-consequence authority. A small number of policy regulars can shape language that later determines whether million-dollar resources can move between registries. Silence can be read as absence of objection. Repeated participation can be mistaken for representation. Procedure can become a substitute for consent.
For inter-RIR transfer politics, the issue is acute because transfer policy affects existing resource holders, not only future applicants. It can reduce liquidity, change exit rights, convert legacy status, alter bargaining power and decide whether a resource can be sold to a global buyer. That is economic governance. An open list may inform such governance. It should not be enough to impose it retroactively on holders who did not meaningfully participate or who were represented only by a thin set of process actors.
Lu's public notes repeatedly attack the idea that "community" can become a legal principal. In "Running-Code Betrayal", he argues that rough consensus and running code were supposed to discipline institutions in favour of operational reality, not give a small policy room authority over live networks. In "Policy Mirror", he proposes a scope test: policy should protect uniqueness, accuracy, fraud prevention, contactability, routing-adjacent coordination, security integrity, transfer recording, dispute-state recording and operational continuity. Rules whose primary effect is to regulate pricing, leasing, customer geography, capital movement or commercial morality should not be mandatory without the consent of affected principals or public-law authority.
That is a strong claim, but it captures the institutional problem. A transfer policy can be written as if it were about technical order while doing the work of capital control. The mailing list may discuss it in the vocabulary of stewardship and fairness. The economic effect may be to reduce the exit value of existing resources. That gap is where legitimacy leaks.
The appeal structure can also favour insiders. AFRINIC's manual allows appeals against chair action, but the appeal must be supported by three persons from the working group who participated in discussions and must be filed within a defined period. That may be a sensible filter against frivolous process challenges. It is also a barrier for an operator that notices the economic impact late, after the thread has become difficult to reopen. The people most affected by the rule may be least able to use the process at the moment the process says it matters.
In a scarcity market, this creates a dangerous equilibrium. The registry invokes community process. Market actors invoke operational reliance. Policy regulars invoke openness. Absent operators pay the cost. Courts then inherit disputes that could have been avoided if the policy process had required direct notice, economic-impact analysis, minority reports and independent review for transfer rules affecting existing assets.
Transfer politics therefore cannot be fixed only by better text in the policy manual. They require a better political economy of text production. A rule that controls exit should not be adopted as if it were a housekeeping edit.
Registry self-interest is not the same as stewardship
Every institution has self-interest. A nonprofit registry has staff, revenue, status, legal exposure, peer relationships and a desire to survive. That is not scandalous. It becomes a governance problem when institutional self-interest is presented as if it were identical to stewardship.
Inter-RIR transfers are especially revealing because outgoing transfers can reduce a registry's member base, fee base and policy reach. A resource that moves to another RIR no longer sits inside the same annual-fee and compliance environment. A registry that restricts outbound movement can say it is protecting regional development. It is also protecting its own perimeter. Those motives may overlap. They are not identical.
Lu's "Running-Code Betrayal" note highlights a sentence he says appeared in the financial assessment of the later AFRINIC transfer framework: because AFRINIC-pool resources can only be transferred in-region, AFRINIC would not lose current resource members through outgoing transfers. That short sentence matters because it makes visible a registry interest normally hidden behind public-interest language. If a rule preserves members for the registry, the rule should be evaluated as both policy and institutional self-protection.
The point is not that AFRINIC is uniquely self-interested. All RIRs face similar incentives. A registry with no member exit is structurally tempted to expand its remit. It can justify fees, programmes, meetings, enforcement, audits and policy machinery because members need the ledger. The broader the registry's control, the more indispensable it appears. The more indispensable it appears, the easier it is to justify broader control. That is how a service layer becomes a political layer without a single dramatic decision.
AFRINIC's fee schedule shows the ordinary, legitimate side of the revenue model. Annual membership categories are tied to aggregate billable resources, and transfer parties must be in good standing. The registry needs money to operate. It must maintain staff, systems, security, support, reverse DNS, RPKI, WHOIS, RDAP and meetings. But fees tied to non-portable registry recognition can start to feel like a tax on continuity. If holders cannot leave, the discipline on fees is weaker. If holders can leave, the registry must price and behave like an accountable service bureau.
This is why inter-RIR portability is an accountability tool. It does not require abolishing AFRINIC. It requires making AFRINIC good enough that members stay because it is reliable, not because exit is blocked. A registry that fears transfer-out may respond by improving service, lowering transaction cost, strengthening data quality and publishing clear metrics. A registry that can prohibit transfer-out may respond by defending the prohibition.
The same logic applies to governance. If board control affects transfer mobility, board elections become economically charged. The Register's 2025 reporting after the election of eight directors said AFRINIC had a chance to convene a board for the first time since 2022, but also faced critics, court risk, Smart Africa-related concerns, a government-ordered investigation and ongoing dispute with Cloud Innovation. In such a climate, a transfer policy ratified by a newly restored or contested board will not be read as neutral administration. It will be read as a move in the distribution of power.
The remedy is not to assume bad faith. It is to design rules that survive bad incentives. A registry should publish why an outbound transfer was denied, under which objective criterion, within what service period, subject to what appeal and with what continuity protection. It should publish aggregate transfer statistics and reasons for rejection. It should separate fee recovery from policy ambition. It should let members see when stewardship is really recordkeeping and when it is institutional self-preservation.
Stewardship is credible when it restrains power. When it expands power and prevents exit, it needs a different name.
"Africa first" can become a poverty penalty
The strongest argument for restricting outbound AFRINIC transfers is regional justice. Africa received a small share of global IPv4. Demand continues. The remaining pool is limited. If unrestricted transfers let global buyers acquire African-issued resources, local networks may be priced out while richer markets gain supply.
The problem is that transfer restriction is a blunt and often self-defeating answer. It does not create more addresses. It changes where holders may sell. If the rule lowers the exit value of AFRINIC resources, it reduces wealth held by African networks. If it discourages inbound transfers, it reduces future supply. If it drives leasing into opaque structures, it weakens accountability. If it gives the registry broad discretion, it raises the value of political influence. If it creates legal uncertainty, it raises the cost of every transaction.
Lu's "Poverty Penalty" and "Policy Mirror" notes make this argument in forceful terms: a region is not enriched when its assets are trapped, discounted and made harder to finance. One need not accept every phrase to see the economics. Capital controls often claim to protect local interests. They can also produce lower asset values, thinner markets and more insider advantage. The people with lawyers and cross-border structures find workarounds. The small operator experiences the rule as friction.
A better Africa-first strategy would focus on lowering transaction costs for legitimate African buyers. That could mean transparent transfer records, standard contracts, financing support, clean leasing records, objective anti-fraud checks, reliable abuse-contact data, open market information and registry service levels. It could mean encouraging global holders to import resources into the AFRINIC region by guaranteeing outbound freedom later. It could mean making the region a trusted place for address capital, not a jurisdiction where address capital fears entrapment.
The IGP article made a related point from a different angle. It argued that the future of Africa's internet development would not be greatly affected by reserving IPv4 addresses to regional use, because the remaining AFRINIC stock was too small to sustain the continent's future growth. African connectivity will need IPv6 and, during transition, access to the global IPv4 market. The conclusion is uncomfortable for conservation politics: the region's problem is not solved by holding a small remaining pool tighter. It is solved by making the region's networks more capable of obtaining, financing and using addresses from anywhere.
This is not a call to ignore abuse. A global market can attract speculation, hijacking, dirty reputation, shell entities and unequal bargaining power. Krebs's reporting on alleged African address misappropriation is a reminder that high market value creates incentives for fraud. But the response to fraud should be targeted: verify control, identify real organisations, maintain transfer history, freeze disputed claims and create independent review. A ban on exit punishes clean holders as well as suspected bad ones.
Nor is it a call to treat IPv6 as irrelevant. IPv6 deployment remains essential. AFRINIC's own public materials and The Register's 2026 APRICOT report show registry voices wanting the remaining IPv4 conversation to end so attention can move to IPv6. The difficulty is that IPv6 does not remove the near-term value of IPv4. Dual stack, translation, hosting compatibility, customer expectations, abuse systems and legacy applications keep IPv4 commercially relevant. A policy that assumes moral victory for IPv6 before operational reality has arrived will misprice the transition.
"Africa first" should therefore mean African network resilience, not registry territorialism. If a rule makes African-held resources less valuable, discourages inbound supply and increases discretion over local operators, its development rhetoric should be questioned. A region can protect itself by owning more options. Lock-in is the opposite of options.
Litigation made the transfer rule a credibility test
AFRINIC's transfer politics cannot be separated from litigation, but litigation should not be allowed to obscure the underlying economics. The Cloud Innovation dispute began with contested claims about use, regional connection and the scope of AFRINIC's Registration Service Agreement. IGP's 2021 account said AFRINIC overreached in its attempt to reclaim resources and that Cloud Innovation's legal response was also excessive. The same article stressed that whether AFRINIC misread its policies or Cloud Innovation violated contracts was for the court to decide. That caution remains useful.
The public record since then has become more complicated. The NRO welcomed receivership as a route to restore functional governance. The Register reported election suspension, annulment, later board formation, ICANN letters, winding-up intervention, takedown-order litigation around Larus and Cloud Innovation statements, and continuing fights over bylaws and member status. AFRINIC has accused its antagonists of trying to paralyse it. Cloud Innovation, Larus and NRS have denied or reframed key allegations and have argued that the real problem is concentrated registry power. Courts have not turned all of those claims into settled conclusions.
For transfer economics, the key point is credibility. A transfer market needs a registry whose decisions are boring. AFRINIC's decisions are not currently boring. A buyer, seller or financier looking at an AFRINIC-origin block must ask whether a transfer policy is settled, whether board authority will be challenged, whether a court order affects the resource, whether a registry communique signals risk, whether a public campaign may affect member votes, and whether ICANN or a revised RIR lifecycle framework could change the environment. That is a large uncertainty premium.
The uncertainty affects both sides of the dispute. AFRINIC needs to restore normal service and escape a pattern in which every institutional act becomes litigation. Resource holders need assurance that recovery will not become an opportunity to harden registry control over transferability, leasing and exit. If AFRINIC uses recovery to narrow outbound movement, holders will see institutional repair as a threat. If holders use litigation to block ordinary registry functions, AFRINIC and its supporters will see market freedom as institutional sabotage. The cycle feeds itself.
This is why objective transfer architecture matters more after crisis, not less. During stable periods, ambiguous rules can survive because trust fills the gaps. During contested periods, every gap becomes a weapon. A transfer denial must be tied to an enumerated reason. A resource hold must distinguish fraud, duplicate claim, payment issue, court order, documentation defect and policy dispute. A completed transfer should have finality absent fraud or adjudication. An appeal should be fast enough to matter commercially. RPKI and reverse DNS should not become leverage in non-technical fights.
ICANN's intervention in the winding-up matter illustrates the delicate line. Its stated concern, as reported by The Register, was to help the court understand AFRINIC's unique role and the nature of numbering resources, including that those resources are not AFRINIC assets available for distribution. That is a valid continuity concern. But continuity of the registry function is not the same as endorsement of every transfer restriction. Protecting the ledger from liquidation should not become protecting the gatekeeper from accountability.
The same distinction should guide courts. A court does not need to decide internet industrial policy to preserve service continuity. It can require status preservation, record integrity, board repair and clear authority while leaving broader market design to a legitimate and bounded process. Conversely, if a registry threatens to impair live resources on contestable policy grounds, courts may be the only available check. Litigation is not healthy as a routine governance tool. It becomes inevitable when the registry's own review architecture is weak.
AFRINIC's transfer rule has thus become a credibility test. If the registry can show that transfer decisions protect record integrity rather than institutional control, it can rebuild trust. If transfer politics remain a proxy for litigation strategy, every block recorded in AFRINIC will carry the price of that doubt.
A neutral transfer architecture is possible
The choice is not between a borderless free-for-all and a regional fortress. A neutral inter-RIR transfer architecture can protect the registry's legitimate functions while allowing asset mobility. The first principle is source certainty. The registry should verify that the transferor is the recognised holder or lawful representative, that the resource is uniquely identified, that no duplicate claim, fraud hold, court order or independent adjudicative hold blocks the record change, and that the transfer instrument is real. This is registry work.
The second principle is recipient reality, not recipient morality. The transferee should be identifiable, contactable and capable of maintaining accurate registry records. It should accept defined service obligations for record publication, abuse contact, reverse DNS, RPKI and dispute handling. It should not have to persuade the registry that its price, geography, leasing intention, customer base or business model is virtuous. If a separate law prohibits a transaction, that law belongs to the competent authority. The registry should not invent a commercial planning jurisdiction.
The third principle is status preservation. Legacy or inbound resources should not lose economically meaningful status merely because the registry records a transfer, unless the holder expressly agrees or a clear legal rule requires it. Status can be described without creating unlimited property. The record can say what the registry recognises, what it does not adjudicate, and what legal or contractual claims remain outside its competence. The key is to avoid using transfer as a forced conversion event.
The fourth principle is finality with narrow reopeners. Once a transfer is recorded, the recipient must be able to rely on it. Reopeners should be limited to fraud, material misrepresentation, duplicate claim, binding court order, explicit abandonment, security-integrity emergency or defined non-payment after notice and review. Later disagreement with commercial use should not unsettle the record unless the conduct violates a clear rule within the registry's scope.
The fifth principle is dispute isolation and independent review. If one resource is contested, unrelated resources should continue. If one contact object fails validation, the block should not disappear. Transfer denial, resource hold, status conversion, revocation threat, RPKI impairment and reverse-DNS removal should not depend only on the same institution whose interpretation is contested. Review needs competence, speed, published standards, interim continuity and authority to order record correction.
The sixth principle is transparency. AFRINIC can publish aggregate transfer statistics without exposing sensitive transaction data: received, approved, rejected, pending, average processing time, reasons for rejection, court holds, fraud holds and appeal outcomes. That information would show whether policy is protecting integrity or suppressing mobility, and it would reduce the rumour premium around AFRINIC-origin resources.
This architecture would not prevent all conflict. It would lower the value of conflict. If transfers are objective, board control becomes less economically decisive. If status is preserved, inbound supply becomes less afraid. If review is independent, courts become less necessary. If records are clear, buyers and sellers pay less for legal uncertainty.
Most importantly, such an architecture would be compatible with African development. It would not require AFRINIC to celebrate speculation or ignore misuse. It would require AFRINIC to distinguish misuse from mobility. Fraud is a registry problem. Duplicate claims are a registry problem. Inaccurate contacts are a registry problem. A lawful holder monetising a scarce asset through transfer or lease is an economic fact. The registry can record that fact cleanly or drive it into more obscure arrangements.
The safest database is the one the market wants to use. That requires trust, not territorial walls.
The next watchpoints
The first watchpoint is implementation of the later transfer framework reported in 2026. The Register has described AFRINIC as having adopted a policy that, in many circumstances, prevents members from transferring IPv4 assets assigned by AFRINIC outside the region. Lu's notes read the framework as classifying resources by origin and making AFRINIC-pool resources effectively in-region for transfer purposes, while treating legacy and inbound transferred resources differently. Watch actual processing behaviour: what requests are accepted, which are rejected, what reasons are given, and whether rejection rests on objective record risk or regional retention.
The second watchpoint is the treatment of inbound and legacy resources. A region short of IPv4 should want imported supply. If inbound blocks lose status, become hard to export, or face uncertain policy conversion, the import channel will shrink. The old policy's rule that transferred legacy IPv4 resources cease to be legacy should be evaluated against market behaviour: does it clean records, or does it discourage legitimate transfers?
The third watchpoint is litigation spillover. The Cloud Innovation, Larus and NRS disputes should be followed through court orders rather than press claims alone. AFRINIC's allegations of paralysis, Larus's denials about court endorsement of leasing, and NRS's claims about member risk are public positions. The market-relevant question is what courts decide, what remains interim, and whether registry services continue without collateral damage.
The fourth watchpoint is member governance. Transfer policy adopted after a contested governance crisis needs stronger legitimacy than ordinary policy. Watch whether AFRINIC publishes board reasoning, conflict disclosures, membership verification, bylaw clarity and independent review mechanisms. If member rights remain uncertain under Mauritian company law or bylaws, transfer decisions will inherit that uncertainty.
The fifth watchpoint is the revised global RIR lifecycle framework. A mechanism for assisting or derecognising a failing RIR may protect continuity. It may also create a higher-level gatekeeper if triggers are vague. The useful version protects records, escrow, security services and failover. The dangerous version lets the RIR club or ICANN steer regional economic policy under the label of stability.
The sixth watchpoint is leasing recognition. Leasing will not disappear because policy disapproves of it. If AFRINIC offers a clean way to record delegated operational contacts, abuse responsibility and time-limited use without treating leasing as an admission of misuse, the database becomes more accurate. If leasing is treated as evidence of violation, the database will learn less about the market it is supposed to describe.
Finally, watch language. Stewardship should name the invariant being protected. Ownership should identify the legal interest claimed. Community should disclose who participated and who pays. Continuity should mean the ledger, not the gatekeeper.
AFRINIC can still be a strong registry. A strong registry is not one that traps resources. It is one whose records are accurate enough, portable enough, reviewable enough and trusted enough that markets prefer the formal path. Regional stewardship and global asset mobility do not have to be enemies. They become enemies when stewardship is used to close the exit.