A regional ISP preparing to expand at an internet exchange can make the engineering look deceptively simple. The cross-connect is ordered. Two upstreams are willing to quote transit. A route server is available. The network team has an ASN, a traffic forecast, a migration plan and customers tired of paying for a single carrier's congestion. In the rack, the problem is cables, optics, BGP sessions and maintenance windows. In the commercial room, it becomes a test of credibility. The peering coordinator asks for route objects. The upstream asks who is authorised to announce the prefix. The IXP asks for valid contact information and filtering hygiene. An enterprise customer asks whether its allowlists, reverse DNS, geolocation and abuse handling will survive migration.
The network can carry traffic before the market is satisfied that it should carry that traffic. That is the economics of interconnection dependency. A prefix is useful not only because packets addressed to it can move, but because strangers in the routing economy accept a chain of evidence: this organisation is the recognised holder or authorised user; this autonomous system may originate the route; this contact can respond to abuse; this reverse delegation can be changed; this customer can migrate without inheriting a hidden registry dispute. The stronger the chain, the more choice the network has. The weaker it is, the more dependent the network becomes on the one upstream, platform, broker or incumbent willing to tolerate the risk.
AFRINIC makes this dependency visible because it sits at the recognition layer for a region where IPv4 scarcity, governance crisis and address-value disputes have converged. AFRINIC is the Regional Internet Registry for Africa and parts of the Indian Ocean. Its public materials describe a registry for IPv4, IPv6 and autonomous system numbers, with services around WHOIS, RDAP, reverse DNS, an Internet Routing Registry and RPKI. Those are technical functions, but they are also evidentiary functions. They help other networks decide whether an announced route belongs in a filter, whether a customer can bring its own addresses, whether a route-origin authorisation should be trusted, and whether a holder can change the operational facts attached to a prefix.
The official description cannot carry the analysis by itself. AFRINIC has also been the subject of public reporting on alleged historical address-record manipulation, the high-value Cloud Innovation dispute, litigation that affected the institution's finances and governance, years without a normal board, court-appointed receivership, disputed and annulled election processes in 2025, a later board election, and continuing 2026 litigation and ICANN involvement. For interconnection markets, the significance is practical rather than theatrical: registry-layer uncertainty raises the risk premium attached to AFRINIC-administered resources. That premium appears in peering decisions, route filtering, transit terms, customer migration, leasing diligence, BYOIP files, abuse escalation and the ability of small networks to escape dependence on large carriers.
The mechanism is narrow. A registry ledger supplies part of the credit foundation for interconnection contracts. If the ledger is neutral and predictable, it lowers transaction costs. If it becomes a discretionary gatekeeper over commercial use, customer geography or transfers, it raises them and shifts bargaining power upward.
Interconnection is a market in credibility
Interconnection is often described through physical and routing terms: fibre, ports, routers, ASNs, traffic ratios, settlement-free peering, paid transit, route servers and local exchange fabrics. Those terms are necessary, but they are not sufficient. Two networks can be physically connected and still fail to interconnect economically. One may not accept the other's routes. One may refuse to peer unless the route set is documented. One may require route-origin validation, a clean abuse desk, acceptable prefix lengths, stable traffic and proof that the customer behind the announcement is authorised. The technical act of announcing a route is only the beginning of the market act of being believed.
That belief is not sentimental. It is a low-cost substitute for investigation. An upstream carrier cannot reconstruct every customer's historical allocation file before accepting a prefix. An IXP route server cannot litigate the chain of authority behind every route object. A peer cannot call every registry, lessor and enterprise customer before building a filter. A cloud provider cannot manually audit every national corporate registry before allowing a customer to bring an address block into its network. The internet's interconnection market works because it relies on standard evidence: registry records, route objects, RPKI ROAs, RDAP or WHOIS data, abuse contacts, reverse-DNS delegations, letters of authorisation and visible routing history.
Each item is a credibility instrument. It turns a private claim into a fact that counterparties can process. The registry record says who is recognised. The IRR object says which AS may originate a route under the routing policy record. The ROA provides cryptographic route-origin evidence. RDAP and WHOIS expose holder and contact data. Reverse DNS maps address use into names that mail, logging and operational systems can understand. Abuse contacts tell counterparties where to escalate bad traffic. Geolocation correction links address use to customer markets. None of these alone proves perfect legitimacy. Together they reduce the cost of saying yes.
AFRINIC's relevance to interconnection therefore extends beyond allocation. It is the institution whose records and services feed the evidence layer that peers, transit providers and customers consult. A registry that keeps this layer boring makes interconnection cheaper. A registry that is seen as unstable, politicised, discretionary or legally uncertain makes interconnection more expensive even if packets keep flowing. The added cost does not always appear as a published surcharge. It appears as a longer onboarding file, a request for extra indemnity, a refusal to accept a route object, a stricter max-prefix rule, a lower transit discount, a customer's hesitation to migrate, or an incumbent carrier's ability to say that only its address space is safe.
This is why a registry should be understood as part of market infrastructure. It does not sell every transit contract. It does not negotiate every peering session. It does not run every route server. But it supplies the public facts that reduce the need for private bargaining over basic authority. When those facts are strong, smaller networks can use them to negotiate upward. When those facts are weak, larger networks and incumbents can demand bespoke proof, impose risk terms or keep customers inside bundled address arrangements.
Prefix recognition is the first bargaining chip
Before a network can bargain effectively over traffic ratios, ports or transit rates, it must establish that the prefixes it intends to announce will be recognised. This is the first bargaining chip because it determines whether the operator is asking to connect as an independent network or as a customer whose address identity remains dependent on someone else. A network that can show clean, portable prefix authority can shop among upstreams, join route servers, negotiate private peering and promise customers continuity. A network that cannot show that authority must buy credibility from others.
The distinction between provider-aggregatable space and portable space is familiar to engineers, but its economic meaning is often understated. A customer using addresses assigned by an upstream may be easy to connect quickly, but the upstream becomes part of the customer's identity. Leaving the provider may require renumbering, firewall changes, allowlist updates, reverse-DNS changes, geolocation corrections and customer communication. A network with its own recognised prefixes can multi-home, change transit providers, peer locally and preserve customer identity through routing changes. Portability is not a cosmetic convenience. It is an exit right in the interconnection market.
Registry recognition is what makes that exit credible. A route may be technically announced from many places, but counterparties ask whether the announcing AS has authority. They ask because accepting the wrong route can create security, abuse, reputation and contractual risk. If the prefix holder's registry record is current, if route objects are correct, if ROAs match the intended origin, if contacts are reachable and if no unresolved dispute clouds authority, the route is easier to accept. If those signals conflict, the announcement may still propagate through permissive networks, but higher-quality peers and cautious transit providers will add friction.
This point does not require treating IP addresses as ordinary property. The standard registry doctrine that number resources are not simple private property remains central to uniqueness and coordination. But "not property" does not mean "no reliance." A resource holder or authorised user relies on recognition when it signs customers, buys transit, configures RPKI, delegates reverse DNS, builds reputation and promises portability. Interconnection counterparties rely on recognition when they accept routes. The economic right is not the same as ownership of land; it is a recognised authority relationship that reduces the cost of exchange.
AFRINIC's scarcity position increases the value of that relationship. The region entered Soft Landing Phase 1 in 2017 and Phase 2 in January 2020, with small allocation limits and continued need evaluation. Public reporting in 2026 described the remaining unallocated IPv4 pool as modest relative to continental need. In such a setting, a clean recognised prefix is more than an address range. It is optionality: the option to change upstreams, peer at an IXP, lease capacity, support BYOIP, expand customers and retain bargaining leverage.
If registry discretion threatens that optionality, the economic effect resembles an exit toll. A holder may still possess a routeable block, but counterparties discount the ability to move it. An upstream carrier can then offer a simple proposition: use our addresses and avoid registry uncertainty, or use your own and pay for the risk. A large operator may accept the risk because it has lawyers, routing teams and multiple upstreams. A small ISP may not. Prefix recognition therefore sits before price negotiation. It determines who negotiates from independence and who negotiates from dependence.
AFRINIC's crisis added a registry-risk spread
AFRINIC's institutional troubles are often discussed as governance history, but interconnection markets read them as risk spread. A risk spread is the extra caution, discount or procedural burden attached to an otherwise usable asset because the institution around it is uncertain. In the AFRINIC context, the spread attaches to registry-layer recognition: the fear that holder authority, route-origin evidence, transfer status, reverse-DNS control, account standing or public-contact data could become entangled in governance conflict, litigation or discretionary review.
The public chronology supplies enough evidence for that caution. Reporting in 2019 described allegations that valuable African IPv4 blocks associated with dormant or defunct organisations had been manipulated or sold through entities linked to a former AFRINIC staff figure. AFRINIC said at the time that it was investigating. The later Cloud Innovation dispute placed large IPv4 holdings, out-of-region use claims, leasing economics, resource review and registry authority at the centre of litigation. Public analyses described a bank-account freeze and institutional paralysis. AFRINIC then operated for years without ordinary board stability, entered court-supervised receivership, attempted elections, saw the June 2025 process annulled after allegations around voting authority, and later seated a board in September 2025. By 2026, public reporting described recovery efforts, budgets and strategy work, but also continuing lawsuits and ICANN intervention in a winding-up context.
An interconnection counterparty does not need to decide every legal and factual dispute in that sequence. It only needs to observe that registry authority has become commercially salient. If a registry action can trigger litigation large enough to threaten the institution's finances, if an election can become a contest over the body that controls the ledger, if global coordination bodies discuss emergency measures, and if courts in the host state become part of ordinary continuity analysis, then the prefix record is no longer merely clerical in market eyes.
The spread is often quiet. A transit provider may ask for more documentation before accepting AFRINIC-origin space from a customer. An IXP operator may be more conservative around route-server filters. A peer may insist that IRR and RPKI data be perfectly aligned before turning up sessions. A customer may ask for assurances that addresses are not involved in disputes. A cloud onboarding team may require additional proof of authority. A broker or lessor may charge for managing continuity risk. A buyer may pay less for a block whose route-origin and registry status are hard to explain.
The spread can persist even after visible recovery. A new board, budget or strategic plan matters, but markets judge whether routine functions have become boring again. Are route objects updated predictably? Are RDAP and WHOIS data current? Are RPKI services insulated from ordinary commercial disputes? Are reverse-DNS delegations preserved during holds? Are transfer and account-status decisions explained? Are member authority and powers of attorney verified? Can a small network get a clear answer without becoming part of a factional fight?
AFRINIC's defenders can reasonably argue that a registry must protect its records, prevent fraud and preserve the public-resource character of numbers. Critics can reasonably argue that registry discretion over economically embedded resources needs stronger liability, due process and exit protections. Interconnection counterparties do not wait for philosophical resolution. They price uncertainty. If AFRINIC reduces the spread, networks using its resources can bargain on traffic, quality and cost. If it increases the spread, bargaining shifts to proof, indemnity and dependence. The same prefix may still route in both cases. It will not have the same market value.
Peering committees ask ledger questions before engineering questions
Peering is often described as a relationship among networks with comparable traffic interests, but the first questions are frequently about identity and evidence. Who is the autonomous-system operator? Which prefixes will be announced? Are the route objects current? Are ROAs valid? Are contacts reachable? Is the route set consistent with the public record? Is the applicant trying to pass a customer's disputed or leased space through a peering relationship without clear authority? Does the operator maintain a reasonable abuse process? These questions appear procedural, but they decide whether a network enters the peering market on equal footing.
A large network can sometimes overcome uncertainty through reputation. Its brand, legal department, PeeringDB history, NOC contacts and visible traffic can reassure counterparties. A smaller African ISP or edge network has fewer substitutes for evidence. It cannot rely on global name recognition. It needs the registry and routing records to speak cleanly. The ledger becomes its credential.
Selective peering policies make this obvious. A peer may require a minimum traffic level, geographic presence, 24-hour NOC, publicly registered AS, consistent route objects and no default route. It may reject routes that are RPKI-invalid or missing from trusted IRR sources. It may require that a customer provide a letter of authorisation if the origin AS differs from the recorded holder's ordinary pattern. These are not exotic requirements. They are how networks reduce operational risk. A registry record that is current and predictable makes these requirements routine. A registry record that is stale or contested turns them into negotiation.
Open peering does not eliminate the issue. Even a route server at an IXP must decide which routes to distribute. Many route servers build filters from IRR data, RPKI status, prefix-length limits and member-declared route sets. If the applicant's route objects are missing, stale, inconsistent or tied to a questionable holder relationship, the physical port does not deliver the expected reachability. The network may be present at the exchange but invisible in the routes that matter.
This is where AFRINIC's role is specific. It is not only the region's allocation body. It is part of the evidence infrastructure that peering committees and route servers consult. Stable database, IRR, RDAP/WHOIS, reverse-DNS and RPKI services lower the cost of admitting members and routes. Account holds, resource disputes or governance uncertainty make peering decisions more conservative.
The distributional effect is severe. Large carriers can bypass route servers through private bilateral sessions, pay for bespoke review or rely on existing trust. Smaller networks depend more on standard processes. If standard evidence is degraded, they lose the cheaper path into interconnection and may remain connected only through the upstream willing to sponsor, aggregate or tolerate their announcement. Peering committees do not set out to protect incumbents, but weak evidence can produce that result.
Registry neutrality therefore has peering value. A narrow evidentiary ledger lets peering committees ask ledger questions and receive predictable answers. A gatekeeping registry whose future view of use may change forces committees to wonder whether today's answer will remain tomorrow's evidence. The uncertainty hardens peering requirements and narrows the path for edge networks.
Route filtering converts database quality into reachability
Route filtering is the point where registry data becomes a packet-level consequence. A route filter may be generated from IRR objects, RPKI validation status, customer declarations, maximum prefix limits, route-set expansion, bogon lists and internal policy. To the affected network, the result can look binary: the route is accepted or rejected. Behind that binary result sits a chain of database quality.
A missing route object can keep a route out of an upstream's prefix list. A stale object can authorise the wrong origin or fail to authorise the new one. An RPKI ROA that lists an old origin can make a legitimate migration appear invalid. A ROA with an overly strict maximum length can break a de-aggregation needed for traffic engineering. A WHOIS or RDAP record with stale contacts can delay correction because the provider cannot confirm authority. A reverse-DNS delegation tied to an old operational account can create reputational friction during migration. A resource hold can make counterparties reluctant to accept updates even when the route works technically.
This is why IRR, RPKI, RDAP, WHOIS and reverse DNS should not be analysed as isolated services. In interconnection markets they are evidence channels inside filtering systems and contracts. The route filter does not care that the registry service is conceptually separate from the transit contract. It consumes the evidence and makes a decision. The customer does not care that reverse DNS is separate from route-origin validation. It experiences both as part of migration success. The peering coordinator does not care that an internal registry dispute concerns policy rather than routing. It sees uncertain authority and applies caution.
AFRINIC's public materials describe registry services that support exactly these evidence layers. The economic issue is not whether a particular AFRINIC service has failed in a dramatic outage. The issue is whether governance and registry discretion can make evidence less predictable. If a transfer is delayed, route objects may not be updated. If member authority is contested, ROAs may be difficult to change. If a dispute flag is vague, counterparties may overreact. If account standing is used broadly, technical services may appear vulnerable to administrative leverage. The route filter then transmits institutional uncertainty into reachability.
For a multi-homed network, route filtering is also a bargaining device. If both upstreams accept the same clean route set, the customer can shift traffic, negotiate rates, and survive outages. If one upstream accepts the route only because it has a special relationship with the customer or lessor, while another rejects it for evidence reasons, the customer is multi-homed in diagram but dependent in practice. Its ability to exit the first upstream is constrained by the evidence expectations of the second.
Route filtering therefore gives incumbents leverage without requiring bad faith. A large carrier can say that it is willing to provide addresses from its own pool, originate routes under its own policies and manage records for the customer. That offer may be efficient. It also locks the customer into provider-controlled identity. If the customer's own AFRINIC-recognised resources are difficult to document, the carrier's bundled addresses become more attractive. The weaker the registry evidence chain, the stronger the upstream's position.
The answer is not to lower filters and accept unsafe routes. That would undermine routing security and punish careful networks. The answer is to make the evidence chain more reliable. A registry should preserve last verified status, maintain accurate public records, support predictable route-origin changes, keep IRR data clean, distinguish dispute types and avoid broad administrative holds that make unrelated routing evidence suspect. Better evidence allows stricter filters with less exclusion. Poor evidence forces operators to choose between safety and reachability.
In an interconnection market, database accuracy is the difference between a prefix that can move and a prefix that remains beholden to the one path that still accepts it.
RPKI, IRR and RDAP are contractual evidence channels
Technical communities sometimes discuss RPKI, IRR, RDAP, WHOIS and reverse DNS as though they belong in separate silos. Interconnection contracts collapse those silos. A transit agreement, peering policy, IXP route-server rule, customer migration plan or BYOIP process usually asks for evidence of the same underlying fact: who is authorised to use and announce this prefix, and who will be accountable if something goes wrong?
RPKI supplies one answer by linking a prefix to an authorised origin AS through a route-origin authorisation. IRR supplies another answer through route and route-set objects used by filters. RDAP and WHOIS supply public registration and contact information. Reverse DNS supplies operational naming and reputation continuity. Abuse contacts supply escalation. Letters of authorisation bridge cases where the holder, operator and origin AS are not the same entity. A clean interconnection file aligns these signals. A weak file leaves them to contradict one another.
Contracts turn contradiction into risk allocation. An upstream may accept a customer's prefix only if the customer warrants that it is authorised to announce it and maintains route objects. A cloud platform may require the customer to prove control before allowing customer-supplied address use. An enterprise customer may require its provider to maintain stable IP identity, valid reverse DNS and reachable abuse handling. A peering agreement may permit route rejection if routes become RPKI-invalid or inconsistent with policy. The technical evidence is not advisory. It becomes a condition of commercial performance.
AFRINIC's registry layer matters because it is one of the sources from which these contractual signals derive. If AFRINIC's recognised holder data, account authority and route-origin support are stable, counterparties can rely on standard terms. If they are exposed to discretionary holds, unclear dispute states or governance uncertainty, counterparties add bespoke terms. They may require indemnities for registry action, reserve rights to stop announcing routes, demand additional letters from the recorded holder, or refuse customer-provided addresses until the registry issue is cleared.
This is not a pure RPKI story. A valid ROA can coexist with stale abuse contacts. A correct RDAP record can coexist with a missing route object. A route object can persist after a transfer and become stale. Reverse DNS can remain under the wrong operational control even when routing changes. Abuse contacts can be current but lack authority to solve route-origin questions. The market needs the bundle because each evidence channel answers a different counterparty fear.
The bundle is especially important in leasing and delegated-use arrangements. A holder may remain the registry-recognised resource member while a downstream network originates the route, serves customers and handles operations. That can be legitimate and efficient. It also requires clarity. Who can request a ROA? Who maintains IRR objects? Who receives abuse complaints? Who controls reverse DNS? What happens if the lease ends? What if the holder enters a dispute with the registry? Without clear evidence, counterparties may treat the arrangement as shadow control or hidden transfer, even if the commercial structure is ordinary.
The registry should not need to approve every private contract to support this market. It does need to make operational responsibility legible enough that interconnection counterparties can rely on it. That means stable public records, clear delegated-contact mechanisms, predictable route-origin procedures, precise dispute flags and service-specific holds. A transfer question should not automatically impair RPKI. An abuse-contact correction should not become a business-model audit. A payment issue should not make peers wonder whether route-origin evidence will disappear.
The discipline is evidentiary. AFRINIC should ask what fact the market needs to know, what proof establishes it, what service depends on it, and what remedy follows if it is wrong. When the registry stays at that level, it strengthens contracts. When it moves from evidence to permission, it makes every contract price the possibility of institutional intervention.
Upstream dependence rises when prefix mobility falls
A network's dependence on an upstream is not measured only by the number of transit contracts it signs. It is measured by how easily it can move traffic, customers and address identity if one contract becomes too expensive, unreliable or strategically constraining. A network with two upstreams but only one acceptable route-origin story is not truly independent. It has redundancy in ports, not necessarily in bargaining power.
Prefix mobility is the centre of this problem. If a network can originate its own recognised prefixes through more than one upstream, it can compare prices, shift traffic, negotiate service quality and survive disputes. If the network uses addresses controlled by one upstream, migration becomes costly. If its own addresses are registry-recognised but route objects, ROAs, contacts or dispute status are unclear, the migration may be blocked by the receiving provider's filters or legal policy. In both cases, the incumbent gains leverage.
AFRINIC's scarcity and institutional uncertainty make prefix mobility more valuable and more fragile. IPv4 scarcity means a growing network cannot assume that it will obtain new provider-independent space quickly. It may lease, transfer, acquire a customer block, use provider-assigned space or intensify NAT. Each choice has an interconnection consequence. A leased block may require proof from the holder. A transferred block may wait on registry approval. A customer block may need new ROAs and reverse DNS. Provider-assigned space may lock the network into the provider. NAT may solve address count while worsening customer experience or traceability. The address plan becomes an upstream strategy.
The market impact is clearest when a small ISP wants to escape a single upstream. The incumbent may be costly or congested, but it controls the addresses currently visible to customers. A new upstream offers better rates, but requires clean route authority for the ISP's independent or leased block. The IXP route server will not distribute the route until filtering data aligns. Enterprise customers ask whether their services will keep stable IPs. The ISP's engineers can configure BGP, but the business cannot move until the evidence chain is accepted. If AFRINIC-related records are uncertain or slow to update, the old upstream remains powerful.
Large carriers benefit from this structure even without deliberate exclusion. They can bundle transit, addresses, reputation management, routing support and customer assurance. The bundle is attractive when independent address use is administratively risky. A small operator may prefer its own identity in principle, but choose the carrier's space because it lowers immediate friction. Over time, the carrier's address bundle becomes a switching cost. The customer is buying reachability and surrendering future exit.
A neutral registry lowers that switching cost by making independent and delegated address authority easier to prove. It lets a small network say to both upstreams: here is the recognised holder, here is the authorised origin, here are route objects, here are contacts, here is reverse DNS, here is dispute status, and here is the process if anything changes. The upstream can then compete on price and quality rather than on its ability to substitute for registry trust.
A discretionary registry raises the switching cost. If the upstream fears that a holder's commercial use, geography or account status could be revisited unpredictably, it may refuse the route or demand stronger terms. If the customer fears that registry review could disrupt route-origin evidence, it may stay with the incumbent. If the lessor fears that delegated use could be recast as suspect, it may restrict customer mobility. Upstream dependence rises because prefix mobility falls.
That is the uncomfortable institutional point. Ambiguous control over address movement has downstream market effects even when justified as stewardship, conservation or anti-abuse. It can still strengthen incumbents by making independent mobility harder. A registry that wants competitive interconnection should treat prefix mobility as a public-interest outcome, not as suspicious commercialisation by default.
IXPs lower costs only when the address story is clean
Internet exchanges are designed to reduce dependence on expensive upstream transit by allowing networks to exchange traffic locally. For regional ISPs, content networks, universities, hosting firms and public networks, an IXP can improve latency, resilience and bargaining power. A port at an exchange changes the economics only if routes are accepted. A network can plug into a switch and still fail to obtain the full benefit if its prefixes do not pass the evidence tests used by peers and route servers.
The IXP model relies on trust compression. Many networks connect to a shared fabric. They cannot all negotiate deeply customised trust arrangements for every route. Route servers, route policies and membership rules therefore use standard evidence to reduce risk. Prefixes must be documented, origins must be credible, contacts must work, max-prefix settings must be sane, and route filters must be maintainable. The exchange is a low-cost interconnection marketplace only because participants can treat much of this evidence as routine.
AFRINIC's registry functions feed that routine. A member's ASN and prefixes, public contacts, route objects, RPKI status and reverse-DNS arrangements help an IXP decide whether the network is a normal participant. If those records are stale or contested, the IXP may still admit the member physically but restrict route-server participation or leave peers to decide bilaterally. That weakens the exchange's value for the very network trying to reduce transit dependence.
The problem is especially acute in edge markets where one IXP may be the only realistic local peering venue. A large carrier can build private interconnects or pay for multiple exchange presences. A small ISP may have one exchange, one cross-connect budget and a narrow window to convince customers that local traffic will improve. If one missing route object, stale RDAP contact or unresolved holder question keeps its routes out of the route server, the IXP no longer lowers costs as promised. The ISP remains dependent on transit for traffic that could have stayed local.
This is not an argument for IXPs to weaken filters. Weak filtering would create route leaks, hijack risk and mistrust. The argument is that the registry should make good filtering easier. It should provide accurate records, stable route-origin support, clear delegated-use pathways and precise status labels. It should not force IXPs and peers to guess whether a prefix is clean, disputed, under review, frozen for unrelated reasons or merely waiting for an ordinary update.
AFRINIC's governance history raises the stakes because IXPs are not just technical switches in regional development rhetoric. They are instruments for lowering dependence on international transit and large carriers. If registry-layer uncertainty makes local peering harder, the cost reduction promised by IXPs is partly clawed back. The region may invest in exchange fabrics, training and community meetings while smaller networks still struggle to prove route authority in a way peers accept.
A narrow ledger helps IXPs fulfil their economic function. It gives route servers reliable inputs, members a common evidentiary baseline and peers the confidence to focus on traffic rather than registry rumours. It allows small networks to use standard processes instead of bespoke negotiations with every larger participant. A gatekeeping registry does the opposite, turning each route into a possible policy question and each IXP into a place where registry uncertainty is priced by filters. The physical exchange fabric can lower the cost of interconnection only after the evidence fabric has done its work.
Customer portability is an interconnection problem
Customer portability is often discussed in relation to cloud platforms or enterprise migration, but it begins in the interconnection market. A customer that moves from one hoster, ISP or upstream arrangement to another wants more than an account transfer. It wants stable reachability. Its customers, suppliers, banks, security providers, mail systems, firewalls and allowlists may already know a set of addresses. If those addresses cannot move cleanly through new upstreams and peers, the customer is not portable in practice.
IPv4 addresses function as operational memory. They appear in firewall rules, fraud systems, payment gateways, VPN configurations, monitoring tools, mail reputation, DNS, customer documentation and geolocation databases. Replacing them can be costly even when technically feasible. A regional ISP that can offer customers portable, cleanly documented addresses has a stronger product than one that can only resell an upstream's address bundle. A data-centre operator that can help customers keep address identity across transit changes has leverage. A hoster that cannot prove route authority loses customers to larger providers that can.
This is why customer migration files ask interconnection questions. Will the prefix be accepted by the new upstream? Are ROAs ready for the new origin AS? Are IRR objects updated? Does the recorded holder authorise the move? Are reverse-DNS delegations under operational control? Are abuse contacts pointed to the right desk? Will geolocation databases be corrected? Is the prefix under any registry dispute? Can the provider commit that registry services will remain available during the migration?
Cloud bring-your-own-IP is one example, but not the whole story. The same proof problem appears when an enterprise moves from a national carrier to a regional ISP, when a hosting customer changes data centres, when a university multi-homes, when a managed-service provider consolidates networks, or when a content provider localises traffic through an IXP. The customer is not buying a philosophical view of address rights. It is buying continuity. The provider's ability to deliver continuity depends on the evidence chain around the prefix.
AFRINIC's crisis affects customer portability because it can make that evidence chain harder to sell. A customer that reads about registry litigation, election problems or address disputes may not understand the details, but it understands operational risk. It asks its provider for assurances. The provider asks the holder, lessor, broker, upstream or registry. Each layer adds caution. The migration that should have been an engineering project becomes a trust project.
This dynamic strengthens incumbents. The existing provider can say: do not move; keep our addresses; avoid uncertainty; we will handle the registry and routing paperwork. That may be a legitimate service proposition. It also reduces competitive pressure. A market in which customers fear address migration is a market in which incumbents retain customers through switching costs rather than quality.
Registry neutrality reduces switching costs by preserving the customer's address story. It does not need to guarantee every private contract. It needs to support accurate holder records, delegated authority, route-origin changes, reverse-DNS updates, abuse-contact continuity and clear dispute handling. If a prefix is subject to a defined hold, the hold should say what it affects and what remains preserved. If a transfer is pending, counterparties should understand the service state. If a lease or delegated-use arrangement is legitimate, the operational responsibility should be legible enough for migration.
Customer portability is therefore not an argument for unregulated address markets. It is an argument for evidence that lets customers move without forcing every new provider to become a registry-law specialist. The registry should lower the cost of migration by making authority portable and verifiable. If it does not, customers remain dependent on the providers whose address story is easiest, not necessarily whose network is best.
Abuse, reverse DNS and geolocation sustain trust after onboarding
Interconnection trust does not end when a BGP session comes up. Peers, upstreams and customers continue to monitor abuse reports, reverse-DNS consistency, registration data, geolocation accuracy, routing stability and contact responsiveness. A prefix that passes onboarding can later become commercially costly if these operational signals decay. The registry's evidence role therefore extends into the life of the interconnection relationship.
Abuse contacts are the most direct example. When a peer or upstream sees spam, scanning, bot traffic, copyright complaints, phishing or other unwanted activity, it needs an accountable desk. If the RDAP or WHOIS contact is stale, if the abuse mailbox points to a holder with no practical control, or if delegated-use arrangements hide the operator, the counterparty must choose between tolerating risk, escalating through informal channels, filtering traffic or terminating the relationship. A stale contact is not a minor data defect. It can change the willingness of others to carry traffic.
Reverse DNS has a similar but quieter role. Mail systems, logging tools, network diagnostics and customer support often expect consistent PTR records. During migration, leasing or transfer, reverse-DNS control can lag behind route changes. The result may be mail deliverability problems, fraud flags, customer confusion and support costs. If reverse-DNS updates depend on a registry account caught in a broader dispute, the operational harm can reach customers who have no role in the dispute.
Geolocation is less formally tied to the registry but still depends on credible public evidence and operational continuity. Address geolocation databases are built from many signals, including registry records, routing, provider data, customer input and commercial datasets. When a prefix moves between networks, countries or customer contexts, wrong geolocation can break streaming rights, fraud controls, payment flows, content localisation and public-sector access. Correcting the error often requires proof that the new operator has authority and stable use. Registry uncertainty makes that proof harder.
These services affect peering and transit because they shape perceived operational quality. A peer may not depeer immediately over one stale contact, but repeated unanswered abuse reports change the relationship. An upstream may not reject a customer because of one reverse-DNS delay, but it may impose stricter terms or refuse future delegated-use arrangements. A customer may not leave after one geolocation error, but it may hesitate to move more workloads. Interconnection is renewed continuously through operational trust.
AFRINIC's role should be to make these maintenance signals reliable without turning them into leverage. If a member is under review for a transfer issue, existing abuse contacts should not become collateral damage unless the review directly concerns contact fraud. If a payment dispute exists, reverse-DNS and RDAP continuity should be treated separately from new resource movement where possible. If a holder's authority is contested, the last verified operational contacts should remain visible while the dispute is classified. The goal is to preserve accountability while preventing broad holds from harming innocent downstream users.
Mandate expansion can occur here as well. A registry may begin with a legitimate need for accurate abuse contacts and then use contact accuracy as a basis for broader business-model review. It may begin with reverse-DNS hygiene and move into questions about customer geography, or begin with RDAP accuracy and expand into suspicion of leasing. Each expansion may sound like responsibility, but the interconnection effect is to make routine maintenance feel risky. Operators then delay updates, hide arrangements or rely on intermediaries who promise to manage the registry relationship.
The better rule is service specificity. Abuse contact problems should lead to contact remedies. Reverse-DNS problems should lead to delegation remedies. Geolocation evidence should be supported by accurate public data. Route-origin questions should be handled through routing evidence. Broad punishment should be reserved for clear fraud, abandonment, court order or severe breach. Trust improves when each evidence channel is kept accurate and bounded.
Scarce IPv4 turns evidence defects into capital-control risk
The phrase capital control can sound exaggerated in an internet-number context, but the analogy is useful if kept precise. A registry does not control currency. It does, however, influence the mobility of scarce address resources by deciding what evidence is sufficient for transfer, delegation, route-origin changes, account standing and operational updates. When those decisions are narrow and predictable, they protect the ledger. When they are broad and discretionary, they can trap economic value inside administrative permission.
In a scarce IPv4 market, mobility is value. A block that can be transferred, leased, routed through multiple upstreams, accepted at IXPs, supported in RPKI, updated in RDAP and sold to customers has higher value than a block that is technically routeable but institutionally hard to move. Evidence defects therefore become capital defects. A missing route object can limit transit options. An unclear holder record can block a sale. A stale contact can damage reputation. A disputed account can make peers hesitate. A delayed ROA update can slow migration. Each defect reduces the resource's ability to move toward productive use.
Registry discretion can magnify this effect. If the registry asks whether a holder is the recognised source of a transfer, that is ledger protection. If it asks whether the holder's customer base is morally or regionally acceptable under a vague theory, that is economic licensing. If it preserves last verified route-origin evidence during a dispute, it protects running networks. If it withdraws or threatens operational evidence as leverage in a commercial disagreement, it turns interconnection services into control instruments. If it marks a precise dispute category, it informs markets. If it creates ambiguous fear around a whole account, it imposes a liquidity discount.
AFRINIC's policy environment contains the ingredients for this risk because IPv4 scarcity, regional development rhetoric, transfer limits, leasing disputes and governance conflict overlap. Regional stewardship can be legitimate when it prevents fraud or preserves the remaining free pool for real network needs. It becomes mandate laundering when the language of stewardship is used to supervise lawful commercial movement after reliance has formed. The registry's narrow mandate is to keep uniqueness, authority, contacts and routing evidence credible. It is not to become a private central planner for every downstream customer relationship.
The interconnection market prices the difference. A transit provider does not need to know whether a policy debate is called conservation, regional development or anti-abuse. It asks whether accepting the route creates future risk. A peer does not need to adjudicate the economics of IPv4 leasing. It asks whether the origin AS is authorised and accountable. A customer does not need to decide whether number resources are property. It asks whether its migration will survive registry review. If broad discretion is possible, these actors add caution.
This caution can produce the same economic pattern as capital control: reduced liquidity, wider spreads, greater dependence on approved intermediaries and lower bargaining power for smaller holders. Addresses remain inside the system, but movement becomes expensive. A policy presented as protection for regional resources may make those resources less useful for the regional networks that need them.
A registry can avoid this by classifying evidence defects narrowly. Fraud, forged authority, duplicate claims and court orders justify strong holds. Missing or stale contacts justify correction. Payment problems justify billing remedies and defined service consequences. Ambiguous commercial-use questions require prospective rules, due process and preservation of running networks. Treating all defects as possible grounds for broad account pressure is what turns recordkeeping into capital control.
Small ISPs and edge networks pay the highest price
Small ISPs and edge networks are most exposed to interconnection dependency because they have the least redundancy in both infrastructure and evidence. A global carrier can hold address space across regions, hire counsel, maintain route-policy staff, negotiate with many upstreams, build private interconnects and survive a long registry ticket. A small regional ISP may have one IXP, two upstream quotes, one engineer who handles routing and customer support, and a few prefixes that determine whether it can sell service independently. One unresolved record can change its whole bargaining position.
Consider the operational sequence. The ISP wants to multi-home away from a single upstream. It obtains or leases a prefix. It creates or requests a route object. It prepares a ROA for its ASN. It updates contacts and reverse DNS. It applies for IXP route-server acceptance. It asks enterprise customers to migrate services. If every evidence channel aligns, the ISP can negotiate. If one channel fails, the fallback is often the incumbent upstream's addresses or the lessor's preferred carrier. The ISP is not failing at engineering. It is failing to buy trust at a price it can afford.
The fixed-cost nature of evidence makes the inequality worse. A large network can distribute route-policy, legal and registry tasks across departments. A small network pays the same types of fixed costs over fewer customers. It may not know which IRR database a peer trusts, how a route server expands AS-SETs, whether a ROA maximum length will conflict with traffic engineering, which registry document proves delegated use, or how to correct geolocation quickly. It learns by delay, and delay is expensive.
Stale contacts are particularly punitive for small operators. A big carrier's NOC may be known even when public records are imperfect. A small operator's public record may be its only credential. If abuse contacts fail, peers assume risk. If RDAP data is outdated, upstreams ask for more proof. If reverse DNS still points elsewhere, customers doubt the migration. If route objects are missing, filters reject announcements. Each defect pushes the small network back toward dependence on an intermediary with better-known credentials.
This is why discretionary gatekeeping is regressive in interconnection markets. Large networks can price, litigate or route around discretion. Small networks must comply, wait or buy from incumbents. A gatekeeper may claim to protect the region from extraction, but if its procedures raise the cost of independent prefix use, it can entrench the largest carriers and platforms. The rhetoric of fairness can produce a market structure in which small networks have fewer paths to autonomy.
The practical design objective should therefore be small-network legibility. A small ISP should be able to know which evidence is needed for a prefix to be accepted by common upstreams and IXPs, how to update route-origin evidence, how to keep contacts current, what happens during a hold, and how long ordinary changes should take. It should not need insider knowledge or legal escalation to prove a routine authority chain.
AFRINIC's institutional legitimacy will be judged partly by these operators. If they can use its records to reduce upstream dependence, the registry strengthens regional interconnection. If they must rely on incumbents because registry evidence is too uncertain, the registry becomes another layer of dependency.
Neutral ledgers reduce incumbent leverage
Registry neutrality is often defended as a governance principle. In interconnection markets it is also a competition principle. A neutral ledger lowers incumbent leverage by giving smaller networks and customers a credible way to prove authority without buying identity from a larger provider. A discretionary gatekeeper increases incumbent leverage by making independent address use harder to explain.
The mechanism is straightforward. Incumbents can always sell simplicity. Use our addresses. Use our transit. Use our abuse desk. Use our reverse DNS. Let us originate the route. Avoid the registry file. That bundle may be valuable, especially for customers that do not want operational complexity. It becomes anti-competitive in effect when independent alternatives are made unnecessarily risky by registry ambiguity. If the customer's own or leased prefix can be documented cleanly, the incumbent competes on network quality. If it cannot, the incumbent competes on the customer's fear of evidence failure.
Neutrality does not mean the registry is indifferent to fraud. A neutral ledger is strict about facts. It verifies holder authority, marks disputes precisely, keeps contact data current, supports route-origin evidence, processes legitimate transfers, and protects operational services. Its neutrality lies in refusing to convert those factual checks into broad judgment about business models, customer geography, political respectability or the desirability of IPv4 monetisation.
AFRINIC's address-value disputes have made this path harder but more necessary. When a registry is accused of overreach, its defenders may respond by asserting institutional authority. When a large holder resists review, critics may respond by calling for stronger controls. Both reactions can expand the gatekeeper prize. The more power the registry has to decide commercial outcomes, the more every faction wants to control it. The more every faction wants to control it, the less neutral the ledger appears. Incumbents and large carriers can then monetise the loss of neutrality through managed dependence.
A neutral ledger also reduces the need for private substitutes. If AFRINIC's records are reliable, peers and upstreams do not need to maintain extensive private assurance files for every AFRINIC-origin prefix. If the records are not reliable, large networks build their own trust systems. Those private systems may be operationally sensible, but they are less accessible to small networks. They can become hidden gates: accepted by one carrier, rejected by another, corrected through private contacts rather than public process.
The same principle should guide emergency and oversight mechanisms. ICANN, the NRO, courts or receivers may need to preserve continuity in extreme cases. Their purpose should be to preserve last verified records and service continuity, not to choose winners in address economics. Otherwise the system merely moves discretion upward and leaves the interconnection market pricing a different authority risk.
Neutrality is not institutional weakness. It is the condition under which networks can bargain without first seeking permission from the strongest one in the room.
Institutional legitimacy is measured by interconnection costs
A registry's legitimacy can be discussed through bylaws, elections, board seats, court orders and recognition frameworks. Those are necessary, but they are not the only test. For networks, legitimacy is measured by whether the registry lowers the cost of interconnection. A legitimate registry makes it easier for unrelated parties to rely on the same facts. An illegitimate or unstable one forces them to build private workarounds.
The measurable signs are practical. How long does it take to update a route-origin authorisation after an authorised migration? Can a route object be corrected without a broad account review? Are RDAP and WHOIS contacts current enough for peers to escalate abuse? Does a route server accept the member's prefixes using normal filters? Can a customer bring its addresses to a new provider without extra legal drama? Are dispute flags precise enough that counterparties know what is actually affected? Are reverse-DNS delegations preserved during unrelated litigation? Can small networks understand the process without hiring specialists?
These indicators matter more than institutional rhetoric. A board elected after years of receivership is significant because it can restore ordinary authority, budgets and management. It becomes legitimacy only when ordinary registry functions become predictable. A court order preserving the institution is significant because it prevents collapse. It becomes legitimacy only when the preserved function is narrow and trustworthy. ICANN or NRO involvement may be necessary to protect global coordination. It becomes legitimacy only if it preserves the ledger without expanding unaccountable discretion.
AFRINIC's 2025 and 2026 recovery context should be read this way. The later board election and reported budget or strategy efforts are positive if they reduce the registry-risk spread. Continuing litigation and public disputes are damaging if they keep counterparties unsure whether records, services and policy interpretations are durable. The issue is not whether every conflict disappears. Infrastructure institutions often operate amid disputes. The issue is whether the disputes are isolated from ordinary evidence channels.
Legitimacy also requires admitting uncertainty in controlled ways. A registry that hides disputes creates false confidence. A registry that labels too broadly creates panic. A market-grade status system would distinguish ordinary update pending, transfer review, payment issue, documented authority dispute, court hold, fraud investigation, security lock and completed record. Each status should state service effects: whether routing evidence remains valid, whether reverse DNS can change, whether RDAP contacts remain current, whether transfer is paused, and whether existing interconnection should be disturbed. That precision lowers the need for rumours.
Due process is part of legitimacy because it lowers litigation incentives. A holder, lessee or customer is more likely to cooperate with a record review if last verified operational status is preserved, the issue is bounded, timelines are known and appeal is real. A peer or upstream is more likely to continue carrying traffic if it sees that the registry issue does not affect route authority. A customer is more likely to migrate if it knows how disputes are handled. Procedure is not a legal ornament; it is an interconnection cost reducer.
AFRINIC's institutional legitimacy will not be restored by declaring that the region needs a registry. The region does need one. Legitimacy will be restored when networks can use AFRINIC records to reduce dependence on intermediaries and negotiate interconnection without an added uncertainty premium.
A narrow evidentiary ledger is the design answer
The design conclusion follows from the market mechanism. AFRINIC should be a narrow evidentiary ledger for interconnection, not a broad gatekeeper over the interconnection market. Its essential task is to preserve unique, accurate, current and operationally useful records that counterparties can rely on when accepting routes, joining IXPs, negotiating transit, migrating customers and escalating abuse. The registry should be strong in evidence and restrained in economic permission.
Operational preservation is the starting point. If a prefix, holder, contact, ROA, route object or reverse-DNS delegation was valid before an ordinary dispute, it should remain preserved unless the specific dispute concerns fraud, duplicate authority, security compromise, court order or another defined emergency. Running networks should not lose credibility because an unrelated commercial, payment or governance question is open.
Route-origin evidence also has to be predictable. Operators should know how to create, change, delegate and revoke ROAs and route objects; what proof is required when holder and origin AS differ; how leases or customer assignments are represented; and how transfers affect existing routing evidence. RPKI and IRR services are safety and credibility tools, not bargaining chips. Their service states should be documented, auditable and insulated from ordinary institutional politics.
Holds need boundaries. A hold should identify the affected resource, the reason, the evidence threshold, the service consequences, the cure path and the expected review time. Transfer holds should not automatically become RPKI holds. Payment holds should not automatically disrupt reverse DNS. Abuse-contact correction should not become broad commercial-use review. Severe actions that could impair interconnection should require notice, reasons, proportionality and appeal. Courts remain available for legal disputes, but routine evidence disagreements should not become institutional warfare.
Recordkeeping must remain separated from market control. AFRINIC may record holder identity, contact authority, dispute status, routing evidence and service responsibility. It should not use those records to decide whether a lawful customer mix, leasing model, transit strategy, BYOIP use, upstream change or commercial monetisation is virtuous unless a clear, prospective and narrow rule tied to registry function applies. Anti-fraud work should not become a licence to supervise business plans. Regional development language should not become a way to immobilise scarce resources.
Portability should be treated as a legitimacy objective. A registry that supports competitive interconnection makes it easier for networks to move prefixes across upstreams, IXPs and customer relationships while preserving accountability. Portability does not mean anarchy. It means evidence moves with the recognised relationship and counterparties can verify authority without relying on incumbents. In a scarce IPv4 market, portability is one of the main ways small networks avoid dependence.
These disciplines do not weaken the registry. They make its authority more defensible. A registry that confines itself to evidence can be strict without being feared as a private economic regulator. It can reject forged documents, preserve court orders, correct stale contacts, clean up old records and maintain routing security while assuring the market that ordinary commercial movement will not be punished through ambiguous discretion.
Return to the regional ISP at the IXP. The engineer can connect the cable and configure BGP. The business still needs others to believe the prefix. If AFRINIC's ledger is narrow and reliable, the ISP can show clean evidence to both upstreams, join the route server, migrate customers, correct geolocation, maintain abuse contacts and negotiate from independence. If the ledger is uncertain or discretionary, the ISP may still connect, but its bargaining power leaks away to the upstream, broker or incumbent that can provide a safer story.
That is the economics of interconnection dependency. Addresses are not abstract assets floating above the network. They are the credit foundation of interconnection contracts. AFRINIC's task is not to decide the destiny of every address-based business model. It is to make the evidence credible enough that networks can interconnect without asking permission from the strongest carrier, the loudest faction or the most risk-averse gatekeeper. The narrower the ledger, the wider the market for independent interconnection.