The sea is a tariff. It charges for every repair crew that must fly in, every router card that waits in customs, every cable fault that has no terrestrial detour, every hotel booking system that cannot afford a failed payment window, and every government service that must keep working when the weather has closed the airport. For an island network, distance is not only measured in kilometres. It is measured in spare parts, engineer hours, insurance premiums, procurement exceptions, satellite minutes, and the number of credible people who can change a registry record at two in the morning.
That is why island connectivity should not be treated as a small version of mainland connectivity. A rural operator on a continent may face low density, long access lines and thin revenue. A new entrant may face capital scarcity and a difficult wholesale market. An island operator may face all of that, and also a sharper problem: the fallback options are fewer, slower and more expensive. The landing station, airport, main data centre, tourism economy, port, customs platform, banks and public agencies may all depend on the same small collection of carriers, engineers and international routes. A failure in one layer quickly becomes a test of the others.
AFRINIC sits inside that problem in a way that is often underplayed. The African Network Information Centre is the regional internet registry for Africa and parts of the Indian Ocean. Its service perimeter includes island economies such as Mauritius, Seychelles, Madagascar, Comoros and Reunion alongside continental markets. It is registered in Mauritius. It administers IPv4, IPv6 and autonomous system number resources, and provides the public records and technical services that help other networks understand who holds a resource, who can be contacted, what names are delegated, and what routing evidence should be trusted.
Those functions can sound clerical until an island has to use them in anger. An island trying to add a second upstream needs an address and routing story that the new upstream will accept. A port moving a customs system needs reverse DNS, contacts and route evidence that do not contradict the migration plan. A hotel group moving booking infrastructure needs stable public identity during the peak season. A government disaster-recovery site needs a way to become reachable when the main carrier, cable or data centre is impaired. In each case the registry is not the whole infrastructure. It is the recognition layer that lets infrastructure become a credible option.
The island question is therefore not whether AFRINIC can describe itself as a regional steward, nor whether every current dispute around the registry has one simple villain. The question is narrower and more economic: does the registry make island redundancy more bankable, or does it make geography even more expensive? If routine record changes are slow, if scarce addresses are hard to plan around, if routing authority is unclear, if public contact records are stale, or if institutional turbulence makes counterparties nervous, island networks pay twice. They pay the ordinary island premium of cables, labour and logistics. Then they pay a registry-certainty premium on top.
Islands convert administrative uncertainty into balance-sheet risk
On a large mainland network, administrative delay is irritating but often survivable. Work can be moved to another metro area. Staff can be borrowed from a neighbouring city. A supplier can drive replacement equipment overnight. A second carrier may be physically independent. A data-centre cutover can be postponed without affecting the port, the airport, the tourism ministry and most of the banking sector at the same time. The geography contains slack, and slack is an economic asset.
Island networks have less of it. The same operator may serve government offices, hotels, payment terminals, school networks, ferry operators, airport services and local enterprises. The same landing-station corridor may carry the traffic of several sectors. The same small engineering labour pool may maintain BGP sessions, fibre rings, generators, power systems and emergency communications. A registry ticket that looks small in an administrative queue can become material when it sits inside a maintenance window that must align with air freight, customs clearance, a foreign vendor visit, a public-service cutover and the start of tourist season.
This is the first island-specific economic point: uncertainty does not remain in its original category. A delayed ASN request becomes a postponed multi-homing project. An unresolved reverse-DNS change becomes a problem for a bank, payment processor or security vendor. A stale RDAP or Whois contact becomes a reason for foreign networks to slow an abuse or routing investigation. A disputed resource record becomes a financing question for a data centre whose customers want to know whether their public endpoints will remain portable.
The damage is not always a dramatic outage. It is more often a widening margin in every plan. Project managers add weeks. Procurement officers demand stronger warranties. Banks ask for clearer continuity evidence. Public agencies choose the incumbent because the incumbent has the least complicated address story. A smaller ISP holds back from expansion because the address plan cannot be settled before a contract deadline. A hotel group tolerates avoidable lock-in because renumbering during high season would be reckless.
AFRINIC's role is not to remove island geography. It cannot shorten the sea lane or conjure a second landing station. Its practical duty is to avoid making geography harder to insure. A registry that publishes clear records, handles routine changes predictably and preserves live reliance during disputes reduces the number of variables that island risk managers must price. A registry that appears discretionary or unstable turns paperwork into a balance-sheet item.
AFRINIC's Indian Ocean perimeter is economically central
The Indian Ocean is sometimes treated as a footnote to African internet governance. That shorthand misses the structure of the problem. Mauritius, Seychelles, Madagascar, Comoros and Reunion are not merely points on a service-region map. They are economies in which law, logistics, infrastructure and market scale are compressed. The distance between a registry record and a public-service consequence is shorter than it appears from a policy meeting on the mainland.
An island market does not experience number-resource uncertainty in the same way as a large continental market. The continental operator may worry about long-haul fibre, cross-border terrestrial routes, tower economics, spectrum, rural access subsidies and domestic wholesale regulation. The island operator worries about those where relevant, but also about cable landings, ocean weather, port logistics, airport capacity, import duties, insurance exclusions, a thin vendor bench, and whether a second upstream is truly independent or merely another contract on the same physical dependency.
This does not make island networks special in a sentimental sense. It makes them analytically useful. They reveal what a registry is worth when the economy has limited room to absorb mistakes. AFRINIC's public functions--number-resource management, public registration data, reverse DNS, IRR and RPKI services, policy implementation and member support--become inputs into tourism, customs, government-service continuity, banking operations, remote work and disaster planning. The registry's quality is not judged only by whether packets move today. It is judged by whether the next network change can be explained, financed and executed without heroic persuasion.
AFRINIC's own legal home reinforces the point. The organisation is registered in Mauritius, an island jurisdiction whose courts and corporate-law mechanisms have been central to parts of AFRINIC's recent institutional history. That does not mean Mauritian law is a problem. Courts are often necessary when private institutions fail to govern themselves cleanly. But it does mean that an island legal vessel carries a regional technical function. Local procedure can have regional operational consequences.
For island members, that is not an abstraction. They already live with jurisdictional layering. A cable may be owned by a consortium, land in one territory, connect to another, be financed partly abroad and serve services hosted in several places. A customs platform may depend on local law, foreign software, international banks and cloud infrastructure. AFRINIC fits this pattern: a local corporation, a regional service region, global recognition of addresses, and operational dependence by networks that cannot simply choose a different ocean.
Cable landings create capacity, not independence
Submarine cables are the most visible symbols of island connectivity. They come with ceremonies, maps, landing photographs and promises of lower prices. They can transform latency and wholesale bandwidth. They can make local hosting more plausible, support new data-centre investment and give public agencies a stronger basis for digital services. But a cable landing is not the same thing as network independence.
The difference lies in control. A local ISP connected to two cable-backed carriers may look redundant. If its public addresses are assigned by one of those carriers, its ability to move traffic, change commercial terms or survive a dispute remains constrained. The physical route may be diverse while the customer's public identity remains captive. Provider-assigned space can be efficient for ordinary access services. It is far more dangerous when used by customers whose continuity matters: ports, ministries, banks, hospitals, airport systems, emergency services, large hotel groups and data-centre tenants with many downstream customers of their own.
The same is true for data centres. A facility near a landing station can advertise low latency and resilient power. Enterprise customers will still ask whether they can bring addresses, maintain route-origin evidence, preserve reverse DNS, keep security allowlists intact and move between upstreams without renumbering. Those questions are not decorative. They determine whether the data centre is a genuine platform or merely a well-powered building attached to one dominant address story.
The registry makes cable capacity economically usable when it allows networks to establish identities that can survive supplier changes. An ASN, clean public contact records, portable address space where justified, credible route-authorisation material and predictable reverse-DNS processes help the customer turn a second route into a real option. Without them, the second cable may lower the price of bandwidth while leaving the switching cost almost untouched.
This matters in small markets because incumbency is sticky. The operator that historically held addresses, government relationships, cable access and registry competence often begins with a large advantage. New capacity does not automatically break that advantage. If challengers and large customers cannot get a reliable public identity of their own, the new landing becomes a capacity expansion for the existing market structure rather than a competitive shock. The development headline says redundancy. The contract detail says dependence.
There is a second-order effect on investment. Investors considering a landing-adjacent data centre, cloud node or enterprise facility ask whether customers can migrate without a painful renumbering exercise. If every prospective tenant must negotiate address recognition through an incumbent, the facility's addressable market shrinks. If tenants can bring or obtain clean network identity, the same facility looks less like a captive annex to one carrier and more like a neutral platform. The registry does not finance the building, but it helps decide whether the building can host independent economic activity.
AFRINIC's policy environment cannot by itself create competition. It can, however, determine whether competition has the address layer it needs. A registry that treats legitimate portability as ordinary reduces the hidden switching costs that protect incumbents. A registry that turns every change into an uncertain negotiation strengthens the very concentration that cable policy is meant to loosen.
The visitor economy makes failures seasonal and public
Tourism gives island network risk a calendar. Demand rises with school holidays, airline schedules, conferences, cruise calls, sporting events, religious periods and high-value travel seasons. Hotels, resorts, booking platforms, card processors, travel agencies, airports, health-entry systems, ride services, ferry operators and customs desks become one visitor-facing network economy. A fault in a quiet month is a repair problem. The same fault during peak season is a national earnings problem.
The registry layer participates in this economy through trust signals. A visitor's bank, card issuer, travel platform, corporate VPN, device-security software or fraud-monitoring provider does not understand local telecom history. It sees addresses, names, contacts, geolocation hints, route evidence, abuse records and reputation signals. If an island network is forced into a hurried address migration, an unclear upstream arrangement or a poorly documented route change, the failure may show up as a blocked payment, suspicious login, delayed confirmation email, broken remote-work session or support escalation at a hotel desk.
IPv4 scarcity sharpens the problem. Many tourism systems still rely on IPv4 in production, even where operators are deploying IPv6. Payment terminals, reservation engines, property-management systems, vendor portals, remote-access appliances and foreign counterparties often operate on inherited assumptions. Clean IPv4 is therefore not a nostalgic preference. It is a working input during a long transition. In a peak week, an address plan that is "good enough eventually" may not be good enough at all.
The reputational effect is harsher for islands than for large countries. A visitor who loses connectivity, cannot pay, or cannot authenticate from a hotel may not blame a particular access provider. They may remember the destination. A port delay, airport-system fault or payments outage is easily translated into a story about national reliability. That translation may be unfair. It is nevertheless how tourism markets work, because visitors experience infrastructure as part of the product.
Registry predictability cannot sell a resort room. It can prevent avoidable friction from damaging the promise that a premium island destination must make: guests can arrive, pay, work, receive help, pass through border systems and communicate during a storm. When registry procedures are slow or unclear, island operators postpone changes until after the season. That may be prudent. It can also freeze a bad architecture for another year, keeping the next season exposed to a known weakness.
The better approach is to treat network identity as part of tourism continuity planning. Ministries and large hotel groups should know which addresses support booking, payments, security monitoring and guest connectivity; who controls them; whether they are portable; which upstreams can announce them; how reverse DNS is changed; and what evidence foreign counterparties will require during a migration. AFRINIC does not run the tourism economy, but its record quality helps determine whether the tourism economy can make those plans with confidence.
Ports, customs and state services price outages in days
Island economies often reveal infrastructure failure at a small number of public choke points. A port cannot clear containers by averaging network uptime over a quarter. A customs single window cannot ask importers to wait while a routing-authority dispute is explained. A tax deadline does not move because a public agency's address records are stale. A hospital, civil registry, court portal or benefits system may have only a few public endpoints, but those endpoints represent a large share of state capacity.
These systems depend on recognition chains. The service may be run by a ministry, hosted by a state-owned carrier, operated by a private data centre, protected by a managed-security firm and monitored by a foreign vendor. Its addresses may belong to a carrier, a public agency, an old government technology unit or a contractor. RDAP and Whois records indicate contacts. Reverse DNS connects addresses with expected names. Routing records and origin authorisations help counterparties decide whether an announcement should be accepted. Abuse contacts show where complaints will go. Each element looks modest. Together they form the envelope around public reachability.
In a large market, a government agency may have several domestic paths to fix this envelope. On an island, the agency may have fewer credible suppliers, fewer specialists with registry experience and a procurement culture that assumes the visible systems integrator owns the whole connectivity problem. The result is fragmented accountability. The ministry owns the policy service. A carrier owns the address space. A data-centre contractor owns hosting. A security company owns monitoring. A registry-facing account may sit with a person who has changed jobs.
That fragmentation becomes visible during an incident. Engineers know the technical answer but not the legal authority. Lawyers know the contract but not the route. Procurement officers know the supplier but not the registry account. The registry sees a member relationship but not the full public-service dependency behind it. Meanwhile citizens, importers, hospitals and tourists experience delay.
AFRINIC can reduce this cost by making public records role-aware and procedures legible. A useful registry file should make it easier to distinguish holder, operator, technical contact, abuse contact, reverse-DNS manager and routing authority when those roles differ. Disputes should be marked without destroying the last verified operational state. Legitimate corporate changes should not require a public agency to rediscover its own network identity from scratch. Emergency escalation should be narrow enough to help continuity without turning the registry into a general-purpose crisis ministry.
For governments, the lesson is direct. Critical digital services should not go live without an address-dependency map. The map should sit beside the cybersecurity plan, disaster-recovery plan and service-level agreement. It should answer who can change the registry record, who can announce the prefix from a backup path, who can update reverse DNS, what happens if the current provider is unavailable, and whether the procurement contract protects public reachability during a commercial dispute.
Satellite backup is expensive insurance, not a second internet
Satellite connectivity has improved. Lower-earth-orbit systems have changed expectations about latency, deployment speed and the usefulness of emergency backhaul. For islands, satellite can be an invaluable insurance policy when a submarine cable is cut, a landing station is impaired, a cyclone damages terrestrial links, or a remote site needs temporary service. It should not be mistaken for a complete substitute for the primary internet economy.
The first reason is cost. Even where equipment is easier to obtain and throughput is better than it was, satellite capacity is still priced like a scarce contingency product compared with high-volume submarine capacity. It may be enough to keep emergency communications, core government functions, basic banking, priority hotel operations or selected enterprise services alive. It is not usually a cheap way to carry an entire island's normal traffic profile, particularly during a high-demand event.
The second reason is operational selectivity. Someone must decide what moves to satellite when the cable fails. Emergency services? Government portals? Bank clearing? Airport systems? Hotel booking? Mobile core functions? Cloud access for the largest enterprises? Public Wi-Fi? Residential broadband? Those decisions become harder if the address plan is unclear. If critical services are buried behind provider-assigned addresses, carrier NAT, stale reverse DNS or uncertain route authority, moving selected services cleanly to a satellite path becomes slower and riskier.
The third reason is reputation. Satellite failover may change paths, latency, geolocation signals, security assumptions and traffic patterns. Counterparties may need to know that the traffic still belongs to the same island service. Banks and government systems may need allowlists updated. Remote vendors may see new source characteristics. Abuse and incident contacts must still work. The registry layer is not the satellite link, but it helps make the failover legible to the rest of the internet.
This is where registry certainty and disaster recovery meet. A cable-fault playbook should not merely say "use satellite". It should identify which prefixes move, which ASNs announce them, what route-origin material supports the change, who can publish or update records, what reverse-DNS changes are needed, which contacts will answer abuse or security queries, and how the island will return to normal service without leaving a trail of inconsistent records.
AFRINIC's contribution is to keep these mechanisms ordinary. In a real disaster, no island wants to negotiate first principles with its registry, upstreams and counterparties. It wants tested procedures, current records and reliable evidence. Satellite is the expensive insurance policy. Registry clarity is the policy wording.
That clarity also prevents over-promising. A minister can announce that satellite backup exists; a carrier can list the capacity in a resilience slide; a hotel group can tell shareholders that an emergency path has been procured. None of those statements answers which services will fit through the backup, how traffic will be prioritised, or whether foreign counterparties will keep trusting the changed path. A sober registry and address plan forces the hard conversation before the cable alarm sounds. It replaces a heroic rescue narrative with a rationing schedule that can be defended.
Small markets buy redundancy through procurement
Redundancy is often described as an engineering property. In small island markets it is also a procurement property. The technical ability to add a second provider matters less if contracts, tenders and evaluation criteria reward the cheapest bandwidth while ignoring address control. A service can appear redundant on paper while leaving the customer unable to change upstreams without renumbering, reputation repair, reverse-DNS work and a round of explanations to foreign counterparties.
Public procurement is especially exposed. A ministry may buy a platform, hosting contract or managed network service without specifying who controls the addresses, how changes are made, how route evidence is maintained or what happens to public reachability when the supplier changes. The resulting contract can be perfectly legal and operationally fragile. The state has bought a service but not the exit rights needed to keep the service alive under stress.
Private procurement has the same weakness. A resort group may choose a provider for price and coverage, only to discover later that payment systems, guest services and remote corporate access are entangled with that provider's address block. A bank may accept a data-centre migration plan without testing how quickly public endpoints can move to a second upstream. A shipping agent may rely on a managed service whose registry account is controlled by a reseller several commercial layers away. A small ISP may lease addresses without enough clarity over update rights, dispute procedures or continuity if the lessor's own standing changes.
The registry can help by making good procurement easier to verify. Clear public records reduce the need for every buyer to become a registry expert. Predictable transfer, leasing, reverse-DNS and contact-update procedures make contract clauses more credible. Published service expectations allow buyers to ask suppliers the right questions. Proportionate dispute markers help buyers understand risk without panicking over live service.
Insurance markets should care as well. Cyber and business-interruption policies often ask about backups, vendors, incident response and disaster recovery. They less often ask whether a critical service can preserve public reachability when a provider changes or a cable fails. In island economies, that omission is expensive. An insurer that understands address control can distinguish between a customer that owns a usable continuity option and one that has merely bought a second invoice. Better registry evidence can therefore lower due-diligence cost beyond the telecom sector.
Regulators and competition authorities should also widen their lens. They often examine retail prices, wholesale access, landing-station openness and dominant-carrier behaviour. They should ask whether address dependence is being used as an invisible switching cost. A supplier that offers low initial bandwidth while keeping a customer inside provider-bound numbering may be selling a cheap entry and an expensive exit. A supplier that supports portable identity, clean route authority and documented migration rights may be delivering more resilience even if its headline price is higher.
In small markets, the address plan should be scored like power resilience or cybersecurity. If a bidder cannot explain who controls public addresses, how backup announcements work, and how records are changed during provider transition, the bid contains a hidden continuity cost. AFRINIC cannot write every tender. It can maintain the public infrastructure that lets tenders test the right thing.
Address scarcity has a different arithmetic offshore
IPv4 exhaustion is a global fact, but its local arithmetic varies. AFRINIC's post-exhaustion regime means that new IPv4 allocations are constrained and justified through formal process. That rationing may be necessary. Scarcity, however, has different consequences in a small island market than in a large mainland economy with deeper address inventories and more specialised intermediaries.
An island operator often cannot solve scarcity by buying time. The business case may be tied to one cable landing, one hotel season, one public-service deadline, one donor-funded digitisation project or one data-centre tenancy. If the IPv4 plan is not settled when the project window opens, the opportunity may not return in the same form. A mainland provider may re-sequence work across regions. The island provider may have only one realistic window before equipment, visitors, budgets and contractors disperse.
Scarcity also favours incumbents. The operator with legacy IPv4 holdings can make commitments that a newer entrant cannot. It can offer business customers stable public endpoints, easier reverse-DNS handling and a simpler compliance story. A new entrant may have better radio design, better customer service or a more efficient operating model, yet still lose because its public-numbering position looks uncertain. The registry has not set out to protect the incumbent. Scarcity and delay can protect it anyway.
Leasing and transfers can soften the constraint, but only if the rights are clear. If a lessee cannot be confident about update authority, dispute handling, route evidence, reputation history and registry recognition, leased addresses come with a risk discount. The same is true for transfers. A small island operator considering a purchase needs to know not only the market price of the block but also whether the registry process is predictable enough for financing, due diligence and customer commitments. Uncertainty reduces liquidity and raises the effective price of expansion.
IPv6 adoption is essential, but it does not eliminate the short-term premium on clean IPv4. Public agencies, hotels, banks and foreign partners may still depend on IPv4 reachability. Many security and compliance processes remain tuned to IPv4 identities. Islands cannot simply declare the transition finished before their counterparties have done so. The realistic policy is dual: push IPv6 hard while treating clean IPv4 evidence as a scarce production input that must be allocated, transferred and documented without theatrics.
AFRINIC's best scarcity posture would be austere and predictable. Ask for evidence that supports uniqueness, need, contactability and responsible operation. Publish clear service expectations. Avoid discretionary rhetoric that turns rationing into moral judgment. Scarcity already raises the value of every decision. The registry should not add unnecessary ambiguity to the price.
Evidence is not paperwork when markets are thin
In registry discussions, evidence can sound like a burden: incorporation documents, utilisation data, contact updates, route descriptions, abuse contacts, customer plans, transfer papers and billing records. For island networks, evidence is also a market instrument. It is how an operator convinces a bank that a loan supports durable service. It is how a public buyer believes a provider can survive a cable fault. It is how an upstream accepts a route. It is how a hotel group, port authority or data-centre tenant avoids reintroducing itself to every foreign counterparty during a migration.
The evidence stack is layered. Public registration data says who holds the resource and who can be contacted. The ASN establishes a network identity. Reverse DNS helps link addresses with expected names. IRR records, where used, help filters and peer onboarding. RPKI can support route-origin decisions. Abuse contacts show where complaints should go. Reputation and geolocation systems are not registry services, but they often look to registry and routing evidence when deciding whether a change is legitimate.
None of these mechanisms is perfect. Each is part of a lower-cost explanation. That matters in a thin market because there are fewer specialists to translate ambiguity. A large multinational can hire registry counsel, routing consultants and reputation teams. A small island provider may have one senior engineer, one commercial manager and a relationship with an upstream. The registry record must carry more explanatory weight because the local market has fewer people available to supply missing context.
The history of alleged African IPv4 record abuse shows why evidence quality matters. Once IPv4 has a secondary-market value, the record is part of the asset. Manipulated or poorly governed records can create private gains and public uncertainty. Island networks may hold modest blocks in global terms, yet those blocks may support critical national services. The market value and the public-service value are not the same. A block that looks small on a broker's spreadsheet may be large in the life of a port, hospital or government platform.
Evidence must therefore be strong without becoming punitive. A registry should investigate fraud, stale records and false claims. It should not make ordinary business evolution feel like an accusation. Island networks change because hotels open, ports modernise, governments digitise forms, cables land, storms reveal weak points and satellite options improve. The evidence system must be able to record those changes calmly.
The useful test is whether the evidence requested improves the record. Who is responsible? What service depends on the resource? Which ASN will originate it? Which contacts will respond? How will abuse be handled? What backup path exists? Which customers are affected by a change? Those questions protect the public record. Broad inquiries that drift into business ideology or institutional defensiveness do not.
Institutional turbulence increases the cost of contingency
AFRINIC's institutional difficulties should not be used as a universal explanation for island network weakness. Islands have their own commercial constraints, engineering limits, regulatory choices and weather risks. But the registry's recent turbulence is too relevant to ignore. Allegations of record abuse, disputes over resource governance, litigation in Mauritius, periods of impaired governance, receivership, election controversy, board-restoration efforts and outside concern over registry continuity all shape how markets price reliance on AFRINIC-administered resources.
The effect is not always visible as a direct fee. It appears as caution. A lender asks for stronger evidence before financing a data centre. A public buyer chooses a larger incumbent because its address story seems easier to defend. A lessor adds conditions. An upstream asks more questions before accepting a route. A customer demands warranties over transferability and update rights. A project manager builds extra time into a migration. Each action is individually rational. Together they create a registry-certainty premium.
The premium is highest where local slack is lowest. A global carrier can keep teams in several regions, maintain multiple address positions and absorb a slow process. An island operator may not have that option. If a registry issue delays a project by a month, the project may miss a tourist season, a government budget year, a donor milestone, a vendor visit or a cable cutover. Institutional risk becomes embedded in ordinary timing.
Recovery language must therefore be judged by operational evidence. A restored board, a budget process, public statements about morale or strategy, and engagement with the wider internet community may all be useful signs. They do not by themselves reduce the island premium. Island operators need to see routine tickets handled predictably, records preserved during disputes, published service commitments met, emergency escalation understood, and registry services insulated from unrelated governance fights.
The most important distinction is between institutional survival and network continuity. Institutional survival means the registry continues as a corporate body. Network continuity means live networks, customers, public services and security records remain usable while disputes are resolved. Those goals overlap, but they are not identical. A registry that protects the last verified operational state during a dispute behaves like infrastructure. A registry that allows institutional conflict to cloud live resource status behaves like another hazard.
For island networks, the distinction is practical. They cannot afford to make every continuity plan depend on a favourable interpretation of a registry's latest governance episode. They need a record layer designed to continue calmly even when the institution around it is under strain.
Mauritius shows the legal vessel matters
Mauritius is not merely a headquarters line. It is the legal vessel through which AFRINIC operates. That matters because legal process can affect bank accounts, corporate authority, receivership, elections, contracts and the recognition of who may act for the organisation. When a regional technical function is housed in a domestic corporation, domestic law and regional continuity are linked whether policy communities like it or not.
This is not an argument against courts. A registry that controls valuable resources and membership relationships must be subject to law. Courts may be the only forum capable of dealing with corporate paralysis, contractual disputes, alleged misuse of authority or contested elections. The danger is not legal oversight. The danger is a continuity design so weak that ordinary legal process forces networks to wonder whether essential records and services will remain stable.
Island economies understand this pattern. A cable consortium may be governed by foreign contracts while landing in local territory. A tourism platform may process payments abroad while serving local hotels. A shipping system may depend on foreign banks, local customs and global software. A national data centre may host public services whose vendors sit in several jurisdictions. Legal and operational geographies rarely match cleanly.
AFRINIC's Mauritian history is a larger version of the same problem. Its records serve a region. Its company is local. Its recognition function is global. Its disputes may be heard in a particular court. Each layer is legitimate in its own domain, but continuity requires the layers to be designed for conflict. Courts should be able to act without unintentionally threatening live registry services. The registry should be able to respect legal orders without treating every dispute as a reason for operational uncertainty. Operators should know what remains stable while lawyers argue.
The practical response is not to demand that Mauritius surrender authority, nor to pretend AFRINIC is an ordinary company whose failure would have no systemic effect. It is to separate the continuity of data and services from the outcome of any one governance dispute. Versioned records, independent audit trails, clear authority succession, service failover, published emergency procedures and non-destructive dispute handling would reduce pressure on courts and markets alike.
Island networks should watch this carefully because their own dependencies often have the same shape. The contractor, cable owner, cloud region, satellite provider, port system and public agency may each sit in a different legal frame. Good continuity planning does not erase jurisdiction. It makes jurisdiction survivable.
Upstream concentration turns address control into market power
In an island market, upstream concentration can hide inside ordinary contracts. A customer may see two suppliers, two invoices and two router sessions. The underlying physical path, international ownership, landing-station access or address dependence may still leave the customer with one practical exit. Address control is one of the easiest forms of concentration to miss because it sits behind technical terminology and procurement shorthand.
If a carrier supplies both connectivity and the addresses, the customer may become dependent on that carrier for more than bandwidth. Changing provider can require renumbering, reputation repair, reverse-DNS changes, security reviews, customer notices and new allowlists. For a small enterprise this is irritating. For a port, bank, hospital, public agency, mobile operator or major resort group, it can be a board-level risk. The address block becomes part of the supplier's leverage.
This does not mean provider-assigned addressing is always wrong. It is normal and efficient for many retail services. The problem is silent misclassification. Critical services are sometimes treated like ordinary access customers because no one in procurement asked whether the address plan created lock-in. The supplier may not even need to behave badly. The architecture itself gives the supplier an advantage in every renewal and dispute.
Provider-independent resources and clean routing authority are therefore competition tools. They allow customers to take cable diversity seriously, move between upstreams, add disaster-recovery paths and test supplier claims. They do not remove all switching costs, but they make the cost visible and manageable. In small island markets, that visibility can discipline prices and improve resilience.
AFRINIC's role is to make legitimate independence possible without making it casual. The registry should require evidence, maintain accurate records and prevent fraudulent claims. But it should recognise that portability is not merely a technical convenience. It is a market-structure safeguard. A public agency or large economic node that cannot change upstreams without changing its public identity is not fully protected by a second cable, a second router or a second service-level agreement.
Competition authorities should incorporate this into market reviews. Landing-station access, wholesale pricing and retail concentration remain important. Address dependence should sit beside them. A provider that helps customers document portable identity, route authority and continuity rights may be contributing to market resilience. A provider that keeps critical customers in provider-bound numbering while advertising redundancy may be selling a partial truth.
Disaster recovery depends on boring records
Disaster-recovery plans often overvalue equipment and undervalue records. A router in a box, a satellite terminal in a storeroom, a second data-centre rack and a backup fibre path are useful only if the affected services can be made recognisable to the outside world. In an island incident, the question is not simply whether bits can leave the shore. It is whether the right bits can leave with an identity counterparties will accept.
The disaster plan should therefore begin with a service map. Which prefixes support emergency communications, hospitals, ports, customs, airports, government identity systems, banks, payment processors, mobile cores, hotel booking, education platforms and major data centres? Which are provider-assigned? Which are portable? Which ASNs originate them in normal conditions? Which upstreams can announce them in backup conditions? Who controls reverse DNS? Which route-origin material is published? Which contacts will answer abuse, security and operational questions during the event?
These questions are not glamorous. That is their virtue. In a crisis, glamour is useless. A boring registry record can be read by an upstream engineer, a security team, a public official, an auditor and a foreign counterparty without a conference call about institutional politics. It shows what is supposed to happen and who can authorise it.
AFRINIC can strengthen disaster recovery by making the boring path easy. Routine contacts should stay current. Reverse-DNS processes should be reliable. RDAP and Whois services should be resilient. IRR and RPKI publication should have continuity arrangements. Account authority should survive staff turnover and corporate change. Dispute handling should avoid collateral damage to live emergency services. Published escalation paths should distinguish between genuine incident response and ordinary queue-jumping.
Island governments and operators have equal duties. They should rehearse address failover, not just power failover. They should test whether a backup upstream will accept announcements, whether security vendors will tolerate changed paths, whether public records remain consistent, and whether the right people can update them under stress. A plan that depends on one engineer's memory or one provider's goodwill is not a plan. It is a hope with a diagram.
The discipline is modest but powerful: make the record as resilient as the route. An island that has spent millions on a cable, satellite contract or data-centre backup should not let the continuity plan fail because nobody can update a registry contact or prove authority over a prefix.
The registry should be narrow, portable and non-punitive
The best registry posture for island markets is not dramatic. It is narrow, portable and non-punitive. Narrow means the registry concentrates on uniqueness, accuracy, contactability, delegation, routing-support publication, transfer recognition and proportionate compliance. It does not try to become an economic planner, moral arbiter or substitute telecom regulator. The more functions it claims, the more expensive its uncertainty becomes.
Portable means the registry recognises the economic value of exit. Legitimate networks should be able to add suppliers, change providers, move data centres, reorganise corporate structures and support disaster recovery without unnecessary renumbering. Provider-bound space will remain common, but critical services should not be trapped in it by accident. Transfers, leases and changes in authority should be documented well enough that counterparties understand what has changed and what has not.
Non-punitive means disputes should be handled with precision. If fraud is suspected, investigate the relevant resource and authority. If payment is late, use remedies proportionate to the risk. If a transfer is disputed, mark the dispute and preserve the last verified operational state where possible. If contacts are stale, require correction without turning every dependent customer into collateral. In a small island market, destructive remedies can travel quickly from a registry action to innocent users because sectors share the same infrastructure.
This posture does not weaken the registry. It strengthens it. A registry that can explain its actions in narrow operational terms is easier to defend than one that uses broad discretion. It can say what record it protects, what evidence it needs, what service it preserves and why a remedy is proportionate. Operators may still disagree, but they can price the process. Markets can live with rules more easily than with institutional mood.
AFRINIC can still train operators, support IPv6 adoption, convene members and contribute to internet development. Those roles are useful when they strengthen the region's capacity. They become dangerous if used to justify unpredictable control over scarce resources. Island networks need development support, but they need restraint even more. The registry's greatest contribution to island economies may be to make itself less exciting.
That is not a small ambition. Boring infrastructure is difficult to build because it requires discipline during conflict. It requires the institution to preserve records when tempted to signal power, and to ask for evidence without turning evidence into theatre. For island networks, that discipline is worth more than a speech about stewardship.
What island operators should demand before the next fault
The next cable fault, cyclone, landing-station incident, routing error, procurement failure or institutional dispute will not wait for a perfect governance settlement. Island operators, governments and large customers should make a registry-layer checklist part of ordinary resilience planning now.
The first demand is an inventory. Identify the prefixes and ASNs that support critical economic and public services. Map them to ports, customs, airports, hospitals, government portals, payment systems, banks, tourism platforms, mobile networks, data centres and emergency communications. Record who holds each resource, who operates it, who can update contacts, who controls reverse DNS, which upstreams announce it, and what route-origin material exists.
The second demand is classification. Separate provider-assigned services from portable services. Provider-assigned addressing may be acceptable for many access customers. It should be a deliberate decision for critical services, not a hidden default. If a public or commercial system cannot tolerate renumbering during a provider change, its address plan should say so.
The third demand is tested authority. A registry account that no one has used in years is not a continuity asset. Operators should test whether the current authorised contacts can update records, whether corporate changes have been reflected, whether abuse contacts are monitored, whether reverse-DNS changes can be completed within a defined period, and whether backup announcements will be accepted by upstreams. Testing should occur before the storm, not during it.
The fourth demand is procurement language. Contracts for connectivity, hosting, cloud access, data centres, public platforms and disaster recovery should specify address control, update rights, cooperation during transition, route evidence, reverse-DNS obligations, dispute handling and customer continuity. A low price should be discounted if the supplier leaves the customer with an expensive exit. A higher price may be justified if it buys real portability.
The fifth demand is registry service assurance. AFRINIC should publish the operational information that island risk managers need: routine ticket categories and timing, emergency escalation paths, continuity arrangements for RDAP, Whois, reverse DNS, IRR and RPKI services, rules for preserving live resources during disputes, and guidance for corporate changes affecting critical services. Perfection is not required. Legibility is.
The final demand is cultural. Boards and ministers should stop treating addresses as a technical footnote. In island markets, addresses are part of continuity, competition and national reputation. A board that approves a data-centre move without seeing the address plan has not approved the whole risk. A ministry that tenders a digital service without asking who controls public reachability has left a gap in the contract.
The island premium should not be enlarged by the registry
Island economies already pay for geography. They pay through cable consortia, landing stations, maintenance zones, fuel, freight, import delays, limited labour pools, smaller customer bases, weather risk and the need to keep public services working with fewer fallback options. Some of that premium is unavoidable. The registry premium is not.
A clean registry record cannot stop a ship from cutting a cable. It cannot make satellite capacity cheap. It cannot guarantee that every hotel, port, ministry or carrier will procure wisely. It cannot remove the commercial power of incumbents or make IPv4 abundant again. But it can reduce the number of avoidable uncertainties that island networks must carry. It can make address identity portable where justified. It can keep public records current. It can make routing evidence easier to trust. It can preserve live reliance during disputes. It can publish service expectations so small markets can plan around them.
The opposite path is costly. If registry procedures are slow, opaque or politicised, islands respond rationally. They favour incumbents with older address holdings. They overbuy redundancy they do not fully control. They delay migrations. They pay lawyers and consultants to explain what a clear record should have explained. They accept supplier lock-in. They treat satellite as a magic escape when the address plan is not ready to move. They make digital public services depend on people and contracts whose authority has never been tested.
That is the economics of island network dependency. It is not a plea for exemptions. It is a demand for institutional modesty in a place where modest failures can have large effects. The sea already imposes a tax on capacity, labour, logistics and repair. AFRINIC's task is not to add a second tax through uncertainty.
For island networks, the highest form of regional stewardship is practical: keep the record clean, keep procedures predictable, make legitimate portability ordinary, preserve operational reliance during disputes, and let operators buy real redundancy instead of buying explanations. The ambition sounds small. In an island economy, it is the difference between a second route and a second dependency.