ARIN and the economics of island network dependency

Summary

• ARIN-region island networks face a distinctive dependency problem: a few cable landings, a small carrier set, exposed ports, storm-prone power and repair windows, and a customer base that cannot tolerate identity instability during outages.
• Registry evidence becomes valuable because upstream choice is thin. Whois/RDAP, RPKI, reverse DNS, transfer records and stable holder identity help carriers, enterprise customers, hospitals, ports and public agencies trust a network while its physical path changes.
• ARIN should be judged as a narrow ledger and evidence service, not as an island-development authority. Resilience comes from separable layers: cable path, upstream contract, power, routing security publication, public-number records, dispute isolation and portability.

The planning room before the storm

Imagine the room in late May, a week before the Atlantic season begins. It is not a distant boardroom where connectivity is treated as an abstract utility. It is a practical room: an island operator, a hospital administrator, a port technology lead, a public-safety office, a hotel group, a fuel supplier and a carrier account manager are comparing lists. The lists are not glamorous. Generator hours. Diesel delivery slots. Which landing station has a hardened feed. Which upstream contract allows emergency rerouting. Which cloud edge has enough local cache. Which enterprise customers need fixed public addresses for payment networks, remote maintenance, medical devices, customs data flows and guest Wi-Fi authentication. Which support desk can still prove who holds which Internet number resources if the main fibre path is down and customers are calling from satellite phones.

This is the economics of island dependency. It is not merely that islands are far from the continent. Distance matters, but the sharper issue is concentration. A continental operator often has several neighbouring metros, several long-haul paths, a broad wholesale market and a queue of construction crews within driving range. An island operator may have one or two international cable paths, a handful of landing sites, a small set of wholesale carriers, a port that can be closed by weather, and repair economics shaped by vessel availability rather than a truck roll. Every redundancy decision is therefore expensive. Every delay has a cash cost. Every numbering decision has a continuity cost.

ARIN sits in this story at a narrow but important layer. Its region page places the United States, Canada, many Caribbean islands and North Atlantic territories inside its service region. That list is not a development plan; it is a map of registration responsibility. Anguilla, Antigua and Barbuda, the Bahamas, Barbados, Bermuda, the Cayman Islands, Dominica, Grenada, Guadeloupe, Jamaica, Martinique, Montserrat, Puerto Rico, Saint Kitts and Nevis, Saint Lucia, Saint Martin, Saint Pierre and Miquelon, Saint Vincent and the Grenadines, Turks and Caicos, the British Virgin Islands, the U.S. Virgin Islands and other territories do not become easier to connect because a registry lists them. But when networks in those places need portable identity, clean records and recognized evidence during a crisis, the registry layer is part of the continuity stack.

The word "dependency" can sound pejorative. In island networking it is mostly descriptive. A hospital depends on the island operator. The operator depends on power, landing access, upstream carriers, spare routers, fuel and the ability to prove its network identity to counterparties. A port depends on customs links and shipping data. Hotels depend on booking and card networks. A government office depends on identity services, public alerts and cross-border communications. The island economy is therefore not just connected by broadband. It is tied to a small number of physical and contractual chokepoints.

The seasonality of that risk is plain. The National Hurricane Center's tropical cyclone climatology describes the Atlantic season as running from June 1 to November 30 and the Atlantic basin as including the Atlantic Ocean, the Caribbean Sea and the Gulf region; for the 1991-2020 climate period it gives an average season of 14 named storms, seven hurricanes and three major hurricanes. Those figures are factual exhibits, not a forecast of any one island's year. A single landfall, near-miss, shore-end cable fault or port closure can define the season for a small network. For an island operator, climatology is therefore a calendar for cash reserves, spares, maintenance freezes, satellite contracts, generator testing, customer notices and escalation paths.

The usual connectivity debate tends to ask whether an area has broadband. Island dependency asks a harder question: what remains legible when the primary path fails? If a provider moves traffic from one upstream to another, can customers still identify it? If a cable outage forces traffic over a constrained backup, can counterparties distinguish congestion from impersonation? If a transfer or merger changes the holder of number resources, can the records support trust rather than confusion? If a hotel chain, hospital or port authority is asked to accept a degraded service level for a week, can the operator show that the network has not become a ghost behind borrowed addresses and stale contact records?

This is where public-number evidence has economic value. Not because a registry confers resilience by decree, and not because a monopoly over uniqueness creates sovereignty over local development. The value lies in reliable coordination. Number uniqueness is a narrow coordination function. When the same address space cannot be held by two unrelated parties at once, when a holder record is public enough for counterparties, when routing security statements are linked to the actual holder, when reverse DNS is maintained, and when transfers are recorded through known procedures, the island operator can keep commercial identity steady while changing physical paths. That is not a cure for a broken cable. It is a way to reduce the secondary damage that comes when a broken cable also breaks trust.

Geography priced as concentration

Geography is often described in romantic language: the island as a gateway, a bridge, a maritime crossroads. The network economist sees something plainer. Geography is an input cost. Water creates distance. Distance raises build cost. Build cost reduces the number of independent paths. Fewer paths increase the bargaining power of the owners of those paths. The result is not simply higher prices; it is a different structure of risk.

The Caribbean and North Atlantic islands in the ARIN region are diverse. Puerto Rico is not Bermuda; Barbados is not Saint Pierre and Miquelon; Jamaica is not the Cayman Islands. Some have multiple international connections, data-centre projects, larger domestic markets and stronger carrier competition. Others have smaller populations, fewer landing options and a thinner wholesale market. Yet they share a family resemblance that matters for network dependency. The international link is often a submarine link. The landing point may be near coastal infrastructure exposed to wind, surge, flooding and power failure. The route away from the landing may pass through limited ducts, road corridors or facilities. The port and airport that bring in repair equipment may themselves be disrupted. A backup path may exist, but at a capacity and price that makes it a safety valve rather than a full substitute.

Submarine cable economics are lumpy. A cable is not bought in increments that match the first thousand customers on a small island. It requires planning, permits, marine survey, construction, landing agreements, maintenance arrangements and capital that must be recovered across traffic and time. For a large continental market, a new path can sell capacity into many carriers, cloud providers, content platforms and enterprises. For a smaller island, demand may be essential but not deep enough to justify many parallel facilities. This is the central paradox: the social value of redundancy may be high precisely where the private revenue base for redundancy is thin.

The same logic applies to landing diversity. A second cable landing on the same beach or into the same power-exposed district is not the same as a second path with separable risks. But true separation costs more. It may require another coastal site, another power feed, another terrestrial route, another maintenance arrangement and another set of local permits. Each layer of separation raises resilience and expense together. The island operator therefore faces an uncomfortable portfolio choice. Over-invest and customers complain about prices in normal years. Under-invest and customers discover, during a storm or cable fault, that "connected" meant connected through a narrow neck.

Counting cable names can therefore flatter the real position. Several cables may share a narrow coastal corridor, a common shore-end risk, the same landing building, the same duct leaving the beach, the same power substation or the same port logistics chain for repair. The expensive question is not how many lines appear on a map, but how many failures are actually independent. An island can look well connected in a regional cable diagram and still have a practical dependency on one fenced compound, one access road, one diesel supply route, one customs office or one marine repair contractor.

Repair economics make the distinction sharper. A terrestrial fibre cut in a continental city may be ugly but familiar: locate the break, dispatch crews, expose the duct, splice, test and restore. A submarine fault near an island adds vessel scheduling, fault localization, permits, weather windows, sea state, shore access, port clearance, spare cable, jointing expertise and the priority queue of other regional faults. A repair vessel cannot be in two places at once, and it cannot work safely in every sea condition. The waiting period is an economic period, not dead time: banks, ports, clinics, schools, customs platforms, hotels and public agencies are deciding which traffic matters most while the repair calendar is governed by weather and maritime logistics.

The constraint is not only physical. It is contractual. A small operator may buy international capacity from a larger carrier that controls or influences landing access. If the same carrier also competes downstream, the island operator's bargaining position is weakened. If the alternate upstream is technically available but commercially unattractive, redundancy exists on a diagram but not on the income statement. If transit diversity requires prepayment, minimum commits, cross-connect fees and equipment that sits idle for most of the year, the spare path becomes a standing option premium.

Economists are comfortable with options. A spare path has option value because it can be exercised when the primary path fails. But unlike a financial option, a network spare path must be maintained before it is exercised. It must be monitored, tested, secured, documented and integrated into routing practice. If it is left untouched until the storm arrives, it may fail at the moment it becomes valuable. Island dependency therefore converts redundancy from a capital question into an operational discipline.

This is why comparisons with continental connectivity are often misleading. A mainland business park may complain about carrier choice while still sitting near fibre owned by several providers. An island hospital may have nominal options but rely on one economically viable international path most of the time. A continental operator can sometimes buy emergency capacity from a neighbouring metro. An island operator may be waiting on a vessel, a customs clearance, a generator delivery or a weather window. Latency is not just a measurement in milliseconds; it is a market signal about where traffic must go before it returns. Repair time is not a support ticket; it is a constraint on the island's ability to trade, heal, govern and reassure visitors.

The public-number layer does not change this geography. ARIN cannot make a cable land in a different bay. It cannot force a carrier to price spare capacity cheaply. It cannot turn a satellite backup into fibre economics. What it can do is maintain the evidence that an island network is the recognized holder or authorized user of the number resources it announces, that contact and organization records are current, that routing security publication matches the intended origin, and that reverse DNS does not become another source of suspicion. In concentrated geographies, such evidence is more valuable because the market has fewer informal cross-checks. When everyone knows everyone, a phone call may work during normal times. During a regional outage, public evidence travels farther than personal trust.

Upstream concentration and the value of public evidence

Upstream concentration changes the meaning of trust. In a large market, if one carrier relation deteriorates, a network can often lean harder on another. The commercial relationship may be painful, but the network's public identity can remain stable across a wider set of counterparties. On an island, upstream concentration can make identity feel partly captive. The operator may fear that changing transit providers will require renumbering, customer disruption, routing delays, new filters, new reputation checks and new commercial due diligence. If public-number records are weak, portability becomes theoretical.

The trust problem appears in several forms. A hospital wants to know that the address space used for remote radiology, lab results, device management and patient portals is not a temporary workaround that will be withdrawn next week. A port wants customs and shipping partners to allow traffic after an emergency reroute. A hotel group wants card processors, booking platforms and fraud tools not to treat degraded island traffic as suspicious simply because routing changed. A government office wants emergency notices, domain services and authentication endpoints to remain reachable under a different path. In each case, the physical route may change, but the counterparty wants continuity of identity.

ARIN's Whois/RDAP material is useful here as a factual exhibit. It describes public access to registration data for Internet number resources, organizations, contacts and related records. The point for island networks is not that every field is perfect or that a registry page settles every dispute. It is that public, machine-readable, routinely used evidence reduces the cost of verification. When a customer, upstream, security team or enterprise buyer needs to check who is associated with a number resource, the registry record is a known place to look.

That evidentiary role matters more when upstream choice is thin. In a concentrated market, the incumbent carrier may be the easiest path for a customer to understand. The challenger operator must prove that it is not merely reselling an uncertain service with fragile identity. Clean registry records help the challenger tell a different story: here is the holder identity; here are the contacts; here is the routing security posture; here is the reverse DNS plan; here is the transfer history if addresses changed hands. The proof is not a marketing deck. It is public coordination evidence.

Emergency rerouting tests that evidence under pressure. If an island operator normally exits through one carrier but must shift a hospital, port or government service through another path, the commercial question and the trust question arrive together. Will the alternate upstream accept the announcements quickly? Will remote security teams understand the changed path? Will filters, reputation tools and customer allowlists treat the traffic as legitimate? Will the incumbent exploit confusion by presenting itself as the only safe identity? Clean registry evidence cannot force cooperation, but it reduces the room for a physical bottleneck to become an identity veto.

The value is also defensive. Island operators are often asked to explain incidents that have multiple causes: a cable fault, a power interruption, a transit filter, a cloud edge issue, a content cache miss, a damaged tower, a customs delay for replacement parts. In such conditions, customers may conflate every failure into a single suspicion about the operator. Public-number evidence helps separate layers. A routing change can be shown as a routing change, not as a change in resource holder. A temporary upstream path can be shown as temporary, not as a hidden sale of the network. A transfer can be shown as a recorded event under known policy, not as a private scramble.

The RPKI page provides another exhibit. It explains that resource holders can make cryptographically verifiable statements about which ASN should originate a prefix, and that operators can compare global routing announcements with validity data. In island settings, this is not a theoretical security nicety. A storm week is a bad time for a network to discover that its new emergency path is rejected because its routing security publication is stale or incomplete. The backup must be not only physically available but also believable to routing counterparties.

Reverse DNS has a similar, quieter role. ARIN's reverse DNS guidance notes the use of PTR records and the need for holders to maintain them. Reverse DNS is not the centre of resilience policy, but poor maintenance can cause practical frictions in mail handling, logging, troubleshooting and trust checks. When an island operator is already asking customers to accept congestion, rerouting or failover, avoidable reputation failures are costly. A small administrative weakness can become an economic penalty because the market has little patience during an outage.

The same point applies to transfers. ARIN's transfer page describes conditions for transfers due to merger, acquisition, reorganization, specified-recipient transfers within the region and inter-RIR transfers under compatible policy. Those procedures matter because IPv4 scarcity has turned address space into an asset with timing risk. If an island operator is acquired, spun out, recapitalized or forced to buy scarce IPv4 space, customers need continuity that survives the legal change. A transfer record cannot guarantee good service, but it can reduce ambiguity about who is responsible for the numbers after the corporate event.

In concentrated island markets, ambiguity is expensive. It raises the cost of insurance, procurement, interconnection and emergency coordination. It gives dominant upstreams an advantage because they can present themselves as the safer identity. It burdens smaller operators that may have technically sound networks but weaker administrative posture. Registry evidence therefore has distributional effects. It can lower entry and switching costs, not by subsidizing a competitor, but by making identity portable and verifiable.

That is the correct way to understand ARIN's importance in the island context. It is not a planner of island telecom markets. It is not a funder of submarine routes. It is not an arbiter of local carrier strategy. It is a public ledger for scarce, unique Internet number resources and related evidence. Where geography concentrates physical bargaining power, a clean ledger gives operators and customers a modest counterweight: the ability to separate who they are from which path they must use today.

Storm recovery turns registry timing into continuity

Storm recovery compresses time. In ordinary months, a network has the luxury of planned maintenance windows, staged customer notices, contract review and slow registry cleanup. During a storm, tasks collapse into hours. A primary path degrades. A landing station loses commercial power. A generator has fuel for a fixed period. A hotel group asks whether guest connectivity can be throttled to preserve emergency traffic. A hospital asks whether remote specialists will remain reachable. A port asks whether its customs platform can keep moving. A government office asks whether a public notice site can stay online. The operator is no longer selling bandwidth. It is allocating continuity.

Seasonal outlooks can be quiet in probability terms and still leave an island exposed to severe single-event risk. The practical point is that preparation must not wait for a named storm. For island networks this translates into registry timing. Records, contacts, routing security data, reverse DNS and transfer documentation must be ready before the event. They are not glamorous disaster supplies, but they shape the operator's ability to make credible claims when the event arrives.

The reason is that storm recovery is full of temporary substitutions. A cable path may be replaced by another cable path at lower capacity. Traffic may be shifted toward a different upstream. A cloud edge may absorb local demand while international bandwidth is rationed. Enterprise traffic may be prioritized above entertainment traffic. Satellite terminals may be deployed for hospitals, public-safety offices or remote communities. Mobile networks may rely on generators and microwave links. None of these substitutions erase the need for stable public identity. If anything, they increase it.

The practical test is whether the network can change path without changing face. During recovery, packets may leave by a different shore, a different carrier, a different satellite-backed tunnel or a temporary mainland hub. The customers, however, still need to see the same accountable operator behind the service. That is the registry-layer meaning of continuity: not a promise that the route is unchanged, but evidence that the holder identity, contacts, routing authorization and reverse-DNS posture still describe the operator responsible for the traffic. In a small island market, losing that evidence can be as damaging as losing bandwidth, because the next call from a bank, hospital vendor or logistics platform may be a refusal rather than a support request.

The most dangerous period is often not the first hour of failure, when everyone knows a storm is responsible. It is the third day, the tenth day or the second week, when customers are tired, enterprise counterparties are applying normal security policies, and remote support teams outside the island have lost patience with exceptions. At that point the operator must show that the changed path is still the same network in the relevant public sense. Whois/RDAP contact data must lead to the right people. RPKI statements must match the intended emergency origin. Reverse DNS must not advertise stale names that confuse logs and filters. Transfer or holder records must not leave ambiguity after a recent corporate transaction.

Registry continuity therefore means continuity of records, services, security publication and customer evidence. It does not mean institutional immunity. A registry can fail operationally, a record can be wrong, a holder can neglect maintenance, and a policy process can be slow. The point is more modest: when the registry layer works well, it prevents the outage from becoming an identity crisis. It gives the island operator a stable reference while the physical layer changes.

This is also why the quality of contact records has economic value. In normal times, stale contacts may appear to be housekeeping. During recovery, they are an impediment. An upstream attempting to validate an emergency change may reach an old technical contact. A customer may escalate to an address no longer monitored. A security team may hold traffic because a record does not match the claimed operator. A transfer that was not completed cleanly may cause a delay precisely when delay is most expensive. The administrative layer becomes a bottleneck because every other layer is already stressed.

Island dependency also exposes a public-private timing mismatch. Public agencies often plan around emergency response phases. Carriers plan around contracts, maintenance and service levels. Customers plan around business continuity. Cable owners plan around vessel availability and repair priority. The registry layer has its own timing: account authority, record updates, transfer review, routing security publication, reverse DNS changes. If those timelines are ignored until the storm, they collide. Good resilience practice brings them forward into the pre-season checklist.

The registry layer is not the only pre-season item, and it should not be elevated above fuel, spares or landing security. But it is often cheaper to fix early than late. Updating contacts costs less than buying emergency satellite capacity. Cleaning routing security publication costs less than losing an enterprise customer to distrust. Documenting holder identity costs less than a procurement dispute after a merger. Making portability real costs less than accepting a poor upstream contract because renumbering would be too disruptive.

Storm recovery also shows why portability is a safety valve rather than a political slogan. A portable number resource does not free an island operator from cable dependency. It does not guarantee cheaper transit. It does not force another carrier to sell capacity. But it changes bargaining. If the operator can move upstream without renumbering major customers, the incumbent's leverage is reduced. If the operator can preserve customer identity across failover, the spare path has more value. If the operator can prove its holder status after a transaction, financing and emergency procurement become easier.

The opposite condition is lock-in by administrative friction. A network may be technically capable of changing upstreams but practically trapped because renumbering hospitals, port platforms, government services, hotel payment gateways and enterprise VPNs would be too disruptive. The incumbent then benefits not only from physical concentration but from identity concentration. The policy aim should be to avoid that compounding effect. Island operators cannot remove water from the map, but they can reduce the number of ways a physical dependency becomes a contractual and identity dependency.

Spare-path economics: options, not miracles

The phrase "redundant connectivity" is often too easy. It implies that a second path solves the problem. In island economics a spare path is more like insurance with deductibles, exclusions and capacity limits. It is valuable, but only if its limits are understood.

Start with the second cable. A genuinely independent cable path can transform the risk profile of an island. It can reduce outage probability, improve bargaining power, support data-centre investment and give enterprise customers confidence. But the second cable has to be independent in the right ways. A second strand on a related route may not be enough. A second landing in the same exposed zone may share power and flood risk. A second wholesale provider that ultimately relies on the same upstream group may not create commercial independence. Resilience is not counted by diagrams; it is measured by correlated failure.

Then consider alternate landing and terrestrial diversity. The landing station is a critical point because it joins the marine and land portions of the network. If the landing is hardened but the inland route is exposed, the path remains fragile. If the inland route is diverse but both paths depend on a single power feed, the improvement is partial. If the port needed for repair equipment is closed, even a well-designed site may wait. Each added layer of separability reduces correlated risk but adds capital, permitting, maintenance and coordination cost.

Microwave and fixed wireless links can help across smaller distances or between neighbouring islands. They can provide local resilience, connect secondary facilities and support emergency traffic where terrain and weather allow. Yet they are not a universal substitute for high-capacity international fibre. Towers need power. Antennas face wind loading. Spectrum and line-of-sight constraints matter. Storms that damage cable landing infrastructure may also damage towers and roads. The economic role of wireless redundancy is therefore targeted: it can preserve specific flows, not erase island dependency.

Cloud edges, caches and local interconnection points are another tool. If popular content, public information, software updates, local media and government pages can be served locally, international capacity pressure falls during an outage. A local exchange can keep domestic traffic on-island rather than sending it abroad and back. Caches can make degraded international connectivity feel less severe for some use cases. But local content does not replace the need for global reach. Hospitals, ports, banks, hotels, universities and government offices depend on counterparties outside the island. Caching buys breathing room; it does not replace public identity or international paths.

The island-specific value of a cache or exchange is not that it makes the island self-sufficient. It changes the shape of the emergency queue. If software updates, streaming demand, local media, government advisories and school platforms can be served locally, scarce international capacity can be rationed toward clinical traffic, customs links, payment authorization, hotel operations, relief coordination and carrier command channels. A modest local IXP can also make domestic traffic less absurd: two institutions in the same town should not have to traverse a distant mainland path merely to exchange local packets during a cable incident. The hard part is governance and maintenance. Caches need power, refresh policy, content-provider relationships and security. IXPs need neutral rules, participants, equipment spares and routing discipline. They are resilience institutions, not decorative racks.

The Nautilus research paper on submarine cable and IP-link mapping notes that submarine cables are a backbone of the Internet and that failures are hard to repair in remote ocean settings. That framing is useful for island operators because it captures the cross-layer nature of the problem. A cable fault is physical, but its consequences appear in routing, application reachability, customer trust, commercial contracts and public confidence. The more concentrated the island's options, the more likely a physical event becomes an economic event.

Spare capacity also has a moral hazard problem. Customers rarely want to pay for capacity that is idle in normal times. Regulators and public agencies may demand resilience after a disaster but resist tariffs that fund resilience before one. Operators may understate risk to win customers or overstate resilience to satisfy procurement language. The result is a market in which true spare-path economics are hidden. Everyone wants redundancy; few want to carry its full cost.

This is where public evidence can discipline claims. An operator that says it has independent upstreams should be able to show routing practice, not merely a sales statement. An operator that says it can fail over should be able to show tested procedures, current RPKI publication and contact records. A customer that demands resilience should be able to specify which services must retain public identity, which can tolerate private addressing, which can be cached, which can move to satellite, and which must fail closed. The registry layer does not certify the whole design, but it gives several pieces of objective evidence.

The option value of a spare path is also shaped by IPv4 scarcity. A backup provider may be able to carry traffic, but customer equipment, remote firewalls, payment processors and legacy applications may be tied to known public IPv4 addresses. If failover requires address changes, the spare path is less valuable. If addresses are portable and routing security publication is prepared, failover becomes more credible. If the island operator relies on upstream-assigned space and cannot take it elsewhere, the spare path may become a partial backup for generic browsing but not for critical enterprise services.

Thus the economics of redundancy and the governance of number resources meet in a practical place. The question is not whether ARIN should make island networks resilient. It cannot. The question is whether ARIN's policies and services allow operators to convert scarce public-number resources into portable, verifiable continuity. In island settings, that conversion has unusual value because every other spare option is expensive.

The mistake would be to treat spare paths as a binary metric. A better approach is to rank them by service class. Emergency command traffic, hospital traffic, port traffic, public information pages, payment traffic, hotel guest traffic, school traffic, entertainment traffic and bulk software updates do not need identical treatment. Some need stable public IPv4. Some can use IPv6. Some can tolerate CGNAT. Some can be cached. Some can pause. Some can move to satellite. Resilience planning is the art of matching the economic value of each flow to the right continuity layer.

Satellite backup is insurance, not substitution

Satellite connectivity has changed the island resilience conversation. Older satellite service was often too expensive, high-latency and limited for broad substitution. Newer low-earth-orbit options have improved latency and made emergency deployment more practical. For a clinic, emergency office, remote settlement, vessel, school or temporary operations centre, satellite can be the difference between isolation and partial connectivity. But the central economic point remains: satellite is backup, not a substitute for submarine capacity and public-number continuity.

The distinction matters because satellite can be oversold in public debate. After a storm, a visible terminal on a roof is easy to understand. It gives officials and customers a symbol of recovery. It can restore messaging, video calls, logistics dashboards, emergency forms and limited office work. It can support first responders and temporary command posts. But the aggregate demand of an island economy is larger and more varied. Hotels, hospitals, banks, ports, schools, call centres, media outlets, government offices, cloud services and households all compete for capacity. Satellite can prioritize essential flows; it cannot cheaply absorb every normal broadband habit at fibre scale.

Satellite also has its own dependencies. Terminals need power, safe mounting, clear sky view, account management, traffic policy and local distribution. Weather can affect performance. Regulatory approval and import logistics may matter. If terminals are not pre-positioned, they must arrive through the same disrupted ports and airports as other emergency equipment. If the local network behind the terminal is poorly designed, capacity is wasted. If critical services depend on fixed public IPv4 addresses, satellite access may require careful tunnelling or other arrangements to preserve identity. The terminal is therefore one component of a resilience design, not the design itself.

Low-earth-orbit service improves latency, but it does not repeal arithmetic. Backhaul from a terminal to the rest of the island still needs power, Wi-Fi, fibre, Ethernet, mounting hardware, local routing and someone with authority to decide which packets deserve the scarce path. Contention can rise when a whole region is damaged. Public IPv4 presentation may differ from the ordinary cable path. Geolocation, fraud scoring and remote-access controls can behave oddly. A satellite link that works well for a command post may be the wrong answer for a whole resort district, an overloaded emergency shelter or a port community trying to move every commercial document at once.

For island operators, the best use of satellite is often selective. A hospital may need a satellite-backed path for clinical coordination and administrative access, not for every guest device in the building. A port may need shipping documents, customs data and vessel coordination, not recreational traffic. A public-safety office may need messaging, maps and incident management, not unrestricted streaming. A hotel may need card processing, reservations and emergency communication before it needs full guest bandwidth. A public agency may need alert publication, relief registration and cross-border coordination. Each use case has different address, latency, throughput and security needs.

This is why public-number planning remains relevant even when satellite is available. Critical services may be tied to known addresses allowlisted by external parties. A sudden move to a satellite path can break assumptions about origination, geolocation, reverse DNS, fraud scoring and routing security. Some services can be placed behind tunnels or application gateways. Others require direct reachability. Still others should have been designed with domain-based access rather than address-based trust. The registry layer will not solve these architectural choices, but clean records reduce confusion when a temporary path is used.

Satellite also changes bargaining in a subtle way. Even if it cannot replace submarine capacity, it can reduce the absolute power of a single cable-dependent upstream for the most critical flows. A hospital with a tested satellite fallback is less hostage to the first hours of a cable outage. A public agency with pre-arranged terminals can keep core coordination alive. A small operator with satellite for command-and-control can manage its network even when customer traffic is degraded. That safety valve can improve negotiations, but only if it is real before the outage.

The economics are again about option value. A satellite subscription, pre-positioned terminal, mounting kit, power plan and tested traffic policy cost money in normal times. The value appears during rare but severe events. If procurement rules judge connectivity only by normal-month cost per megabit, satellite looks expensive. If they value avoided hospital disruption, port delay, tourism reputational damage and emergency coordination failure, it looks different. Island dependency requires the second lens.

Yet satellite backup can also create false comfort. If policymakers believe that a few terminals remove the need for cable diversity, landing hardening, local caching, power investment and public-number portability, the island becomes more fragile. The correct conclusion is layered resilience. Satellite belongs in the layer for emergency reachability and selected continuity. It does not belong in the layer for bulk international capacity. It does not erase the need for cable repair arrangements. It does not make IPv4 scarcity irrelevant. It does not replace RPKI, reverse DNS, Whois/RDAP accuracy or transfer records.

There is also a fairness problem. Satellite backup is often easiest for well-funded institutions: private hospitals, major hotels, ports, banks, large government offices and international aid groups. Smaller clinics, local media, schools and small businesses may be left with degraded connectivity. If the public-number and routing layer is neglected, the best-prepared actors can build private workarounds while the wider market remains confused. A registry-led continuity baseline helps reduce that divide by making basic identity evidence available to all operators, not just to the best-funded buyers of backup technology.

The lesson is not anti-satellite. It is pro-realism. Satellite is a powerful resilience instrument when matched to the right traffic, powered correctly, pre-positioned, tested and integrated with identity planning. It is an expensive disappointment when treated as magic. Island networks need both humility and precision: humility about what any single technology can do, precision about which flow needs which path and which public identity.

IPv4 scarcity and the timing of identity

IPv4 scarcity has a special meaning in island networks because it turns time into strategy. ARIN's IPv4 addressing options page records that its free pool was depleted on September 24, 2015, and points networks toward reserved cases, the waiting list, transfers and IPv6 adoption. For a large buyer, scarcity may be a procurement problem. For an island operator, it is also a continuity problem.

The reason is compatibility. Many critical island services still encounter counterparties that expect IPv4. Hospitals use vendors with legacy remote-access assumptions. Ports connect to shipping, customs and logistics platforms that may not be fully IPv6-ready. Hotels deal with booking platforms, payment processors, fraud tools, guest-device behaviour and support vendors. Government offices interact with foreign agencies, contractors and document portals. Small businesses use point-of-sale providers and cloud dashboards built for the lowest common denominator. In such settings, IPv4 is not a sign of technical virtue; it is a compatibility tax.

Scarcity raises the cost of that tax. A new island provider cannot simply ask for large fresh IPv4 holdings and build a clean numbering plan. It may need transfers, upstream-assigned addresses, CGNAT, IPv6, careful prioritization and customer education. Each option has trade-offs. Transfers require money and process. Upstream-assigned addresses reduce portability. CGNAT can complicate logging, abuse handling, inbound reachability and some applications. IPv6 is the right long-term direction but does not eliminate all IPv4 dependencies. Scarcity therefore becomes a timing constraint: the operator must decide which customers receive scarce public IPv4, when to seek transfers, when to renumber, and how to preserve identity during growth or recovery.

Timing matters most around corporate events and outages. Suppose an island operator buys another network, merges with a local provider or takes investment that changes control. If the address resources are not transferred and documented cleanly, customers inherit uncertainty. Suppose the operator needs more IPv4 space for a hospital, port and hotel segment but waits until storm season to secure it. The market may not cooperate. Suppose the operator relies heavily on upstream-assigned space and then tries to change transit providers after a pricing dispute. Renumbering critical customers may make the switch commercially impossible. In each case, scarcity turns identity into a planning variable.

ARIN's transfer procedures matter for this reason. They do not create abundant IPv4. They define how scarce resources can move under known rules. The transfer page describes specified-recipient transfers and inter-RIR transfers under compatible needs-based policy, as well as requirements around current registered holders, disputes, minimum transfer size and documentation. For island networks, the value is not merely legal compliance. It is the ability to show customers and counterparties that a scarce identity asset has moved in a recognized way.

That evidence also affects financing. An investor evaluating an island operator will look at cables, contracts, churn, revenue, equipment and regulation. It should also look at number-resource posture. Are IPv4 resources held directly or assigned by upstreams? Are contacts current? Are transfers complete? Is RPKI publication aligned with intended routing? Are reverse DNS delegations maintained? Are critical customers dependent on addresses that cannot move? A weak answer can reduce valuation because it increases the cost of changing upstreams, serving enterprise customers and surviving a crisis.

Scarcity also affects competition. If a dominant carrier controls the most usable public IPv4 inventory directly or through customer assignments, smaller operators may face higher switching costs. If enterprises cannot move without renumbering, they may stay with the incumbent even when service is poor. If a new entrant must rely on CGNAT for most customers while the incumbent can offer clean public IPv4 to lucrative enterprise accounts, market power deepens. A registry cannot and should not allocate addresses as an industrial-policy tool. But it can maintain procedures that make transfers, holder clarity and portability credible.

IPv4 scarcity should therefore be discussed without nostalgia. The goal is not to preserve IPv4 as a permanent privilege. The goal is to manage the transition while real dependencies remain. In island economies, a failed compatibility assumption can have direct consequences: a clinic cannot reach a vendor; a port cannot clear a shipment; a hotel cannot process a booking; a public agency cannot update an emergency page; a maintenance contractor cannot reach a device. These are not ideological objections to IPv6. They are reminders that transition costs fall unevenly.

The cleanest policy line is this: IPv4 scarcity should not become identity captivity. If scarce addresses are needed for critical compatibility, their registration and routing evidence should support portability where policy allows. If addresses move through transfer, records should be timely and clear. If addresses are upstream-assigned, customers should understand the lock-in. If CGNAT is used, logging and customer expectations should be explicit. If IPv6 is deployed, it should be treated as production infrastructure, not a checkbox. The island operator's task is to build a mixed reality honestly.

That mixed reality is why public evidence has more value in island markets than its administrative appearance suggests. A public IPv4 address tied to a hospital service is not just a technical label. It is part of a web of firewall rules, vendor contracts, remote access, reputation, DNS, logs and customer expectations. Changing it in calm weather can be difficult. Changing it during storm recovery can be dangerous. Registry continuity does not freeze the address forever, but it gives the operator a way to manage change without making every counterparty start from zero.

IPv6, CGNAT and the stubborn compatibility edge

IPv6 and CGNAT are practical tools, not moral categories. Island networks need both, and they need to be honest about what each can and cannot do. IPv6 expands address availability and should carry as much native traffic as possible. CGNAT allows many customers to share scarce IPv4 addresses and can be essential for consumer broadband economics. But neither tool removes the compatibility edge where critical services still depend on stable public IPv4 or on counterparties that treat address identity as part of trust.

For residential browsing, CGNAT may be acceptable most of the time. Many applications are outbound, encrypted and tolerant of shared public addresses. For a hotel guest watching video or checking email, the user may not know whether CGNAT is present. For some gaming, remote work, camera, VPN and small-business use, it can be more intrusive. For abuse handling and law-enforcement requests, shared addressing requires careful logging and time correlation. During a storm, when logs may be needed to distinguish abuse from recovery traffic, poor CGNAT practice can raise operational risk.

For institutions, the trade-offs are sharper. A hospital may need inbound connections from vendors or fixed tunnels to partners. A port may have equipment and logistics connections designed around fixed public IPv4. A government office may have foreign-agency allowlists that change slowly. A hotel group may have payment and security vendors that expect known addresses. A bank or insurer may reject traffic from unexpected shared pools. These are not always best practices, but they are real. An island operator that ignores them will discover that technical elegance does not clear the queue at customs or keep a clinic's vendor link alive.

IPv6 reduces this pressure when counterparties support it. It permits cleaner addressing, simpler end-to-end reachability and less dependence on scarce IPv4 pools. It also helps island operators avoid building a future entirely around borrowed scarcity. But deployment is uneven across customers, vendors and application providers. An island operator cannot mandate that every foreign platform, payment provider, remote-support vendor and government partner be IPv6-ready by the next storm. The operator must therefore run a dual reality: push IPv6 where possible, preserve IPv4 continuity where necessary, and avoid letting either side become an excuse for weak planning.

This dual reality has procurement implications. A public agency buying connectivity should ask not only for bandwidth but for address posture. Which services will receive public IPv4? Which will use IPv6? Which will sit behind CGNAT? How are logs retained? How are emergency changes handled? What happens if the upstream changes? Can critical addresses be routed over a backup path? Are RPKI statements prepared for that path? Are reverse DNS delegations maintained? Who has authority to update records if the technical lead is unreachable after a storm?

Enterprise customers should ask similar questions. Many island businesses treat Internet access as a commodity until an outage reveals hidden dependencies. A hotel may discover that guest Wi-Fi is less important than payment authorization and property-management access. A clinic may discover that cloud applications work but a device vendor's remote access fails. A port may discover that one logistics partner can tolerate a changed path while another cannot. These are not bandwidth problems alone. They are identity, addressing and compatibility problems.

CGNAT also affects market transparency. A provider with limited IPv4 may rely heavily on shared addressing while advertising speeds comparable to a provider with more public IPv4. For many customers the difference is invisible. For critical users it is not. If procurement does not ask about it, the cheaper offer may win until the first crisis. The better market outcome is not to demonize CGNAT but to label it. Residential service can use it. Critical service may need exceptions. Small businesses should know what they are buying. Emergency offices should not learn their address posture during landfall week.

The registry layer helps by giving direct holders a clearer way to manage scarce public resources and associated records. It does not solve every downstream assignment. It does not know every hotel firewall or clinic vendor. But it supports a chain of accountability. The direct holder can keep records current, publish routing security data, maintain reverse DNS and document transfers. Customers can ask whether their provider holds resources directly, uses upstream-assigned space or relies on shared addressing. Counterparties can verify public claims. The market can price identity rather than hiding it.

There is a temptation to treat IPv6 as a reason to stop caring about IPv4 portability. That would be premature in island settings. The correct approach is transition with evidence. Publish IPv6. Use it seriously. Move traffic where counterparties allow. Reduce unnecessary dependence on IPv4. But for the services that still require IPv4, maintain clear records and avoid captive arrangements where a physical upstream also controls identity. Scarcity should be managed down, not ignored.

The stubborn compatibility edge is where policy debates become operational. It is easy to declare that the future is IPv6. It is harder to keep a port open after a storm when one legacy platform remains IPv4-only. It is easy to say that CGNAT is efficient. It is harder to explain to a small business why inbound access broke. It is easy to say that registry data is administrative. It is harder to regain trust when a customer cannot tell whether an emergency reroute is legitimate. Island dependency lives in these hard edges.

ARIN's proper role: ledger, not development gatekeeper

ARIN's importance in island dependency should be framed narrowly. A registry that controls uniqueness over Internet number resources has a duty to be accurate, available, predictable and fair within its remit. It does not thereby become a sovereign planner of island connectivity. Monopoly over uniqueness creates duty, not sovereignty.

Heng Lu's public notes on number-resource portability, decentralising Internet governance, protecting the registry ledger rather than the gatekeeper and structural risk in the registry layer are useful doctrine context for this distinction. Read in the island setting, they point to a simple institutional rule: continuity should attach to the ledger, evidence and portability of number resources, not to the comfort of any single registry institution, incumbent carrier or preferred physical path. The notes should not be treated as substitutes for operational evidence. They are a reminder that network identity becomes fragile when the same gatekeeper controls both scarcity and the proof needed to escape scarcity.

The distinction is important because islands need development, but not every important institution should become a development gatekeeper. Cable diversity may require public finance, regional cooperation, anchor customers, multilateral funding, private capital and telecom regulation. Power resilience may require grid investment, generator policy, fuel logistics and building codes. Port repair may require maritime planning. Satellite backup may require procurement and licensing. Local interconnection may require carrier cooperation. ARIN's layer is different: it is the public-number ledger, related registry services and the policy framework for number resources.

This narrow role should be defended because it makes accountability clearer. If an island lacks a second cable, the registry should not be blamed for failing to build it. If a carrier charges too much for transit, the registry should not be treated as a tariff regulator. If a government has weak emergency procurement, the registry cannot fix it. But if holder records are hard to update, if transfer procedures create avoidable uncertainty, if service availability is poor, if routing security tools are unclear, or if portability is undermined by administrative friction, the registry layer is accountable.

The ledger role has several components. First is holder clarity: the public should be able to determine which organization is associated with a number resource, subject to appropriate privacy boundaries for individuals. Second is contact utility: the records should support operational escalation, not merely legal formality. Third is routing security linkage: resource holders should have practical ways to publish cryptographically verifiable origin information. Fourth is reverse DNS authority: delegations should be manageable and maintained. Fifth is transfer legibility: scarce resources should move through known procedures that customers and counterparties can understand. Sixth is service continuity: the registry's own public services must be resilient enough that island operators are not left without evidence during a regional incident.

This is a high bar, but it is not an imperial one. It treats ARIN as part of the Internet's coordination machinery, not as a development ministry. The political economy is healthier when each layer is judged by its actual function. Cable operators should be judged on route diversity, maintenance and pricing. Carriers should be judged on interconnection and failover. Public agencies should be judged on preparedness and procurement. Enterprises should be judged on continuity planning. ARIN should be judged on whether the number-resource evidence remains accurate, portable and usable.

The same narrowness protects island autonomy. If ARIN were framed as an island-development gatekeeper, it would invite claims that a Virginia-based registry should approve or prioritize local carrier strategy. That would be wrong. Island markets have their own regulators, governments, operators, customers and political constraints. Registry policy should not decide whether a territory needs a new cable, which carrier should land it, or how a public agency should buy satellite backup. It should ensure that number uniqueness and evidence do not become avoidable obstacles to resilience.

The ledger role also avoids a common misunderstanding about registry continuity. Continuity is not institutional self-protection. It is not a claim that the registry can never fail, should never be criticized, or deserves immunity because the Internet needs unique numbers. Continuity means that records, services, security publication and customer evidence must remain dependable enough that network operators can preserve identity across physical and commercial disruption. The beneficiary is not the institution as such; it is the customers and counterparties that rely on stable number-resource evidence.

In island settings, this view has a practical policy consequence: ARIN should make portability and record hygiene easy before crisis. Operators should not need heroic manual intervention to keep identity stable when switching upstreams, completing a transfer or preparing emergency routing. Documentation should be clear. APIs and account authority should be manageable. Support should understand that small island operators may have thin staffing and high storm exposure. Security should be strict without being brittle. A locked account during storm recovery can be as damaging as a stale record.

This does not mean giving islands a separate entitlement to scarce IPv4. Scarcity is real, and the registry cannot create what does not exist. It means recognizing that scarcity interacts with dependency. Where policy allows transfers, the process should be transparent. Where waiting lists exist, expectations should be realistic. Where IPv6 adoption is encouraged, support should be practical. Where routing security is promoted, tools should be usable by smaller operators. Where reverse DNS is delegated, shared authority and maintenance obligations should be understood.

The phrase "ledger, not gatekeeper" therefore has two sides. It limits ARIN's power and sharpens ARIN's duties. It says that ARIN should not decide island development. It also says that ARIN cannot dismiss registry-layer friction as minor paperwork. In an island economy, paperwork can become downtime, and downtime can become lost trade, delayed care and weakened trust.

Policy conclusion: resilience through separable layers

The policy answer to island network dependency is not a single institution, technology or funding line. It is separability. A resilient island network separates the layers that fail, bargain and recover on different timelines. Cable path should be separable from upstream relationship. Upstream relationship should be separable from public-number identity. Power should be separable from the landing building's weakest feed. Emergency traffic should be separable from entertainment traffic. Routing security publication should be separable from a single staff member's laptop. Registry evidence should be separable from local commercial disputes. Portability should be separable from incumbent goodwill.

This layered view changes what should be measured. Instead of asking only how many megabits are sold, policymakers should ask how many independent international paths exist, whether landings share power and flood risk, whether backup capacity is contractual or merely aspirational, whether critical institutions have tested failover, whether local caching reduces international load, whether satellite terminals are pre-positioned, whether public IPv4 dependencies are documented, whether IPv6 is actually used, whether CGNAT is labelled, whether RPKI is current, whether reverse DNS is maintained and whether Whois/RDAP contacts reach the right people.

The measures should be service-specific. A hospital does not need the same continuity profile as a beach bar. A port does not need the same profile as a streaming household. A hotel does not need all traffic protected equally. A public-safety office may need low bandwidth but high reliability. A school may tolerate a pause in nonessential traffic but not lose access to core learning platforms for weeks. The island economy contains many continuity tiers. Treating them as one broadband number hides the real dependency structure.

Public agencies can help by procuring continuity rather than slogans. Contracts should define which services must remain reachable under cable failure, power failure, upstream change and storm recovery. They should require tested failover, not just named backup. They should ask how addresses are held, whether critical services depend on upstream-assigned space, how emergency routing changes are authorized, and how registry records are maintained. They should fund pre-season exercises. They should not demand resilience while refusing to pay for idle options.

Operators can help by making identity planning part of customer design. When onboarding hospitals, ports, hotels, banks and public agencies, they should document which services require stable public IPv4, which can use IPv6, which can sit behind CGNAT, which can fail over to satellite, and which need domain-level rather than address-level trust. They should maintain contact records, routing security publication and reverse DNS as resilience assets. They should test backup paths before storm season. They should disclose meaningful limits rather than selling redundancy as magic.

Regional cooperation can help because islands are not isolated in economic life even when they are physically separated by water. Neighbouring territories can share lessons on cable landing regulation, satellite procurement, emergency mutual aid, caching, local exchange development and public-number hygiene. Larger markets in the region can act as recovery hubs, but the relationship should not become dependence by another name. A hub is useful when it provides optionality. It is dangerous when it becomes the only place traffic, evidence or support can go.

Investors and insurers should price registry posture explicitly. An island operator with portable resources, current records, clean routing security publication, documented transfer history and tested failover has lower continuity risk than one with opaque upstream assignments and stale contacts. That difference should affect valuation, loan terms and insurance analysis. The market already prices physical assets. It should price identity assets as well.

ARIN's contribution should remain focused. It should keep the registry layer clear, usable and resilient. It should support small operators in maintaining records and adopting routing security. It should make transfer and account authority procedures predictable. It should avoid becoming an arbiter of island development, while accepting that its own layer has material consequences for island continuity. A clean ledger is not a luxury in concentrated geographies. It is part of the evidence that lets networks move without disappearing.

The last test is practical rather than rhetorical. It arrives at the landing station, the fuel dock, the customs desk, the hotel front office, the clinic's vendor portal and the routing table at the same time. Water remains water. Cables remain expensive. Storms remain seasonal. Repair vessels remain finite. Satellite remains bounded. IPv4 remains scarce. Upstream concentration remains a bargaining reality. But the damage can be reduced if dependencies do not compound. A broken cable should not also break identity. A carrier dispute should not also trap customers in address captivity. A storm reroute should not also trigger avoidable distrust. A transfer should not also create confusion about who is responsible for critical numbers.

For ARIN-region islands and offshore territories, resilience is therefore a portfolio of separations that must be bought, tested and evidenced before it is needed. Separate path from identity. Separate evidence from marketing. Separate backup from substitution. Separate registry duty from development power. Separate scarcity management from lock-in. That portfolio is not cheap, but the alternative is to let geography decide too much. In island network economics, the most valuable infrastructure is often the piece that prevents one failure from becoming many.