Summary

  • The post-exhaustion generation is the cohort of networks that could not obtain an ordinary, scalable initial IPv4 allocation from a free regional pool. Small final-pool or reclaimed allocations do not recreate the abundance-era compact.
  • Regional stories differ, but the direction is common: ARIN's free pool reached zero in 2015, the RIPE NCC ended ordinary /22 distribution in 2019, LACNIC reached exhaustion in 2020, and APNIC limits new and existing members to a small amount of final pool before directing larger needs to transfers.
  • Entry now requires a portfolio of substitutes: purchased or leased IPv4, provider-assigned space, CGNAT, IPv6, cloud capacity, and waiting-list uncertainty. Each adds financial costs, dependency, complexity, reduced portability, or customer friction that incumbents can often treat as an optimisation rather than an entry condition.
  • The RIPE NCC's own analysis of the waiting list showed both genuine use by new entrants and the limits of account-based statistics. By 2023, it had issued 5,572 /24 reclaimed allocations, while a substantial number of recipients or waiting list members also used transfers or multiple accounts.
  • Formal equality is not enough when inherited address stock, policy expertise, voting participation, and transaction familiarity are distributed by date of entry. Legitimacy requires evidence on who bears each rule, not a presumption that every member had the same historical opportunity.
  • Reforms should protect predictable transfer recognition, readable lease authority, portable records, first-time entrant guidance, cohort impact analysis, and meaningful representation without confiscating valid holdings of incumbents or reviving discretionary allocation from an empty pool.
  • A 2015-2027 cohort study should combine allocation and transfer records with carefully sampled entrant interviews, comparing launch cost, time to independent routing, address dependency, funding, customer constraints, and participation in registry decisions.

The title describes a cohort, not all late-founded companies

"Never received an initial allocation" requires precision because registry terminology did not stop at exhaustion. The RIPE NCC calls its reclaimed /24 a waiting-list allocation. APNIC still allows a maximum /23 from the final pool under current policy. ARIN maintains limited reserved categories and a waiting list fed by returns or revocations. A recently founded company can acquire a larger portfolio by transfer, merger, or corporate reorganisation.

The relevant distinction is between an ordinary allocation path and a residual or market path. Before exhaustion, an eligible operator could demonstrate projected need under regional policy and receive a block sized according to the prevailing allocation rules, then request more if justified. The registry controlled supply and charged membership or administrative service fees rather than a negotiated scarcity price for addresses.

After exhaustion, a first block may be a small ration designed as a foothold, not a growth portfolio. Anything larger comes from another holder, a lessor, an upstream provider, or an acquisition. The registry retains important recognition and verification functions, but it does not supply the economic quantity the operator needs from a common inventory.

This creates a cohort whose defining experience is market dependency from the start. Some cohort members now have substantial addresses. Wealth or a successful acquisition does not change how they entered. Others started just before an official exhaustion date but entered under severe final-pool rationing and immediately depended on transfers. The boundary must therefore be tested using actual resource history rather than the year of founding alone.

A useful classification would identify the date of the organisation's first independent numerical resources, the source and size of its first IPv4 block, whether that block was a normal free-pool allocation, a final-pool ration, a waiting-list recovery, a transfer, a merger, a lease, or an upstream assignment, and when it first needed more. The cohort is an empirical group, not a political label.

This definition also avoids exaggeration. The post-exhaustion generation has not been excluded from the Internet. It built networks using available options. The question is whether the institutions that govern these options have adapted their legitimacy claims to the new compact.

September 2015 is a useful milestone, not a global midnight

ARIN's announcement on 24 September 2015 provides a clear starting point for the requested period. It indicated that the last addresses from ARIN's IPv4 free pool had been assigned. Future approved requests would depend on the waiting list for unmet requests or on the transfer market, while reclaimed and returned space would feed the list.

For an entrant in the ARIN region, the announcement changed the outside option. Before exhaustion, a qualified request could result in supply by the registry, subject to gradual restrictions and availability. After, approval did not mean the requested addresses existed in inventory. A transfer required a source holder and a transaction. A waiting-list position created uncertain delay. Reserved categories applied only for specific purposes.

The global story is staggered. APNIC activated final /8 rationing earlier and now states that new and existing members may receive at most a /23 from that pool, directing larger needs toward transfers. The RIPE NCC distributed a /22 equivalent under its final /8 policy until 25 November 2019, then directed new entrants toward a reclaimed /24 via a waiting list. LACNIC's final exhaustion phase ended in 2020 after a rationing period for new members. AFRINIC's inventory and institutional conditions followed a different path.

These dates mean that "post-exhaustion" must be regional and functional. A provider founded in one region in 2016 may have had no ordinary ARIN pool, while an operator elsewhere still had access to a small regional ration. Neither had the same path as a 2005 operator whose first allocation could support launch and expansion with significantly lower scarcity.

The milestone still has analytical value because 2015 begins a decade in which market dependency became normal in several regions. By 2020, three large service regions had officially exhausted ordinary available supply, and APNIC had reduced its final-pool maximum. Operators born between 2015 and 2027 have learned scarcity as a permanent design condition rather than an approaching event.

The institutional debate often treats this as a technical timeline. It is also a constitutional timeline. The date at which the registry can no longer repeat its founding distributive act divides members between those who experienced allocation and those who experienced recognition after private acquisition. This difference should appear in governance evidence.

An initial allocation brought more than addresses

The value of an initial allocation in the abundance era was not limited to the number of addresses. It provided a set of institutional benefits.

First, it offered a low-cost path to independent addressing space. The operator paid fees and bore compliance costs but did not have to finance the scarcity value of each address through a seller. Capital could be used for fibre, radios, routers, personnel, customer acquisition, and resilience. Address requests were part of operational training rather than a separate asset transaction.

Second, it established portability. Provider-independent or registry-assigned space, subject to regional policy, made it easier to change upstream connectivity and maintain an independent routing identity. Provider-assigned space could still be useful, but it tied renumbering and continuity more tightly to the upstream relationship.

Third, the allocation produced organisational evidence. A registry decision, a public record, account history, and recognised contacts helped the operator prove it was a real network. This evidence could be presented to transit providers, peers, customers, auditors, and investors. A new network did not need to explain a chain involving a seller, a lessor, or prior reputation history before establishing basic authority.

Fourth, the allocation created an expectation of growth. Additional requests were not guaranteed and conservation rules tightened. Yet the ordinary purpose of the institution included evaluating future needs and providing eligible growth as long as inventory permitted. The operator learned to plan around policy and utilisation thresholds rather than a private market price and seller availability.

Fifth, the allocation shaped policy identity. Members could regard themselves as entities in a common resource-distribution system. Debates over conservation, aggregation, minimum sizes, and needs testing affected access to the shared inventory. The legitimacy story was reciprocal: the community accepted constraints because an institution applied them to a pool held for regional distribution.

After exhaustion, most of this bundle comes apart. Recognition remains public, but supply comes from a private holder. Portability depends on the acquired or leased arrangement. Growth is priced by a market. Membership may still be required for reception or registrations, but joining does not replicate the historical opportunity. A post-exhaustion institution that speaks as if the old bundle remained intact misunderstands the new member's experience.

The abundance generation did not receive a moral entitlement

Recognising a historical advantage does not imply that earlier holders acted wrongfully. Networks requested addresses under the policy of their time, built services, and incurred costs. Many used their space efficiently. Some returned or transferred surplus. A block allocated decades ago may now support critical infrastructure, millions of customers, or a complex internal design that cannot be changed cheaply.

Moreover, not all early entrants received a large or easy allocation. Needs testing, slow administration, regional differences, and changing minimum sizes mattered. Small networks could struggle even during nominal abundance. Some incumbents later bought additional space at market prices and deployed expensive address-sharing systems. Date of entry is an exposure variable, not a complete moral biography.

The error would be to convert historical receipt into a permanent presumption of institutional neutrality. Earlier members received an option that later members cannot exercise: the opportunity to establish an address domain through ordinary distribution from the free pool. This option may have produced no surplus for some and enormous strategic value for others. It still existed.

Valid holdings should not be confiscated simply to create symmetry. Forced redistribution would disrupt operations, weaken trust in registry records, and create discretionary policy power over which uses deserve continuity. A legitimacy reform that starts with arbitrary taking would reproduce the control problem it claims to solve.

The appropriate response is to make the current market and registry environment contestable. Transfers must be predictable. Lease roles must be legible. First-time entrants should not face unnecessary evidentiary hurdles designed around established accounts. Fees should pay for services rather than hide scarcity rents. Governance bodies must measure who bears proposed rules. Portability and appeal should limit institutional dependency.

The abundance generation can support these reforms without abandoning its legitimate resources. Indeed, incumbents benefit from reliable transfers, clean records, safer leasing, lower litigation risk, and credible new counterparties. The problem is not intergenerational punishment. It is whether the institution can adapt after its original distribution function has become historical.

Entry now requires a portfolio, not a single request

The RIPE NCC's 2024 report on obtaining addresses captured the multiplicity of current choices. Informed by transaction data, market prices, and interviews with ISPs, mobile operators, transfer brokers, officials, and network consultants, it identified IPv6 allocation or lease, provider-independent IPv6 assignment, an IPv4 waiting-list request, IPv4 lease, IPv4 transfer, address optimisation, and CGNAT among the practical paths.

This list is useful precisely because no single item replaces the old initial allocation. A new ISP may request an autonomous system number and an IPv6 allocation, lease a small IPv4 block for launch, use CGNAT for residential customers, buy space after funding, maintain provider-assigned addresses for part of the network, and stay on a waiting list for reclaimed space. Each choice solves a different part of the problem.

The portfolio has sequencing risk. A lease may support launch but may not satisfy a lender seeking long-term control. A transfer purchase offers durability but consumes capital before customer revenues. Provider-assigned space reduces entry cost but increases dependency on change. A waiting-list block may arrive after the network has already built around another arrangement. CGNAT supports scale but alters support and logging burdens. IPv6 is strategically necessary but does not eliminate the demand for accessibility to IPv4-only services and users.

The entrant must coordinate these components while building the network. The incumbent can often progressively optimise an existing address domain. It can reclaim underutilised blocks, add translation, selectively buy, or use its history to negotiate. The entrant faces the portfolio before having operational scale, transaction experience, or a customer base.

This is the defining institutional difference. An allocation-era applicant was primarily concerned with whether it was eligible under policy and how much inventory the registry could assign. A post-exhaustion entrant asks how to finance, combine, verify, and exit multiple dependencies. Registry rules still count, but they fit into a broader market design that the registry does not fully control.

Policy must be evaluated against this portfolio, not against an imaginary applicant awaiting a single decision. A modest rule in isolation can become burdensome when it adds to transfer diligence, lease evidence, routing authorisation, geolocation correction, reputation remediation, and customer assurance. Cohort analysis makes the cumulative cost visible.

The waiting list is a foothold and a data set

The RIPE NCC waiting list demonstrates both the value and the limit of residual allocation. Each eligible LIR that has not yet received an IPv4 allocation from the RIPE NCC can request a /24 from reclaimed space. A /24 can support infrastructure, translation gateways, business services, or an initial independent routing presence. It is not trivial.

It is also only 256 addresses and arrives on an uncertain schedule. In March 2026, the RIPE NCC reported 794 LIRs on the list and a 477-day wait for the first LIR in the queue. During the same month, it reported over 1.5 million IPv4 addresses transferred in February. The contrast is instructive: reclaimed rationing provides small blocks slowly while the transfer system moves much larger quantities required by active markets.

The RIPE NCC's 2023 analysis provides more detail. As of 7 July of that year, it had made 5,572 /24 allocations via the list, totalling 1,426,432 addresses, and 1,025 LIRs were waiting. Among single‑LIR members already allocated, 1,596 appeared to depend solely on the waiting‑list block and were routing it. 499 others had also received transfers. Among 749 single‑LIR members still waiting, 29% had already received at least one transfer.

These figures reject two simplistic stories. The list is not meaningless; many recipients route the block and seem to depend on it. The list is not a complete entry system; many entities acquire transferred space before or after the ration. The registry account is also an imperfect proxy for an independent business because members may hold multiple LIRs and historical rules allowed strategic account structures.

The success of the waiting list must therefore be measured by recipient outcomes, not by the number of allocations alone. Did the block help a genuinely new network start independently? How long before it needed a transfer or lease? Did it remain routed? Did the company survive? Did it deploy IPv6? Was the block transferred after a holding period or membership closure? None of these outcomes proves intent by itself, but together they show whether the ration supported entry.

The queue is a public option on the margin. It can reduce early dependency for some networks. It cannot justify governing the entire post-exhaustion generation as if a /24 recreated the ordinary initial allocation.

Transfers are now part of the entry constitution

The transfer market is sometimes described as an external commercial layer that begins after registry policy. For post-exhaustion entrants, it is rather a constitutional entry path. The ability to find a seller, agree on terms, verify authority, satisfy registry conditions, and receive a clean record determines whether a network can obtain durable, large-scale IPv4.

Empirical research on reported transfers found that markets were indeed moving resources to use. A 2017 study of ARIN, APNIC, and RIPE NCC records reported that transferred blocks tended to be routed after transfer, utilisation increased, and 63% of transferred space in its dataset came from historical holdings. It found no evidence that buyers were broadly hoarding acquired space. It also found concentration among large entities and noted that public transfer records omit part of the economic picture.

This omission matters more for first-time entrants. Registry logs typically show address ranges, dates, and organisations, but not the price, financing, broker fees, failed transactions, reputation remediation, or contractual risk. A transfer may be publicly complete while the entrant paid a premium for a small block, accepted onerous closing conditions, or delayed launch.

The OECD had warned before the market matured that acquiring scarce addresses would add transaction costs for new entrants and make experimentation more expensive. This prediction should now be tested with current cohort evidence rather than repeated as theory. The cost includes more than the price per address. Legal review, registry familiarity, diligence on prior use, sanctions checks, routing history, geolocation correction, blacklist remediation, financing, and timing all count.

Predictable registry regulation can reduce these costs without controlling the private price. The institution must publish evidence requirements, processing time percentiles, common reasons for delay, appeal outcomes, and aggregated failure categories. It must separate fraud protection from discretionary judgement about the buyer's business model. It must coordinate dependent services such as reverse DNS and routing authorisation at effective time.

A transfer market with reliable regulation is not a betrayal of stewardship. After free-pool exhaustion, it is one of the main ways stewardship remains relevant for entry. The registry's legitimacy depends on administering recognition narrowly enough that newcomers can use the market without becoming supplicants to opaque discretionary power.

Leasing reduces the cost of capital while adding dependency

Leasing can be rational for a new operator. It aligns expenditure with revenue, avoids a large initial purchase, allows a business to test demand, and provides addresses faster than a queue. A lessor with available space can utilise it while retaining recorded holding. In illiquid or fast-moving markets, leasing may be the only practical bridge.

The bridge creates divided control. The recorded holder may control records and RPKI authorisation while the lessee operates the route and serves customers. The lessor may need to approve origin changes, reverse DNS, or abuse contacts. A sublease or intermediary can add another party. If the contract ends, the lessee may have to renumber customers or acquire replacement space quickly.

Reputation follows the addresses. A block may carry geolocation errors, a blacklist history, or prior abuse associations. The lessee may suffer customer complaints while lacking full authority to correct every public record. Conversely, misconduct by a lessee can damage the holder's portfolio and make future use harder. Both parties need clear evidence, monitoring, and termination rules.

Leasing also affects independence. A new ISP that cannot carry its address capacity to another infrastructure provider may be less able to change transit, hosting, or registry support. If the lessor is linked to an upstream provider or a competitor, the commercial dependency may exceed the lease price. A customer may see a working service without knowing how much continuity depends on renewal by another holder.

The response is not to ban leasing or claim that every operational use must involve a full transfer. It is to make authority legible. Public records should distinguish the recorded holder from validated operational contacts where policy allows. Routing authorisation should reflect the actual origin. Abuse accountability must be reachable. Contracts should provide notice, cure, data export, and transition periods proportional to the customer's dependency.

Post-exhaustion legitimacy requires taking leasing seriously because entrants use it seriously. An institution that recognises only the holder's identity while ignoring stable operational roles leaves the new generation to prove its legitimacy through private documents that outsiders cannot easily assess.

CGNAT converts scarcity into an operational burden

Carrier-grade network address translation (CGNAT) allows many customers to share a smaller public IPv4 pool. It is an essential response to scarcity and can be well designed. It is not a free substitute for addresses.

The operator must purchase and maintain translation capacity, design redundancy, manage port allocation, retain logs in accordance with applicable law, respond to abuse reports, diagnose applications, and offer exceptions for customers needing inbound accessibility. Support teams must distinguish home-router translation from provider translation. Law enforcement requests may require time, address, and port correlation. Outages can affect many subscribers sharing the same public infrastructure.

Customers experience the burden unevenly. Ordinary web browsing may work without visible difference. Gaming, peer-to-peer applications, self-hosting, certain VPN uses, and inbound services may be constrained. Shared reputation can mean one user's behaviour affects others. Business customers may demand a dedicated public address, pushing the operator toward a purchased or leased offering.

For an incumbent operator, CGNAT can be a tool to extend a large existing domain. For a new ISP, it can be part of the minimum viable architecture before the first customer arrives. The same technology therefore has a different impact. One network chooses an efficiency layer; another cannot launch without it.

This does not make address-sharing policy illegitimate. It means institutional claims must count the cost. A small waiting-list allocation that supports a translation platform may enable thousands of customers, but that result depends on capital equipment, operational skill, and acceptance of reduced end-to-end accessibility. Reporting only the number of allocated addresses underestimates what the entrant provided itself.

Registry and policy communities should hear these operational experiences when evaluating minimum allocation sizes, lease evidence, abuse contacts, and IPv6 transition. Those closest to scarcity are not just ration applicants. They are designers of infrastructure that converts scarcity into usable service.

IPv6 is the destination, not a response to historical impact

IPv6 provides the scalable addressing space needed for future growth. New networks should deploy it from the start when equipment, upstream providers, and customers allow. Registries can allocate IPv6 under current policy without recreating the IPv4 scarcity market. A post-exhaustion entrant may therefore be better positioned architecturally than an incumbent carrying older IPv4 assumptions.

This advantage does not erase the transition burden. A new ISP still serves users who need access to IPv4 destinations. Enterprise customers may require public IPv4. Devices and applications vary. Peering and content availability differ by region. The operator may need dual-stack, translation, or purchased IPv4 even when most internal planning is IPv6-oriented.

The burden is asymmetric because the incumbent already owns the bridge. It can deploy IPv6 while continuing to use a historical IPv4 domain. The entrant must deploy IPv6 and finance or design the IPv4 bridge simultaneously. Telling both to 'adopt IPv6' gives correct strategic advice while ignoring different starting positions.

IPv6 should not be used to dismiss demands for fair transfer, leasing, or waiting-list administration. Better IPv4 market rules do not harm IPv6; they make transition less dependent on incumbent favour. Scarce IPv4 should also not become an excuse to postpone IPv6 indefinitely. Purchased addresses are a transition input, not a source of renewed abundance.

The menu of the RIPE NCC's 2024 report is useful because it places IPv6 and IPv4 options together. The choice is not binary. A credible entrant can obtain a portable IPv6 allocation, use limited IPv4 leasing, run CGNAT, and buy a block later. Governance must support this combination while keeping the long-term direction clear.

The legitimacy question is therefore not whether post-exhaustion entrants deserve IPv4 over IPv6. It is whether institutions recognise that entrants bear the transition cost in a way that abundance-era members did not bear at their formation. A fair institution can strongly promote IPv6 while measuring this difference.

Formal membership equality can preserve historical inequality

Regional registries and their communities often rely on open participation, member voting, elected councils, and public policy discussions. These mechanisms are valuable. They do not automatically equalise cohort influence.

Abundance-era members have longer institutional memory. Their staff may know policy terminology, mailing list practices, election history, and registry contacts. Large incumbents can assign specialists to meetings. Their address domains make changes to transfers, fees, leasing, and certification materially important, giving them reason and ability to participate.

New entrants are busy building networks. They may join primarily to obtain an autonomous system number, IPv6 resources, a waiting-list slot, or transfer recognition. The registry is one dependency among others. A founder managing capital, upstream providers, customer installation, and scarce IPv4 may not spend days tracking a policy debate whose vocabulary assumes prior experience.

Rules of 'one member, one vote' or 'one account, one voice' may still be formally neutral. The asymmetry appears before the vote: agenda setting, proposal drafting, participation in consultations, candidate recruitment, and ability to interpret effects. Multiple accounts can still complicate the denominator when organisational identity and account identity diverge.

This does not mean new entrants always share a single viewpoint. Some favour liberal transfers; others worry about price and concentration. Some lease; others buy. Some are IPv6-first; others serve markets with strong IPv4 demand. A cohort seat should not become an instruction to vote as a bloc.

The institutional obligation is to make absence visible. Policy impact notes should identify how a proposal affects organisations without historical IPv4 allocation, first-transfer recipients, lessees, waiting-list members, and small regional networks. Consultations should actively sample these groups. Councils should receive evidence even when participation is low. Representation is not achieved by declaring the door open.

Legitimacy after exhaustion depends on the institution's ability to hear members whose first encounter is a market gate rather than a distribution advantage. If their only recognised role is to comply with rules written around prior holdings, formal equality becomes inherited governance.

A generational compact should not become discretionary preference

Once cohort inequality is recognised, the temptation is to create special allocations or subjective preferences for 'deserving' newcomers. This would be dangerous. An empty pool cannot sustain broad promises, and discretionary selection would encourage lobbying, identity gaming, and unstable expectations.

The better compact concerns institutional conduct. New entrants should receive predictable and reviewable recognition for legitimate transfers. They should have clear paths to document leases and operational authority. Waiting-list rules should honestly state quantity and uncertainty. First-time entrants should receive guidance that does not assume institutional memory. Fees should match services. Appeals should be accessible before a transaction or network launch collapses.

Policy changes must include cohort impact. A holding period may deter quick resale but also lock a new business into a mistaken purchase. A minimum block size may support routing aggregation but increase entry capital. A needs test may limit speculation but require forecasting from a business without operational history. Fee scales based on address quantity may charge historical holdings more, while flat fees may weigh more heavily on a small entrant. There is no neutral design without distributive effect.

The compact also protects incumbents. Record integrity, anti-fraud checks, and dispute preservation remain necessary. A newcomer should not bypass evidence simply because it entered late. Transfers should not erase legitimate restrictions. Leasing should not hide abuse accountability. Portability should not create duplicate authority.

The difference is that controls must be tied to the risk they address. Newness itself is not evidence of fraud. The absence of historical use is inevitable for a business that has not yet launched. A market purchase is no less legitimate because an earlier member received space from a pool. A forecast can be tested proportionally without requiring the entrant to mimic an incumbent's history.

This approach replaces preference with contestability. The institution does not choose winners. It removes avoidable advantages built into opaque regulation, inaccessible information, and non-portable records. Scarcity remains, but scarcity is not allowed to justify every inherited rule.

The missing evidence is a longitudinal cohort of entrants

Allocation and transfer data reveal movement but not the full experience of entry. A serious 2015-2027 study should build a longitudinal cohort across regions.

The sample should start with organisations that obtained their first autonomous system number, IPv6 allocation, small final-pool block, waiting-list block, or inbound IPv4 transfer during the period. Business identity must be carefully resolved so that additional accounts and subsidiaries are not automatically counted as independent entrants. A comparison group should include similar organisations that entered before regional exhaustion.

Registry records can show dates, resource types, allocation sources, transfers, mergers, account continuity, and some public contacts. BGP observations can show whether and when the network originated routes, with the usual collector limitations. RPKI records can show authorisation. Business registers and public service evidence can help distinguish operational networks from dormant accounts.

Interviews are necessary because many decisive costs are private. Entrants should be sampled by region, size, access technology, business model, and survival outcome. Questions should cover launch plans, address source, quoted and final acquisition cost, legal and broker fees, lease terms, reputation remediation, waiting-list assumptions, CGNAT design, IPv6 readiness, funding, procurement, customer demands, registry support, and participation in decisions.

The interview method must avoid survival bias. Failed or delayed entrants are harder to find but may reveal the greatest burden. Brokers, upstream providers, investors, public buyers, and consultants can identify abandoned transactions, while confidentiality protections can encourage candour. Institutional staff should be interviewed separately and their claims compared with user experience.

Results should not imply causation from entry date alone. Access to capital, national regulation, backhaul cost, market size, technology, and management quality all affect survival. The study should use matched comparisons when possible and publish uncertainty. Its purpose is to determine whether post-exhaustion status adds a measurable burden after accounting for obvious differences.

The RIPE NCC's 2024 market report based on interviews is a useful foundation because it confirms that operators use a portfolio of acquisition and optimisation methods. The next step is to follow entrants over time and link these choices to institutional outcomes.

Legitimacy needs cohort metrics, not just anecdotes

A governance dashboard for the post-exhaustion generation should publish a small set of sustainable metrics.

The first is time to independent routing: from organisational formation or first resource request to visible origination of portable address space. The second is source of first IPv4 capacity: ordinary allocation, residual ration, transfer, lease, upstream assignment, or merger. The third is effective address cost, including transaction and remediation expenses when entities consent to anonymous reporting.

The fourth is dependency. How many entrants depend on a single lessor, a single upstream provider, or a single provider-controlled address arrangement? How much notice do they have before loss of capacity? The fifth is transition architecture: IPv6 origination, customer adoption when available, use of translation, and demand for dedicated IPv4.

The sixth is registry friction: processing time, repeated evidence requests, withdrawals, refusals, appeals, and cancellations for first-time recipients compared with experienced entities. The seventh is market quality: block size distribution, concentration, interregional compatibility, prior use issues, and small-buyer access.

The eighth is representation: response to consultations, meeting attendance, candidacy, and voting by resource-history cohort without exposing individual votes. The ninth is business outcome. Did the network launch, grow, remain independent, merge, close, or become permanently provider-dependent? The registry should not grade business success, but cohort-level outcomes may reveal whether its rules add avoidable barriers.

The tenth is customer consequence: procurement exclusions, static address demand, support burden, and service limitations reported via structured surveys.

No single metric proves legitimacy. A high number of transfers can coexist with poor small-buyer access. A short registry processing time may follow months of private search. An IPv6 route can coexist with high IPv4 dependency. A waiting-list allocation can be valuable but too small for growth.

Published together, the metrics would replace mythology with impact. Abundance-era members could see which advantages remain historical and which services they now share with entrants. New operators could assess realistic paths. Policymakers could test proposals against observed cohorts rather than the loudest case.

The NRS should start from post-exhaustion reality

The Number Resource Society (NRS) can make a positive contribution if it designs membership for networks that never received the old compact. It should not promise a new free pool or imply that recognition creates addresses. It should make scarcity less institutionally extractive.

First, the NRS should recognise multiple legitimate entry paths. A member may own transferred space, lease capacity, use provider-assigned addresses, operate IPv6-first, or combine them. Rights and evidence should match the role without pretending every arrangement is identical.

Second, portable records should reduce dependency on change. Identity evidence, authorisation history, uncontested status, and service data should be transferable between qualified providers under clear conditions. A member should not lose its continuity simply because it changes registry support, lessor, infrastructure provider, or commercial arrangement.

Third, market recognition should be narrow and predictable. The service verifies authority, uniqueness, scope, restrictions, and effective change. It does not decide whether the entrant's business model is sufficiently fashionable or whether a seller deserves the scarcity value. Hard cases receive reasons and review.

Fourth, governance should publish impact by cohort. The NRS can distinguish members by resource entry history for analysis without attributing political virtue. Consultations can reserve structured evidence sessions for first-transfer recipients, lessees, IPv6-first operators, and small networks. Council documents can explain who bears each proposal.

Fifth, pricing must avoid converting administrative dependency into a private scarcity tax. Members should pay for verification, publication, security, correction, and continuity. Fees must be transparent and challengeable. Optional advocacy or commercial services should not be bundled with the ability to maintain essential records.

The NRS will not remove the advantage of incumbents created by history. It can ensure that a late entrant's evidence is intelligible, that its provider is replaceable, and that its voice is not treated as less informed because it lacks abundance-era memory. This is a credible, non-promotional argument for the Society: it starts from the market that entrants actually face.

The 2027 test is whether rules can outlive their founding moment

The last year of the study horizon should be used to test institutional adaptation, not to announce a future conclusion.

By 2027, each regional service organisation should be able to publish how many new operational entrants entered via residual allocation, transfer, lease-supported routing, and IPv6-first designs. It should distinguish organisations from accounts and identify data limitations. It should report processing and appeal outcomes for first-time recipients separately from those of experienced entities.

Policy proposals affecting transfers, holding periods, fees, leasing, routing authorisation, and waiting lists should contain cohort impact notes. At least a portion of consultation evidence should come from entrants launched after exhaustion, including businesses that struggled or failed. Public reports should explain how evidence did or did not change the decision.

Transfer and waiting-list statistics should be read together. If hundreds of networks wait over a year for a /24 while millions of addresses circulate via transfers, governance should focus on regulatory quality and small-buyer access rather than claim that the queue is the main entry path. If APNIC's /23 remains useful, outcomes should show what recipients build and when they need more.

Leasing should become more legible without turning registries into supervisors of every private contract. Operational authority, abuse contact, routing authorisation, and transition rights can be improved while recorded holding remains clear. IPv6 progress should be measured alongside the cost of ongoing IPv4 bridging.

The strongest test is voice. Do post-exhaustion operators appear on councils, policy groups, consultations, and evidence panels in proportion to their role in current network growth? If not, has the institution changed the cost of participation or merely repeated that meetings are open?

These questions do not guarantee a single policy answer. They require the institution to demonstrate that rules written during abundance have been reconsidered under scarcity by people who live under scarcity from day one.

Conclusion: the institution must remember what the entrant cannot

The post-exhaustion generation does not await the return of the old Internet. It buys, leases, shares, translates, and deploys IPv6. Its networks are real. Its customers do not care whether a service was built from inherited addresses or a carefully financed portfolio, as long as connectivity works.

Governance cannot be as indifferent. Entry history shapes cost, dependency, and policy experience. An abundance-era member may remember the registry as a policy-constrained source of addresses. A post-exhaustion member knows it as a verifier, fee collector, waiting-list administrator, transfer settlement point, and controller of dependent services. Both views may be accurate, but they do not create the same legitimacy claim.

The evidence already shows change. ARIN directed unsatisfied demand to the waiting list and transfers after September 2015. The RIPE NCC moved from a final /22 ration to a reclaimed /24 and, in March 2026, reported a queue whose first member had waited 477 days while over 1.5 million addresses circulated via transfers the previous month. APNIC tells networks needing more than a /23 to find a transfer source. Interview-based research describes a portfolio of purchase, lease, sharing, and IPv6 options rather than a single allocation path.

Reform is not retroactive moral judgement. Earlier holders need stable recognition. Scarcity is real. Anti-fraud checks remain necessary. IPv6 is the long-term path to growth. Reform is institutional modesty: not applying abundance-era assumptions forever, not confusing an account with an entrant, not calling a small ration equal opportunity, and not treating market dependency as a private matter when registry rules determine whether the market clears.

A legitimate post-exhaustion institution measures impact by cohort, makes transfers predictable, gives lease roles usable evidence, protects portability, offers review, and reduces participation cost. The NRS can embody these principles if it treats them as enforceable limits rather than a pretence to replace one controller with another.

By 2027, the decisive question should not be whether new networks have adapted. They already have. It should be whether institutions formed in an era of initial allocation have learned to govern a generation that never received one.

Sources and scope

The cohort proposed here is narrower than all companies founded after 2015 and broader than members currently on a waiting list. Regional dates and rationing policies differ. Public registry and BGP data cannot establish private price, lease terms, financing, customer loss, or reason for business closure. These questions require carefully sampled interviews with surviving, delayed, and failed entrants. Governance proposals concern current institutional impact; they do not assert a right to confiscate valid holdings, create duplicate authority, or obtain IPv4 that no free pool can supply.