Summary

  • On 3 February 2011, the final-five-/8 rule gave one /8 to each of the five Regional Internet Registries. It exhausted IANA's ordinary central unallocated pool but left substantial regional inventory, previously allocated holdings and continuing IPv4 demand.
  • The event ended one kind of discretion: IANA could no longer satisfy ordinary regional IPv4 demand with another full /8. It enlarged another: each RIR's policy, membership, transfer and registration choices became more important to the economic usability of scarce holdings.
  • Scarcity did not make registry records optional. Uniqueness, accurate holder information, fraud control, reverse DNS and later routing-security assertions became more valuable. The governance question is whether those necessary functions justify control over price, commercial purpose, geography or the merits of a transaction.
  • The secondary market was not a later aberration. A documented 2011 bankruptcy sale and empirical research covering transactions through mid-2012 show that address value and reassignment were already becoming visible as the central pool closed.
  • The five equal final allocations did not create five equal regional experiences. Different inventories, demand, transfer rules and exhaustion dates converted a uniform top-level act into distinct regional entry costs and recognition risks.
  • A defensible post-scarcity settlement would preserve an auditable, interoperable record while making the administrator replaceable and its refusal power narrow. Number Resource Society can contribute as an advocacy and member-representation organisation: not by promising new IPv4 supply or operating a registry, but by campaigning for a clear separation between necessary verification and discretionary scarcity control.

A ceremony that closed a warehouse, not a network

The cameras in Miami captured a clean ending. Representatives of ICANN, the Number Resource Organization, the Internet Architecture Board and the Internet Society marked the allocation of the last five large blocks in IANA's central unallocated IPv4 pool. The contemporaneous announcement used a warehouse analogy: the last crates had left the central warehouse for regional distribution centres. It was a useful image because it distinguished the top of the hierarchy from the regional shelves below.

It was also an image with a limit. A physical warehouse closes when the goods leave. IPv4 did not. Existing addresses continued to support networks, new services still required compatibility with IPv4-only users, and organisations with more addresses than they needed could transfer control to organisations with too few. The manufacturing line had ended, but the installed stock remained productive. From that day forward, administration concerned not only the last free inventory but also the movement, recognition and reliability of assets already embedded in operating networks.

The arithmetic was dramatic. A /8 contains 16,777,216 address values. Five of them contain 83,886,080. The IANA registry records the February 2011 designations as 102/8 for AFRINIC, 103/8 for APNIC, 104/8 for ARIN, 179/8 for LACNIC and 185/8 for RIPE NCC. That equality was a global trigger, not a measurement of equal regional need, equal prior holdings or equal future growth. It divided the final central stock by institution.

The trigger itself came after two further /8 allocations to APNIC earlier that week. The Address Supporting Organization minutes recorded that the APNIC allocations activated the final-five rule, that the registry had been updated and that the public event was intended to carry the exhaustion message beyond specialist circles. This was planned administration, not an accidental discovery that the cupboard was empty.

That distinction matters for accountability. No individual at the ceremony chose which operator would receive the last address. A previously adopted global rule performed a top-level division. The economically consequential choices came afterward, as each RIR decided how its remaining inventory would be rationed, which transfers would be recognised, which applicants would qualify and how disputes over control would affect the public record.

The central ceremony therefore should not be read as the moment authority vanished. It was the moment one visible form of authority reached its terminal condition. The next phase was less theatrical and more consequential: regional policy clauses, membership accounts, transfer desks, supporting documents, holding periods, waiting lists and registry status changes began to allocate the practical costs of scarcity.

There were at least three scarcity clocks

Public discussion often compresses IPv4 exhaustion into one date. That obscures the institutional sequence.

The first clock was architectural. IPv4 uses a 32-bit address field, which creates a finite mathematical space. Reservations, special-purpose uses and historical allocation patterns mean the globally usable supply is smaller than the headline total of roughly 4.3 billion values. That constraint existed long before 2011.

The second clock was IANA's ordinary central pool. It measured full blocks not yet allocated at the top level to an RIR or reserved for another purpose. That is the clock that reached its final-five trigger in February 2011. It did not measure how many addresses remained unassigned inside each RIR, how many assigned addresses were actively used, how much space could be recovered, or how much demand could be met through transfer and sharing.

The third clock was regional. Each RIR had inventory accumulated under earlier allocations and its own demand pattern. APNIC, RIPE NCC, LACNIC, ARIN and AFRINIC entered their restricted or exhausted phases at different times and under different policies. RIPE NCC, for example, began allocating from 185/8 in September 2012 and did not report final exhaustion of its available pool until November 2019. One global ceremony therefore began five regional countdowns rather than ending all allocation at once.

There was also a market clock. It measured how quickly operators, buyers, sellers, lenders and service providers began treating recognised IPv4 control as valuable. That clock did not wait for every RIR to exhaust its inventory. In 2011, the Nortel bankruptcy sale to Microsoft publicly priced 666,624 addresses at $7.5 million, or $11.25 each. Geoff Huston's later analysis used the transaction as a rare public reference while warning that confidential contracts made a reliable market price difficult to observe.

The clocks interacted but were not identical. Central exhaustion increased expectations of future scarcity. Regional inventory delayed the point at which some applicants had to buy. Transfer recognition affected whether a seller could convert unused stock into cash and whether a buyer could obtain a clean record. Routing and customer compatibility affected the operational value after the record changed. IPv6 deployment and address sharing affected the buyer's alternatives. No single date settled all of those variables.

The authority question becomes clearer once the clocks are separated. IANA controlled the second clock through global allocation rules. RIRs increasingly controlled the administrative junction between the third and fourth: the point where depleted regional supply met transactions in existing stock. The market could agree a price, but registry action could still determine whether the new holder appeared in the recognised record and could use associated services with confidence.

What IANA stopped deciding

It is easy to overstate IANA's pre-2011 role by projecting the later value of IPv4 backward. IANA did not approve every network's allocation, inspect every transfer or determine every route. It managed the top of a hierarchy.

RFC 7020, published in 2013 as a description of the number registry system then in operation, identifies IANA as a role rather than an organisation. It places that role at the root of the IP address and AS number allocation hierarchies, serving RIRs, which in turn serve LIRs and other customers. The same document names allocation-pool management, hierarchical allocation and registration accuracy as distinct goals. It also states that actual route announcements are operational matters outside the registry system's scope.

Before the central pool closed, IANA's recurring IPv4 decision concerned whether an RIR qualified for another allocation under global policy. Utilisation and projected need informed the flow of full blocks. The global mechanism constrained discretion, but the top-level supply relationship still mattered: an RIR that met the criteria could receive fresh inventory from the common pool.

After February 2011, IANA could not answer ordinary regional IPv4 demand with another unallocated /8. That removed the most important replenishment option. It did not erase IANA's number role. It continued to maintain authoritative registries, support reverse DNS at the top level and later distribute recovered space under a different global policy. Yet the scale and logic changed. The post-exhaustion mechanism concerns smaller recovered holdings and equal scheduled allocations, not a return to an expanding primary pool.

The 2012 post-exhaustion global policy notice makes the distinction explicit: IANA could make allocations smaller than /8 from recovered space. The recovered-space registry records returns and later allocations. Recovery kept the top-level record alive; it did not reverse scarcity.

IANA also ceased to sit above regional allocation appeals in the way earlier guidance had contemplated. RFC 7020 notes that RFC 2050 had allowed a final appeal to IANA after other avenues were exhausted, then says regional consensus-based appeal policies had developed and further appeal to IANA was no longer appropriate. The change cannot be attributed to the 2011 ceremony alone, but its timing illuminates the same institutional direction: regional bodies were not temporary counters beneath a central allocator. They had become the final recognised authorities for many decisions affecting holders in their service regions.

What stopped in 2011 was therefore narrow but decisive. The central role could no longer dilute a regional scarcity dispute by supplying more ordinary IPv4 stock. If a regional rule restricted an applicant, delayed a transfer or required additional evidence, there was no meaningful top-level reservoir to make the conflict disappear. Authority over scarcity became harder at the edges because abundance no longer softened its consequences.

What the five RIRs inherited

The RIRs inherited more than five blocks. They inherited the burden of deciding how a finite regional remainder would meet claims that could not all be satisfied.

In the allocation era, a need test answered two linked questions: is the request justified, and if so, how large should the allocation be? Once a final-pool cap applies, those questions separate. An applicant can demonstrate a need for far more than the cap and still receive only the standard ration. Need becomes an eligibility filter, while policy fixes quantity. That is a shift from case administration toward distributive design.

Different rules can distribute the same scarcity differently. A registry can favour the order of applications, equal blocks per account, equal blocks per legal organisation, established holders with documented growth, entirely new entrants, networks with IPv6 plans, critical infrastructure, auction bidders or applicants that pass a transfer needs test. Each rule has a plausible rationale and a constituency that bears its cost. The address length does not choose among them.

The five RIRs also inherited responsibility for the recognised identity of already allocated space. As the free pool diminished, that function became more important than issuance. A buyer needed confidence that the seller controlled the range, that no conflicting claim would prevent the update, that the new record would be accepted by counterparties and that supporting services could follow. A seller needed a credible route from commercial agreement to final recognition. Lenders and acquirers needed to know whether the record could survive default, merger, insolvency or reorganisation.

The administrator's refusal power therefore acquired economic weight. A refusal might protect against fraud or duplicate claims. It might also delay deployment, financing or a company sale. A request for better evidence could improve accuracy; an open-ended request could create hold-up risk after the buyer had committed funds. A regional transfer boundary could preserve a policy preference; it could also strand stock away from higher-valued use. Scarcity did not make every refusal abusive, but it made the reason, timing and remedy material.

The RIRs' own language often described their task as stewardship. That word can express care for continuity and accuracy. It becomes dangerous when it blurs the line between keeping a record and controlling the asset economy around the record. A custodian of uniqueness needs authority to prevent two simultaneous recognised holders. It does not follow that the custodian should decide which commercial model deserves the address, whether a buyer's growth plan is socially preferable, or whether a region has a political claim to value embedded in a globally routable identifier.

The inherited duty was real. So was the temptation to convert necessary verification into a broader scarcity mandate.

Equal final blocks created unequal regional power

The final-five rule looks impartial because every RIR received one /8. At the IANA layer, it was simple, predictable and immune to a last-minute race for the remaining blocks. Those are genuine virtues. Equal division also made the exhaustion signal unmistakable.

But equal institutional shares did not make the economic position of each region equal. RIRs differed in the amount and age of previously allocated space administered in their regions, growth rates, member composition, address intensity, transfer readiness and access to legacy holdings. They also differed in how much inventory remained outside the final /8 and how their policies treated that inventory.

The result was not one post-2011 market. It was a set of connected markets with different recognition conditions. An operator's alternatives depended partly on which RIR held the relevant record, whether inter-RIR transfers were compatible, what need or holding conditions applied, and how quickly documents could be reviewed. Regional administration, originally justified as a way to distribute service and policy development, became a potential trade boundary around scarce value.

The 2008 OECD study on IPv4 management anticipated the problem before central exhaustion. It warned that scarcity could raise barriers to entry and strengthen incumbents, and it examined transfer mechanisms as a way to move previously allocated space while preserving technical safeguards. The report also recognised that regional differences and inter-RIR rules could affect market efficiency. Those were not speculative side issues. They became the post-2011 operating environment.

Equal /8s also allocated discretion. Each RIR obtained a finite option to decide how long its last phase would last and who would receive the option value preserved by rationing. A stringent cap could extend formal availability but force growing networks into the market sooner. A looser rule could meet larger current needs but close free-pool entry earlier. A waiting list could preserve a low-cost channel for some entrants while making timing uncertain. None of these effects can be evaluated merely by counting how long the pool survived.

The appropriate denominator is not years until exhaustion. It includes eligible applicants, successful recipients, unsuccessful and withdrawn requests, quantity awarded, quantity demanded, transfer purchases, time to recognition, cost of substitutes and post-allocation use. Without those measures, a long-lived pool may be celebrated even if it supplied too little to support meaningful entry, while a market that moved underused holdings into active networks may be treated as failure because prices became visible.

The central ceremony distributed the same prefix length. It did not distribute the same opportunity, the same burden or the same power.

Scarcity made recognition a productive input

An IPv4 address is not valuable because a row in a registry can be admired. It is valuable because networks, services and customers can use it in a globally interoperable system. Registration contributes to that value by reducing uncertainty.

A reliable record tells counterparties which organisation is recognised in relation to a range. It supports contact, reverse DNS administration, transfer history and later route-origin authorisation. It helps investigators distinguish a stale entry from a contested claim. It allows a buyer to show investors, upstream providers and customers that a transaction has moved beyond a private promise.

Recognition is therefore a productive input, but it is not the whole product. A changed record does not compel other networks to route the range. It does not erase address reputation, configure customer systems, assign addresses to devices or settle every legal question about a seller's assets. RFC 7020 is useful here because it expressly keeps routing decisions outside the registry function. The record should describe and authenticate a control relationship without pretending to manufacture operational truth.

Scarcity increased the value of this input. When addresses could be obtained through ordinary allocation at low incremental cost, a disputed record could sometimes be avoided by requesting different space. As alternatives narrowed, the same dispute attached to an asset with replacement cost. Delay became expensive. Inaccurate identity could impair a sale. An unresolved claim could affect a merger. A frozen account could put customer continuity at risk even before any route changed.

This is why the phrase "mere bookkeeper" can mislead. Bookkeeping is not trivial when the book coordinates unique rights relied upon by a global network. A land register, securities ledger or vessel registry can be narrow and still economically crucial. The safeguard is not to deny the record's value. It is to prevent the record-keeper from using that value to acquire unrelated discretion.

Post-2011 authority should be measured at this junction. Which registry act is strictly necessary to keep the record unique, accurate and secure? Which act instead judges the commercial merits of a transaction? Which delay protects against a specific conflict? Which delay merely preserves administrative leverage? Which data field supports an operational dependency? Which field becomes a pretext for territorial or behavioural control?

The scarcity clock made those questions unavoidable because recognition had become harder to replace.

The market arrived before the institutions finished naming it

The public history is sometimes told as if transfer markets were an unfortunate consequence discovered after exhaustion. The evidence is less tidy.

Commercial reallocation was discussed before 2011, and some RIR transfer policies existed before the central ceremony. The Microsoft-Nortel bankruptcy sale made the price dimension visible in the same year as the final-five allocation. Empirical work by Milton Mueller, Brenden Kuerbis and Hadi Asghari covering November 2009 through June 2012 identified 85 transactions, 208 blocks and more than 6.3 million addresses exchanged. The study's market estimate was incomplete, but incompleteness is itself important: private transactions and imperfect records meant institutional statistics could not capture the whole economy.

Later research by Ioana Livadariu, Ahmed Elmokashfi and Amogh Dhamdhere examined reported transfers and inferred changes from routing data. Their published analysis found that most transferred space was routed after transfer and that utilisation generally increased, while also identifying concentration among major actors and transfers that might not appear in official lists. That evidence does not prove every transfer efficient. It does refute the assumption that trade was merely speculative paper movement with no operational use.

The emergence of price changed the registry's position. Under primary allocation, the institution decided whether an applicant could receive stock on administratively defined terms. Under transfer, buyer and seller negotiated value, but the institution could still decide whether the transfer met regional rules and would be entered in the recognised record. The RIR did not set the private price, yet its approval conditions, delays and cross-region compatibility affected the transaction's expected value.

That is a form of market power even when no fee is charged for the transfer. If a registry can prevent or postpone recognition, parties price that risk into escrow, legal review, broker selection and closing conditions. If its rules prohibit a class of movement, the affected holding may trade at a discount or remain stranded. If it offers a clear, fast and reviewable path, more value can move with less intermediation.

The proper conclusion is not that registries should ignore transfers. A market without an authoritative update path invites duplicate claims, stale records and fraud. The conclusion is that transfer recognition must be treated as a bounded service. The registry verifies identity, authority, the range, conflicting claims and the requested change. It publishes the result and preserves history. It should not use recognition to impose an unpriced option over the holder's commercial decisions.

The market did not abolish the ledger. It exposed how much the ledger's recognition power was worth.

A transfer is not a fresh allocation

Allocation and transfer are often placed under one policy heading because both end with a recognised holder. Economically, they are different.

A fresh allocation distributes inventory that the registry has not previously allocated to another holder. The institution must choose among competing claims to a finite pool. Need, conservation, queue order and caps are ways of managing that distributive problem, even when their incidence should be challenged.

A transfer moves an existing holding between parties. The registry's core task is not to choose who deserves scarce public stock. It is to verify that the transferor controls the recognised interest, that the transferee is correctly identified, that the range and scope are clear, that no unresolved conflict prevents a safe update, and that uniqueness is preserved when the record changes.

Applying primary-allocation logic to a transfer gives the administrator a second allocation decision after the parties have already found each other and agreed value. A needs test can then become a veto over the buyer's capital plan. A regional-use rule can become a border around a global identifier. A holding period can burden speculation, but it can also trap a legitimate reorganisation or change in demand. The policy objective may be defensible; the power still needs evidence, proportionality and review.

The distinction became sharper after the final-five event. IANA no longer had ordinary stock to distribute. Increasingly, the available response to unmet IPv4 demand was reallocation from existing holdings, sharing, leasing, acquisition of an address-rich company, or deeper IPv6 deployment. When an RIR placed allocation-era moral language over those private adjustments, it risked converting coordination into capital control.

Scarcity does not prove that every address should be freely alienable without safeguards. Number resources have routing, reputation and identity externalities. A reckless update can harm third parties. Fraudulent transfers can steal operational infrastructure. Fragmentation can affect routing scale. Those risks justify precise controls tied to the record and technical system.

They do not justify an unlimited mandate. The burden should remain on the restriction: identify the harm, show how the condition reduces it, publish the evidence standard, disclose processing time, provide a cure route and make an adverse decision reviewable. If a narrower control can protect uniqueness and accuracy, the registry should use it.

This is the post-2011 constitutional divide. Allocating the last unallocated stock required distributive rules. Recording later movement requires reliable settlement. Treating both as the same act preserves a scarcity power that the end of the pool should have narrowed.

The hidden option held by an incumbent registry

Every delay or refusal gives someone an option. In a transfer, the buyer may have committed capital but not yet obtained recognition. The seller may remain the holder of record while bearing contractual duties to close. Customers may depend on a planned migration. A lender may wait for proof that collateral value has moved. During that period, the registry controls a decisive condition without bearing the parties' full downside.

The option has several dimensions. The institution can ask for more documents, interpret policy, classify a corporate change, investigate a conflicting claim, screen authority and decide whether a transfer falls within compatible regional rules. Many of those actions are legitimate. What makes them powerful is asymmetric exposure: the registry's cost of another week is administrative; the parties' cost may include financing, lost deployment, contract breach or customer uncertainty.

Good governance prices this asymmetry into service design. Evidence requirements should be published and connected to decision-relevant facts. Routine cases should have stated service targets. Complex cases should receive a reasoned explanation of what remains uncertain. Requests for additional proof should identify the issue they are meant to resolve. Parties should know whether a hold prevents only the record change or also affects existing operational services. Appeals should be independent enough to correct the original desk.

Data should reveal the tail, not only the median. A registry can process most transfers quickly while a small set of high-value or contested cases remain unresolved for months. Those are the cases in which its option power is largest. Reporting should separate routine changes, mergers, insolvencies, inter-RIR transfers, suspected fraud, conflicting authority and legal restraint. It should disclose withdrawals and refusals as well as successful updates.

The 2011 ceremony is relevant because it removed the easiest institutional answer to delay: use different free stock. Once replacement required a market purchase or complex renumbering, the record-keeper's timing became part of the asset's cost. Scarcity translated administrative latency into economic exposure.

An accountable registry should not profit politically from that exposure. It should not cite the value of the addresses as a reason for broader authority. Value raises the required quality of verification and remedy; it does not enlarge the purpose of the institution.

Recovered space did not restart abundance

The recovered IPv4 pool can create confusion about whether the 2011 event was truly final. Addresses have been returned to IANA, and IANA has made later allocations to the RIRs under the 2012 post-exhaustion policy. The correct conclusion is neither that the ceremony was false nor that every later allocation was trivial.

The ceremony was final for the ordinary pool of unallocated /8s. Recovered space is different in origin, scale and distribution. It consists of ranges returned or otherwise made available after prior administration. The global policy allocates from that pool on a schedule and by formula. IANA's public registry preserves the chain.

This mechanism demonstrates the continuing value of a top-level ledger. Without an authoritative record, a returned range could remain ambiguous or be claimed twice. Equal distribution from the recovered pool also prevents a race in which one RIR captures all returns. Those are legitimate coordination functions.

But the recovered pool does not make IANA the ordinary supplier again. The quantities are too small relative to continuing demand, and the allocations are not generated by new protocol space. Each recovered address has a history. It may require quarantine, investigation, reputation repair or more careful deployment. Regional policy still determines how the received range reaches eligible holders.

Recovery therefore reinforces the distinction between creation and recognition. IANA can recognise that space has returned, record it and reallocate it under the global formula. It cannot manufacture more IPv4. RIRs can distribute their shares under applicable rules. They cannot eliminate the scarcity cost borne by networks that need larger, cleaner or immediately usable blocks.

The post-exhaustion mechanism also shows why portability matters. Returned space should not become a permanent justification for institutional monopoly. The record and formula should be reproducible, auditable and capable of continuity if an operator fails. The public interest lies in the integrity of the allocation history, not in the corporate survival of any particular administrator.

The scarcity clock kept running because recovered fragments entered a world of continuing demand. Administration remained necessary, but its legitimate scope became more exacting.

Incumbents and entrants faced different versions of the same rule

Scarcity policy often speaks in universal terms: every applicant follows the rule, every LIR receives the cap, every transfer satisfies the same test. Equal clauses can have unequal effects because actors arrive with different starting positions.

An incumbent with a large historical holding can meet growth from inventory, improve utilisation, deploy sharing gradually, transfer space among affiliates or sell surplus. A new network begins with no such option. If the regional free-pool ration is small, it must purchase, lease, obtain space from an upstream provider, redesign service around sharing, or accept a narrower launch. The incumbent experiences scarcity as portfolio management. The entrant experiences it as an up-front financing condition.

Transfer rules amplify the difference. A holder with experienced legal and registry staff can prepare documents, manage account history and wait through a review. A smaller buyer may depend on a broker, external counsel and borrowed capital. A delay of the same length consumes different proportions of each party's resources. A fixed compliance burden is regressive when one actor can amortise it across millions of addresses and another across a first modest block.

This is why conserving a residual pool cannot, by itself, prove that entry was preserved. A token allocation may be useful for transition, infrastructure or a small service. It may also be far below the quantity needed for competitive scale. The right evaluation asks what the recipient could do after receiving it, how soon additional addresses were needed, and under what terms those addresses could be obtained.

The transfer market can reduce incumbency by allowing old holdings to move. It can also expose wealth differences because entry now requires capital. Both statements can be true. The relevant comparison is not market versus a costless abundance that no longer exists. It is market plus bounded, reliable recognition versus rationing plus trapped or informally moved stock.

An effective registry does not promise equal commercial outcomes. It makes the conditions of recognised control clear, protects against fraud and avoids adding discretionary barriers unrelated to those functions. Public policy can address competition and connectivity through institutions that have the mandate and tools to do so. A private number registry should not become an unaccountable substitute regulator merely because incumbency is real.

The final-five event made historical allocations more valuable. That distributional inheritance cannot be undone by declaring all later applicants equal at the desk.

Technical safeguards are the strongest case for a narrow ledger

Criticism of gatekeeping can become careless if it treats all registry conditions as rent seeking. Some controls protect the very market and continuity that critics want.

Uniqueness is non-negotiable. Two recognised holders cannot safely control the same range at the same time. Identity and authority checks are necessary because valuable accounts attract fraud. Change history matters because a buyer needs to understand how control moved. Public contactability supports incident response. Reverse DNS delegations must follow legitimate authority. RPKI services require careful control because an erroneous or malicious certificate action can affect route-origin validation.

Accurate records also constrain incumbent power. An old holder should not be able to sell the same range twice. A registry insider should not be able to rewrite control without evidence. A buyer should not have to accept a seller's private spreadsheet as the only proof. A court, insolvency practitioner or corporate successor may need a stable record against which to present an order or transaction history.

These functions require expertise, security and continuity funding. A thin registry is not a cheap website. It needs strong authentication, separation of duties, immutable logs, tested recovery, documented evidence standards, secure signing, public interfaces and an independent correction path. It should be able to survive the loss of key staff, premises, systems or legal authority without corrupting the record.

The narrow-ledger argument is stronger because it funds these duties seriously. It rejects a false bargain in which low fees excuse weak security or poor appeals. It also rejects the opposite bargain in which technical importance is used to justify control over commercial purpose, geography, price or speech.

Each restriction should pass a necessity test. Does it protect uniqueness, registry accuracy, security, fraud prevention or continuity? Is the harm evidenced? Is the rule the least restrictive means? Are reasons and time limits published? Can an independent reviewer correct error? Does the institution bear consequences for negligent action? Can the service and record move to a successor?

If the answer is yes, the condition belongs in a credible registry. If the answer is that the institution prefers a particular market structure, believes an operator's use is morally inferior or wants to preserve regional control over value, the condition requires a different mandate.

Scarcity raises the value of safeguards. It does not erase the boundary around them.

The accountability statement that should have followed Miami

The 2011 announcement clearly told the world that central abundance had ended. It did not provide an equally clear constitutional statement for the scarcity era. Such a statement would have identified the authority that moved, the decisions it covered and the limits that would apply.

First, it would have separated IANA exhaustion from RIR exhaustion. Every public update would show central stock, regional available stock, reserved space, recovered space and distribution rules without presenting one number as the whole supply picture.

Second, it would have named transfer recognition as a core service. The statement would distinguish private agreement, legal control, registry recognition and operational routing. It would explain which evidence the registry verifies and what it does not decide.

Third, it would have narrowed policy claims. Conservation rules for unallocated inventory would not automatically govern already allocated holdings. Regional service boundaries would not be described as ownership boundaries. Community deliberation would create administrative rules within a defined mandate, not title to scarce value.

Fourth, it would have required measurable due process. Applicants and transfer parties would receive published evidence lists, service targets, reasoned decisions, cure periods and independent appeal. Emergency holds would have scope and expiry. Aggregate reporting would include denials, withdrawals, tail latency and reversals.

Fifth, it would have treated interoperability as an obligation. RIR records and transfer rules should be compatible enough that a legitimate movement does not fail merely because two regional institutions use incompatible policy tests. The global value of uniqueness should not be converted into five private trade barriers.

Sixth, it would have required continuity beyond the incumbent. Authoritative records, signed histories, interfaces and operating documentation would be exportable to a qualified successor under tested conditions. Reserves would finance handover as well as survival. The institution would not be able to argue that the importance of its own ledger made replacement impossible.

Finally, it would have acknowledged market evidence. Price and transfer volume would not be treated as embarrassment. They would be analysed alongside IPv6, sharing, routing and competition. Official transfer lists would be understood as records of recognised changes, not a complete account of private value or operational use.

Such a statement would not have solved scarcity. It would have made power legible at the moment power changed form.

What rights-based advocacy can contribute

The Number Resource Society should not answer the 2011 event by promising to recreate a free pool or presenting itself as a successor registry. It is neither. Its useful contribution is advocacy: represent members who experience registry decisions, publish evidence-led comparisons and campaign for rights and continuity rules that stop any record-keeper turning scarcity into permanent discretionary control.

A positive model begins with the operator. The holder should have a clear right to an accurate record, to transfer subject to proof of authority and conflict checks, to receive reasons for adverse action, to correct error, to preserve operation during a dispute where technically safe, and to move the record service if the incumbent fails. Those rights should be contractual and auditable rather than dependent on institutional goodwill.

The common layer that NRS can advocate remains thin. Authorised registry operators would specify and implement identifiers, evidence classes, signed change events, conflict status, timestamps, public and protected fields, and interoperability between recognised registries. That operational layer should not decide whether a buyer's business is worthy, assign regional moral ownership to addresses or use a policy room to re-price private capital.

NRS can also improve market transparency without receiving confidential contracts. It can research public records, document member experience and advocate aggregate reporting by the RIRs and qualified independent researchers of recognised transfer quantity, processing time, refusal reason, dispute status and reversal. Price reporting can be voluntary, aggregated or drawn from independently verified samples held by those authorised collectors. The aim is not to become a broker, record-keeper or auditor. It is to make recognition risk visible enough that buyers and sellers do not pay an avoidable opacity premium.

Portability is the central discipline. If a registry operator becomes insolvent, captured or persistently unreliable, a verified holder should not lose the ability to prove its history. A successor should be able to restore the authoritative record from signed evidence while conflicting claims remain labelled rather than silently overwritten. No administrator should possess an unreviewable power to strand the asset merely because it hosts the current record.

This settlement must still confront hard cases. Courts may issue inconsistent orders. Corporate authority may be disputed. A transfer may be fraudulent. Routing may continue under a party no longer recognised. Sanctions or law may constrain service. NRS should not pretend that advocacy or cryptography eliminates judgment. It should campaign for authorised RIR decision-makers to give bounded reasons and for genuinely independent reviewers or competent courts to provide review, while keeping those judgments separate from commercial merit.

The model is positive because it values the registry enough to constrain it. Accurate records, fraud control, security and recovery deserve stable funding. Advocacy, conferences and wider policy ambitions should not be financed through a scarcity toll attached to indispensable recognition. The holder pays for trustworthy coordination, not for the privilege of being judged by a territorial monopoly.

The 2011 ceremony showed that a global rule can execute predictably at a known threshold. The next institutional advance is to apply the same discipline to the rights of the people and organisations whose networks carry the consequences.

The clock that matters now measures refusal power

Fifteen years after the ceremony, the most important scarcity measure is not the number of full /8s left at IANA. That number reached its terminal condition in 2011. Nor is it simply the count of addresses in a recovered pool. The live measure is how much economic and operational dependence sits behind a registry decision, how long that decision can remain unresolved and what remedy exists if it is wrong.

The Miami event deserves its place in Internet history because it made a finite architectural constraint visible. Its deeper consequence was institutional. A replenishing hierarchy became a recognition hierarchy. IANA's ordinary allocation discretion diminished; regional authority over the remaining stock and existing holdings became more consequential. The scarcity clock did not stop. It moved from a central counter to regional ledgers, transfer desks and the balance sheets of operators.

The response should not be nostalgia for abundance or denial of technical limits. It should be a narrower and stronger settlement. Keep the identifiers unique. Keep holder records accurate. Protect accounts and signing systems. Record transfers promptly. Isolate disputes. Preserve reverse DNS and routing-security continuity. Publish reasons and service performance. Make appeal real. Make succession possible.

Then stop.

Do not turn a service region into title. Do not turn conservation of a vanished free pool into permanent supervision of private exchange. Do not turn the value created by operators into a mandate for the administrator. Do not confuse a ceremony attended by institutions with consent from every network that later had to buy, lease, share or defend scarce addresses.

IANA's last-five act closed a chapter of distribution. It did not settle who should govern the value that remained. That question belongs to the holders, operators and users who bear the consequences, protected by a ledger designed to serve them and disciplined enough to remain ordinary.

Sources