After IPv4 exhaustion, conservation language changes character. It no longer mainly describes the careful rationing of a remaining pool. It becomes a way of deciding how much friction, suspicion and institutional judgement should attach to movements of an asset whose original distribution is already settled. For APNIC, that distinction is more than semantic. In the Asia-Pacific region, where older address holdings, fast-growing access markets, national registry structures, cloud hubs, island economies and late-building networks coexist, the words conservation, stewardship, need, anti-hoarding, fairness and community protection can do very different kinds of work.
Some of that work is indispensable. A registry must keep number resources unique. It must know who holds a block, who is accountable for it, and whether a transfer would leave the operational record false. It must prevent fraud, resolve disputes, keep reverse DNS and route authorisation intelligible, and protect bounded residual pools from immediate arbitrage. These are not decorative concerns. They are the reason a registry exists.
But the same vocabulary can also become distributional cover. It can make a delay look prudent when it is merely costly. It can make a documentation burden look fair when it favours firms with staff and lawyers. It can make hostility to leasing sound like protection of the commons when it mainly protects permanent holders. It can make broker suspicion sound virtuous when it raises search costs for small buyers. It can make a late entrant prove its virtue while the incumbent's historical stock remains normal, quiet and already recorded.
The practical question is therefore not whether IPv4 is scarce. That has been settled. Nor is it whether APNIC's policy documents contain recognisable public-interest language. They do. Current APNIC number resource policy uses the familiar vocabulary of uniqueness, registration, aggregation, conservation, fairness and minimised overhead. Post-exhaustion material describes a final 103/8 pool, returned non-103 space, small residual delegations, transfer paths and IPv6 as the long-term answer. Transfer conditions require recipients to show need, require registration updates, impose limits in disputed cases and restrict transfers of certain final-pool delegations for a period after original delegation.
The issue begins after those facts are accepted, not before them.
Those official statements are useful exhibits. They show the institutional grammar in which decisions are made. They should not be mistaken for proof that the grammar is economically neutral. The harder question is when a phrase that once protected future access now protects inherited advantage; when a rule that once prevented premature depletion now suppresses useful movement; and when a registry that thinks of itself as steward of a public resource becomes, in practice, a gatekeeper over liquidity, documentation and legitimacy.
The answer is not to abolish stewardship. It is to make stewardship narrower, more auditable and less moralised. APNIC's post-exhaustion role is strongest when it protects the ledger and weakest when it lets old scarcity language do the work of new distributional argument.
Conservation after the pool
Conservation began as common sense. IPv4's 32-bit design created a finite address space, and early allocation practice was generous by later standards. Classful addressing, large historical delegations and rapid Internet growth made waste visible. CIDR, route aggregation, private addressing and network address translation stretched the system, but none of them created new public IPv4 space. Registry conservation, in that period, had an intelligible object: slow depletion of a pool that still existed.
Need-based allocation also had a practical logic. If a registry held unallocated addresses and many operators wanted them, some rationing rule was unavoidable. First-come, first-served without scrutiny would have rewarded speed, paperwork capacity and confidence. Auctions would have rewarded capital. Political allocation would have rewarded influence. Need review at least tried to connect delegation to actual network deployment. It was imperfect, but its economic purpose was visible.
The pre-exhaustion argument for conservation also had an operational side. Address policy was never only about counting endpoints. Fragmented delegations could make routing more expensive. False records could slow abuse handling and incident response. Abandoned space could be hijacked or disputed. Aggregation, registration quality and accountable holders mattered regardless of scarcity. Scarcity simply made these disciplines easier to defend.
In the APNIC region, the older conservation case was especially plausible. The region contained fast-growing economies that were adding users, mobile networks and enterprise services at a scale that made "future entrant" a concrete category rather than a slogan. A final pool reserved for small bootstrap delegations could help a new network connect rather than leaving the last available addresses to the best-informed incumbents. A cap on final-pool use could be defended because the protected object was specific: a known block, a defined purpose and a bounded class of applicants.
That history matters because the original conservation claim should not be caricatured. The Internet did not scale by treating number resources as ordinary office supplies. The registry disciplines that accompanied scarcity were real. But exhaustion changes the object of conservation. Once the broad free pool is gone, the main policy problem is no longer how to stop premature depletion. It is how to treat transfers, leases, returns, legacy holdings, record updates and residual pools after the original distribution has already produced winners and losers.
This is where rhetoric becomes dangerous. A sentence that was reasonable when a registry still held a large common pool may become misleading when the registry mostly controls recognition and friction. The word conservation survives. The economics underneath it do not.
The moral vocabulary of scarcity
Scarcity language allocates moral status before it allocates addresses. A network asking for resources may be described as a builder, a hoarder, a speculator, a late entrant, a victim of history or a risk to the community. A seller may be described as releasing idle space or monetising a public resource. A broker may be market infrastructure or an opportunist. A lease may be efficient temporary access or a way of detaching numbers from need. The official record may be a ledger or a moral instrument. These choices shape policy even when no one says they are making distributional choices.
The most attractive words are often the most ambiguous. "Fairness" can mean equal treatment of applications, protection of late entrants, resistance to market concentration, respect for historical delegations, or adherence to community procedure. These meanings conflict. "Stewardship" can mean reliable record-keeping. It can also mean broad institutional discretion over behaviour that staff or policy participants consider improper. "Need" can mean proof that a scarce pool should be consumed. It can also become a test of whether a buyer's business model looks familiar enough to be trusted. "Anti-hoarding" can target genuine warehousing. It can also stigmatise inventory, transition buffers and lease supply that make operational sense after exhaustion.
The vocabulary is powerful because it sounds non-commercial. Conservation appears nobler than price. Community protection sounds more civic than liquidity. Stewardship sounds safer than market recognition. Yet post-exhaustion IPv4 is already commercial whether policy likes it or not. Addresses have market prices because they remain useful and cannot be newly produced in sufficient quantity. Operators acquire them, lease them, return them, route them, use them as part of mergers and value them in transactions. A registry may disapprove of certain conduct, but disapproval does not remove the market. It only decides whether the market will be visible, clean and accountable, or informal, opaque and expensive.
The central error is to let official vocabulary define the conclusion. APNIC's policy language records what the institution says it is doing. It does not prove that every use of conservation reduces waste, that every need test improves fairness, or that every delay protects the community. For economic analysis, the test is incidence: who pays, who waits, who benefits from friction, and what risk the rule actually reduces.
This is not a claim about bad faith. Institutions preserve old language because old language is safe. It reassures participants that a public resource is not being abandoned to pure commerce. It also simplifies debate. A rule described as conservation starts with moral credit. A rule described as liquidity restriction must defend itself. That difference is precisely why the vocabulary needs discipline.
Post-exhaustion conservation should have to name the thing being conserved. Is it a residual APNIC-managed pool? Uniqueness? Accurate registration? Routing hygiene? Anti-fraud assurance? A time-limited final-pool purpose? If so, conservation has a concrete object. If the protected object is simply institutional comfort with slow movement, then conservation has become a veil.
From rationing to recognition
The decisive institutional shift after exhaustion is from rationing to recognition. Before exhaustion, the registry was deciding whether to delegate addresses from a pool it controlled. After exhaustion, much of the registry's power lies in deciding whether and how the public record will recognise a movement between parties. That is not a minor administrative step. Recognition affects route filtering, abuse handling, RPKI, reverse DNS, reputation, insurance, financing, due diligence and the willingness of counterparties to transact.
Yet recognition is economically different from allocation. If APNIC is distributing addresses from a residual pool, it is choosing among potential consumers of a common resource. If it is recording a transfer of space already in circulation, the resource has already been delegated. The registry still has legitimate interests. It should verify that the source holder is real, that the block is not in unresolved dispute, that the recipient can be identified, that the transfer does not violate a specific pool restriction and that the records will become more accurate rather than less. But the case for broad judgement over the recipient's business plan is weaker than it was in the allocation era.
This difference often disappears inside the word need. Need review made intuitive sense when a common pool would be depleted. In transfer recognition, need review becomes a test attached to the ledger. It may still reduce fraud or sham demand, but it also becomes a fixed cost. It asks the late entrant not merely to buy addresses, but to describe itself in the form the institution recognises.
That cost is not evenly distributed. Large operators can maintain compliance staff, produce forecasts, pay advisers, tolerate delay and absorb a failed transaction. Smaller operators often need addresses because demand is immediate. They may lack long utilisation histories precisely because they are new. Their business may be volatile. They may need a small block for CGNAT pools, multihoming, customer-facing services or migration. A data-centre provider may need addresses to win contracts, not only to serve customers already signed. A cloud service may have elastic demand that does not fit traditional allocation narratives. An enterprise may need addresses for security architecture or staged transition rather than simple endpoint growth.
If the evidence model recognises only tidy, traditional demand, it favours incumbents and mature business models. The burden appears procedural, but it is economic. A requirement to produce a detailed plan may be low-cost for a regional carrier and high-cost for a small provider trying to close a modest transfer. A delay that is tolerable for a buyer with surplus inventory may be fatal for a new operator whose customer launch depends on address availability. A discretionary request for more information may sound cautious while moving bargaining power to sellers and experienced buyers.
The better distinction is between risk verification and economic approval. APNIC must verify identity, authority, dispute status, policy restrictions and record quality. It may require evidence that a transfer is not a sham. But the more it uses need to judge the attractiveness of a recipient's business model, the more it should admit that it is making distributional policy. That policy may be defensible. It should not be hidden inside a conservation sentence.
Recognition should be fast when the risks are standard. Where the source is legitimate, the block is clean, the recipient is accountable and the record update reduces operational ambiguity, the presumption should favour clean recognition. The registry's power should rise with concrete risk, not with general discomfort that addresses have acquired a price.
The region makes neutrality hard
APNIC's geography is not a background detail. The region includes advanced carrier markets, enormous mobile populations, developing broadband systems, export-oriented data-centre hubs, national registry structures, small island economies and operators whose customers have very different ability to pay. A single policy grammar can be formally equal while economically uneven.
Formal equality is tempting in such a region. It avoids explicit preference among economies, business models and stages of development. It gives the appearance of one community standard. But rules are not measured only by their text. They are measured by who can comply cheaply. A documentation request that is trivial for a large incumbent may be a barrier for a small ISP. A transfer fee that is minor in a large acquisition may be material for a /24 or /23 purchase. An approval delay that a cloud platform can manage with inventory may force a local provider into heavier CGNAT or lost sales.
The region's late builders face a double disadvantage. They missed much of the era when IPv4 was easier to obtain, and they now acquire addresses in a market whose price is set by global demand. When a small provider in a lower-income market buys address space, it is not competing only with local peers. It is exposed to valuations shaped by hosting firms, security services, cloud regions and enterprises with higher revenue per address. Registry friction adds another layer to that price.
National Internet Registry arrangements add complexity. NIRs can make administration more local, accessible and responsive. They can understand language, domestic corporate identity and local operator culture better than a regional secretariat can. But they also create another layer where conservation language is interpreted. The same principle, such as demonstrate need or preserve community resources, may be applied with different levels of formality, speed and local expectation.
None of this means APNIC should try to equalise every market condition. It cannot. A registry is not a development bank, competition authority or universal subsidy mechanism. But it can avoid pretending that conservation language is neutral just because it is uniform. In a heterogeneous region, every fixed burden has incidence. Every discretionary rule rewards familiarity. Every slow process protects someone who already has addresses.
Precision therefore matters. If a rule protects the final 103/8 pool, say that. If it protects record accuracy, say that. If it prevents immediate resale of a specially delegated block, say that. If it is designed to prevent fraud, identify the fraud risk. If it slows transfers because a faster market feels uncomfortable, that discomfort should be argued as policy, not smuggled into the word stewardship.
In the Asia-Pacific setting, conservation cannot be a mood. It must be an identified mechanism. Without that discipline, the same language that once protected future entrants can burden the very networks that arrived too late to benefit from abundance.
Need and the tax on late entry
Need is the most durable bridge between allocation policy and transfer policy. It appears fair because it asks for a relationship between addresses and use. It resists the idea that number resources should be accumulated merely because an organisation has capital. It gives the community a way to say that IPv4 remains part of a public numbering system rather than a free-floating commodity.
The difficulty is that need is not a natural fact waiting to be observed. It is an administrative category. Someone decides what counts as need, which documents prove it, how far into the future demand may be projected, whether customer growth must be signed or merely credible, whether transition needs count, whether leasing demand counts, whether CGNAT pools count and how much uncertainty is tolerable. Those decisions distribute access.
Late entrants are especially exposed. An established network can show historic utilisation, customer curves and existing assignments. It may have address management systems that produce reports in familiar forms. A new entrant may have contracts in negotiation, a buildout plan, a customer pipeline and a financing window. Its demand may be real but not yet legible to an old allocation template. If the template treats past utilisation as the safest proof of future need, it rewards those who had addresses earlier.
The problem is not that every forecast should be accepted. IPv4's price creates incentives to exaggerate. Fraud exists. Shell entities can be used to acquire space. Disputed corporate histories can be exploited. But risk control should be explicit. If a question is asked to test identity, say so. If it is asked to detect sham demand, define the warning signs. If it is asked because the address block comes from a protected residual pool, confine the rule to that pool. If it is asked because the registry does not like a business model, the community deserves to know that judgement is being made.
Need review can also become pro-cyclical. The operators with the strongest cash positions often have the best documentation and the greatest ability to wait. The operators for whom address scarcity is most binding often have the least administrative capacity. A rural access provider, a small hosting company, a local enterprise network or a mobile challenger may be more constrained by IPv4 than a large platform, yet less able to satisfy a polished evidentiary standard. A test designed to prevent hoarding can therefore reward organisational maturity rather than social usefulness.
The sensible reform is not to abandon need altogether. It is to segment it. Transfers of already-circulating space should emphasise legitimacy, accountability, clean records and specific risk indicators. Small routine transfers should have predictable evidence paths. Use cases should be recognised in their own terms: access networks, CGNAT, cloud services, data centres, enterprise migration, critical infrastructure and exchange points do not all demonstrate demand in the same way. Where a claim is unusual, the review can become deeper. Where a claim is standard and the source is clean, the path should be swift.
Need is legitimate when it prevents the ledger from being used as a laundering device for false demand or when it protects a clearly bounded residual pool. It becomes a tax on late entry when it asks networks to compensate, through documentation and delay, for the historical fact that they did not receive addresses earlier.
Transfers and the price of procedural virtue
Transfers are the afterlife of IPv4 allocation. They are how space moves from lower-value or unused holdings to higher-value uses after the registry can no longer satisfy demand from a free pool. A transfer market is not a betrayal of conservation. It is the predictable consequence of exhaustion. If the resource remains useful, cannot be newly produced and is unevenly held, movement will occur through sales, mergers, returns, lease-like arrangements, internal reorganisations or informal workarounds.
APNIC recognises transfers. Its public materials distinguish transfers from simple name changes and describe categories such as mergers, historical resources and unused IPv4. Current policy contains intra-region and inter-RIR transfer provisions, source and recipient conditions, minimum sizes and compatibility requirements. Public transfer logs and broker-related material exist because the market is no longer an exception. It is part of the operating environment.
The hard question is how much procedural virtue the market should be made to buy. Some friction is necessary. A registry that rubber-stamps every claimed movement would invite fraud, hijacking, disputed authority and dirty records. But friction should be priced against the risk it reduces. If a source holder is verified, the block is undisputed, the recipient is known, the transfer respects pool-specific limits and the records can be updated cleanly, the registry-integrity case for extended delay is weak.
Slow transfers have distributional effects. Sellers prefer buyers who can close. Buyers with cash, counsel and prior experience can absorb uncertainty. Smaller buyers may pay more to compensate sellers for process risk or may lose deals to larger counterparties. If inter-registry compatibility narrows available supply, regional buyers face a thinner market. If transfer fees are fixed or lumpy, small transactions become more expensive in proportion to their size. If documentation standards are ambiguous, institutional familiarity becomes an asset.
This is where liquidity control appears as a consequence rather than the main thesis. The issue is not that APNIC is a financial regulator in disguise. It is that transfer rules determine how easily address rights, operational control and registry recognition can move. A liquid market lowers search costs and helps late entrants obtain resources. An illiquid market raises effective prices and protects holders who already possess stock. Conservation rhetoric can make illiquidity sound responsible because slowness is confused with prudence.
The economics are sharper than the language. An extra week of review may look harmless to a committee, but it can change a deal's risk allocation. A request for additional evidence may be sensible in a complex case and wasteful in a routine one. A broad warning about speculation may deter abuse, or it may deter the very intermediaries and small buyers who need a visible market. The same tool can protect the ledger or protect incumbency.
APNIC's task is not to make transfers frictionless. It is to make friction legible. Every delay should map to a risk category: source legitimacy, recipient identity, disputed status, fee standing, final-pool restriction, NIR coordination, inter-registry compatibility, record cleanup or suspected fraud. A market can tolerate rules better than it can tolerate ritualised uncertainty. Procedural virtue becomes expensive when no one can tell which virtue is being purchased.
Leasing and the discomfort of temporary access
Leasing is where conservation rhetoric becomes most moralised. A sale can be recorded as a change of holder. A lease or lease-like arrangement gives operational use to one party while another may remain the registered holder or retain contractual control. That structure can create real operational problems. Records may not show who is using the addresses. Abuse contacts may be stale. Route authorisation may be unclear. A lessee may announce space without sufficient authority. A lessor may fail to supervise downstream conduct. Termination may leave routing and contact data confused.
Those are registry concerns. They are not excuses. They are precisely the sort of risks a number registry is competent to identify.
But the existence of risk does not settle the policy. Leasing also exists because permanent purchase is not always efficient or affordable. A small ISP may need temporary growth capacity. A hosting provider may need addresses for customers whose duration is uncertain. A content or security service may need migration capacity. A company may need IPv4 reachability while moving systems to IPv6. A network may need public endpoints for CGNAT infrastructure but not enough to justify buying a large block. Leasing turns a capital cost into an operating cost. For late entrants, that can be the difference between market access and exclusion.
Hostility to leasing often borrows anti-hoarding language. It suggests that leasing detaches addresses from need, commodifies a public resource and rewards historical holders. Sometimes that is true. A lessor may warehouse space, extract scarcity rent and provide little operational accountability. But a blanket suspicion can punish the wrong side of the transaction. The late entrant leasing a modest block because it cannot buy at current prices is not the historical holder that obtained abundant space in an earlier era. Temporary access can be an adaptation to inherited inequality, not evidence of moral failure.
The registry question should be operational: can temporary use be made visible enough, accountable enough and reversible enough? If records are misleading, there is a problem. If the lessor cannot provide route authorisation, there is a problem. If abuse handling fails, there is a problem. If termination produces ambiguous authority, there is a problem. But if route origin authorisation, abuse contacts, lessor accountability and termination procedures can be made clear, it is harder to claim that the payment schedule itself offends conservation.
A mature post-exhaustion policy would not romanticise leasing. It would not ask APNIC to become a commercial lease registrar for every downstream assignment. It would define minimum operational truth. Who is authorised to originate the route? Who receives abuse reports? Who can revoke use? Who remains accountable if the downstream user disappears? Which record must be updated when use changes? Those questions are closer to stewardship than general condemnation.
Conservation should not mean forcing every temporary need into permanent ownership. Sometimes conserving accuracy means acknowledging temporary use rather than pushing it outside the visible record. A registry that refuses to see a market reality does not make the reality disappear. It makes the ledger less useful.
Brokers, suspicion and market infrastructure
A post-exhaustion transfer market requires search, pricing, diligence, escrow, reputation checks and knowledge of registry procedure. These functions do not perform themselves. Buyers and sellers are fragmented. Blocks vary in reputation, route history and legal clarity. Counterparties may sit in different jurisdictions. A buyer may not know whether a block is clean, whether a seller is authorised, how long recognition will take or which registry conditions matter. Intermediaries emerge because the transaction is specialised.
APNIC's public material acknowledges this environment by listing registered IPv4 brokers under conditions of conduct while making clear that listing is not an endorsement. That posture is sensible. It neither turns brokers into public officials nor pretends they do not exist. The question is whether conservation rhetoric will treat intermediaries as a moral category or as a set of functions that can be regulated for transparency.
In a thin market, hostility to intermediaries often favours insiders. Large platforms, established carriers and experienced networks can find counterparties through reputation and existing relationships. Smaller operators, especially in developing markets, may need help discovering sellers, understanding price, screening disputes and navigating transfer procedure. If brokers are stigmatised rather than disciplined, the visible market shrinks and the private market becomes more valuable.
There are genuine risks. A broker can misrepresent policy, pressure inexperienced sellers, obscure beneficial interests, steer buyers toward dirty blocks or exploit information asymmetry. The answer is conduct regulation and transparency, not rhetorical suspicion. Clear process steps, public transfer logs, standardised warnings, dispute checks, clean statements of restricted pools and consequences for dishonest conduct are better tools than treating market intermediation as inherently anti-community.
This is one reason conservation language must be careful. A registry can say that the numbering system is not ordinary property without implying that every market service is corrupt. It can say that public records must remain accurate without implying that private negotiation is improper. It can require good-faith representation of policy without implying that price discovery is shameful.
Brokers are not proof that conservation has failed. They are evidence that post-exhaustion distribution requires infrastructure. The policy question is whether that infrastructure reduces information costs and improves record accuracy, or whether it obscures risk. A registry that keeps this distinction clear will be more effective than one that lets unease with commerce substitute for analysis.
How rhetoric hardens
Policy language does not become powerful only when it appears in formal text. It hardens through repetition: meeting debates, consultation documents, staff explanations, slide decks, mailing-list exchanges, implementation notes and routine support interactions. A phrase such as community protection can begin as a reasonable objection to waste and later become a default response to any proposal that would make transfers faster or temporary use more visible.
This is a channel, not the main story. The institutional economics lies in the incentives created by the rule; the policy process explains how the rhetoric becomes durable. Registries are consensus institutions. They rely on community legitimacy. That makes them cautious. A familiar phrase with moral weight is easier to use than a fresh distributional argument. Participants who benefit from the status quo rarely need to say that they benefit. They can invoke conservation, prudence or fairness. Late entrants, by contrast, must usually describe a concrete cost.
The asymmetry matters. Incumbency often speaks in abstractions because it already has the resource. New demand speaks in particulars: customer growth, NAT pressure, address prices, transfer delay, missing documentation, lost deals. The abstraction sounds more principled. The particulars sound self-interested. Yet economics often lives in the particulars. A fixed fee, a two-week delay, an ambiguous leasing stance or a narrow need template can do more distributional work than a formal policy statement.
Language also creates priors for staff and participants who are not trying to redistribute anything consciously. If a proposal is introduced as market liberalisation, listeners look for abuse. If the same proposal is introduced as record accuracy, they look for operational benefit. If a transfer applicant is described as seeking inventory, the word may suggest speculation; if the same inventory is described as resilience, migration space or a buffer against customer churn, the claim sounds more responsible. None of these labels proves the underlying facts. They decide which facts must be proved first. That is why post-exhaustion rhetoric deserves economic scrutiny even when the written policy appears balanced.
Rhetorical hardening also makes adjacent risks harder to separate. Scope creep is one such risk. A registry may start with a clear operational task and gradually move toward broader judgements about acceptable market conduct. That is a real concern, but it is not the central point here. The deeper issue is that post-exhaustion language can keep the institution's self-image stable while its economic role changes. The registry may still say conservation, but it is now deciding how costly it is for addresses to move.
Good governance requires periodic deflation of old words. The community should ask not only whether a proposal is consistent with conservation, but what conservation means in that specific case. Which pool? Which record? Which risk? Which operator cost? Which incumbent benefit? Without these questions, rhetoric becomes a ratchet. It expands easily, contracts rarely and is treated as neutral because it sounds inherited.
NIRs and the geography of discretion
National Internet Registries are one of the distinctive features of the APNIC region. They can make registry services more local and more usable. They can understand domestic legal forms, language, operator culture and market structure. In large economies, national mediation can be practically necessary. In smaller or linguistically distinct markets, local knowledge can reduce friction rather than increase it.
The same structure can also multiply discretion. A regional conservation norm may be interpreted through national procedures. A request can involve local records, APNIC second opinions, transfer categories, domestic expectations and cross-registry coordination. The official principle may be the same, while the applicant's experience differs.
APNIC's NIR policy framework requires national registries to implement applicable APNIC address management policies and maintain records, while APNIC remains open to direct membership rather than handing sole national control to an NIR. That balance matters. It prevents the regional registry from disappearing behind national monopolies. But it does not eliminate the economic effects of layered process.
Consider the phrase demonstrate need. In one setting it may be handled pragmatically because the local registry knows the operator and the use case. In another it may become a formal exercise with multiple rounds of evidence. In a market dominated by a few incumbents, a language of national stewardship may align with the comfort of existing holders. In a market of small providers, the same language may become a brake on expansion. Neither result is inherent in the policy text. Both can emerge from how discretion is used.
NIR mediation also affects transparency. Whois responses, data-source indicators, reverse DNS, route objects and contact records must remain intelligible to operators outside the national context. A local process that improves access but leaves the wider record confusing has not conserved the Internet's operational fabric. Conversely, a strict regional process that ignores local realities may conserve procedural neatness while raising real costs.
The answer is auditability across layers. If a request is delayed, reduced or refused, the affected network should know which criterion is at issue. If additional evidence is required, the reason should be specific. If APNIC's view changes a local assessment, the difference should be explained. If a final-pool restriction applies, it should be named rather than folded into general stewardship. If cross-registry compatibility is the obstacle, the parties should know which condition is binding.
National mediation is not the enemy of conservation. Untraceable discretion is. The geography of APNIC makes local administration valuable; it also makes transparent reasoning essential.
Fees and the politics of carrying costs
Fees are often described as administration. In a post-exhaustion address economy, they are also distribution. Membership fees, transfer fees, annual carrying costs and processing charges influence who can hold addresses, who can receive transfers and whether clean registration is cheaper than informal workaround.
There is no serious argument that registry services should be free. APNIC maintains databases, security systems, RPKI services, reverse DNS, support, training, policy meetings and operational infrastructure. Transfer review consumes staff time. Membership provides portability and a relationship with the registry. Cost recovery is legitimate.
The distributional issue is how the costs scale. A fixed charge that is modest in a large transaction can be significant in a small one. A fee due before completion may be routine for an established operator and a cash-flow burden for a new entrant. Annual fees tied to holdings can encourage return of unused space if calibrated well; they can also push small users into leasing or provider-assigned space if calibrated poorly. Low carrying costs for large dormant holdings can preserve old stock. High transaction costs for small clean transfers can make the ledger less accurate by making unofficial arrangements more attractive.
Fee debates rarely sound ideological, which is why they deserve scrutiny. They sound like budgets, categories and cost recovery. But in a market where IPv4 addresses have high value, a budget rule can shape market structure. If the cost of becoming a recognised holder is too high, small operators may remain dependent on upstream space, delay multihoming, use heavier CGNAT or accept less transparent temporary arrangements. That affects resilience, competition and customer experience.
The registry should ask empirical questions. Does a fee recover actual review and record-maintenance cost, or does it create a barrier beyond that cost? Does it scale sensibly by block size and transaction complexity? Does it encourage accurate registration? Does it make clean small transfers more attractive than opaque leases? Does it place any meaningful cost on large idle holdings? Does it publish enough information for the community to judge the incidence?
Conservation should not mean making every movement costly. Movement from unused or lower-value holdings to active use can itself be conservation, if it is accurately recorded. Sometimes a fee supports that by funding clean process. Sometimes it impedes it by adding a regressive fixed burden. The difference cannot be settled by invoking stewardship. It has to be measured.
IPv6 does not erase residual IPv4 power
IPv6 is the only durable technical answer to IPv4 exhaustion. That is true, and no serious post-exhaustion policy can ignore it. A future in which IPv4 is less necessary would reduce the rents attached to historical holdings and weaken the distributional power of transfer rules. APNIC's role in training, measurement and IPv6 allocation remains important.
But the long run has been long. IPv6 adoption is uneven across access networks, enterprises, devices, applications, content, security tools and regional services. Connectivity is bilateral. A provider cannot unilaterally decide that its customers no longer need to reach IPv4-only destinations. A network can deploy IPv6 aggressively and still require IPv4 for translation, customer support, business services, hosting, security architecture and legacy reachability.
This makes "deploy IPv6" a double-edged sentence. As technical advice, it is sound. As a response to transfer friction, it can be an evasion. Late entrants are often among the strongest IPv6 adopters because they lack abundant IPv4. Their continuing IPv4 demand is not necessarily evidence of backwardness. It is evidence that the rest of the Internet has not completed the transition.
CGNAT shows the problem plainly. It conserves public IPv4 by sharing addresses among users, but it shifts cost to operators and sometimes to customers. It requires hardware, logging, port management, abuse handling, troubleshooting and operational expertise. It can complicate lawful access requests and degrade applications that assume end-to-end reachability. For large mobile networks it may be routine; for smaller providers it can be expensive and fragile.
If conservation rhetoric treats CGNAT as proof that late entrants can simply manage with less IPv4, it undercounts the cost. Historical holders with ample space can offer simpler connectivity. Late entrants engineer around scarcity. If policy then makes it difficult to acquire modest public space for translation pools, business customers or operational resilience, conservation becomes a second burden on the same operators.
The moral hazard runs both ways. If transfers are too easy and profits too comfortable, some holders and users may delay IPv6. If transfers are too hard, smaller networks may be trapped in worse workarounds while larger incumbents enjoy the legacy map. The correct question is not whether IPv4 need should disappear. It is whether a specific rule accelerates genuine transition or merely raises the cost of incomplete transition for those least able to bear it.
A registry can promote IPv6 without falsifying present demand. It can make IPv6 easy to obtain, support training, publish adoption metrics and encourage architectures that reduce future IPv4 dependence. It can also recognise that, during transition, clean IPv4 record movement is part of operational hygiene. Punishing IPv4 dependence by making the ledger harder to update is not transition policy. It is frustration masquerading as conservation.
History as an invisible allocation rule
Every post-exhaustion IPv4 debate has an unspoken participant: the historical address map. Current holdings reflect decades of allocation norms, institutional access, business timing and regional development. Some networks received large blocks when the Internet was smaller and expectations were different. Others arrived when strict caps, transfer prices and documentary review had become normal. Present scarcity is therefore not just a technical condition. It is a distribution inherited from time.
This is especially important for APNIC because growth in the region did not align neatly with early Internet allocation. Some Asia-Pacific networks became central early. Many others expanded later through mobile broadband, domestic cloud services, e-commerce, outsourcing, enterprise digitisation and mass consumer connectivity. The faster a later economy grows, the more it feels the cost of a map drawn under earlier conditions.
Historical holdings are not illegitimate merely because they are old. Many built real networks. Renumbering is costly. Stability matters. A registry cannot casually seize or rewrite settled delegations without damaging trust and operations. The ledger must have continuity.
Continuity, however, is not moral innocence. A historical holder's current advantage is not proof of superior need. It is often proof of timing. Conservation rhetoric can obscure this by focusing scrutiny on present demand. The late entrant must prove need, avoid the appearance of hoarding and justify market acquisition. The incumbent's inventory is already normal because it is already in the record.
This creates a narrative asymmetry. A block held by an old network for future flexibility may be called reserve. A block acquired by a new network for future growth may be called speculation. The technical difference may be small. The rhetorical difference is large. One is history; the other is a transaction. Institutions are usually more comfortable defending history than defending transactions.
The lesson is not to turn IPv4 into pure private property. That would weaken the registry's ability to maintain records and enforce basic conditions. The lesson is to apply public-resource language to incumbency as well as new demand. If addresses are public resources, stale records, unresponsive contacts, ambiguous routing authority and genuinely unused historical holdings deserve scrutiny. If transfers are permitted, late entrants deserve predictable recognition rather than moral suspicion.
History cannot be undone by language. It can be softened by clean movement, accurate records, accountable leasing, return mechanisms and fee structures that do not reward dormancy. If conservation slows movement while respecting historical stock as settled, it compounds the inequality it claims to manage.
Fairness after exhaustion
Fairness is appealing because it lets a technical institution speak in civic terms. It is also dangerous because it carries multiple meanings at once. Equal treatment, priority for late entrants, protection against waste, resistance to market power, respect for consensus and continuity of records can all be called fair. They are not the same.
Before exhaustion, fairness could plausibly mean that similarly situated applicants should receive addresses according to demonstrated need and compliance with policy. After exhaustion, similarity is harder to find. Some networks have inherited stock. Some have none. Some can buy large blocks. Some can lease only small amounts. Some operate through NIRs. Some serve high-revenue customers; others serve low-margin access markets. Some can wait; others cannot. A single fairness sentence cannot reconcile those conditions.
APNIC's fairness problem is therefore distributional. Who bears the cost of the historical map? If late entrants bear it through high prices, heavy documentation, CGNAT, transfer uncertainty and moral suspicion, then conservation should not call that neutral. If incumbents bear some cost through accurate records, scrutiny of stale holdings, carrying charges and expectations to return or transfer unused space, the burden is more balanced. If brokers are regulated for transparency rather than stigmatised, small buyers gain access to information. If temporary use is made accountable rather than ignored, operators with short-term needs are not forced into permanent ownership or invisibility.
Naming distribution does not require APNIC to become a social planner. It requires institutional honesty. A final-pool transfer lock favours future direct applicants over immediate resale. A need test favours organisations whose demand is legible in accepted formats. A fixed fee favours larger transactions. A slow process favours experienced counterparties. A strict stance on leasing favours permanent purchasers. An NIR process may favour operators familiar with national structures. These effects may be justified, but they should be visible.
The temptation is to avoid this clarity because distributional language sounds divisive. Conservation sounds cleaner. Stewardship sounds less commercial. Community protection sounds more dignified than deciding who pays the fixed cost. But distribution does not disappear when the institution avoids naming it. It defaults to history, capital, familiarity and delay.
Fairness after exhaustion should be framed as fair access to accurate recognition. Networks should not receive addresses merely because they want them. But networks that lawfully acquire or use addresses should have predictable ways to make that reality visible, accountable and operationally clean. The registry should reduce the penalty for arriving late without falsifying the ledger or rewarding fraud. That is a harder fairness ideal than the old conservation vocabulary usually admits.
What conservation can still honestly do
The critique of conservation rhetoric should not become anti-registry rhetoric. APNIC still has legitimate conservation tasks after exhaustion. They are narrower than the old free-pool story, but they are real.
Uniqueness remains absolute. No block should have two legitimate holders. No transfer, lease-like arrangement or sub-use should create ambiguity about who may originate a prefix or who is accountable for it. If conservation means protecting uniqueness, it is essential.
Registration remains central. Whois data, NIR records, reverse DNS, route objects, RPKI and abuse contacts are not bureaucratic decoration. They are part of the Internet's trust fabric. A movement of addresses that leaves stale contacts, unclear authority or unresolved sub-assignments can harm operators far beyond the immediate parties.
Anti-fraud review is necessary. IPv4's price creates incentives for forged authority, hijacked companies, disputed legacy claims, misleading restructurings and abusive routing. A registry that verifies source legitimacy, recipient identity and dispute status is not obstructing the market. It is making the market safer.
Pool-specific restrictions can be legitimate when clear and bounded. Restrictions on the immediate transfer of certain final-pool delegations protect the purpose of that pool: small delegations for operational use rather than instant resale. One may debate the details, but the logic has a concrete object. It identifies a pool, a time window and a behaviour.
Routing and operational hygiene still matter. The registry cannot guarantee global routability, but it can avoid policies that needlessly fragment records, detach registered holders from operational responsibility or make authorisation difficult to interpret. A registry that lets the record drift away from reality is not conserving anything useful.
IPv6 transition remains part of stewardship when it expands future capacity rather than scolding present dependence. Training, easy IPv6 allocation, measurement and operational guidance are conservation in the broader sense of reducing future pressure. Making IPv4 record updates unreliable is not.
Minimised overhead is also a conservation value. APNIC's own policy goals include it for good reason. Excessive paperwork can damage stewardship by pushing transactions outside the record. A clean, fast, predictable update may conserve more operational truth than a slow review that encourages informal use.
Legitimate conservation is therefore less grand than the rhetoric around it. It is a discipline of unique numbers, truthful records, accountable holders, bounded residual pools, clean transfers, anti-fraud controls and transition incentives. Anything beyond that may still be defensible. It should be defended under its real name: distributional policy, competition policy, fee policy or community preference.
Auditability is the line
The line between stewardship and control is auditability. A rule that cannot be audited asks the community to trust discretion. In a post-exhaustion market, that trust is expensive, especially for late entrants.
Auditability begins with criteria. If a transfer recipient must demonstrate need, acceptable evidence should be described in practical terms. Customer demand, CGNAT design, signed contracts, utilisation history, data-centre capacity, migration plans, business services and IPv6 transition architecture may all matter in different cases. Criteria should identify the risk being reduced, not merely the document being demanded.
Time is the second element. Routine cases should have expected processing windows. If a verified source, clean block and accountable recipient still face delay, the reason should be categorised. Missing identity evidence is different from a source dispute. A fee issue is different from an NIR second opinion. A final-pool restriction is different from inter-registry compatibility. "Further review" is too vague when delay changes bargaining power.
Proportionality is the third element. A small transfer to a local ISP should not carry the same evidentiary burden as a complex corporate acquisition involving multiple blocks and historical claims. A temporary operational arrangement should not be forced into a permanent-transfer evidence model if the real risks are abuse contact and route authorisation. A new entrant should be reviewed for reality, not punished for lacking history.
Reversibility is the fourth. Conservation measures are more legitimate when they can be adjusted after evidence of harm. A transfer lock can be reviewed. A documentation requirement can be narrowed. A fee can be rescaled. A leasing interpretation can be updated if better record mechanisms emerge. Irreversible discretion is not stewardship; it is control.
Outcome publication is the fifth. APNIC need not reveal confidential business plans to publish aggregate statistics: processing times, reasons for delay, approvals, withdrawals, reductions, NIR-related timing differences, final-pool cases, inter-registry outcomes and appeal results. Such information would show whether conservation rules target operational risk or merely slow movement.
Equal scrutiny of incumbency is the sixth. Auditability should not focus only on applicants. If public-resource language is serious, stale records, unused historical holdings, unresponsive contacts and ambiguous route authorisation by existing holders matter too. A system that audits new entrants intensely while treating incumbency as natural conserves the past more than the commons.
These tests do not eliminate judgement. Fraudsters exploit rigid rules. Legal histories can be messy. Networks differ. But complexity is not an argument for broad moral language. It is an argument for decision records that show which fact mattered, which policy applied, what remedy was available and whether similar cases were treated similarly.
Genuine conservation survives auditability because it has evidence. Control resists auditability because its value lies in discretion.
A post-exhaustion compact
The economics of conservation rhetoric points toward a modest but stronger compact for APNIC. The registry should protect the ledger rather than perform moral ownership of IPv4. It should conserve what can still be conserved after exhaustion: uniqueness, truthful records, accountable authority, bounded residual pools, routing hygiene, anti-fraud controls and transition incentives. It should avoid using conservation as an all-purpose justification for discretionary control over liquidity.
The compact begins by separating resource types. Final-pool delegations can have special rules because they exist for a specific purpose. Returned non-103 space can have its own allocation logic. Historical and transferred resources already in circulation require a different emphasis: record accuracy, source legitimacy, recipient accountability and clean operational handoff. Treating these categories under one moral umbrella confuses policy.
Need review should become more risk-specific. For transfers, evidence should be standardised by use case and scaled by size. CGNAT pools, access networks, cloud services, hosting, enterprise migration, critical infrastructure and exchange points do not demonstrate demand identically. Where the source is legitimate and the block is already in circulation, the default should favour recognition unless a specified risk appears.
Temporary use should be brought closer to operational truth. Leasing and lease-like arrangements should not be celebrated uncritically, but neither should they be pushed into invisibility. Minimum expectations for route authorisation, abuse handling, lessor accountability and termination would protect the ledger better than general suspicion. The aim should be accurate responsibility, not ritual discomfort with non-permanent use.
Transfer transparency should improve. Public transfer logs already acknowledge market reality. Aggregate process data would show whether conservation protects integrity or suppresses useful movement. Fee incidence should also be reviewed through the lens of small transfers and late entrants. Cost recovery is legitimate; regressive fixed burdens that push clean transactions into informal channels are not.
Policy language should be disciplined. Conservation should be used when a rule protects a defined pool, prevents waste of APNIC-managed residual space, preserves accurate records or reduces an operational risk. Fairness should be accompanied by an explanation of who gains and who pays. Stewardship should not expand discretion without measurable criteria. Community protection should not become a veto on market mechanisms that help disadvantaged buyers.
Above all, late entry should be treated as a structural fact rather than an applicant defect. Many networks did not miss the age of abundant IPv4 because they were careless. They arrived later because economies, technologies and user demand developed unevenly. APNIC cannot rewrite that history. It can avoid compounding it. It can make recognition predictable, keep records clean, support IPv6, regulate conduct rather than roles, and ensure that old scarcity language does not become a ceremonial defence of incumbency.
The strongest version of APNIC's role is not that of a moral tribunal over every IPv4 transaction. It is the disciplined bookkeeper of a public numbering system whose entries have become economically valuable because the protocol ran out of room. That role is powerful enough. It requires restraint precisely because every sentence about conservation can move money, delay networks and protect old advantages.
IPv4 conservation once meant slowing depletion of a pool. In APNIC's post-exhaustion world it should mean something narrower and more demanding: preserve the ledger, make movement honest, keep residual allocations bounded, expose the distributional costs of friction and give late networks a fair path to recognised use. Anything less lets the old language do too much work. Anything more risks turning the registry from custodian into gatekeeper.