Summary

  • APNIC's current IPv4 regime is not a simple line outside a closed shop. It is a rationing system built from small-pool limits, returned-space treatment, a future first-come waiting list, and a transfer market that exists whenever administrative supply is too small, too slow or too uncertain.
  • The economic test is whether APNIC can keep queue discipline narrow: publish clear state transitions, match block sizes honestly, prevent gaming without business-plan control, and remain a recordkeeper of scarcity rather than a gatekeeper over who deserves scarce capital.

The useful scene is not a public conference microphone. It is a finance meeting at a regional network operator that has already committed to routers, access circuits, customer installations and cloud interconnection. The network can live with IPv6 for part of the design. It can run private addressing internally. It can buy managed NAT capacity. It can sign upstream contracts. But a slice of the customer base still needs public IPv4 reachability, and not in a theoretical way. Payment providers want stable endpoints. Enterprise customers want allowlists. Older equipment wants direct addressing. Abuse teams want attribution that survives a complaint. The planned expansion needs more than a token block, but the administrative channel can provide only a small amount and only if the applicant fits the current rule.

That is where the waiting-list economy begins, even before the queue formally fills. APNIC's present policy architecture still allows new and existing members to receive a small amount of IPv4 space from the remaining pool. The ordinary ceiling is a total /23 from the 103/8 pool, after the 2019 reduction from the earlier /22 limit. The minimum delegation size is a /24. Recovered non-103/8 addresses no longer sit behind the old separate waiting list; since July 2019 they have been treated as part of the remaining IPv4 pool. APNIC's policy text also says that once all IPv4 addresses run out, requests from new applicants will go onto a first-come waiting list, and the single pool of recovered and IANA-delegated addresses will feed those waiting-list requests as addresses become available.

This sounds tidy because it uses familiar administrative words: pool, request, first come, available, recovered. But a queue is not only a procedure. It is an allocation price expressed in time and uncertainty. If the operator needs a /21 and the pool can ration a /23, the queue does not solve the engineering plan. It provides a small anchor around which the operator must buy, lease, translate, delay, split customers, or redesign. If the operator needs addresses by the next procurement milestone and the future queue cannot promise delivery, the queue is a planning risk. If returned space arrives in odd fragments, a place in line may not equal the block size the network can use. If anti-gaming rules make eligibility depend on who has already received what, the queue becomes a strategy game around corporate accounts, timing and transfer restrictions. If the market can deliver a block at a known price while the queue cannot deliver a known date, the market becomes the real outside option.

Waiting lists are often defended in the language of fairness. That is the weakest defence. Fairness is too abstract to discipline the institution. A queue can be fair in appearance while creating high hidden costs. It can be first come but not useful. It can treat unlike applicants alike while rewarding the applicant whose project can wait. It can avoid explicit price but impose a working-capital cost on every network that keeps equipment, customers and financing idle. It can prevent one obvious form of speculation while encouraging less visible forms of account structuring. It can preserve a public story of equal access while telling the market that serious demand must move elsewhere.

The better defence of a queue is narrower. Where a registry still holds a small administrative pool, it must decide how to distribute it without becoming a price regulator or a planner of business models. A waiting list can be legitimate if it performs a modest function: it turns tiny, irregular, returned supply into predictable recognition events; it avoids discretionary favour; it gives small applicants a transparent path to a minimal public block; it records why a request was complete or incomplete; it tells the market what the administrative channel can and cannot do. The queue is not legitimacy by itself. It is a test of whether the registry can ration scarcity without pretending to own it.

The Returned Block That Does Not Fit

Imagine the returned block as a used machine coming back to a dealer. It may be real. It may be valuable. It may not be immediately fit for resale. APNIC's own explanatory material distinguishes available space from reserved space and notes that reserved space may include voluntarily returned space undergoing quality checks or reclaimed space awaiting administrative clearance. That distinction matters. Returned IPv4 is not automatically clean inventory. It can carry routing memory, reputation damage, old reverse DNS habits, stale customer references, corporate authority questions and registry-state residue. A recovered prefix can be technically unique yet commercially awkward.

The applicant in the queue does not experience that as institutional nuance. It experiences delay, mismatch and uncertainty. Suppose a /24 becomes available but the next applicant's business plan needs 700 addresses in the first year. A /24 may be enough to begin, but it may not satisfy enterprise product promises. Suppose a /22 appears but policy, previous receipts or conservation rules prevent giving the whole block to the next applicant. Splitting it may create more beneficiaries but less usefulness. Suppose a returned range needs a quarantine period because of abuse reputation. The queue position is then not a delivery date; it is a contingent claim on usable inventory.

That is why block-size mismatch is central to waiting-list economics. IPv4 is not perfectly divisible in practice. Routing tables, minimum transfer sizes, customer segmentation, reverse DNS, RPKI, geolocation and reputation all create lumpy value. A /24 has special significance because it is the common smallest globally routable unit in many operational contexts. Smaller fragments may be policy-visible but less useful in the routing economy. Larger fragments may be more valuable because they reduce operational complexity, but rationing rules may prevent one applicant from receiving the entire quantity. The queue therefore allocates not only addresses but fit.

Fit is where time becomes capital. A network that receives too little must add another source. It may buy a transfer block. It may lease. It may use CGNAT for residential customers and reserve public addresses for business customers. It may keep a product in pilot mode until the address plan is credible. It may launch in one city before another. It may promise customers later migration. The cost is not merely the market price of addresses. It is the cost of holding the whole commercial plan in a half-deployed state.

This is why APNIC's /23 cap has an economic meaning beyond conservation. The cap keeps small quantities available for more account holders, especially after the final /8 policy architecture. It also makes clear that APNIC's administrative channel is not a growth supply channel. An applicant that needs more than a /23 is directed toward transfers. That is a remarkably important institutional statement. It means the queue and the market are not substitutes of equal power. The queue supplies a minimal foothold. The market supplies scale.

The difference should be made explicit, because confusion benefits no one. If APNIC's administrative supply is a starter block, the queue should be evaluated as a starter-block mechanism. It should not be sold as an answer to growth demand. If the transfer market is the scale mechanism, APNIC's duty is to keep the transfer record accurate, predictable and neutral, not to pretend that a waiting list can defeat scarcity. The danger is not that the queue is small. The danger is institutional romance about smallness.

Time As The Quiet Price

Markets show price with money. Queues show price with time. That makes queues politically attractive and economically dangerous. A money price can be attacked, audited, compared, financed or hedged. A time price hides inside procurement slippage, customer churn, delayed revenue, duplicate architecture and the option value of patience. It is paid unevenly because not all networks have the same ability to wait.

The waiting-list applicant pays before receiving anything. It must prepare documentation. It must keep account standing. It must respond to questions. It must maintain a plausible deployment plan. It must decide whether to purchase equipment before address certainty. It must decide whether to sign customers before public-address certainty. It must decide whether to use leased addresses as a bridge and then migrate. If the queue moves slowly, the applicant has bought an option with uncertain exercise date. If the queue moves suddenly, the applicant must be ready to accept, configure and pay for a block that may no longer match the revised plan.

This is not the same as the new-entrant disadvantage problem. A future article can examine how incumbents begin with inherited address stock, history and bankability. The queue article has a narrower object: how the rationing device prices scarcity through timing. An incumbent can also face queue timing when a business unit wants a small additional block. A new entrant can avoid the queue by buying or leasing. A public institution can prefer waiting because procurement rules dislike informal market sourcing. A cloud customer can bypass the problem by renting platform public addresses, while losing portability. The queue is not an entrant story. It is a time-price story.

Time price matters in Asia Pacific because APNIC's region contains very different timing regimes. Some applicants operate inside mature markets with transfer brokers, legal counsel, predictable banking and cloud alternatives. Others operate in economies where foreign-exchange approval, import cycles, local-language documentation, NIR coordination or public procurement can make a month of uncertainty much more expensive. A first-come rule looks equal on the registry page. It does not make the cost of waiting equal.

The strongest queue therefore needs time transparency. It should publish enough state information to let applicants model the option. How many completed requests are ahead? What block sizes are likely? How much returned space is available, reserved, under quality review or administratively blocked? What happened to recent returned blocks? What reasons caused applications to be treated as incomplete? How often does a place in line translate into a usable delegation? The purpose is not to create a trading screen for administrative supply. It is to prevent the queue from becoming a fog machine.

Opacity creates two bad incentives. First, applicants overbuild around hope. They wait too long because the official channel might deliver. Second, applicants abandon the queue because they cannot price waiting. In both cases, the registry has failed to tell the truth about scarcity. A disciplined registry would say, in effect: this pool can offer small, uncertain quantities; the queue will be processed by clear state rules; anything beyond that belongs to transfer, lease, IPv6 transition and private planning. Such honesty may sound less generous, but it is more useful.

First Come Is A Rule, Not A Theory Of Justice

First-come ordering is attractive because it is easy to understand and hard to defend against only when manipulated. But first come is not a theory of economic justice. It rewards early readiness. Readiness is partly effort, partly information, partly administrative capacity, partly luck. The applicant that learns the rule first, assembles papers fastest, keeps the right account status, and understands how to make a request complete may beat the applicant with more urgent service demand. That is not necessarily wrong. It is simply what the rule does.

The alternative is not obviously better. If the registry ranks by need, it becomes a business-plan judge. If it ranks by public value, it becomes a development agency. If it ranks by willingness to pay, it becomes an auctioneer. If it ranks by technical sophistication, it favours incumbents and consultants. If it tries to rotate by economy, it imports national politics into a regional registry. A narrow first-come list may be less ambitious because ambition at this layer becomes discretion.

The practical question is therefore what first come means. Does time begin when the applicant clicks a form, when APNIC receives all required evidence, when APNIC determines the request is complete, when fees are paid, when account status is clean, or when the applicant passes a review? Each choice allocates value. A completeness rule discourages low-quality speculative filings, but it gives APNIC staff power over the moment the clock starts. A submission-time rule is easy to timestamp but encourages thin filings made only to reserve a place. A payment-time rule creates cash-flow and banking bias. A review-completion rule favours applicants with experience answering registry questions.

The queue can be legitimate only if these moments are publicly defined. It should not be enough to say first come. The clock must have a visible start. Incomplete applications should receive reasoned notices. Cure windows should be short and predictable. Rejection reasons should be categorized. Refiling should not become a way to jump. Staff should have room to correct errors, but not to invent priority. If the registry cannot explain why applicant A is ahead of applicant B without revealing confidential business data, the queue is too discretionary.

This matters because a queue for scarce IPv4 has a market value. A place in line can influence whether a buyer delays a transfer, whether a lender funds a rollout, whether a supplier offers payment terms, or whether a customer signs. Even if the queue position is not transferable, it affects private decisions. Anything that affects private decisions becomes a governance surface. The standard should be boring precision.

The first-come rule is also vulnerable to account strategy. If each account holder can receive only a capped quantity from the administrative pool, corporate structure matters. A group with subsidiaries, affiliates, joint ventures or country units may have more ways to present demand than a single-company operator. APNIC's policies and account rules can restrict obvious abuse, but anti-gaming rules can easily become corporate-control review. The line is thin. The registry may need to verify that an applicant is real, distinct, responsible and not using shells to multiply access. It should not become a corporate strategy tribunal.

Anti-gaming design should focus on objective facts: common control, shared infrastructure, prior receipts, address-use declarations, transfer history, and account relationships. It should publish the categories of facts that matter. It should avoid moral speculation about whether a business model is worthy. The test is not whether APNIC likes the applicant's commercial plan. The test is whether the queue is being used to defeat the rationing rule.

The Transfer Market Outside The Door

APNIC's exhaustion page is blunt about the scale problem: members can still receive small amounts, but if they need more than a /23, they should consider an IPv4 transfer. The transfer page describes transfers as movement of number resources from one legal entity to another. It lists unused IPv4 and AS number transfers, merger or acquisition transfers, and historical-resource transfers. The current policy says APNIC will process and record IPv4 transfers subject to conditions, maintain a public transfer log, and impose limits such as the /24 minimum transfer size and the five-year non-transferability of 103/8 delegations.

That transfer market is not an embarrassment to the queue. It is the reason the queue can remain narrow. If APNIC tried to make the waiting list the main answer to demand, it would have to decide who deserves scarce growth capacity. If it tried to suppress the market because the queue looks fairer, it would trap demand behind administrative scarcity. If it tried to price the market, it would become a regulator of private capital. The cleaner settlement is to recognize the difference: the queue allocates residual administrative supply; the market reallocates already-held space; the registry records transfers so the public record remains accurate.

The outside option changes the applicant's calculation. Waiting may be rational if the project can start with a small block, if the applicant's address need is modest, if procurement rules prefer APNIC delegation, or if market prices are temporarily unattractive. Buying may be rational if timing is decisive, if the needed block is larger, if clean history matters, or if a board wants a capital asset rather than a service dependency. Leasing may be rational if demand is temporary or uncertain, though it creates renewal and control risk. CGNAT may be rational for some subscribers, while imposing logging, support and reputation costs. IPv6 deployment may reduce future dependence, while not eliminating near-term IPv4 reachability needs.

The queue is one option among these, not the master option. That is the economic truth. The more APNIC recognizes it, the more legitimate its queue becomes. The less APNIC recognizes it, the more the queue becomes a symbolic promise that serious networks route around.

Transfers also discipline queue rhetoric. If a block has a market price and the queue offers a tiny administrative allocation at ordinary registry cost, the queue confers value. That value need not be corrupt. It can be a deliberate policy choice to give each account holder a small chance at basic public IPv4 continuity. But the five-year non-transferability rule for 103/8 delegations shows the concern: without restrictions, a low-cost administrative block could be captured for resale rather than use. Anti-flipping rules are therefore part of queue economics.

The challenge is to prevent flipping without freezing useful movement. A five-year restriction on final-pool delegations is easy to understand. It protects the starter-block purpose. But any restriction also creates edge cases. What if the recipient merges? What if the business fails? What if the network pivots? What if an address block is more useful elsewhere before five years have passed? Policy can require return to APNIC where the original reason no longer exists. It can define merger treatment. It can permit narrowly supervised changes. But it should keep the reason visible: the restriction protects the rationing design, not a theory that the registry owns the economic future of the block.

The Subsidy Should Be Named

A waiting-list allocation from a scarce administrative pool is a subsidy, even when nobody writes the word. The subsidy is not necessarily paid from APNIC's budget. It is paid by the difference between the cost of receiving a small delegation through the registry channel and the market value of comparable IPv4 space. The recipient receives more than a database update. It receives a scarce input that other networks may have to purchase, lease or work around. Pretending that no subsidy exists makes the policy harder to evaluate.

Naming the subsidy does not mean condemning it. A regional registry may reasonably preserve a small administrative path so that a network can obtain a minimal routable block without immediately entering the transfer market. That can support basic multi-homing, public service reachability, transition architecture and operator autonomy. It can also give a new account holder enough public identity to participate in routing, abuse handling and customer proof. The problem begins when a small subsidy is treated as if it can solve a large supply problem.

The subsidy has three design questions. Who is eligible? What size is useful? How is resale prevented without trapping legitimate change? APNIC's current architecture answers those questions partly through account-holder limits, the /24 minimum, the /23 ceiling, use-plan requirements, return obligations and five-year transfer restrictions on 103/8 delegations. Those tools are not merely administrative details. They are the price controls around a scarce benefit.

If eligibility is too broad, the subsidy is diluted or captured. If eligibility is too narrow, the registry becomes a judge of acceptable business forms. If the size is too small, the allocation becomes symbolic and pushes serious demand back to the market. If the size is too large, the pool drains quickly and the first applicants receive a windfall. If resale restrictions are too weak, the queue becomes a low-cost acquisition channel for later monetization. If resale restrictions are too strong, a block can become stranded in a failed or changed business while other networks need it.

The hard part is that these trade-offs cannot be solved by moral vocabulary. "Fair access" does not tell APNIC whether a /23 is better than a /24, whether a returned /22 should be split, or whether an applicant with related companies has crossed the line into gaming. "Conservation" does not tell a network whether it should wait six months, buy a transfer block, or use CGNAT for one more product cycle. "Community benefit" does not reveal the opportunity cost of an address block sitting in a queue while a buyer and seller could close a transfer.

The economic answer is to make the subsidy small, explicit and bounded. Small means the administrative path is a foothold, not a substitute for the market. Explicit means applicants and observers can see the rule, the queue state and the reason for restrictions. Bounded means the registry does not extend its subsidy logic into transfers, leases, financing, customer geography or product strategy. A queue subsidy can be legitimate when it is narrow. It becomes dangerous when it becomes a general claim that the registry should decide how scarce IPv4 value moves.

This is also why the queue should avoid theatrical scarcity. APNIC does not need to dramatize exhaustion to justify careful allocation. Scarcity is already evident in the /23 cap, the transfer guidance, the public transfer log and the fact that recovered space matters. The serious task is operational precision. How many small blocks can be distributed without creating perverse incentives? How long should a recipient be restricted from transfer? What evidence shows real use without inviting intrusive review? How should a failed business return or transfer a block without wasting it? Those are not slogans. They are the accounting of a subsidy that happens to be denominated in addresses rather than cash.

The Conservation Doctrine After Exhaustion

Conservation once had a simple administrative function. When a free pool existed, a registry needed criteria to avoid waste. Applicants asked for addresses; the registry checked need; the common pool lasted longer. That logic was imperfect, but it had a concrete task. After exhaustion, conservation changes character. It no longer allocates abundance. It manages scarcity after market value has emerged.

This is the heart of the Heng Lu doctrine that matters here: scarcity should narrow registry power, not expand it. A registry can protect uniqueness, accuracy, transfer records, security assertions and continuity. It should not turn scarcity into institutional rent. It should not use a queue to decide which business models are morally acceptable. It should not treat the absence of free-pool abundance as a reason to become a gatekeeper over capital.

APNIC's waiting-list architecture is a live test of that discipline. The policy can be read in two ways. The narrow reading is defensible: APNIC has a small remaining pool; it caps ordinary delegations; it combines recovered space into one pool; it will create a first-come list when the pool is exhausted; it points larger demand to transfers. That is a modest rationing architecture. The broader reading is dangerous: because IPv4 is scarce, APNIC may decide who gets operational capital, slow or shape market behaviour, and preserve administrative authority under the language of community stewardship. The facts do not force the broader reading. Institutional incentives may.

The conservation doctrine becomes harmful when it forgets the difference between a free-pool allocation and a market transfer. An applicant receiving a scarce APNIC starter block from the remaining pool can reasonably face rationing conditions. A buyer paying market price for already-held space is different. The buyer's willingness to pay and to bear operating risk is a strong signal of need. APNIC can require proof of control, accurate records, no unresolved dispute, account standing and a use plan where policy requires it. It should not convert transfer recording into a judgment about whether the buyer's growth plan is worthy.

The queue also should not be used to shame the market. Market transfers are not a failure of the registry. They are what happens when a scarce production input is already distributed and demand changes. A network that buys addresses is not necessarily hoarding. A network that waits for a small delegation is not necessarily virtuous. A network that leases is not necessarily evasive. Each choice reflects timing, block size, capital, risk tolerance and customer promise.

Queue legitimacy therefore depends on humility. APNIC should be able to say: this is what the administrative pool can do; this is what it cannot do; these are the facts that determine order; these are the safeguards against account gaming; these are the reasons a block is available, reserved, under review or reclaimed; these are the transfer paths for demand that exceeds the pool. That humility makes the queue stronger because it refuses to pretend that procedure abolishes scarcity.

Asia Pacific Makes The Queue Harder

APNIC's region is not a single market. It contains large developed networks, fast-growing mobile markets, island economies, public-sector networks, university and research systems, hosting firms, cloud users, NIR-served economies and cross-border operators. A queue rule that looks simple in one jurisdiction can be expensive in another. The region's heterogeneity is not background colour. It is the mechanism that makes waiting-list rationing difficult.

Consider block-size demand. A dense urban access provider may need addresses for customer-premises equipment, enterprise static services, support segmentation and abuse attribution. A small island network may need fewer addresses in count but more certainty because upstream alternatives are limited. A fintech or hosting company may need reputation-clean space because payment and security systems are sensitive to address history. A public-sector network may need continuity more than volume. An Internet exchange may need special treatment because the value lies in the peering fabric rather than ordinary subscriber count. The same /24 can have different economic force in each case.

Consider timing. A mobile rollout tied to a handset subsidy cycle, a public tender, or a submarine-cable landing cannot always wait for an uncertain administrative queue. A university or public agency may wait longer because procurement is slow anyway. A hosting firm may buy immediately because customers can leave next month. A rural operator may use CGNAT because revenue per user cannot support market purchase at the desired scale. The queue does not merely distribute addresses. It selects for projects whose timing can tolerate it.

Consider documentation. In a region with multiple languages, legal systems and NIR channels, the definition of a complete request can carry hidden burdens. A completeness rule may be necessary to stop speculative filings, but it should be treated as economic power. If the applicant must translate corporate documents, reconcile NIR records, prove upstream use, demonstrate an immediate /24 need, or show a plan for a /23 within a year, the queue position is partly a function of paperwork capacity. That capacity often sits outside engineering merit.

The answer is not to remove documentation. Fraud, shells and speculative filings are real risks when a scarce block has market value. The answer is to keep documentation tied to registry facts. Who is the applicant? What account is responsible? Has the applicant already received its capped delegation? Is the requested size within policy? Is there immediate need where policy requires it? Is the deployment plan specific enough to show that the request is not a resale strategy? Are there common-control facts that indicate gaming? These are legitimate registry questions. Whether the applicant's retail product is socially attractive is not.

The Queue Position As A Financial Option

A waiting-list place is not property in the ordinary sense. It should not be transferable like an address block. Yet it behaves like a financial option because it affects future access to a scarce input at a registry-defined cost. That option can influence investment decisions even if it cannot be sold.

If the option is likely to deliver a /24 within a predictable period, the applicant may delay a market purchase and finance the rest of the network. If the option is likely to deliver nothing useful for years, the applicant may buy, lease or redesign. If the option may deliver a block whose history is uncertain, the applicant may reserve a reputation-clean transfer block for sensitive services and use the APNIC block for lower-risk functions. If the option is vulnerable to incompleteness disputes, the applicant may hire consultants. Every one of those choices has a cost.

This is why queue information is not a courtesy. It is market infrastructure. APNIC does not need to publish confidential applicant details. It can publish aggregate queue depth, recent movement, block-size categories, available/reserved/review states, and reason-code statistics. It can separate ordinary pool movement from special reserves. It can explain how returned blocks move through quality checks. It can state whether a request was skipped because of size mismatch, incompleteness, ineligibility or applicant non-response. These data points reduce speculation without turning the queue into a market.

The queue option also interacts with transfers. Suppose an applicant can buy a /22 today but expects a /23 from APNIC later. It may buy less now, lease temporarily, or split products. Suppose a seller knows many buyers are waiting for administrative supply. The seller may lower price if queue movement is visible or hold price if APNIC supply is opaque. Suppose a lender values a network's address plan. A clear queue position may support bridge financing; a vague expectation will not. The registry may insist that it is not pricing addresses, but its queue transparency still affects prices.

This is not an argument for APNIC to manage market price. It is the opposite. The best way not to manage price is to publish objective registry facts and then step back. Price discovery belongs to buyers, sellers, lessors, lenders and operators. Registry facts belong to APNIC. Confusing the two is how a recordkeeping institution drifts into economic control.

Anti-Gaming Without Suspicion As Governance

Scarce administrative supply invites gaming. A rational institution has to anticipate it. Applicants may file early without serious plans. Related companies may split requests. Holders may try to obtain a low-cost block and later monetize it. Consultants may sell queue navigation. Applicants may overstate immediate need. Returned-space expectations may encourage strategic timing. The existence of a public queue can turn policy literacy into an advantage.

The wrong response is suspicion as a governance style. If every applicant is treated as a possible abuser, the queue becomes slow, intrusive and discretionary. Staff ask for more evidence than the decision requires. Applicants learn to write for institutional taste. Those with counsel do better. Those with real but messy demand do worse. The queue becomes not first come, but first to satisfy an unwritten standard.

The right response is rule-based anti-gaming. If the concern is multiple accounts under common control, define common-control evidence. If the concern is resale, define transfer restrictions and return obligations. If the concern is speculative requests, define completeness and cure. If the concern is inflated plans, require a bounded deployment plan tied to the size requested. If the concern is returned-space contamination, use published quality states. If the concern is staff discretion, publish reason codes and appeal paths.

There is a crucial difference between verifying facts and judging motives. APNIC is competent to verify account status, prior delegations, policy limits, contact authority, dispute status and documented deployment claims. It is much less competent to decide whether a company's business model is admirable, whether a market purchase would be more efficient, whether a customer segment is deserving, or whether the applicant's patience is virtuous. Scarcity tempts institutions to climb from fact verification into motive judgment. That climb is where legitimacy leaks away.

Anti-gaming should also be symmetric. If an applicant cannot game the queue, the institution should not game the applicant. Rules should not change retroactively in ways that strand completed requests. Staff should not delay clarity to manage optics. Reserved blocks should not remain indefinitely mysterious. Exceptions should be logged. If a block is skipped because it does not match the next request, the reason should be visible at category level. A queue that asks applicants for discipline owes discipline in return.

What A Good APNIC Queue Would Prove

A good APNIC waiting list would not prove that administrative rationing is superior to markets. It would prove something narrower and more valuable: that a regional registry can handle the tail of scarce supply without turning scarcity into discretionary power.

It would prove that returned-space state is legible. Available, reserved, quality review, reclaimed and administratively blocked should not be vague buckets when applicants are making capital decisions. It would prove that block-size matching is honest. If the pool mostly contains small fragments, the queue should say so. If a /23 cap means most serious expansion must use transfers, the queue should say so. If a particular returned range has reputation or routing-history issues, APNIC need not shame a prior holder, but it should have a process for condition assessment.

It would prove that first come has a clock. The start event should be objective. Completeness decisions should be reasoned. Cure periods should be clear. Applicants should know when a request is alive, paused, rejected or satisfied. It would prove that anti-gaming is rule-based. Common-control review, previous-delegation checks, non-transferability periods and return obligations should be explicit. It would prove that appeals are possible without converting every queue dispute into litigation. A registry that cannot correct a queue error cheaply will create expensive private hedges.

It would also prove that APNIC understands the transfer market as a complement. The queue should not be a moral weapon against buyers and sellers. It should not imply that market supply is illegitimate because administrative supply is fairer. It should not pretend that waiting solves block-size demand. The transfer market is where many networks will find scale. APNIC's role is to process and record transfers accurately, maintain the public record, and avoid using transfer conditions as a hidden continuation of free-pool control.

Finally, a good queue would prove restraint in language. Scarcity is often surrounded by words that inflate institutional authority: stewardship, community, conservation, fairness, protection. Some of those words have legitimate historical uses. None should obscure the present task. The registry has a book of unique number-resource records. It should keep the book accurate, prevent duplicates, record transfers, preserve security-adjacent facts, and distribute the last administrative remnants by rules that can survive scrutiny. It need not become the conscience of the address economy.

The Real Legitimacy Test

The real legitimacy test is what happens when the queue disappoints. A popular queue at the moment of promise is easy. A legitimate queue after years of scarcity is harder. Applicants will be told that the block is too small, the pool is empty, the request is incomplete, the returned range is not ready, the applicant has already received enough, the transfer restriction applies, or the waiting list cannot promise timing. Those disappointments are where institutional character appears.

If APNIC explains disappointment with objective rules, applicants can adapt. They may buy. They may lease. They may defer. They may use CGNAT. They may deploy IPv6 more aggressively. They may redesign products. They may challenge a decision. They may decide not to enter a market. These are costly choices, but at least they are choices made against a visible constraint.

If APNIC explains disappointment with broad discretion, applicants hedge against the institution itself. They buy more than needed because the administrative channel cannot be trusted. They build account structures to preserve options. They hire intermediaries for process knowledge rather than technical value. They discount APNIC-registered resources in diligence. They treat a registry request as political exposure. That is how a queue intended to preserve legitimacy can consume it.

APNIC's situation is more disciplined than many alarmist accounts suggest. The policy record contains limits, dates, pool treatment, transfer paths and a future waiting-list design. It is possible to read the architecture as a rational attempt to stretch a small pool, reserve a minimal entry path, and send larger demand to transfers. But discipline is not self-executing. As scarcity deepens, every queue decision becomes more valuable. The temptation to defend institutional authority in the name of conservation will grow. The market outside the door will grow more important. The pressure to distinguish real applicants from strategic applicants will increase. The cost of delay will become more visible.

That is why waiting-list rationing belongs in the RIR Watchdog series. It is not a procedural footnote. It is a compact model of post-exhaustion governance. Inside it are the old allocation era, the new transfer market, the politics of fairness, the economics of time, the lumpy reality of block size, the risk of account gaming, the need for public state data, and the question of whether a registry can remain a ledger when the thing it records has become capital.

The small network at the planning table does not need a sermon about scarcity. It needs to know whether the administrative path can produce a usable block in time. It needs to know whether a returned prefix is clean enough for customers. It needs to know whether a place in line is worth waiting for or whether the transfer market is the real supply channel. It needs APNIC to be precise about what it can do and modest about what it cannot.

That modesty is not weakness. It is the only durable source of authority left to a registry after exhaustion. A queue can ration. It can signal. It can reduce arbitrary allocation. It can preserve a minimal path to public IPv4. But it cannot repeal scarcity, and it cannot substitute for capital markets at scale. The moment it pretends otherwise, it stops being a queue and becomes a gate.

APNIC should keep the queue a queue.

Sources and Further Reading