Summary

  • Needs-based review began as a way to ration addresses from a shared unallocated pool. In a paid secondary transfer, the resource is supplied by an existing holder and the buyer bears a market price. Applying the old allocation test after price agreement changes the registry's function from conserving a pool to deciding whether the buyer may deploy capital in the chosen quantity.
  • Current regional rules reveal that the needs test is a policy choice, not a technical necessity. ARIN generally requires documentation of at least 50% use of the requested block within 24 months for larger specified-recipient transfers. APNIC requires a detailed 24-month use plan and, for existing holders, evidence about past use and policy compliance. LACNIC requires an inbound recipient to justify an initial or additional allocation under current rules. RIPE removed general IPv4 need documentation after depletion, while retaining a 50%-within-five-years plan for inbound transfers from regions that require compatible needs-based policies.
  • Price is not a complete anti-hoarding safeguard. A well-funded speculator can buy inventory, and concentrated holdings can reduce future supply. Yet a registry forecast of customer growth, equipment timing or network design is also a weak anti-hoarding test: it favours incumbents with familiar utilisation histories, penalises new models, invites inconsistent judgment and can force buyers to disclose commercially sensitive plans.
  • Legitimate controls should target observable conduct: authenticated parties, no conflicting claim, exact resource identification, transparent holding periods for recently received space, restrictions on rapid resale of subsidised pool allocations, beneficial-control disclosure for linked accounts, concentration reporting and reasoned investigation of sham use. They should not let a registry choose which ordinary network expansion deserves financing.
  • Pre-approval can move uncertainty ahead of price negotiation, and both ARIN and APNIC offer a 24-month form of it. That reduces settlement risk but does not answer the constitutional question. A predictable early gate is still a gate; it should be bounded, appealable and based on objective evidence.
  • Number Resource Society provides a positive direction: a portable ledger service that verifies control, uniqueness and finality, publishes narrow anti-hoarding rules in advance and leaves price, financing, customer strategy, deployment sequencing and ordinary routing choices to holders and their counterparties.

The registry enters the deal after the hardest number is already known

A buyer has agreed to pay for a specified IPv4 block. The price has survived negotiation. The seller has decided that the consideration exceeds the value of retaining the addresses. The buyer has decided that the expected benefit of deployment exceeds the purchase price, financing cost, transfer expense and technical transition risk. Lawyers may have signed. Escrow may be funded. A broker may have earned a conditional commission. Engineers may have reserved time with upstreams.

Then the registry asks the buyer to prove it needs the block.

That request can sound routine because needs tests are older than the transfer market. For decades, an applicant seeking addresses from a registry's available pool had to explain projected use. The registry was allocating a scarce common inventory at an administrative price. If one applicant received more than it could use, later applicants faced earlier depletion. Forecast review was part of the rationing method.

A paid transfer changes the institutional facts. The RIR is not selecting the buyer from a queue for a free or heavily subsidised block. The source is an existing holder. The buyer has exposed itself to a scarcity price. The seller has an alternative use for the asset or proceeds. The registry's record remains essential, but the economic allocation has occurred through a bargain.

When the RIR can reduce or reject the quantity after that bargain, it is exercising a capital-allocation right. It decides that the buyer may not invest in the asset on the terms it selected, even if the seller is authorised, the resource is uncontested and the record could be updated accurately. The decision affects financing, launch schedules, customer capacity and the market value of the seller's inventory.

This does not make every needs test irrational. It changes the burden of justification. A rule designed to protect an unallocated pool cannot be carried into a secondary market by habit alone. The registry must identify the harm it is preventing, use evidence connected to that harm, and stop before review becomes a general judgment on the buyer's business plan.

Needs review was built for an allocation queue

Traditional allocation policy tries to solve a familiar common-resource problem. A registry has a finite stock. Applicants have better information about their own demand than the administrator. Each has an incentive to ask for more because unused capacity may have option value and the administrative fee does not fully price scarcity. The registry therefore asks for network plans, utilisation of earlier blocks and a forecast period.

The test performs two functions. It screens exaggerated requests and determines size. Those functions are connected to the pool: approving less for one applicant leaves more for others. The decision can still be imperfect, but its institutional entity is visible.

Exhaustion breaks that connection. Once a region no longer has enough ordinary inventory to satisfy demand, a transfer buyer is not taking the requested block from the registry's shelf. The block moves from a willing source. Refusing the buyer does not automatically return it to a neutral pool or award it to the next applicant. The seller may seek another buyer, retain it, lease operational use, reorganise the holding or wait.

The old test can then produce the opposite of conservation. If formal transfer is too uncertain, parties may prefer arrangements that do not change the top-level registration record. The addresses can still be used under a service, lease, letter of authorisation or corporate structure, while public records remain less informative. APNIC's historical discussion of transfers recognised this trade-off: needs rules may deter stockpiling, but barriers can also encourage unregistered activity that weakens registry accuracy.

The economic distinction is not between public good and private greed. It is between two rationing systems. An allocation queue uses administrative judgment to distribute inventory. A transfer market uses price and contract to move existing holdings. A registry may add narrow safeguards to the latter, but if it recreates the former in full, the market price becomes only the first of two permissions.

ARIN makes the capital screen explicit

ARIN's current Number Resource Policy Manual gives the clearest formal example. For specified-recipient transfers, the receiving organisation must enter the relevant registration agreement and use transferred resources on an operational network. An organisation automatically qualifies for the minimum transfer size if it has no ARIN IPv4 allocation. For a larger initial block or an additional block, it may qualify by documenting use of at least 50% of the requested block within 24 months.

ARIN also looks backward. An organisation with existing ARIN IPv4 allocations generally must have efficiently used at least 50% of its cumulative holdings before receiving more through transfer. An alternative path can use 80% utilisation of current space and includes additional limits. These rules make the registry an assessor of both forecast and historic deployment.

The policy is predictable enough to describe, and ARIN offers transfer pre-approval based on projected 24-month need. Once granted, the public guidance says a qualifying transfer submitted within the approval period is not re-verified on the same need. That service can substantially reduce the danger of agreeing price first and discovering the quantity is ineligible later.

Yet the decision right remains significant. Suppose a buyer can finance a /17, expects a volatile customer launch and wants reserve capacity to avoid returning to a thin market in six months. If its documentation supports only 50% of a /18 within the period, the registry may approve a smaller quantity than the bargain. The RIR has not merely checked identity or title. It has rationed the scale of investment.

The test can favour familiar forms of demand. A mature ISP can show existing customer assignments and utilisation history. A new cloud, security, access or translation service may depend on a launch whose customers cannot be named in advance. A business buying for acquisition integration may know the strategic need while deployment depends on systems not yet consolidated. A network improving resilience may need parallel capacity before old space is released, making temporary under-utilisation part of prudent migration.

ARIN's rules should therefore be evaluated as capital policy, not treated as neutral clerical procedure. Their thresholds, evidence, approval distribution, review time, reduction rate and appeals matter to market access.

APNIC combines a forward plan with a backward record

APNIC's current policy requires transfer recipients that do not already hold IPv4 resources to provide a detailed plan for use within 24 months. Existing holders must provide the same forward plan, show their past usage rate and provide evidence of compliance with APNIC policies for previous delegations. The source must be the registered holder and the resource must meet transfer conditions.

This design tries to answer a real concern: a party with a history of receiving space and leaving it idle should not obtain more merely by presenting another optimistic forecast. Past use can discipline the narrative. The rule also protects APNIC's conception of transfers as movement from parties that no longer need addresses to those that do.

But the backward test can entrench historical advantage. An incumbent with large, old holdings has extensive data and known assignment patterns. A new entrant has no APNIC usage history. A company changing business model may have records that look inefficient under the old model precisely because it is buying for a new one. A buyer that acquired space in another region may present evidence that does not fit APNIC's categories or an NIR's administrative conventions.

APNIC, like ARIN, offers pre-approval. Its guide says recipient accounts can have need evaluated before finding a source; approval is valid for 24 months, and a transfer within the approved size can proceed without resubmitting the justification. This is a sensible transaction-risk service. It moves the registry decision to a point when no seller's block is locked and no price has to be unwound.

APNIC's 2017 discussion of a proposal to remove transfer need included unusually useful operational claims. Supporters said removal could save around 800 emails a year and that only one transfer had been refused for inability to demonstrate need in the preceding 12 months. Opponents warned of speculation, hoarding and abuse.

Those statements came from a policy discussion summary, not an independent performance audit, but together they expose the central problem: a test can impose recurring documentation on many transactions while formally rejecting few, yet supporters of the test may value its deterrent effect rather than its refusal count.

That deterrent cannot be measured by successful transfers alone. APNIC would need to know how many potential buyers reduced requests, delayed deals, used another region, chose an unregistered arrangement or did not apply. Without that denominator, a low refusal count can mean either that the screen works smoothly or that it filters before observation.

LACNIC retains needs justification while separating commerce

LACNIC's transfer policy requires an organisation in its region to justify the IPv4 resources it seeks as an initial or additional allocation under the policies in force. It also verifies the holder and checks for dispute, and it updates the registration record once the transfer is complete. For intra-regional transfers, the parties provide a signed legal document supporting the transaction.

At the same time, LACNIC tells market entities that it does not intervene in their commercial operations. This is a valuable separation, but it produces a visible tension. The registry stays outside price and broker terms while retaining authority to decide whether the buyer has justified the quantity. It is not a party to the bargain, yet its decision can determine whether the bargain closes.

The tension should be resolved through a narrow definition of the needs review, not through commercial intervention. LACNIC can publish the evidence categories, forecast horizon, treatment of existing holdings, acceptable transition capacity, response times and review route. It can offer advance assessment so the registry decision precedes seller commitment. It can report aggregate reductions and refusals without exposing applicants.

Regional context matters. Buyers in Latin America and the Caribbean can face capital costs, currency risk, import delays and uneven access to large blocks. A forecast that assumes stable equipment delivery or customer financing may not fit that environment. If the registry requires certainty that the market itself cannot provide, the test can ration against precisely the operators facing the highest external uncertainty.

The correct response is not to invent relaxed forecasts for favoured sectors. It is to reduce subjective prediction and focus on observable safeguards. A regional institution should not have to guess which operator's business will succeed.

RIPE's no-need turn shows the test is not inherent to registration

The RIPE region supplies the most important counterfactual. Policy proposal 2013-03, accepted in 2014, removed the general need-based requirement for IPv4 delegations after depletion. Its rationale said need documentation made sense while lower-level demand consumed a remaining public pool, but that the logic changed when the pool could no longer satisfy demand. It also argued that removing bureaucracy could improve registry data by giving organisations more incentive to report transfers.

Current RIPE-807 allows legitimate holders to transfer resources under stated restrictions and requires the RIPE NCC to update registration records. It imposes a 24-month holding period on scarce resources after receipt, with defined treatment for business changes. For ordinary transfers within the RIPE service region, it does not recreate ARIN's 24-month quantity forecast.

This proves a narrow proposition: accurate registration does not technically require a general needs judgment. The RIPE NCC can authenticate parties, apply transfer restrictions, update records, publish a log and provide RPKI and database services without approving the buyer's ordinary business forecast.

The inter-RIR edge reveals how one region's gate can travel. RIPE-807 says that when a transfer enters from an RIR region that requires the receiving region to have needs-based policies, the recipient must provide a plan to use at least 50% of the transferred resources within five years. This is not RIPE's general domestic test. It is a compatibility accommodation to another region's rule.

The result is institutional spillover. A buyer can face different review depending not only on its own registry but on the source region and reciprocal policy requirements. A longer five-year plan may be easier than a 24-month forecast, yet its existence shows that needs policy can segment a global asset market. The recordkeeping act is technically similar; the permission changes with the pairing of institutions.

RIPE's experience does not prove that speculation vanished or that every transfer improved use. It demonstrates that anti-hoarding and registry accuracy can be pursued through tools other than a general forecast. That is enough to shift the debate from inevitability to design.

Price is evidence of commitment, not proof of beneficial use

Critics of needs review sometimes make too much of price. They say a buyer willing to pay market value must need the addresses. That conclusion is unsafe. Price proves willingness and ability to pay under the buyer's expectations. It does not prove immediate deployment, socially beneficial use or absence of a speculative strategy.

A buyer may expect appreciation and hold inventory. A large platform may buy defensively to prevent rivals from obtaining capacity. A financial investor may acquire through an operating entity. A broker may warehouse blocks to improve matching. A resource holder may prefer unused optionality to present routing. Market price rations by wealth and expected return, not by a public definition of productive use.

Price nevertheless carries information that an administrative allocation fee did not. The buyer risks capital. Holding unused space has financing and opportunity costs. A seller choosing among buyers has reason to evaluate execution. Lenders and investors scrutinise the use of funds. Those mechanisms can make a transfer request more disciplined than a near-free pool application, though they cannot eliminate hoarding.

The policy question is therefore comparative. Which instrument detects harmful stockpiling with the least damage to legitimate investment and registry accuracy? A subjective demand forecast may not outperform price plus observable holding rules. It can even create a false assurance: a buyer writes a compliant plan, receives the block and later changes circumstances. The registry has approved the prose, not guaranteed deployment.

The strongest anti-hoarding evidence appears after acquisition: whether the resource remains idle, whether related entities repeatedly buy and resell, whether the holder accumulates beyond plausible operating scale, whether transfers are structured to avoid waiting periods, and whether claimed use exists. Some of that evidence can be observed without asking the registry to approve every customer forecast in advance.

Post-price review changes bargaining power before it changes the record

The registry's decision affects a deal even before refusal. A needs condition changes the contract, pricing and timing.

The buyer may seek a financing commitment sized to the agreed block but risk approval for less. The seller may keep the prefix off the market during review. An escrow agent may hold funds that earn little or create compliance cost. The broker's commission may depend on a completion outside its control. Engineers may reserve migration windows and provider support. Customers may have launch dates dependent on capacity.

These are option costs. The parties pay for uncertainty even when the transfer ultimately succeeds. A longer or less predictable review increases the discount a buyer demands, the deposit a seller requests, the time allowed for closing and the chance that one side walks away. Large firms can absorb these costs more easily than small operators.

Quantity reduction is particularly disruptive. IPv4 blocks are not perfectly divisible for every network design. A buyer negotiating a /16 may need internal segmentation, route aggregation, customer pools and reserve. Approval for a /17 does not simply halve the purchase price; it can change architecture, routing policy and future acquisition cost. The seller may not want to split the block or may have another buyer for the whole.

The test also affects seller liquidity. A block with many theoretically interested buyers may be practically marketable only to those with pre-approval or familiar documentation. The registry thereby shapes the buyer set and price. That is capital rationing even if the RIR never quotes a price.

Public performance measures should capture these effects. Approval count is limited public evidence. Report time to first response, applicant-controlled and registry-controlled days, initial quantity, approved quantity, resubmissions, expired pre-approvals, withdrawals, refusals and review outcomes. Aggregate by applicant type and size where privacy permits. A gate cannot be evaluated only by counting those who passed it.

The test favours incumbents whose past already resembles the form

Needs review tends to reward legibility. Established access providers have customer counts, assignment histories, utilisation tools and staff familiar with RIR expectations. Large cloud companies can produce capacity models. Experienced brokers know how to package a request. These actors may have genuine need, but they also have an administrative advantage.

New entrants present uncertainty. Their customers may not sign until address capacity is secured. A security provider may need clean space before onboarding clients. A hosting company may need inventory distributed across sites to sell service. An operator entering a new country may not have historical utilisation in that entity. A company replacing carrier-grade translation may know the engineering objective while the migration curve remains uncertain.

The circularity is obvious: to prove demand, the entrant may need customers; to win customers, it may need addresses. An incumbent's past use becomes evidence of future entitlement, while the entrant's future plan is discounted because it lacks past use. That can turn conservation into protection of the installed order.

Corporate structure compounds the problem. A group may operate networks through several subsidiaries, acquire companies or centralise address management. Review can focus on the legal applicant while economic use sits across related entities. Allowing every affiliate to aggregate freely creates avoidance risk; refusing to recognise group operations misreads reality. The answer is beneficial-control disclosure and consistent aggregation rules, not discretionary suspicion.

Geography also matters. A buyer may deploy in several regions or use globally distributed infrastructure while holding an account in one RIR. Forecast templates built around one service territory can mistake distributed use for weak local need. The registry should verify the relationship and applicable policy without pretending that an address has a fixed physical nationality.

An anti-hoarding rule that systematically gives established holders easier access than challengers can increase concentration. That outcome deserves explicit competition analysis, not a claim that every approved applicant met the same form.

Forecasts are weakest where investment is most consequential

A 24-month forecast appears objective because it uses a number and a date. Its reliability depends on assumptions the registry cannot fully test: customer acquisition, equipment supply, product launch, merger integration, traffic growth, IPv6 adoption, translation architecture, provider contracts and macroeconomic conditions.

Some demand is lumpy. A mobile or broadband operator may need a block before subscriber migration. A cloud provider may open a site only if network capacity is secured. A buyer can rationally acquire a larger aggregate to avoid fragmentation and repeated transaction cost. The utilisation curve may be slow and then steep, not linear.

Resilience creates deliberate slack. Parallel migration requires old and new address plans to coexist. Disaster recovery capacity should not be fully consumed in ordinary operation. Security services may reserve clean ranges to move customers under attack. An efficient forecast that assumes every address must be assigned quickly can punish operational prudence.

IPv6 makes prediction harder, not irrelevant. A buyer may expect IPv6 traffic to grow while still needing IPv4 for customers, translation pools or compatibility. If IPv6 adoption exceeds the forecast, IPv4 use may be lower than planned. That is a desirable technical outcome, not necessarily evidence that the original transfer was abusive. A rigid needs test can create a perverse incentive to maximise IPv4 utilisation to validate the purchase.

Registries can review arithmetic, but they cannot become venture investors with superior knowledge. If staff decide that one product's conversion rate is credible and another's is not, they are making sector judgments without bearing the investment result. Consistency becomes difficult to audit because confidential files hide comparators.

The more consequential and innovative the investment, the less likely its future demand will resemble a stable historical series. That is precisely where capital rationing by administrative forecast is least defensible.

Commercial confidentiality is part of the cost

To prove need, a buyer may disclose network diagrams, customer projections, expansion plans, utilisation by site, supplier schedules and corporate transactions. These can be competitively sensitive. They may reveal where the buyer plans to build, which customers it expects, how constrained its current network is and when it must acquire capacity.

Registries can impose confidentiality controls, and responsible staff may handle data carefully. The structural risk remains: the institution asks for information beyond what is necessary to identify the holder and keep the record accurate. More collected data creates more security, access, retention and litigation questions.

The burden is unequal. A large firm can provide aggregated evidence and negotiate disclosure. A small operator may have only a few identifiable customers. A start-up's forecast may effectively reveal its entire strategy. Cross-border transactions can place documents under different privacy and discovery regimes.

Data minimisation should follow the policy purpose. If the concern is rapid resale, collect holding history and beneficial control. If it is a sham entity, verify legal and operational existence. If it is concentration, collect controlled holdings at appropriate bands. Do not request customer names and detailed equipment plans merely because those fields were once used for free-pool allocation.

Where a forecast remains, the registry should publish a minimum evidence standard, accept several proof types, separate reviewers from commercial conflicts, limit retention and provide an applicant with the material findings. Confidentiality must not make the decision unauditable to the person affected.

Anti-hoarding is a legitimate objective that needs a better proxy

Hoarding can reduce available supply, raise prices and allow a holder to extract rents without improving network use. Because IPv4 space is finite and globally useful, concentrated speculative inventory can affect more than the buyer. A complete rejection of anti-hoarding policy would ignore that externality.

The first task is to define the conduct. Holding reserve for a planned network is not the same as buying solely for resale. Keeping capacity during migration is not the same as warehousing. A broker temporarily controlling a block for settlement is not necessarily an investor. A large operator may have extensive holdings because it has extensive customers.

The second task is to choose an observable proxy. Forecast prose is indirect. A buyer can produce a detailed plan and still speculate. An innovative operator can present an uncertain plan and still deploy. The registry should prefer evidence tied to behaviour: time since acquisition, subsequent transfer, common control, routing or assignment activity where relevant, repeated unused accumulation and attempts to evade holding rules.

Holding periods are blunt but legible. RIPE policy generally prevents scarce resources from being transferred for 24 months after receipt. APNIC prevents addresses from its 103/8 free pool from being transferred for five years after original delegation. ARIN restricts source eligibility and access to its waiting list in specified circumstances. LACNIC applies transfer waiting rules to certain resources. These tools target rapid recycling and subsidised inventory rather than every buyer's business forecast.

They can still cause harm. A holder facing insolvency or a genuine business change may need an exception. A long lock can trap underused space and reduce liquidity. Related entities can attempt avoidance. The rules therefore need defined exceptions, beneficial-control aggregation and public rationale.

Concentration reporting is another tool. Publish holdings by anonymised bands, acquisitions over time and shares of transferred volume, with safeguards against exposing network security or customer data. A regulator of record integrity need not set a hard cap immediately; transparency can reveal whether a cornering risk is real.

Anti-hoarding policy becomes credible when it names conduct, measures it and uses a proportional response. "We asked for need" is not a result.

Pre-approval reduces settlement risk but cannot legitimate every test

Pre-approval is the most practical improvement within a needs-based regime. The buyer presents its request before choosing a seller or fixing a final price. The registry approves a maximum quantity for a defined period. The buyer can then negotiate within that envelope, and the seller knows the principal recipient test has been addressed.

ARIN and APNIC both describe 24-month pre-approval arrangements. APNIC says a transfer within the approved size can proceed without the recipient providing justification again. ARIN says a qualifying request submitted within the approval period is not subject to re-verification of that need. These commitments make the approval a usable market instrument rather than an informal opinion.

Good pre-approval needs portability across specific transactions. It should not be tied to one broker or seller. It should state the quantity, relevant conditions, expiry and events requiring update. It should not be withdrawn merely because market price changes. If policy changes during the period, transition treatment should be clear.

The service also needs timing and review guarantees. A pre-approval that takes months or can be reduced without reasons moves uncertainty rather than removing it. Buyers should be able to test an adverse decision before committing capital. Aggregate approval and reduction data should show whether access differs by firm size or request type.

Most importantly, pre-approval is procedural, not constitutional. It answers "when should the registry decide?" It does not answer "what may the registry decide?" Moving an overbroad business-merit judgment earlier makes it less disruptive but not necessarily legitimate.

The long-run design should use pre-clearance for objective record and anti-abuse facts: identity, account standing, controlled holdings, transfer-lock status and maximum amount under a published rule. Ordinary deployment judgment should remain with the buyer.

A boundary test can separate restraint from obstruction

A transfer condition should survive five questions.

First, what specific shared harm does it prevent? Preserving unique registration, preventing forged authority, avoiding duplicate transfer and stopping rapid resale of subsidised allocations are concrete. Disliking speculation in the abstract is not enough.

Second, is the evidence closely connected to that harm? Corporate authority connects to forged transfer. Holding history connects to rapid flipping. A two-year customer forecast is only loosely connected to future resale and may be weaker than observed control and transaction history.

Third, can the condition be applied deterministically or with bounded judgment? A clear 24-month holding period is predictable. A staff conclusion that a product launch is insufficiently credible is not. Some discretion is unavoidable, but reasons and comparable standards can constrain it.

Fourth, is there a less obstructive instrument? Pre-approval can avoid post-price surprise. Disclosure can address concentration before a cap. A holdback or staged record change can address uncertain quantity. A targeted investigation can address linked-account avoidance. The registry should not default to full refusal.

Fifth, who reviews error and who bears delay? A consequential denial needs reasons, a prompt independent review and a defined clock. If the RIR's mistake destroys a bargain, merely inviting a new application months later is not an adequate remedy.

Under this test, record authentication clearly survives. A general authority to decide whether the buyer's investment is economically wise does not. Anti-hoarding controls survive only to the extent that they target observable acquisition and disposal conduct proportionately.

Staged recognition can address uncertainty without choosing winners

Some buyers genuinely cannot prove full near-term deployment, while some large requests create legitimate concern. The choice need not be approve everything or deny the bargain.

One option is staged closing across divisible prefixes. The parties contract for a larger total, but registration changes in agreed tranches under objective dates or conditions. Price and escrow can be allocated correspondingly. This reduces the quantity held before deployment without requiring the RIR to judge the business's ultimate merit.

Another is a disclosed reserve category. The buyer identifies a portion for migration, resilience or launch contingency, subject to a longer holding period rather than a fictional immediate-use claim. The category should not become a privilege available only to favoured sectors. Any buyer meeting the same objective conditions should qualify.

A third is a safe harbour based on capital at risk and conduct. An authenticated buyer acquiring within a concentration band, accepting a no-resale period and meeting record obligations could receive recognition without a granular forecast. Larger or linked accumulations would trigger enhanced disclosure. The threshold should be reviewed against actual market data.

Temporary transfers can also serve real needs, as RIPE policy recognises, but they should not be forced on a buyer that seeks permanent control. Temporary registration changes the risk allocation and may weaken financing or customer confidence. It is an option, not a substitute for a permanent market.

Staging must preserve aggregation and routing feasibility. A registry should not split a block into pieces that create inefficient announcements or unusable network design merely to display conservative approval. The parties and operators understand the technical consequence; the registry protects minimum transfer size and record coherence.

These alternatives share a principle: regulate observable exposure, not the imagined virtue of the buyer.

Refusal data is necessary to know whether the gate works

Needs policy is often defended without a complete denominator. Published transfer logs show approvals. They do not show potential requests abandoned after advice, pre-approvals reduced, files withdrawn, deals expired, sellers lost or buyers diverted into other arrangements.

A serious account would begin with inquiries, formal pre-approval requests and transfer submissions. It would record requested and approved quantities, applicant type, prior holdings band, source region, principal reason for reduction or refusal, applicant-controlled time, registry-controlled time, review and final result. Confidential business data can remain protected.

Reasons need a stable taxonomy: limited public evidence forecast, prior utilisation, missing authority, dispute, transfer lock, prohibited resource, incomplete documents, policy incompatibility or applicant withdrawal. Combining them into "not approved" makes a business-plan gate look like identity protection.

The distribution matters. If almost every established ISP receives full approval while new entrants receive smaller amounts, the policy may be protecting incumbency. If large speculative buyers repeatedly pass because they can prepare documentation, the test may not address hoarding. If approval is nearly universal but imposes long delay, the cost lies in administration rather than exclusion.

Review outcomes are equally important. A high reversal rate suggests inconsistent first decisions. No reviews may mean decisions are sound or that challenge is too costly. The registry should publish access, time and outcomes.

Evidence can change the policy. A holding period might prove more effective than forecast review. A concentration risk might appear only in related entities. A low refusal rate paired with substantial delay may justify a safe harbour. Without data, the needs test survives as inherited ritual.

Inter-RIR compatibility can export the strictest gate

An IPv4 block is globally routable, but transfer permissions remain regional. Inter-RIR rules often require reciprocal or compatible policy. If one RIR insists that its counterpart maintain a needs-based condition, a no-need region may add a special plan for incoming transactions to preserve compatibility, as RIPE does.

This can create a strictest-gate effect. A buyer in a relatively permissive region faces an extra condition because the seller's region insists on it. A seller's market shrinks if potential destination policies are deemed incompatible. Brokers route transactions through feasible pairs. Price and liquidity vary with institutional boundaries that do not alter the technical resource.

Some compatibility is essential. The RIRs must agree on source authority, exact resource, effective time and record handover. They should not accept a transfer that creates contradictory registration. Those are record invariants.

Business-plan symmetry is different. Two registries do not need identical views of customer forecasts to exchange an accurate record. Requiring policy likeness can turn coordination into a private trade barrier. The receiving registry can apply its own lawful holder conditions without making the source region adopt the same economic philosophy.

The proper compatibility standard is functional: can both sides authenticate the transition, preserve uniqueness, apply known locks, produce one final state and support correction? If yes, differences over ordinary deployment should not prevent the record from moving. Anti-hoarding controls can follow the holder under the receiving region's published rules rather than being imposed through institutional reciprocity.

This separation would widen lawful transfer paths while making the registry layer thinner. It would not remove accountability. It would focus accountability on the act the two registries actually share.

Number Resource Society points to non-gatekeeping finality

Number Resource Society offers a positive institutional direction because it begins with the ledger service rather than with inherited territorial permission. The common task is to preserve a unique, authentic, auditable record of control and transfer. Everything placed above that task needs separate justification.

An NRS transfer design would verify source control, recipient identity, exact resource, signatures, conflicting claims, applicable holding restrictions and effective time. It would publish deterministic rules and standard evidence. A holder could obtain advance confirmation of objective eligibility without disclosing a customer list or persuading staff that its business model deserves capital.

Anti-hoarding restraints would be explicit. A recently received block could carry a visible no-resale period. Related control could be aggregated. Large concentration could trigger public banded reporting or a defined enhanced review. Exceptions for insolvency, merger or technical necessity would have stated criteria. The service operator could not invent a new merit test inside an individual case.

The ledger state would be portable. Holders would not lose recognised history merely because they changed service provider or rejected an optional institutional programme. Independent verification and succession testing would reduce the incumbent's leverage over finality. Correction would be possible for fraud or demonstrable error through a bounded process, not open-ended discretion.

Price and financing would remain with counterparties. Deployment timing and routing would remain with operators. NRS could support proof of control and route authorisation without dictating which upstream must accept an announcement. The narrow service would be valuable precisely because it claims less.

This is not laissez-faire disguised as technology. Deterministic uniqueness, authentication, conflict handling, audit history and proportionate anti-abuse rules are strong controls. They are stronger than subjective review in one important sense: a entity can know the rule before committing capital and can challenge a decision against visible criteria.

Portability and finality turn restraint into structure. The recordkeeper remains accountable to the ledger instead of using the ledger to govern the market.

The price agreement should change the burden of proof

Once a buyer and seller have agreed price, the registry stands before a completed economic choice. It may still prevent a forged transfer, enforce a published holding period, resolve a conflicting claim and require an accurate recipient record. It may not assume that its inherited allocation questionnaire is automatically superior to the parties' capital judgment.

The market does not guarantee beneficial use. Nor does a needs form. Price can support speculation; forecasts can be written to pass. The mature answer is to compare instruments and target conduct. Holding periods, beneficial-control aggregation, concentration data, advance objective clearance, staged recognition and reasoned review can address risk with less obstruction.

The boundary is easiest to see by asking who bears the decision. The buyer bears the purchase price, financing and operating result. The seller bears the opportunity cost and delivery obligations. Upstreams bear routing relationships. The registry bears the integrity, consistency and continuity of the recognised record. Power should follow those burdens.

A registry that chooses a buyer's quantity without bearing the investment has become a capital rationer. Sometimes public institutions ration capital under an enacted mandate, published objectives and full accountability. An RIR cannot acquire that role simply because an old form is available.

The more defensible future is narrow and demanding. Authenticate well. Prevent contradictory state. Publish finality. Measure delay. Expose concentration. Constrain rapid flipping where evidence supports the rule. Give reasons and review. Then record the bargain.

IPv4 scarcity makes this restraint urgent. Every additional permission layer carries a market price of its own: delay, uncertainty, disclosure, lost entry and institutional leverage. Number Resource Society's promise is not that scarcity disappears. It is that the ledger stops pretending to decide who deserves scarce capital.

Sources