Summary

  • Before 2022, many ARIN end users paid annual maintenance fees of USD 150 for each IPv4 block, USD 150 for each IPv6 block and USD 150 for each ASN, while internet service providers paid a Registration Services Plan fee based on the aggregate size of their address holdings. Similar-sized holdings could therefore receive similar registry services at radically different prices.
  • ARIN's 2022 harmonization placed end users and covered legacy-resource holders into the same address-size tiers used for the Registration Services Plan. ARIN said about 49% of end users would pay the same, 21% would pay less and 30% would pay more. It also estimated about USD 3.6 million in additional annual revenue if customers paid as modelled.
  • The redesign rewarded some fragmented or multi-resource accounts while charging much more for a single consolidated end-user block. Under the published schedules, a hypothetical end user with one /24 and no other covered resource moved from USD 150 to USD 250; one /22 moved from USD 150 to USD 500; one /20 moved from USD 150 to USD 1,000; and one /16 moved from USD 150 to USD 4,000.
  • A typical end user with one IPv4 block and one ASN could pay less: ARIN illustrated a move from USD 300 to USD 250 for a small account. This is why the reform cannot be described simply as an increase. It changed the unit of charging from number and customer status to the larger fee tier produced by aggregate IPv4 or IPv6 holdings.
  • In 2024, ARIN completed harmonization by moving ASNs into the same schedule. ARIN estimated that 6,800 single-ASN holders would move from USD 150 to USD 250 annually, while about 313 ASN-only customers with multiple ASNs would save between USD 50 and USD 1,750. Fifteen existing Registration Services Plan customers were expected to move into a higher tier because of their ASN holdings.
  • Harmonization improved horizontal equity and made every covered resource holder eligible for the same membership path. It did not prove vertical fairness. A defensible schedule also needs retrospective incidence reporting by organisation type, resource composition, size, location, membership status and service use, plus a cost standard tied to registry work rather than the market value of scarce IPv4.

The invoice changed its unit of account

Fee reform sounds procedural until a network receives a different bill for the same addresses. ARIN's redesign between 2021 and 2024 was not a routine percentage adjustment. It changed the unit by which annual registry cost was distributed. The old system asked whether a customer was an internet service provider, an end user, a legacy-contract holder or an ASN-only organisation, then applied different charging rules. The new system increasingly asked what resources sat under the organisation and which aggregate tier those resources reached.

The reform had a strong fairness argument. ARIN maintained the same recognised registry, support systems and security services for organisations whose historical labels could differ even when their holdings looked similar. A hosting company classified as an ISP could pay by size while a hosting company classified as an end user paid per block. Business models had also changed. Enterprises connected customers, cloud providers operated like carriers, universities served affiliates, and organisations used assignments in ways that no longer fit an old distinction.

A fee system built around labels could reward the accident of original classification.

ARIN made the disparity concrete in its 2021 fee consultation. It said two hosting companies with the same 65,000 IPv4 addresses, roughly a /16, could find that one paid more than 25 times as much as the other despite receiving the same services. That is a serious horizontal-equity problem. If the registry does equivalent work for equivalent holdings, a 25-fold difference requires more than history as justification.

But removing an old disparity does not remove distribution. It creates a new one. A holdings-based schedule shifts cost toward organisations with larger aggregate blocks, including enterprises, universities, public bodies and other end users that once paid a small per-resource maintenance charge. It shifts cost away from some organisations that held many separately charged resources. It can reduce a bill for a small dual-resource network while greatly increasing a bill for a large consolidated block.

The phrase "fee harmonization" captures the administrative goal and obscures the incidence. Harmonization tells the public that categories became uniform. It does not say which organisations paid more, how much more, whether they could pass the cost to customers, whether they had a vote, or whether the higher charge tracked the cost of serving them. Those questions require reconstruction at invoice level.

The 2021 starting point

Before the change, ARIN used two important annual charging logics. Internet service providers and organisations that had chosen the Registration Services Plan paid a tiered annual amount based on address-space size. Many end users instead paid USD 150 for each assigned IPv4 block, USD 150 for each assigned IPv6 block and USD 150 for each ASN. The historical fee page and ARIN's 2021 member-meeting explanation both preserve these amounts.

The two logics valued different things. The tiered plan treated aggregate address capacity as a proxy for the organisation's share of registry cost or ability to support the shared service. The end-user schedule treated each resource record as a unit of maintenance. One large block could therefore be cheap, while many small blocks could become expensive. Under the tiered plan, aggregation increased the bill as the total crossed larger address categories; the number of separate blocks mattered mainly through their cumulative size.

Neither logic is a pure measure of cost. Registry work has fixed and variable components. Every organisation needs account administration, authentication, billing, support, public registration and security. A transfer, recovery or disputed authority request can require substantial work regardless of block size. Larger holdings can create greater consequence and sometimes more complex records, but one /16 does not necessarily generate 16 times the annual service work of one /20. Per-resource billing also has weak cost logic because ten simple /24 records may not cost ten times one /20 record.

The old distinction persisted because ARIN's institutional history separated providers receiving allocations for downstream customers from end users receiving assignments for their own networks. That difference once connected to policy and service expectations. Over time, the services converged. ARIN pointed to the Internet Routing Registry, Resource Public Key Infrastructure and DNSSEC as investments used across customer types. The case for charging different prices solely because of an inherited label weakened.

The redesign was therefore a choice between imperfect proxies, not a move from politics to mathematics. ARIN chose aggregate holdings. That made bills easier to compare and reduced disputes about whether an organisation behaved enough like an ISP. It also converted address scale into the dominant annual charging variable. Scarce-number holdings became not only an entity recorded by the registry but a measure of how much the holder financed the institution.

The 2022 schedule

The consultation opened on 9 April 2021 and closed on 10 May. ARIN proposed moving end users from annual per-resource maintenance fees to the Registration Services Plan, moving covered legacy holders into the same structure while retaining a rising annual cap, adding organisation-creation and recovery charges, raising source transfer fees to USD 500, and creating a temporary IPv6 waiver for the smallest tier. The Board adopted the schedule on 23 June 2021 for implementation on 1 January 2022.

The 2022 Fee Schedule set eleven address tiers. A 3X-Small account paid USD 250 for aggregate IPv4 holdings of /24 or smaller, or IPv6 holdings of /40 or smaller. A 2X-Small account paid USD 500 for more than a /24 through a /22, or more than a /40 through a /36. The next tiers doubled: USD 1,000 through a /20, USD 2,000 through a /18, USD 4,000 through a /16, USD 8,000 through a /14, USD 16,000 through a /12, USD 32,000 through a /10, USD 64,000 through a /8, USD 128,000 through a /6, and USD 256,000 above a /6.

For an organisation with both IPv4 and IPv6, ARIN used the service category corresponding to the larger of the two aggregate holdings, not the sum of dollar charges for both protocols. ASNs held by an organisation already in a plan were covered, but ASN-only organisations remained outside the tier structure until 2024. This detail explains why some end users paid less even while the reform raised substantial revenue.

At the ARIN 48 member meeting, President and CEO John Curran described a typical end user with one IPv4 block and one ASN. Under the old per-resource schedule, that organisation paid USD 150 for the block and USD 150 for the ASN, or USD 300. If its block fit the smallest category, the 2022 plan could reduce the annual charge to USD 250. The transcript records his statement that many such organisations dropped by USD 50.

ARIN also disclosed the wider distribution. In the ARIN 48 fee-harmonization update, Curran said 49% of end users would pay the same, 21% would pay less and 30% would pay more. Some would pay "remarkably more," and ARIN had contacted holders of very large blocks. He estimated that, before behavioural changes such as consolidation or movement, the schedule would add about USD 3.6 million in annual revenue, moving ARIN from roughly USD 22 million to more than USD 24 million.

Those are unusually useful incidence figures. They prevent a false binary. The reform was not merely a fee increase because seven in ten end users were modelled to have no change or a reduction. It was not revenue-neutral because the expected increase was material. It was a redistribution that corrected one form of inconsistency by charging a defined minority more.

Eight reconstructed bills

The following cases apply the published pre-2022 USD 150 per-resource amounts and the 2022 aggregate address tiers. They are schedule simulations, not claims about a named organisation's actual invoice. Agreement status, legacy caps, billing date, waivers, returned resources and resources held under separate organisation identifiers can change a real result.

Hypothetical covered holdings Pre-2022 end-user method 2022 RSP tier Change
One IPv4 /24 only USD 150 USD 250 +USD 100; +66.7%
One IPv4 /24 plus one ASN USD 300 USD 250 -USD 50; -16.7%
One IPv4 /22 only USD 150 USD 500 +USD 350; +233.3%
One IPv4 /22 plus one IPv6 /48 USD 300 USD 500 +USD 200; +66.7%
One IPv4 /20 only USD 150 USD 1,000 +USD 850; +566.7%
Ten separate IPv4 /24 blocks only USD 1,500 USD 1,000 -USD 500; -33.3%
One IPv4 /16 only USD 150 USD 4,000 +USD 3,850; +2,566.7%
One IPv4 /8 only USD 150 USD 64,000 +USD 63,850; more than 425 times the old bill

The first two rows show why resource composition mattered. A very small end user with only one /24 paid more because the new minimum exceeded one old maintenance charge. Add one ASN, and the old method charged twice while the new plan still fit inside one minimum tier. The same network scale could therefore produce either a rise or a fall depending on whether it held another separately billed resource.

The /22 example appeared in public criticism during the consultation. One commenter noted that an end user with a /22 and a /48 could move from USD 300 to USD 500. The objection was not that uniformity had no value. It was that small end users could subsidise a redesign framed as equity. ARIN responded at institutional level by retaining a new USD 250 minimum tier and a temporary IPv6 waiver that let qualifying 3X-Small organisations receive up to a /36 without moving to the USD 500 category.

The ten-/24 example shows the opposite distortion. Under the old method, block fragmentation generated ten maintenance charges. Under the new method, ten /24s aggregate to more than a /22 but no more than a /20, producing the USD 1,000 tier. The bill falls by one-third. A holder with one consolidated /20 had paid only USD 150 before the reform and moved to the same USD 1,000 tier. Two organisations with comparable aggregate address capacity could therefore approach the reform from opposite directions because one arrived with ten records and the other with one.

The /16 and /8 rows explain ARIN's outreach to large end users. The percentage changes are spectacular because the old denominator was so small. Whether such a holder actually paid the modelled amount depends especially on legacy and agreement status. Many large historical blocks may be legacy resources outside an ordinary agreement or covered by an LRSA cap. The table isolates the schedule's logic for covered end-user resources; it does not assert that every large historical holder received the uncapped invoice.

The micro-simulations reveal the redesign's real principle. ARIN rejected block count and customer label as the primary annual measures. It chose aggregate capacity bands. That improved consistency among equal-sized covered holdings. It also made the carrying charge for a large block rise sharply even when the registry's annual work did not rise in direct proportion.

Who were the thirty percent?

ARIN's public aggregate identifies the 30% who paid more but does not provide a complete public cohort table by organisation type, old bill, new bill and resource composition. The schedule permits a careful inference about likely profiles, while the absence of named billing records prevents a census.

The first likely group is end users with one address block and no separately charged ASN or second protocol block. They had benefited from the simplicity of one USD 150 maintenance charge. Even a /24 moved to USD 250. The larger the consolidated block, the larger the increase. Enterprises, universities, research organisations, public agencies and critical-infrastructure operators can fit this shape.

The second group is address-rich end users. A company holding a /16 for its own infrastructure could move into a USD 4,000 annual plan despite having one block. The redesign treated its aggregate resource capacity like that of a provider holding the same amount. This is horizontal equality by resource size. It is not necessarily equality by margin, service demand or ability to pass cost onward.

The third group is organisations whose IPv6 holdings set a higher tier than their IPv4 holdings. The temporary waiver protected the smallest organisations from one IPv6-related jump, but outside its boundary the larger of the IPv4 and IPv6 categories determined the fee. A network that deployed generously sized IPv6 space could therefore face a tier driven by a resource with no scarcity economics comparable to IPv4. This risk is why fee design must distinguish administrative scale from asset value.

The fourth group is organisations near category boundaries. Aggregate holdings slightly above a /24 moved from USD 250 to USD 500; slightly above a /22 moved from USD 500 to USD 1,000. Returning or reorganising a small amount of space could change the tier, although operational and contractual needs may make that impractical. Step functions create a large marginal fee at each boundary even if the average charge remains modest.

The fifth group consists of covered legacy holders whose cap changed over time, though they require separate treatment. The 2022 schedule retained a cap and increased it by USD 25 annually. The 2024 schedule later limited continuing capped treatment to active LRSAs entered before 1 January 2024; legacy resources brought under agreement afterward entered ordinary RSP pricing. Similar historical holdings could therefore face different annual costs depending on contract timing.

The burden did not necessarily stay with the invoiced organisation. A commercial network could pass it through to broadband, hosting, cloud or managed-service customers. A university or public body might absorb it by delaying another network expense. A profitable enterprise could treat it as immaterial. A small operator with weak bargaining power might accept lower margin. Incidence is not the name on the invoice. It is the final place where price, investment or service changed.

The case ARIN made for the redesign

ARIN's strongest argument was equal service for equivalent holdings. The organisation had invested in registry automation, authenticated routing records, RPKI, DNSSEC, operational redundancy and support used by both providers and end users. Customer labels no longer described service consumption reliably. A schedule that charged one /16 holder more than 25 times another was difficult to defend as stable equitable cost recovery.

The second argument was simplicity. One address schedule reduces classification disputes, training burden, billing exceptions and confusion. Staff no longer need to explain why one organisation pays under an ISP table while a similar organisation pays per resource. Customers can estimate cost from aggregate holdings. The published calculator can show current and proposed bills.

The third was membership equality. When end users entered the Registration Services Plan, they became Service Members and could seek General Membership. ARIN's 2021 membership-structure consultation presented this as opening a governance path previously unavailable to end-user organisations under the old model. Equal fees came with equal institutional standing rather than only a higher invoice.

The fourth was service sustainability. ARIN said increased transfers and organisational changes consumed significant staff resources and that transaction charges could support timely service. It also linked cost recovery to robust redundant infrastructure and routing-security services. A registry should not underprice work until service quality collapses. Users benefit when charges are predictable enough to fund authentication, fraud prevention, record accuracy and continuity.

These arguments have force. A serious critique should not pretend the old schedule was neutral. It rewarded particular historical classifications, created inconsistent bills and could overcharge organisations with many small records. It also risked making end-user status a lesser service category as registry security tools became more important.

The remaining question is proportionality. Equal holdings may deserve equal basic treatment without requiring a fee curve that treats address quantity as the best proxy for all annual cost. ARIN needed a practical schedule, not an impossible invoice for every support ticket. Yet once it chose size bands, it assumed that capacity was a sufficiently fair basis for cross-subsidy. That assumption should be tested against actual costs and affected cohorts.

Horizontal equity is not vertical equity

Horizontal equity asks whether similarly situated organisations pay similarly. By removing the ISP-end-user divide, ARIN improved this dimension. Two covered organisations with the same aggregate /16 now enter the same address tier regardless of how an old request labelled them.

Vertical equity asks whether differently situated organisations bear a defensible share. A regional carrier, a hyperscale platform, a public university and a municipal network can hold the same amount of address space while having very different revenue, customer base, service demand and capacity to influence ARIN. The RSP table treats them alike at the invoice. That may be administratively necessary, but it should not be called complete equity without evidence.

Cost causation provides one test. ARIN should estimate the fixed annual cost per organisation and the variable cost associated with resource records, support, security services, transfers, recoveries and complex authority changes. If a /16 account generates little more annual work than a /20 account, the eightfold fee difference is a cross-subsidy rather than direct cost recovery. Cross-subsidy can still be legitimate. It should be named and justified.

Benefit provides another test. Larger holdings receive a valuable continuity service over a larger operational surface. Accurate records, RPKI access, reverse DNS, routing data and transfer recognition can protect more addresses and more downstream activity. Charging by size can therefore reflect protected benefit even when marginal staff cost is lower. The problem is deciding whether the relevant benefit is registry service or the market value of scarce IPv4.

That boundary matters. ARIN did not create the entire economic value of a /16. The value arises from technical usefulness, scarcity, routability, customer dependence, market demand and decades of network investment. ARIN's record makes control more legible and transactions more reliable. A fee justified by the cost and consequence of maintaining that record remains close to registry service. A fee justified simply because the holder owns a valuable scarce asset begins to resemble a capital levy.

Ability to pay is a third test, but it is hard to administer without turning the registry into a financial assessor. Revenue-based fees would require sensitive data, invite restructuring and move APNIC-style regional coordination toward taxation. ARIN reasonably avoided that path. It can still examine whether its minimum charges and tier cliffs impose disproportionate burdens on small organisations, public networks and Caribbean operators.

The redesign should therefore be judged as a workable proxy subject to correction, not a final theory of fairness. Uniformity solved the old label problem. Retrospective evidence must show whether the new proxy placed cost where ARIN intended.

The 2024 ASN redesign repeated the pattern

ARIN completed fee harmonization two years later. Until 2024, an organisation holding only ASNs paid USD 150 annually per ASN and a USD 550 issuance fee for each new ASN. Organisations with address resources generally had ASNs included within their Registration Services Plan. This preserved another status difference: an ASN could be bundled for one customer and separately charged for another.

The 2023 ASN fee consultation proposed placing ASNs into RSP tiers. One to three ASNs would be 3X-Small at USD 250; four to fifteen would be 2X-Small at USD 500; sixteen to sixty-three would be X-Small at USD 1,000; sixty-four to 255 would be Small at USD 2,000; and 256 or more would be Medium at USD 4,000. The larger fee produced by an organisation's IPv4, IPv6 or ASN holdings would determine the plan.

ARIN disclosed its model in more detail than many institutions would. Using May 2023 customer and revenue data, it estimated that about 6,800 single-ASN holders would pay USD 100 more each year, rising from USD 150 to USD 250. About 313 ASN-only customers with multiple ASNs would pay less, saving between USD 50 and USD 1,750. Only fifteen existing RSP customers were expected to move higher because of ASN count: thirteen from 3X-Small to 2X-Small, and two from the smallest categories to X-Small.

The distribution followed the same logic as 2022. A minimum organisational charge increased the bill for the largest number of very small holders. Bundling reduced the bill for holders with several resources. Four ASNs had cost USD 600 annually under the old per-ASN method and moved to USD 500. Fifteen had cost USD 2,250 and moved to USD 500. Sixteen had cost USD 2,400 and moved to USD 1,000. Scale beyond the first resource produced savings because the redesign charged a band rather than each ASN.

The Board adopted the 2024 Fee Schedule on 15 August 2023. It also eliminated the USD 550 issuance charge for new ASNs. This created a useful distinction between entry and carrying cost. A new single-ASN organisation avoided the former issuance fee, making entry cheaper, while an existing single-ASN holder paid USD 100 more each year. Over time, the annual increase could outweigh the one-time saving for a continuing account; for a new entrant, the reform reduced the first barrier.

ASN-only customers also became Service Members and eligible to seek General Membership. Again, this was a real extension of institutional standing. It did not make the fee increase voluntary. An organisation that needs one ASN for an independent routing policy cannot substitute another ARIN-recognised registry provider within the region. Membership eligibility gives voice; it does not create price competition.

The 6,800 figure should be central to any account of ARIN fee reform. A policy can be coherent in aggregate and still depend on a broad base of small holders paying more. The increased minimum may accurately reflect fixed per-organisation cost. If so, ARIN should publish the cost bridge. The public should not have to choose between accepting the word equitable and assuming the increase was arbitrary.

Tier boundaries change behaviour

A fee table is not only a way to collect revenue. It changes decisions at the margin. Organisations may consolidate resource records, return unused space, delay a request, split functions across legal entities, avoid an additional ASN, reconsider IPv6 size or move resources under different agreements. Some responses improve registry accuracy; others distort network design.

The 2022 address tiers can reward returning enough space to fall below a boundary. That may help recover unused resources. It can also encourage an organisation to design around a billing threshold rather than operational need. The temporary IPv6 waiver recognised this risk at the smallest tier by allowing more IPv6 space without a higher charge. The waiver implicitly concedes that prefix size and service cost do not always align.

The 2024 ASN bands create their own boundaries. A third ASN remains in the USD 250 tier; a fourth moves to USD 500. A fifteenth remains at USD 500; a sixteenth moves to USD 1,000. These jumps may be small compared with network cost, but for a small anycast operator, research network or volunteer-supported service, they are not invisible. The policy should not discourage technically justified routing separation.

Organisation identifiers complicate the incentive. ARIN calculates fees per organisation identifier under the applicable agreements. Corporate groups with several identifiers may face different aggregation outcomes from a single consolidated entity. Legitimate corporate separation should be respected, but the fee model should monitor whether restructuring defeats its equity premise or whether consolidation requests are delayed by avoidable administration.

Legacy status is another behavioural boundary. Under the 2024 schedule, active LRSAs entered before 1 January 2024 retained a USD 200 cap for covered legacy IPv4 resources, rising by USD 25 per year. Legacy resources brought under agreement after that date faced ordinary RSP fees without the cap. A holder deciding whether to sign could compare access to modern services and clearer contractual treatment against a potentially much larger annual bill. Registry security should not become hidden leverage over historical expectations.

Behavioural response also affects ARIN's revenue estimate. Curran cautioned in 2021 that consolidation or movement could reduce the modelled USD 3.6 million increase. This is why pre-implementation estimates need post-implementation reporting. If customers returned resources, split accounts, declined agreements or fell into arrears, the final incidence differed from the calculator's first answer.

Membership was broadened, not completed

ARIN connected harmonization to membership. End users entering the RSP in 2022 became Service Members and could apply for General Membership. ASN-only customers received the same route in 2024. This removed an old governance inequality and strengthened the case that customers paying under one schedule should have equal institutional status.

The distinction between Service and General Membership remains important. Service membership does not automatically cast a vote. An organisation must take the steps required for General Membership, maintain a voting contact and participate sufficiently to avoid later reclassification under ARIN's rules. Many resource holders focus on operating networks, not registry elections. The affected fee base is therefore larger than the active electorate.

That does not invalidate Board authority. ARIN's Board is responsible for the fee schedule and used public consultations before adoption. It does mean that consultation participation should not be mistaken for payer consent. The 2021 consultation ran for one month and generated public discussion. ARIN adjusted the organisation-creation fee from the proposed USD 100 to USD 50 after concern that a higher charge could deter accurate reassignment data. This is evidence that comments mattered.

The decision on the main harmonization still remained with the Board. Organisations that did not follow the mailing list, lacked fee-modelling time or discovered the issue only when invoiced were bound by the result. Smaller entities face a higher participation cost relative to the bill. A carrier may assign policy staff; a municipal network or university may route the notice through procurement and technical teams before anyone understands the effect.

Broadening membership is therefore one remedy, not the complete answer. Fee legitimacy also requires notice directed to affected cohorts, plain-language examples, a calculator that shows old and new bills, and publication of the final distribution. ARIN did several of these things before 2022 and 2024. It should complete the record with actual outcomes.

What a retrospective fee account should show

ARIN had the information needed to estimate incidence before implementation. It said how many end users would pay the same, less or more; it estimated revenue; it counted single-ASN holders, multi-ASN savers and existing customers crossing tiers. A retrospective account could now test those estimates without exposing customer bills.

The first table should show counts and revenue by change band: decrease greater than 50%, decrease up to 50%, no change, increase up to 50%, increase from 50% to 200%, and increase above 200%. Medians alone would hide the large-block tail. Counts without dollars would hide where revenue moved.

The second should identify resource composition. For the 2022 cohort, ARIN should separate one-block end users, multi-block end users, dual-stack holders, address-plus-ASN holders, capped LRSA organisations and existing RSP entities. For 2024, it should show single-ASN, two-to-three-ASN, four-to-fifteen-ASN and larger groups, plus the fifteen address-holding customers expected to move tiers.

The third should describe organisation type in broad privacy-preserving categories: access provider, hosting or cloud, enterprise, university or research, public sector, non-profit, critical infrastructure and other. These labels will not be perfect. They would reveal whether the redesign systematically shifted cost toward organisations unable to pass it on.

The fourth should show geography. ARIN's region includes the United States, Canada, Caribbean economies and North Atlantic islands with very different market scale and payment capacity. USD 250 or USD 500 is not the same burden everywhere. Geographic reporting should use sufficiently large groups to protect customer identity.

The fifth should connect fees to membership. How many newly created Service Members became General Members? How many voted? How many later returned to Service status? If higher or harmonised fees were partly justified by equal governance standing, the public should be able to see whether the new standing became usable participation.

The sixth should show behavioural outcomes: resource returns, organisation consolidation, agreement changes, waivers, arrears, reinstatements and support contacts related to the redesign. None should be interpreted automatically as opposition. Together they reveal friction.

The seventh should bridge revenue to cost. The 2021 estimate of USD 3.6 million additional annual revenue should be compared with actual collections and with the registry capabilities cited in support of the change. ARIN need not assign every engineer to a tier. It should show whether incremental revenue funded core registration, routing security, authentication, support, redundancy, transfer work, reserves or other activity.

Such reporting would not reopen every invoice. It would turn the word equitable into a testable claim. If the result shows that most increases fell on address-rich organisations receiving substantial protected benefit while small networks were shielded, ARIN's case strengthens. If it shows large burdens on small public or research networks with little relation to service cost, the Board could adjust the minimum, boundaries or waivers.

A better standard for the next redesign

ARIN should retain the gains from one understandable schedule. Returning to permanent ISP and end-user labels would recreate the inequity the reform addressed. The next redesign should instead add four disciplines.

First, publish cost floors and cross-subsidies. ARIN should estimate the fixed cost of maintaining one customer relationship and the variable costs associated with records, security services, support and complex changes. The minimum fee can then be defended as a service floor rather than an unexplained number. Higher tiers can be identified as shared-cost contributions rather than fictional marginal cost.

Second, smooth the cliffs where feasible. A banded schedule is easy to administer, but doubling at a boundary creates avoidable distortion. More bands, a base fee plus a gradual resource factor, or capped transition adjustments could reduce sudden jumps. Simplicity has value, so any formula should be accompanied by a public calculator and stable examples.

Third, preserve an entry path. Eliminating the ASN issuance fee in 2024 reduced one barrier even as the single-ASN annual charge rose. The IPv6 waiver similarly prevented address-planning needs from immediately raising the smallest bill. ARIN should review these mechanisms by their effect on new and small operators, not only revenue.

Fourth, isolate the ledger from asset taxation. Larger holdings can reasonably finance more of a shared registry where benefit and consequence rise with scale. The justification should remain tied to accurate records, authentication, security, support and continuity. Scarce IPv4 market value should not become an automatic revenue base. The registry records value; it did not create all of it.

A fifth discipline follows from membership: no fee reform should rely on the claim of community support without showing the affected denominator. Consultation entities, Service Members, General Members, voters and invoiced organisations are different populations. The decision record should say which one it means.

The networks that paid more should remain visible

ARIN's fee redesign solved a real problem. A provider and an end user with equivalent covered holdings should not face a 25-fold difference merely because history placed them in different tables. ASN-only organisations should not pay per identifier while the same identifiers are bundled for address holders without a coherent reason. One schedule is easier to explain, administer and audit.

The reform also had identifiable payers. Thirty percent of end users were modelled to pay more in 2022, some dramatically more. About 6,800 single-ASN holders were expected to pay USD 100 more in 2024. Fifteen existing RSP organisations were expected to move to higher tiers because ASN counts joined the calculation. These are not defects to hide. They are the distributional facts by which the redesign should be judged.

Some increases were defensible. A single-block end user had been paying less than a comparable provider for similar services. A single-ASN account still imposed a fixed organisational cost and gained a path to membership. The new schedule eliminated a USD 550 ASN issuance charge and reduced bills for many multi-resource holders. Equity can require some customers to pay more.

What equity cannot require is silence about who those customers were. A neutral-looking tier can produce non-neutral effects across universities, public networks, enterprises, small operators, island economies and large incumbents. The Board's authority to set fees is strongest when the incidence is published before and after the change, the cost standard is visible, and affected customers can challenge assumptions with evidence.

ARIN should now close the historical account. It should publish actual 2021-to-2024 bill-change distributions, explain how the additional revenue supported registry work, and show whether new Service Members used the governance route offered in exchange for uniform treatment. That would preserve the reform's legitimate achievement without turning harmonization into a word that ends debate.

The invoice names the organisation that paid. Accountability requires one more line: who finally bore the increase, what registry cost justified it, and whether the same objective could have been reached with less distortion. Until those questions are answered, ARIN's redesign remains administratively coherent and economically incomplete.

Sources

The bill examples apply published schedule rules to simplified hypothetical holdings. They do not identify an actual customer or account for every agreement, waiver, organisation identifier, billing anniversary, legacy classification or later resource change. Public sources support the aggregate incidence and fee mechanics but do not reveal private invoices or the complete identities of organisations in each change cohort.