Summary

  • A 20 August revision to Draft Policy ARIN-2025-3 would reduce the in-region IPv4-use threshold for justifying out-of-region use from /22 to /24, moving the test from 1,024 addresses to 256.
  • The revision also says that out-of-region usage justification cannot be used for the Waiting List, the Micro-allocation Pool or the dedicated IPv4 block for IPv6 deployment. Organizations already on the Waiting List would stay under the prior rule.
  • Three same-day responses supported the lower threshold but opposed bundling it with the pool exclusions. Their dispute is about distribution channel and cost incidence, not whether the draft has already changed policy. It has not: ARIN-2025-3 remains under discussion.

A smaller number came with a larger boundary

The cleanest fact in the 20 August revision is numerical. Where the draft had required an organization to use at least a /22 inside the ARIN region before counting out-of-region IPv4 use toward justification, it now proposes a /24.

In CIDR notation, RFC 4632 leaves no mystery in the arithmetic. A /22 covers 1,024 IPv4 addresses; a /24 covers 256. The proposed floor is therefore one quarter of its former size. That matters to operators whose legitimate footprint is distributed across regions but whose North American deployment is too small to clear a /22.

Read only that sentence and the revision looks like straightforward relief for smaller networks. The following sentences change the picture. The draft says out-of-region usage justification may not be used under Section 4.1.8, the Waiting List; Section 4.4, the Micro-allocation Pool; or Section 4.10, the dedicated IPv4 block intended to facilitate IPv6 deployment. It also adds a transition: organizations already on the Waiting List would remain subject to the rules in force when they entered.

The draft thus moves two levers at once. It lowers the amount of in-region use needed to recognize a broader network plan. It also limits the direct registry channels through which that recognized need may be satisfied.

Three pool names describe three different cases

The exclusions should not be compressed into a generic phrase such as “special pools.” Each serves a different operational situation.

The Waiting List allocates recovered IPv4 space to approved requests under a queue. A micro-allocation is designed for qualifying infrastructure cases in which small, distinct blocks and routing treatment matter. Section 4.10 reserves a dedicated block for organizations that need a limited amount of IPv4 to make an IPv6 deployment work. The revised draft does not declare those needs invalid. It says out-of-region usage cannot supply their justification.

That distinction is why the revision is more than a technical adjustment to a fraction. A network may pass the general /24 test and still be unable to use that part of its need at one of these three doors. Its need, its eligibility test and its supply channel are separate questions.

The change was not a drafting accident. A July options discussion explicitly compared a design that left affected organizations to the transfer market with a design that named exclusions for the three pools. It also described excluding out-of-region need from the Waiting List as part of the intent. The August language is a policy choice placed in the open.

Support for /24 did not transfer to the exclusions

The first responses made an unusually clean split between the two choices.

Quantum Networks supported lowering /22 to /24 but opposed the new exclusions. Its comment argued that the added language broadened the proposal beyond its original problem statement and would, in practice, leave Section 8 transfers as the route for organizations relying on out-of-region use. It also pointed to possible overlap with ARIN-2025-8 and asked that the pool issue be handled separately.

Peter Potvin made much the same distinction. He endorsed the lower threshold and rejected the three-pool restriction as a separate policy question. His transfer-only description is an interpretation of the revised architecture, not an adopted finding, but it identifies the economic issue: a valid need can be recognized while its lower-cost administrative supply route is withheld.

Tyler Donia focused on who would bear that difference. He argued that market purchases are difficult for smaller businesses and that the exclusions would create two tiers of valid need. Larger organizations may treat transfer prices as an ordinary acquisition cost; a small operator that otherwise qualifies for a limited direct allocation may not.

These comments do not prove community consensus against the revision. They do prove that agreement with the /24 change cannot be counted automatically as agreement with the exclusions. The two propositions have different beneficiaries, costs and evidence requirements.

Earlier staff questions changed shape, not relevance

A March discussion of staff review concerned an earlier version. It recorded no material legal issue at that stage, while anticipating significant Waiting List volume, ticket workload and possible gaming. It also asked whether regional-use limits were needed.

The August revision partly answers the queue concern by refusing to let the new justification operate in the Waiting List. But that does not make the impact disappear. It moves the impact. Fewer claims may reach the queue; more affected demand may seek transfers, redesign deployment, delay expansion or find another qualifying route.

A refreshed staff and legal assessment would therefore need to examine the new text, not merely repeat the earlier conclusion. The operational questions include how staff distinguish in-region and out-of-region use, how anycast and shared services are treated, what evidence a /24 footprint requires, and how the grandfathering date is proved. The economic questions include the number and size of applicants diverted from each pool and the likely price burden.

Four questions are hiding inside one draft

The debate becomes clearer when four decisions are separated.

First is eligibility: how much in-region use should an organization show before its out-of-region use can count? The revision answers /24.

Second is source: may that need be met from recovered space, a micro-allocation, the IPv6-transition block or only through another channel? The exclusions answer no for the first three when the justification depends on out-of-region use.

Third is geography: what use belongs “within” a service region when routing, customers, infrastructure and control are distributed? That remains an evidence problem, especially for anycast and multinational networks.

Fourth is transition: should a network already admitted to the Waiting List keep the earlier rule while a comparable new applicant faces the revised one? The grandfather clause says yes, creating a date-dependent difference that may be justified but should be measured.

Lu Heng's Policy Mirror offers a useful boundary for reading these choices. A thin registry rule keeps records accurate, documents transfers and makes resource control legible. A thicker rule decides which geographic business plan may use which distribution mechanism. IPv4 scarcity makes some thickness unavoidable, but it does not make the decision invisible.

An independent ARIN explainer describes the registry's allocation and policy role without turning it into ownership of the Internet. That is the right scale for this debate: ARIN is not deciding whether a network may operate abroad. The draft is deciding which uses may justify access to particular number-resource channels.

The status is a draft, and the open question is divisible

The independent proposal index lists ARIN-2025-3 as a Draft Policy last changed on 20 August. It is not current policy. No applicant has lost a pool under this revision, and no transfer-market outcome should be reported as a completed fact.

The immediate policy opportunity is to avoid a false package deal. The /24 threshold can be evaluated on evidence about small distributed networks. Each pool exclusion can be evaluated on its own purpose, demand pressure, abuse risk and alternative cost. Grandfathering can be evaluated as a transition rule.

One revision placed all of them in the same paragraph. That does not make them one decision. ARIN lowered a numerical threshold and, in the same edit, proposed closing three administrative doors to that justification. The next discussion should record whether the community agreed to the first, the second, both or neither.

Sources