Summary

  • The ASO Address Council delivered its recommended RIR Governance Document Version 3 to the NRO Executive Council on 1 September 2026. It is a recommended final draft, not an adopted rule.
  • Article 3.7 would require support from at least two-thirds of the RIRs that submit Recognition Assessments before ICANN recognizes a candidate RIR.
  • ICANN could exclude a negative recommendation from the threshold denominator when the submitting RIR’s service region or revenue would be materially affected by recognition.
  • The draft requires written notice, reasoning, a reasonable opportunity to respond and fair consideration of that response. An excluded assessment may still inform the decision.
  • A public result therefore needs three sets: every submitted assessment, the assessments counted for the threshold and all assessments considered on the merits.
  • A recognition-denominator receipt can preserve those sets without exposing confidential material, changing the threshold or creating a public veto.

A conflict rule that changes the count

Most conflict rules tell a decision maker who must disclose an interest, who must recuse or how an opinion should be weighed. Article 3.7 of the recommended RIR Governance Document Version 3 goes one step further. It can alter the denominator used to decide whether a candidate registry has enough institutional support to be recognized.

The basic threshold sounds simple. ICANN may not recognize a candidate unless at least two-thirds of the RIRs that submit Recognition Assessments support the proposal. The next paragraph makes the fraction conditional. ICANN may exclude a negative recommendation from an RIR whose service region or revenue would be materially affected if the candidate were recognized.

That can be a sensible protection. A new registry might divide an existing service region or remove a stream of revenue from an incumbent. Counting the incumbent’s opposition without examining that exposure could reward an interest created by the proposed change itself. Yet excluding the recommendation is also consequential. The same assessment remains a submitted institutional view, but it stops changing the formal fraction.

The draft does not allow that transition silently. ICANN must notify the RIR in writing, explain its reasoning and materiality finding, give it a reasonable opportunity to respond, and fairly consider the response. The excluded assessment can still be considered for informational purposes. These steps create a procedure. They do not, on their own, give a later reader enough information to reconstruct the calculation.

Submitted, counted and considered are different states

Imagine five RIRs submit assessments. Four recommend recognition and one recommends against it. If all five count, four-fifths supports recognition. If the negative assessment is excluded, the formal result becomes four out of four. In that example the threshold is met either way, but the institutional claim changes: unanimous support among the counted set is not unanimous support among all submitting RIRs.

Other distributions could change the threshold outcome. That is why a public notice should not report only “two-thirds support achieved.” It should show the original submitted set and the counted set after any exclusion. Otherwise the public sees an answer without the variable that produced it.

There is also a third set. The draft says an excluded assessment remains available as information. ICANN might therefore count four assessments for the threshold while considering five when judging operational capability, regional need or weaknesses in the proposal. “Excluded” would mean excluded from one arithmetic function, not erased from the record or declared false.

This distinction protects both sides. The candidate can show that the formal threshold was calculated under the published conflict rule. The affected RIR can show that its assessment was received, answered and retained for its proper evidentiary scope. Other operators can see that the final number did not manufacture agreement by hiding disagreement.

Disclosure is not the same as exclusion

Article 3.3 asks each assessing RIR to identify actual, potential or perceived interests. The listed classes include effects on service region or revenue, the composition of members participating in an internal vote, involvement in the Numbering Community policy process, and relevant family or financial relationships.

That disclosure is an input, not a self-executing penalty. A declared interest does not automatically make an assessment unusable. Nor does a negative recommendation prove that the RIR acted from self-interest. Article 3.7 gives ICANN a narrower task: determine whether the specified service-region or revenue impact is material enough to justify removing a negative recommendation from the threshold.

A strong record must therefore preserve the step between the two provisions. It should identify the interest class, the materiality rule applied, the evidence class used, who initiated the proposed exclusion, when notice was sent and whether the response changed the conclusion. “Conflict disclosed” and “assessment excluded” are not interchangeable states.

Article 3.4 already points toward a public evidence trail. It requires publication of the proposal and assessments, permits redaction of confidential information, and lets ICANN request reassessment while publishing its rationale and the response. Article 3.5 also makes version control essential: a material change normally restarts the process unless all parties agree otherwise. A denominator calculated against one proposal should not migrate invisibly to a materially different one.

The receipt should begin before the fraction

A useful recognition-denominator receipt would be short enough to read beside the decision. It would identify the candidate proposal and exact version; list every RIR that submitted an assessment; record each recommendation and disclosure state; and show the original numerator and denominator.

If exclusion is proposed, a second layer would record the actor, notice date, stated materiality basis, response window, response state and disposition. Protected commercial or personal material could remain redacted while the public record names the evidence class and the rule applied. The final layer would show the counted set, final numerator and denominator, the excluded assessment’s continuing informational status, the recognition decision, reasons, corrections and any superseding proposal.

The receipt would not decide whether an interest is material. It would not disclose confidential revenue figures, rank RIRs or invite an online popularity vote. It would make the authorized decision legible after the fact.

This matters because percentages acquire authority quickly. A press release can say the two-thirds condition was satisfied. A meeting slide can reduce the process to a green threshold. Later commentary can call the result “RIR support.” Without the underlying sets, each retelling can quietly expand a bounded calculation into a claim that the whole number community agreed.

Lu Heng’s critique of the multi-stakeholder mirage supplies the relevant warning: attendance and stakeholder status do not create agency. Article 3.7 counts institutional assessments under a proposed rule. It does not measure the views of every member, autonomous system, customer or affected user. Publishing the denominator would not cure that representation boundary, but it would prevent the count from pretending to measure more than it does.

The document is still a draft

The NRO announced the recommended text and rationale on 1 September. It says further opportunities for input will be announced and that the NRO Executive Council will coordinate next steps. The rationale explains that the threshold was lowered from a stricter approach and that a materially interested negative assessment may be excluded while still being considered informationally.

Those statements describe a proposed design, not a live case. The checked record contains no candidate proceeding under Version 3, no excluded assessment and no ICANN materiality decision. The question is therefore prospective and precise: if this mechanism survives, will the public artifact reveal the original denominator, the exclusion route and the final denominator?

ICANN’s Prep Week schedule places a Regional Internet Registry Governance Update on 6 October. That is a useful checkpoint for the draft’s status and for an explanation of Article 3.7. A presentation, transcript or audience count will not itself make the document operative. The next record should say which body is acting, against which version and under which authority.

Sources

  1. NRO — Recommended Draft RIR Governance Document, Rationale Report, and Next Steps
  2. NRO — RIR Governance Document Version 3
  3. NRO — RIR Governance Document Version 3 redline
  4. NRO — Draft v3 Changes and Rationale
  5. NRO — ASO AC Delivers Recommended Draft RIR Governance Document to NRO EC
  6. NRO — Internet Coordination Policy 2
  7. ICANN ASO — ASO Memorandum of Understanding
  8. ICANN — ICANN87 Prep Week Schedule Now Available
  9. Lu Heng — The Multi-Stakeholder Mirage