Summary

  • The Board's rejection of the ICANN80 Kigali advice item 1.a.ii on 24 October 2024 is documented across four pages: a resolution, a per-item scorecard, approved minutes and a status register.
  • Nine directors voted in favour — exactly 60% of the fifteen directors the same minute accounts for — and no page states the denominator the Bylaws test divides by, or that the test was satisfied.

The clause is short and it does three separate things. Article 12, Section 12.2(a)(x) of the ICANN Bylaws requires that advice of the Governmental Advisory Committee on public policy matters "shall be duly taken into account, both in the formulation and adoption of policies". It then provides that where the Board "determines to take an action that is not consistent with Governmental Advisory Committee advice, it shall so inform the Governmental Advisory Committee and state the reasons why it decided not to follow that advice". And it fences the strongest form of that advice: "Any Governmental Advisory Committee advice approved by a full Governmental Advisory Committee consensus, understood to mean the practice of adopting decisions by general agreement in the absence of any formal objection ("GAC Consensus Advice"), may only be rejected by a vote of no less than 60% of the Board", after which the two bodies "will then try, in good faith and in a timely and efficient manner, to find a mutually acceptable solution." You can read the whole of it quoted on the Board's approved resolutions of 24 October 2024.

A note on where that text sits. The current Bylaws page is dated "As amended 3 July 2026", and it puts Article 11 with the Generic Names Supporting Organization and Article 12 with the advisory committees. The consolidated page was truncated before Article 12 when it was retrieved for this piece, so the operative wording above is the version the Board itself reproduces in its own resolution rather than text lifted from the Bylaws page. That distinction matters for anyone checking the arithmetic later: the institution prints its own rule, in its own decision record, on the day it applies it.

Two instruments meet at this clause, and only one of them counts. The GAC Operating Principles define the committee's method in Principle 47 as seeking consensus, "understood to mean the practice of adopting decisions by general agreement in the absence of any formal objection", and direct that where consensus is not possible the Chair "shall convey the full range of views expressed by members to the ICANN Board". There is no threshold in it, no fraction, no count. Principle 2 describes the committee as advisory and states that it is not a decision-making body; Principle 54 provides that the Bylaws prevail where the two documents conflict. So the numeric discipline belongs to the Bylaws alone, and the switch that turns the 60% sentence on is the committee's own statement, required by that same sentence, that a given item is GAC Consensus Advice. One side's trigger is a vote. The other's is the absence of an objection.

On 24 October 2024 the Board took up two items from the ICANN80 Kigali Communiqué concerning the Applicant Support Program. Resolution 2024.10.24.03 calls both of them "GAC Consensus Advice items", records that the Board accepted the compromise on item 1.a.i, and rejects item 1.a.ii. The scorecard is the per-item record: four printed columns — the advice item, its text, the Board's understanding following the Board-GAC call, and the Board response — and in the 1.a.ii row the action verb and the citation in one sentence. The Board "takes the action to reject the ICANN80 Kigali Communiqué GAC Advice Item 1.a.ii in accordance with Article 12, Section 12.2(a)(x) of the ICANN Bylaws." The reasons are in the same cell, in prose: conflicts of interest, legal challenges, the suitability of an independent third-party evaluator, and delay.

I had expected to find the reasons missing. They are not. That is the first thing I had wrong, and it belongs in the record rather than in a footnote: the duty to inform the committee and state why is discharged on a published page, in a cell that anyone can open. What the cell is not is a field. There is no column headed "reasons why the Board decided not to follow that advice", and no column for the vote. A reader auditing the reasoned part of the relationship reads a paragraph.

The vote appears one document over. The approved minutes, published 15 November 2024, record that nine Board Directors voted in favour of Resolution 2024.10.24.03, that two abstained, that four were unavailable to vote, that no director is recorded against, and that the resolution carried. Nine, two, four: fifteen directors, all accounted for. Nine is exactly 60% of fifteen.

Now the arithmetic that is mine and not the Board's, and it is worth stating precisely because it is unflattering to my own thesis. On any denominator the minute could support — from the eleven directors who participated, nine in favour plus two abstentions, up to the fifteen seats the same paragraph accounts for — nine clears "no less than 60%" of the Board. The missing base does not change the outcome of this vote. It changes whether a reader can check it.

What the minute does not print is the one number the test divides by or the sentence "the threshold was satisfied", and no other page reviewed here supplies either.

The recusal paragraph sharpens the point rather than softening it. The same minute records that four directors recused themselves, and then records two of those same directors as abstaining. The tally already accounts for all fifteen seats. Both statements cannot be complete at once: either the four listed as unavailable are the recused four, or the recused sit somewhere outside the fifteen. Nothing in the resolution, the scorecard or the minute resolves which, and other possible bases — directors eligible to vote, directors present — are not printed anywhere.

This is a tension inside one document, printed on the record, and it is the kind of thing a printed denominator would have settled.

The two registers keep different halves of the same relationship. On the committee's side, the GAC Itemized Advice index prints title, delivery vehicle and first delivery date, and the GAC User Guide sets out the full data model: a detail view adds related topics, the reference number, the consensus label, the advice statement and a rationale. No Board disposition, no implementation status. One such page shows the shape — and shows the consensus field spelled "Consenus" on the live site. On the Board's side, the GAC Advice Status page prints "Advice Item | Phase | Action(s) Taken" and links a report current as of 1 October 2026 whose per-item fields run from action request number through current phase and status update to the resolution link and the scorecard link. Ten fields, none of them reasons, none of them a vote.

That split has a history worth keeping. A 2016 review of the effectiveness of GAC advice found the earlier "GAC Register of Advice" holding 154 entries, 102 of them on new gTLDs, with "the other 8 fields in the database" almost without exception empty for every entry — among them a field named "Board Action (accept/disagree)". A field on a register is not a record. The outcome record did not vanish after 2016; it moved to a different family of documents, where its unit is the scorecard row rather than the register entry, and where the scorecard row is generated by the same rule that makes a rejection an event worth recording.

Phase, action and disposition are three vocabularies, and a 15 March 2023 letter from the Board to the committee's chair shows how easily they are confused. It describes the Action Request Register as five phases — Acknowledge, Understand, Evaluate & Consider, Implement, Close Request — plus Closed, and reports 17 consensus advice items in Phase 3, five in Phase 4, none in Phase 5 and one closed. The same letter heads a section on that phase "(16 Items)". A phase tells you where handling stands, not what the Board decided; the letter's own description of Phase 3 says formal consideration happens "via a scorecard and/or resolution". So the number that would let a reader audit the relationship is not in the phase count, not in the action column, and not in the itemized index. It is in the one place nobody prints it.

Put the requirement plainly. To check this rule yourself you need five things together: the disposition, the reasons, the vote, the denominator, and the threshold. The scorecard carries the first two. The minutes carry the third. The resolution carries the fifth, and the threshold itself. The fourth is carried by nobody. That is the whole finding, and it is one field wide — which is why it is worth printing rather than lamenting. Four pages, four institutions' worth of effort, and the single number that converts a published tally into a verifiable test is the number left to the reader's arithmetic.