Summary

  • The Address Council’s Operating Procedures define quorum as “a majority of Address Council members” and print no number anywhere; the base arrives only from a definitions paragraph that excludes vacated, removed and expired-term seats, so the denominator shrinks with every vacancy.
  • One ballot carries two bases in adjacent subsections — “an absolute majority of the voting members” and “a majority of the votes cast” — with an abstention counted toward the turnout floor and excluded from the vote total.

A tally with a numerator and no base

The minute does not look like a problem. “GV said there were 14 votes recorded in the voting system, as follows:” it reads, then “8 votes in favour of HPH” and “6 votes in favour of WW” (ASO AC minutes, 2 October 2013).

Eight is a majority of fourteen ballots. But the procedure does not test a candidate against ballots. Section 7.2.3 asks for “first-choice votes from an absolute majority of the voting members”. Eight is a majority of the ballots recorded; whether eight is a majority of the voting members depends on a number the minute does not print and the procedure does not define.

Quorum in the same minute was settled by roll call and by nothing else: “Having realized quorum, LL called the meeting to order at 1208UTC.” No count, no base, no dissent recorded.

The number that used to be in the text

Until the Kyoto amendments, quorum was printed. The redline strikes this sentence: “Eight members of the council shall constitute a quorum for the transaction of business providing that there is at least one person present from each of the five (5) RIR Geographic Regions.” What replaces it is a fraction: “A majority of Address Council members shall cons5tute a quorum for the transac5on of business” (Kyoto amendments redline, published December 2023).

I have reproduced that sentence with the retrieval’s character substitutions intact, because “cons5tute” is what came back from the PDF and I will not tidy a document I did not see rendered in full.

A named floor of eight, with a regional presence condition bolted to it, became a majority of members. A fraction is only as computable as its base, and the base sits in Section 1: “All references to members of the Address Council in these Operating Procedures shall mean persons who have been duly elected or appointed to the NRO NC to serve on the Address Council by an RIR or an RIR community, as applicable, who have not vacated or been removed from their seats, and whose terms have not expired” (Operating Procedures).

That is a definition by subtraction. It is exact about who is out — vacated, removed, expired — and silent about how many are in. When a seat empties the quorum falls with it, and no published page records the new total. The threshold moves inside a room the reader cannot see into.

One ballot, two bases

The voting section does something the quorum section avoids: it names two different populations for one act.

Section 7.2.2 cancels a ballot when turnout is too thin — “If fewer than half of Address Council members have voted by the time the voting window closes, the vote will automatically be canceled.” Section 7.2.3 awards the seat on an absolute majority of the voting members. Section 7.2.6 then handles abstention both ways at once: “An abstention shall be counted as a vote for purposes of determining whether half of the Address Council members have voted under Section 7.2.2. However, an abstention shall not be counted among the total number of votes for purposes of determining whether a candidate has received a majority of the votes cast.”

Read as a design, that is coherent, even humane: a member with a conflict keeps the ballot alive without tipping it. Read as arithmetic, the same abstention is a vote and not a vote in the same election, and the winner’s test is written against “voting members” while the abstention sentence speaks of “votes cast”. The procedure defines neither term against the other, and nowhere says the two populations are the same set.

I will not award a winner to either sentence. Two rules for one act inside one instrument is a drafting question, and the document does not settle it.

The room asked the same question

If the missing denominator were an outsider’s complaint, that would be one thing. The Council’s own minutes raise it.

Belgrade, 24 October 2022, one seat vacant: “Seeing that one seat is vacant, Kevin B. asked whether that should be interpreted as 4/5 of 14 or 15 members” (Belgrade minutes, FINAL). In the same discussion a member is recorded saying “If a seat is vacant, it does not count towards quorum”, while the amendment threshold then in force is described as requiring “the votes of 4/5 of the AC (12 members)”. Twelve is four fifths of fifteen; that is my arithmetic, not a printed total.

The Secretariat’s contribution in the same minutes is the most useful line in the file. The Council can hold quorum without one of the regions, but “would not currently have quorum for the procedural review”; with four regions it “can operate a standard meeting but not pass procedural updates to the EC”. One roster, two quorums, depending on the kind of business.

Two positions, same hour, same document: a vacant seat outside the base, and a threshold computed on fifteen.

Four tallies, four shapes, no electorate

The published outcomes agree on exactly one thing: none of them prints the population the rule divides by.

  • 6 August 2009 — “There were 12 members of the ASO AC in favour and none opposed so the Chair declared the motion carried.” A count of members (minutes).
  • 11 December 2008 — “The Secretariat conducted a vote for this motion, with 13 votes in favour and none against. The motion passed unanimously.” A count of votes, called unanimity, with no accounting for who did not vote (minutes).
  • 2 October 2013 — fourteen recorded, eight to six.
  • 4 December 2021 — an amendment result reported to the mailing list: “Votes casted: 13 Yes: 11 No: 1 Abstention:1”, with the proposal said not to have reached “the required support of fourth-fifths of all members of this Council” (ac-discuss thread).

The 2021 item is a list report, not a ratified minute, and I treat it as such. It is still the most explicit record in this set: it states the fraction, states the result, and then works the base in public — one participant writes “You just need 80% majority i.e. 12 votes in favor”, another “12/15 of ALL members”. Eleven affirmative votes against twelve is one short. On a base of fourteen, four fifths would be 11.2, so eleven is short there too. The conclusion survives either base, which is why I can state it without picking one.

Something else moved as well. Section 11 now requires “two thirds of the members of the Address Council approve the amendment by vote”, not four fifths. Results computed under the older fraction and the newer text are not the same measurement, and the records do not always say which text applied.

Two silences, and only one is a gap

There is a second absence in these documents. Conflating it with the first would be the cheap version of this story.

Section 7.2 states that for appointments “voting shall be anonymous, vote counts shall remain internal to the ASO, and only the final decision as to whether a candidate has been successful shall be reported externally”, while in all other voting matters “each member’s vote shall be recorded in the Address Council’s public minutes”.

That is a rule, written on purpose, and it is why the 2026 record of the ICANN Board Seat 9 vacancy reads as it does. The 1 July 2026 minutes record “Akinori M. abstained, noting that he could be considered a potential candidate”, a member declining to comment in chat, no objections, and the decision “The ASO AC agreed to use Section 9.5.3” for the vacancy, with candidate-specific matters taken to closed session (1 July 2026 minutes). The August minutes add that the seat “was being filled using section 9.5.3” and that “results are expected by mid-August” (5 August 2026 minutes).

No tally there is withheld improperly. It is withheld by design, and the design says so in the same subsection that leaves the majority base open. That distinction carries the argument: a body may keep a personnel count confidential and still owe its readers the arithmetic of its own published rule.

Where the reconstruction stops

A stranger holding these documents can reproduce the ballots recorded, the candidate split, whether the leader held more than half the ballots — and the result under whatever base the reader happens to assume.

They cannot reproduce the test the procedure actually states. To check an absolute majority of the voting members they need the number of voting members that day. The minute prints ballots, not members. Section 1 defines members by exclusion. And the standalone selection page does not restate a threshold at all; it cross-refers: “The selection phase shall last at least 14 days ... the Address Council shall meet to evaluate the remaining candidates and vote pursuant to Section 7.2” (ASO AC ICANN Board Selection Procedures).

For ordinary motions the procedures print no fraction of their own. Section 10.1 imports a third-party manual wholesale — “The Address Council shall use the current edition of Robert’s Rules of Order Newly Revised for its parliamentary procedure” — and Section 10.2 makes any special rule that would displace it subject to NRO EC approval.

So the arithmetic ends in a room. Not through concealment, and not because anyone counted wrongly: through one unprinted integer in a document that prints nearly everything else.

What would close it

One number would do. A figure in Section 5.2. A definition of “voting members” in Section 1. A minutes line that prints the electorate beside the tally, the way the 2009 and 2013 minutes print the votes.

None of that changes a past outcome, and I am not asserting one was wrong. It changes what the next dispute can be about. A member who loses a ballot by one vote currently cannot point to a printed denominator; a council that wanted to end that argument could print one and keep it ended.

Quotations above are reproduced from retained snapshots of the documents as returned for this review, including the redline’s encoding substitutions. Five boundaries apply: the procedures’ middle sections (2–4 and 6) were read only in fragments; the closed-session record of the Seat 9 selection was not obtained; the redline came back partially; the Belgrade excerpt arrived with ellipses; and no consolidated PDF, transcript or unredacted annex was obtained.

Directory entry: Address Supporting Organization Address Council