Summary

  • ARIN published four candidate-level Board counts for 2026; the third seat turned on five votes, 569 against 564, a difference of 0.47 of a percentage point.
  • The same election instructions archive the results electronically and permit printing only at the request of the President, Chief Operating Officer or Chief Experience Officer, while tally confirmation runs through the President, Board Election Officer, Chief Legal Officer and the Board.

When the American Registry for Internet Numbers posted its 2026 Board of Trustees results on 2 November, it did what careful registries do and what cautious ones often avoid: it printed the losing margin. Tina Morris took 716 votes and 66.73%. Dan Alexander took 634 and 59.09%. Catherine Middleton took 569, 53.03%, and the third seat. Kevin Blumberg took 564, 52.56%, and no seat (2026 ARIN Elections Results). Five votes and 0.47 of a percentage point separate the last winner from the first loser; that subtraction is mine, not ARIN's. Participation was reported with the same frankness: 1,073 General Members voting through 898 Voting Contacts, 45.7% of those eligible.

That is not the interesting part. The interesting part is what the same election's rulebook says about the paper underneath it.

The franchise is specified with a precision worth stating plainly. A candidate reaches the ballot by self-nomination and questionnaire, an outside assessment firm and Nomination Committee review, with a slate required to exceed the open seats (Bylaws, Article VIII); the alternative is a petition needing at least seven days to declare intent and fourteen more to collect signatures, at a threshold of no less than two percent of eligible General Members and never fewer than one hundred, counted once per member through its voting contact. Members in good standing forty-five days before the election may vote, a five percent turnout supplies quorum, and the ballot is electronic. One Voting Contact speaks for each member; the 2026 ballot let that contact select up to three Board candidates and five Advisory Council candidates; and a contact representing several organisations carries a weight greater than one, applied whole to every candidate it selected and never split (Election Instructions). A member group with affiliates therefore votes once, at full weight, for the same list. That is a design choice, and it makes the weight ledger part of the result rather than an implementation detail.

Now the record. The instructions state that "All results are archived electronically and may only be printed at the request of the President, Chief Operating Officer, or Chief Experience Officer." Voters get a confirmation page and an emailed receipt after submitting. The process document adds that an internal election report goes to the President, the Board Election Officer and the General Counsel, who confirm the tally and the process before the Board confirms the result and the President announces it (Election Processes). The announcement names the President, the Board Election Officer and the Chief Legal Officer as the officers who confirmed adherence to the procedures and the tally, and says the results were verified according to the bylaws. General Counsel in the governing documents, Chief Legal Officer on the results page: two labels, one office, no reconciliation offered.

What the inspected public record does not contain is the thing a member would need to recompute 569 against 564: a cast-vote record, a ballot-level dataset, a cryptographic verification artifact, or an independent election audit. The internal election report exists as a described document; it was not obtained in this review, and nothing here establishes that a member has asked for it and been refused.

This is not a question ARIN has never considered. At its meeting on 10 January 2017 the Board reviewed the 2016 election report and discussed the voting software and the outside vendor; the minutes record a trustee suggesting "that election results be contained in a tamper-proof audit bundle (set of election results and supporting data)," with staff to return with a proposal (minutes of 10 January 2017). Nine years later the public record shows a precise tally and an electronic archive whose printing three officers control. The minute is a benchmark from inside the institution; it is not evidence that a bundle was adopted, or that its absence is concealment.

Peer practice cuts in an unexpected direction. The RIPE NCC voting report for May 2026 names its three Executive Board winners by seat, records the election period, a two-factor login, 27 voter-support requests, and an independent observer: John Sweeting, Chief Experience Officer at ARIN (RIPE NCC voting report, May 2026). The meeting transcript thanks him in that role (General Meeting transcript). The voting presentation documents instant-runoff counting, a ballot receipt, a tracking code and QR access to a verification page showing ballot activity (voting presentation); the Articles require ranking and instant-runoff counting and require minutes to record results, without expressly requiring publication of a per-candidate tally (Articles of Association). Candidate-level numbers did not appear in the text of the report I could retrieve, and the May 2025 results page likewise gives 1,039 votes cast and two winners without candidate totals (May 2025 results, May 2024 report). That is a boundary of this review, not a finding that RIPE NCC never publishes them (voting-report landing page).

So the comparison is not more transparency against less. ARIN publishes the finer number; RIPE NCC publishes the process witness and the voter's own receipt and, in the artifacts I could read, not the finer number. And the office that can authorise a printout of ARIN's results served as the named independent observer of RIPE NCC's vote. That coincidence of titles proves nothing about either count, and it shows how differently two membership organisations deploy the same officer role.

One chronology gap stays open. ARIN's elections calendar lists Board confirmation on Thursday 5 November 2026 and public announcement on 6 November; the results page is marked Posted 2 November 2026 (elections page). The inspected pages do not say whether the calendar went stale or the results went out early.

Read together, the evidence supports something narrower than either cheerleading or suspicion. Granular disclosure is a real legitimacy gain: members learn that the third seat turned on five votes, which is more than most membership ballots ever tell them. It is also a limited gain, because the number arrives as an assertion from the officers who certify it, and the arithmetic a reader can perform tests presentation, not eligibility, weighting or aggregation. At five votes, with a ballot rule that lets one multi-organisation contact carry more than one vote, a recomputable record would have to expose the weight ledger, not just the totals.

The boundary that matters is functional rather than paper against electronic. Interface and internal role labels can be operational choices. Eligibility, ballot weights, nomination access, seat allocation, evidence retention and any route to challenge a tally touch member entitlements, and those are accountability questions. Nothing in the record establishes fraud, alteration or an invalid result; the print restriction is a control point, not proof of concealment.

What would change this reading is specific. A petition that crosses two percent or a hundred supporters and appears on a slate as Advanced Through Petition would show the second route working; no Board candidate carried that label on the September slate (2026 slate, nominations announcement). A published dataset with eligibility and weight fields would move ARIN from the finer number to the checkable one. A named observer with a published mandate would borrow the RIPE NCC instrument outright. And an explanation of 2 November against 5 November would close a gap that sits in the middle of the certification story, where a verification question is not an accusation.

Sources: Bylaws; Election Processes; Election Instructions; nominations announcement; 2026 slate; 2026 results; elections page; Board minutes, 10 January 2017; RIPE NCC voting report, May 2026; RIPE NCC General Meeting transcript, May 2026; RIPE NCC May 2025 results; RIPE NCC voting presentation; RIPE NCC Articles of Association; RIPE NCC May 2024 voting report; RIPE NCC voting-report landing page; ARIN directory entry.