Summary

  • RIPE NCC's Articles of Association require more than 50% of votes cast for any resolution, with abstentions excluded from that divisor and recorded separately, and one vote per unsuspended member. Read as the March 2024 text, Article 18.1 does not leave the threshold to inference.
  • The May 2026 voting report lets a member reproduce its charging-scheme result line by line: 1,547 for Model A, 1,479 for Model B, 23 blank. The 2025 decision that set the EUR 1,800 fee appears in a document titled Draft Minutes as approved, with no resolution-level count present in the reviewed record, and the association's own billing page credits it to a different resolution number.

A rule that does the work

The governing text is unusually specific, and it is worth quoting because most of this article follows from it. Article 18.1 of the RIPE NCC Articles of Association states that "all resolutions shall be adopted by an absolute majority of the votes cast, i.e. greater than fifty percent (50%) of the votes cast." Article 18.6 then decides what counts: abstentions and invalid ballots sit outside the votes cast, and must be recorded regardless. Article 16.7 gives each unsuspended member a single vote and bars candidate members.

That is a checkable rule. It defines its own denominator, so anyone holding the counts can redo the division. The May 2025 meeting deck, a Voting Procedure and Platform presentation dated 14 May 2025 and attributed to Karla Liddle-White, restated the same bar in plainer words — "All resolutions require more than 50% of YES votes to be approved" — offered Yes, No or Abstain on each item, and set electronic-only voting to close at 09:00 (UTC+1) on 16 May 2025.

Two other provisions matter for how a scheme reaches the agenda. Article 7.1 keeps an adopted charging scheme alive until it is replaced, and Article 15.4(b) puts the following year's scheme before a General Meeting on an Executive Board proposal. No numerical attendance quorum appears in Articles 14 to 21 of the text as retrieved. That is a statement about the reviewed pages, not about Dutch company law.

May 2026: the sum adds up

The voting report for the May 2026 General Meeting prints the counts. On Resolution 3, the charging-scheme choice, Model A took 1,547 votes and Model B took 1,479, with 23 blank votes; the displayed shares are 51.12% and 48.88%. Add the two options and the blanks and you reach 3,049 ballots. That total is arithmetic on the published counts, not a figure the report states.

RIPE Labs supplies the electorate in an analysis published 17 June 2026: 19,415 eligible members, 3,421 registered to vote, 3,049 casting votes, a 15.7% turnout, 23 abstentions on the charging-scheme question, and Model A — one LIR account, one fee — as the selected model. The two sources agree on the ballot count, which is the useful part.

So the 2026 result can be checked at a desk. Exclude the 23 blanks and the divisor is 3,026; 1,547 is 51.12% of that. The winning option carried by 68 votes, computed from the counts. And 1,547 affirmative votes equal 7.97% of the 19,415 eligible members, again computed, which is a different number with a different meaning: the rule measures votes cast, not the size of the association.

One discipline before leaving this section. Model B was an alternative charging model placed alongside Model A, not a No vote on the scheme. The report's separate Resolution 2 table does use literal Yes and No — 2,400 to 148 with 501 blanks on the 2025 financial report — and even there the arithmetic sums to 3,049. The report also explains its own vocabulary on page 6: "Abstain votes are recorded as "Blank Votes" in the system for resolutions." The document displays no publication date.

It names John Sweeting of ARIN as independent observer, with his review dated 22 May 2026, and records voting from 18:35 (UTC+1) on 20 May 2026 to 09:00 (UTC+1) on 22 May 2026.

May 2025: a fee, a label, no sum

Now the resolution members actually feel. The Draft Minutes of the May 2025 General Meeting record the meeting opening on 14 May 2025, reconvening for results on 16 May 2025, and then state: "Resolution 9: 'The General Meeting adopts the RIPE NCC Charging Scheme 2026.' The resolution was approved." The same section identifies Resolution 1 as adoption of the Financial Report 2024.

Approved is a result word. It is not a count. The retrieved text of that file exposes no resolution-level tally for Resolution 9 — no Yes, No, abstentions or invalid ballots — and the file remains titled Draft Minutes. RIPE Labs reported the meeting-wide picture on 28 May 2025: "Out of 19,713 eligible members, 1,207 registered to vote and 1,039 cast their votes, resulting in a 5.3% turnout." Those are totals across nine resolutions and an Executive Board election. They are not the denominator for Resolution 9, and no division performed on them produces one.

The association's own May 2025 voting report landing page was located but did not open, and neither did its attachment; indexed content for that page reports 1,207 registered and 1,039 balloting, names Assembly Voting as the report provider, and lists an attachment, Voting_Result_RIPE_NCC_General_Meeting_May_2025.pdf, last modified 16 May 2025. A last-modified date is not a publication date, and an indexed attachment name is not a tally. The honest statement is narrow: the count is not present in the record reviewed here.

Two documents, two resolution numbers

The RIPE NCC Billing Procedure 2026 makes the trace harder in a way that has nothing to do with vote counts. In section 2, on the charging model for 2026, it says: "The contributions were decided by members at the May 2025 RIPE NCC General Meeting (Resolution 1)." The minutes say the charging scheme was Resolution 9, and give Resolution 1 to the 2024 financial report. Two pages of the same association, read directly, point a member at different resolutions for the same money.

I am not going to reconcile them, because the record does not. Either page could be the one that needs correcting, and the reason for the gap is not stated anywhere I read. What can be said without guessing is that the chain from an adopted fee to the member's invoice passes through a cross-reference that does not hold.

What the fee actually is

The schedule itself is unambiguous. The RIPE NCC Charging Scheme 2026 sets EUR 1,800 annually per LIR account, EUR 75 per defined independent Internet number resource assignment, EUR 50 per defined ASN assignment, and a EUR 1,000 sign-up fee — and its comparison table shows the same four amounts for 2025 and 2026. "For 2026, the annual contribution per LIR account will remain at EUR 1,800."

Remain is the operative word. Resolution 9 continued a schedule; the supplied documents do not support saying it raised one. That distinction matters for the argument, because a member who objected to Resolution 9 was objecting to a continuation, not to an increase — and still had no published arithmetic telling them how many colleagues agreed.

Two adjacent documents stay outside the claims above. A numbered Charging Scheme 2027 page, RIPE-867, displays a publication date of 11 September 2026 and an indexed description of a EUR 1,894 annual account fee; the original did not open, so none of its terms is treated as read. And the October 2026 results page, published 30 October 2026, exists here only as indexed content reporting 1,102 votes cast and approval of amendments to the Articles authorising execution before a Dutch civil-law notary; no executed deed or effective date was established, so Article 18.1 is quoted as the 2024 text.

What an uncounted approval costs

The distinction that carries this story is not between honest and dishonest. It is between a fee that is computed and a mandate that is computed. The schedule makes the amount reproducible from four published numbers. Only a resolution-level tally makes the authorisation reproducible, and for the 2025 decision the reviewed record supplies the label instead.

For a member who voted against it, the practical consequence is precise. The rule that binds them is numerical, the fee arrives regardless, and the evidence that the numerical condition was met is not in the public file they can reach. That is not an accusation of a bad count. It is a statement about who currently holds the proof and who does not.

The 2026 report shows the association is capable of closing the loop completely — same institution, same kind of question, every input published. That is why the 2025 gap reads as a record-keeping choice rather than a technical limit.

What would end the argument

Four documents would settle this, and each would settle it differently. A certified Resolution 9 tally showing Yes above half of valid non-abstaining votes, with abstentions and invalid ballots recorded separately, would demonstrate the majority and reduce the issue to publication and cross-referencing. A tally showing Yes at or below that threshold would put the word approved in direct conflict with the association's own Article 18.1. If the listed May 2025 attachment already contains the full count, the criticism narrows to discoverability.

If final approved minutes and a corrected billing page both name Resolution 9, the numbering problem is a clerical error with a fix.

Until one of those exists, the defensible sentence is short. RIPE NCC's records establish what members owe. For the 2026 schedule they do not establish, in public, the arithmetic by which members authorised it.

RIPE NCC, which files its membership and number-resource governance record in the BTW directory, is not a closed institution. It publishes its articles, its schemes and — in 2026 — the counts. That makes the missing 2025 count a smaller and more fixable thing than a scandal, and a more useful one: a single annex would close it.