Summary

  • RIPE NCC’s May 2026 draft-agenda notice said a member resolution needed support from 394 members to be added to the General Meeting agenda by 6 May.
  • Article 15.6 of the Articles of Association sets the rule at two percent of the total number of possible votes, requires the verbatim resolution and gives the application a deadline two weeks before the meeting.
  • The later voting record counted 19,415 eligible members in a staff analysis, 3,421 registrations and 3,049 ballots. These are three different stages and none can silently replace the earlier threshold denominator.
  • A privacy-safe threshold receipt could preserve the membership snapshot, rounding rule, immutable proposal text, additions and withdrawals of support, qualification time and final agenda disposition without naming supporters.

The first vote on a member resolution happens before any ballot exists.

For the RIPE NCC General Meeting in May 2026, that first decision was expressed as an integer. The draft-agenda announcement told members that proposed resolutions would need the support of 394 members by 6 May to be added to the agenda. The page linked a proposal system. The ballot would come later, if the proposal reached it.

That distinction is constitutional rather than semantic. A supporter at the agenda stage is not necessarily voting yes on the resolution. The supporter may believe the subject deserves a formal hearing, want the exact text distributed to all members or prefer that the General Meeting reject it on the record. Admission, deliberation and adoption are separate acts.

The number 394 therefore should not be compared casually with a winning majority. It was a gate. Its job was to decide whether a member-authored question would become an institutional question.

The rule has a strong defence

The gate is not an improvised administrative veto. RIPE-818 places it in Article 15.6 of the Articles of Association. A written application from a group of Members jointly entitled to cast at least two percent of the total number of possible votes adds another subject to the agenda. The application must carry the verbatim text of the proposed resolution and reach the Executive Board at least two weeks before the meeting. The Board must immediately forward that text to all Members.

Article 15.7 completes the notice rule: the General Meeting may vote only on resolutions whose verbatim text was sent to Members in the prescribed way. Article 15.8 then creates a higher rung. Members jointly entitled to at least ten percent of possible votes can require the Board to convene a General Meeting within four weeks and may convene it themselves if the application is not honoured within fourteen days.

This architecture protects more than the Board. It protects every member from an agenda that changes after the preparation window. It prevents one unsupported late motion from forcing the association to decide immediately. It requires the actual resolution, not a slogan about it, to circulate in advance. And it gives members a direct route that does not depend on the Executive Board choosing to sponsor their subject.

The strongest case for the two-percent rule is thus straightforward. A statutory meeting needs an admission rule, enough demonstrated support to justify membership-wide notice, and a fixed text that can be examined before voting. Abolishing the gate would not automatically create more democracy. It could reward surprise, ambiguity and the party best able to mobilise at the last minute.

The accountability question is narrower. If the institution publishes the output of the gate as 394, can a member later reproduce how 394 was calculated and how a particular proposal moved through it?

“Possible votes” is not turnout

Article 16.7 gives one vote to each Member who has not been suspended and denies voting rights to Candidate Members. Article 16.8 caps the votes cast by one proxy at one percent of all possible votes. These provisions make the denominator a membership state, not a forecast of participation.

The later official voting report says 3,421 members registered and 3,049 cast ballots. A RIPE NCC staff analysis gives a wider picture: 19,415 eligible members, 3,421 registrations, 3,049 ballots and 15.7% turnout. It also explains that the voting model remains one member, one vote even when a member has several LIR accounts.

Those figures are useful. They do not answer the same question.

Population What it measures What it cannot replace
Possible votes for Article 15.6 The membership base used to calculate the agenda threshold at its snapshot Later registrations or ballots
3,421 registered members Members who completed the voting-registration stage Members entitled to support an earlier agenda proposal
3,049 ballots Members who ultimately cast a ballot The statutory possible-vote denominator

The dates matter as much as the definitions. The proposal cutoff preceded the meeting. The staff turnout analysis described the later electoral event. Memberships can mature, close, become suspended or change state between two snapshots. The checked sources do not say that this occurred in a way that affected 394. They simply do not expose the snapshot used for that calculation.

It would be equally wrong to treat the 394 supporters as a subset of the 3,049 voters. A supporter might not register for the meeting. A voter might never have supported a proposal. The evidence contains no proposal-level supporter list, and none is needed for the argument.

A moving integer needs a dated denominator

The threshold itself shows that it is not timeless. The May 2025 notice published 395 supporters. The October 2025 notice published 397. The May 2026 notice published 394.

395 → 397 → 394

That sequence is not evidence of instability or error. A percentage applied to a changing membership base should produce changing integers. It does show why the integer must travel with its calculation identity. Which Articles version governed? At what time was the possible-vote population frozen? Which membership states were included? Was the result rounded up, rounded to the nearest whole number or derived by another published convention?

The checked pages do not answer those questions. Nor do they show that RIPE NCC answered them incorrectly. Inferring an exact denominator backwards from 394 would manufacture precision because the rounding rule is not in the record examined here.

The captured post-close member-proposal page says the two-week term has closed and that all members had the Article 15.6 right to propose an agenda topic with an associated resolution. It does not retain, on that checked surface, the threshold calculation, proposals, support history or final qualification states. This may be a presentation choice. It is not proof that internal records do not exist.

Support is a state, not a pile of names

A proposal does not merely have a final count. Support can be added, withdrawn or invalidated when a membership state changes or a duplicate event is corrected. A proposal can cross a threshold before the cutoff and later fall below it. The verbatim text can remain immutable, or a revised text can become a different proposal. The Board can receive a qualifying application, check its form and distribute it.

Each transition needs an owner and a time. Without that history, the final count cannot answer the most difficult but ordinary questions:

  • Did the text supported at 10:00 remain the text sent to all Members?
  • Was the denominator fixed when the proposal window opened, when it closed or at another declared time?
  • If support was withdrawn, did qualification depend on the peak count or the count at cutoff?
  • If a supporter ceased to be entitled to vote, which rule governed the event already recorded?
  • If a correction changed the count, was the earlier public state preserved?

These questions do not imply that any disputed proposal existed in May 2026. They define the information a reusable procedure should preserve before a dispute occurs.

The smallest credible threshold receipt

RIPE NCC does not need to publish supporter identities, portal credentials or a membership register. A compact public receipt can prove the operation without exposing the people behind it.

Receipt field Public value Control function
Meeting and rule Meeting ID, Articles version, opening and cutoff time Fix the procedure
Denominator definition Included and excluded membership states Explain “possible votes”
Snapshot Time, aggregate state counts and digest Preserve the population without names
Calculation Percentage, rounding convention and resulting integer Reproduce 394
Proposal identity Stable ID and hash of verbatim resolution Bind support to exact text
Support events Aggregate valid additions, withdrawals and invalidation classes Reconcile the count
Qualification Count and time at the controlling cutoff Show whether the gate was crossed
Board handling Receipt, form/legal status and distribution time Join qualification to notice
Disposition Added, not qualified, withdrawn, superseded or corrected Close the procedural state
Later ballot Ballot ID and result, if one occurred Keep admission separate from adoption

The digest is important. It allows RIPE NCC to prove that the same protected denominator snapshot was used later without publishing the membership roster. Aggregate exclusions can show how Candidate Members or suspended Members were treated without identifying them. Support events can be counted by reason class while the identity link remains protected for authorised review.

The receipt should also state that agenda support is not substantive endorsement. Otherwise a member who wants debate may be publicly characterised as backing the resolution. Privacy here protects participation as well as personal data.

This control could vindicate the present process. If the threshold was calculated from a stable rule, every valid support event was applied to immutable text and the final agenda matched the cutoff state, the receipt would make that competence visible. The aim is not to lower 394. It is to stop 394 from becoming an orphaned administrative fact.

Sources