Summary

  • On 20 January 2021, AFRINIC’s Board approved proposed mailing-list Terms of Use; the minutes say they were “approved for implementation”.
  • AFRINIC opened public comments on 2 February, called the text a draft and set 28 February as the deadline. On 3 March, the Board approved revised Terms after saying feedback had been considered.
  • The March record describes January as a request for consultation before implementation. That description sits in tension with the January record and must not be treated as though the two texts were identical.
  • A later consultation can identify defects and improve a private platform rule. It cannot retroactively become the authorisation that preceded the earlier corporate act, nor can participation confer sovereign power on a private technical coordinator.
  • The sound remedy is a versioned decision-and-consultation receipt linking the first draft, the exact Board instruction, any implementation hold, comments, changes, reasons, final approval, effective date and subscriber notice.

L3 — The decision before the invitation

At 09:08 UTC on 20 January 2021, AFRINIC’s Board meeting began. Under item 5 of the official minutes, “Mailing Lists Terms of Use”, the Board took note of the Terms presented by the chief executive and approved them for implementation. Resolution 202101.595 recorded the institutional act in slightly more compact terms: the Board took note of the proposed AFRINIC Mailing Lists’ Terms of Use and resolved to approve them. The minutes identify HY as proposer, OO as seconder and the resolution as passed.

That is the opening fact, not a conclusion about motive. It is also not evidence that a warning was issued, a message filtered, a posting privilege revoked or any other sanction applied that day. The sealed record proves a corporate approval and the words “approved for implementation”. It does not prove pre-consultation enforcement. Those are different propositions, and institutional analysis becomes unreliable as soon as a striking sequence is embellished with an event the documents do not establish.

The next public state arrived thirteen days later. On 2 February, AFRINIC published a call for comments. The page called the Terms of Use a draft and invited responses by 28 February. A reader encountering only that page would reasonably understand that a proposal was open for examination. A reader who also had the 20 January minutes would know that the Board had already resolved to approve proposed Terms and that its public minutes used implementation language. The comment window therefore did not open against a blank decision sheet. It opened after a recorded Board act.

Why dwell on sequence when the Board later decided again? Because order assigns a different institutional meaning to the same comment. Before approval, a submission can be evidence placed before the decision-maker while the decision is still unmade. After approval, it begins as a request to retain, amend or reverse an existing corporate position. It may be persuasive. It may expose an error. It may cause a substantial rewrite. Yet it does not travel backwards in time and become the permission on which the earlier resolution depended. Consultation can be consequential without being antecedent authorisation.

The distinction matters particularly when a platform uses the vocabulary of community. Mailing lists can be valuable places of expertise, objection and operational warning. Their users may spot consequences that a Board paper misses. But a list is not a legislature for a continent, its active posters are not an electorate, and those who happen to comment do not acquire authority over absent members, operators or other principals. If an institution describes a later exchange as the source of an earlier mandate, participation has been converted from evidence into something it cannot supply.

The February draft was not trivial housekeeping. According to AFRINIC’s published call, a subscriber would, by subscribing, bind itself to the Terms. The proposed package included adherence to a Code of Conduct and Privacy Policy, use of an everyday-life name supported by accurate and verifiable information, one account per list and restrictions on sharing access. It selected Mauritius law for the subscriber relationship, said AFRINIC could amend the Terms at its discretion and reserved legal or equitable remedies.

Its enforcement section also described recognisable tools of platform administration: written warnings, filtering or removing comments, immediate revocation of posting privileges, and temporary or permanent revocation of posting privileges. These provisions could affect who may speak through an AFRINIC-hosted forum and on what conditions. They therefore deserved clear notice, stable text, intelligible reasons and a reviewable adoption trail. But their presence did not transform the list into a public authority.

A provider’s ability to set conditions for its own communications space is an ordinary private capacity, not a police, regulatory or sovereign jurisdiction.

The distinction between platform rule and public power does two kinds of work. First, it prevents exaggeration in AFRINIC’s favour: maintaining a list does not empower a registry to punish people in the public-law sense, adjudicate public rights, confiscate number resources or regulate the region. Secondly, it prevents exaggeration against AFRINIC: moderating a privately operated forum is not inherently illegitimate, and a usable discussion space may need enforceable rules. The serious question is not whether AFRINIC could have Terms of Use. It is whether the record made the rule’s decision states, versions and consequences legible.

On 3 March, after the advertised deadline, the Board acted again. The official minutes say that consultation had taken place, feedback had been considered and the mailing-list Terms had been amended. Resolution 202103.601 approved the revised proposed Terms. The minutes identify EK as proposer and SI as seconder and record unanimous approval. This is evidence of a distinct later decision and an official claim of revision. It must be given its full weight. There is no proper basis in the sealed record for saying that comments were ignored, that consultation was fake, or that the outcome was fixed.

But the March minutes also contain a recital about January. They say, in substance, that the Board then took note of proposed Terms and requested consultation with members and the Community before implementation. That is not how the operative sequence appears on the face of the January public minutes, which say the Terms were approved for implementation, or in Resolution 202101.595, which resolved to approve them. The two official descriptions are close enough to concern the same act and different enough to resist silent harmonisation.

A later recital may preserve information that the shorter January record omitted. Perhaps there was an oral qualification, a Board paper or a private instruction imposing an implementation hold. Perhaps “approved for implementation” was intended to mean approved as a framework, subject to consultation and a second decision. Those possibilities form a credible benign account. They are not, however, sealed facts. The pre-January draft and Board paper are not available in the evidence package. Nor is an exact contemporaneous instruction saying that no implementation could occur before consultation.

The analyst’s job is neither to dismiss the March recital nor to use it to overwrite the January wording. It is to report both and identify what record would resolve the difference.

The final dated signal in this commission comes on 17 March. In his monthly update, AFRINIC’s chief executive said the Terms had been approved and that implementation work was beginning with the community-discuss mailing list. That statement weighs against a casual claim that the Terms had already been fully implemented across the lists before the February consultation. It also provides a useful institutional state: approval existed, and implementation work was said to be beginning.

Yet the update does not establish the exact effective date, the status of every list, the content of every subscriber notice or the point at which any particular subscriber accepted the final Terms.

The chronology can therefore be stated narrowly and confidently. January produced an approval and public implementation language. February produced a page describing a draft and inviting comment. March produced an official account of considered feedback, amendment and a second approval, followed by an update that implementation work was beginning. The chronology cannot be expanded into a finding that sanctions were used before consultation. It cannot reveal how many comments arrived, who made them, which clauses changed, or which submission caused a revision.

It also cannot tell us whether the later description of January reflected an undocumented contemporaneous hold or a retrospective compression of a more awkward record.

This is more than documentary fastidiousness. Governance operates through states. “Draft”, “approved in principle”, “approved for consultation”, “approved for implementation”, “implementation stayed”, “revised”, “finally approved”, “effective” and “accepted” are not decorative labels for the same condition. Each identifies who can still change what, which text controls, and what another person is entitled to rely upon. When a record collapses these states, the institution may still act reasonably, but outsiders cannot reconstruct that reasonableness without trusting an after-the-fact description.

Suppose, for example, that the January decision really authorised only consultation. Then the precise public record should have said so: approve the attached draft for consultation; no implementation before a second Board resolution. Under that wording, the 2 February call would execute the January instruction, and the 3 March decision would complete an intelligible chain. Suppose instead that January approved a usable rule but the Board voluntarily reopened it after concern. That, too, could be legitimate, but the consultation would then be a reconsideration of an approved position.

The institutional virtue lies partly in knowing which of these two accounts is true.

The available record leaves another important distinction intact: approval is not acceptance. The February draft reportedly treated subscription itself as binding the subscriber. But this commission does not have a sealed subscriber-by-subscriber receipt, a final-version notification, an affirmative click record, an account-state transition or a documented exit path. It does not adjudicate the legal effect of the draft’s binding formulation. That question should not be answered by inference from the existence of a Board resolution or by treating continued presence on a mailing list as obviously informed consent.

Nor should the adoption dispute be converted into a personality story. The initials of proposers and seconders belong to the formal record, not to a theory of individual intent. Nothing in the evidence supports an accusation against a director, executive, moderator, commenter or subscriber. The material is sufficient for a better kind of scrutiny: an audit of the organisation’s publicly recorded sequence and the controls that would make such a sequence dependable.

That audit starts with a simple observation. The 20 January decision was a private corporate act. The 2 February invitation was a public request for information and views. The 3 March decision was another private corporate act, this time approving a revised proposal. Their legitimacy does not require the fiction that mailing-list participants became a sovereign constituent body. It requires the more modest and more achievable disciplines of accurate minutes, preserved versions, fair notice, usable submission routes, reasoned handling and a clear effective state.