Summary

  • AFRINIC 32 demonstrated the practical value of a virtual public policy meeting: a large policy docket could be presented, objections could be heard, unresolved proposals could return to the mailing list, and formal consensus decisions could remain distinct from later discussion webinars.
  • The published minutes leave essential dimensions of the virtual floor unknown, including complete registration and live-attendance totals, unique-person and organisation counts, access failures, a full speaking-queue ledger, any exact policy-floor poll instrument and a clause-level record of objections, responses and chair reasons.
  • A participant, speaking turn or diagnostic poll response can contribute evidence to rough consensus, but none is a vote, a representative sample or proof that absent networks, members, users, organisations or states authorised the people in the room.
  • The proportionate remedy is a versioned virtual-meeting receipt that preserves access states, aligned denominators, queue outcomes, any poll instrument, proposal versions, objections, responses, chair assessments, list-to-floor handoffs and corrections while protecting personal data.

L3 — The room without a denominator

Day 1 of AFRINIC 32 began at 11:24 UTC on 16 September 2020. The next day began at 09:15 UTC, when Moses Serugo opened proceedings and stated that interventions in the queue would be taken first come, first served, with each limited to one minute. Those details place the reader inside an institutional moment more revealing than the familiar description of an online conference. A private number registry had transferred a consequential policy floor into a virtual room. Entry, presence, turn-taking and the surviving record had become part of the governing mechanism.

That transfer was reasonable. A physical meeting in September 2020 would have faced obvious continuity problems. A virtual meeting could allow policy discussion to continue without requiring everyone who wished to contribute to reach one venue. It could reduce travel and venue barriers for at least some participants, bring in people from more locations and preserve a schedule on which proposals, objections and decisions needed attention. The official minutes show that the meeting did continue. They identify six Day 1 agenda items: five policy subjects and the results of a PDP co-chair election.

They describe a two-day forum in which proposals were presented, participant reactions were summarised and chairs made or deferred assessments.

Continuity, however, is not the same thing as representativeness. The useful question is not whether the virtual room was real. It plainly was. The question is what the public record allows anyone to know about access to that room and the relationship between those who appeared in it and the much larger set of people and networks potentially affected by registry policy.

AFRINIC’s introductory account described its policy development process as open, bottom-up and transparent. It said anyone could submit a proposal or participate, required at least four weeks of mailing-list discussion and presented the public policy meeting as a forum for discussion and consensus seeking. The minutes also said that all participants present that day would form part of the Policy Development Working Group. That is a statement about the meeting’s participation rule. It is not an attendance count, an identity record or a demonstration of delegated authority.

The missing distinctions matter. Registration is not attendance. Attendance is not necessarily a unique person. A unique person is not necessarily an authorised representative of an organisation. An organisation is not an affected network. A collection of affected networks is not a denominator for all users, operators, members or states in a region. Each step changes the question being answered. Collapsing the steps lets the visibility of a meeting do argumentative work that the underlying evidence cannot support.

The published minutes do not state how many participants were present or how they were identified. They do not provide a complete registration denominator, a live-attendance count, a unique-person count, an organisation count, a member count, an affected-network count or a geographic distribution. They do not supply a join-and-leave history. Without those measures, a reader cannot tell whether the people recorded as present were a small or large share of registrants, whether multiple sessions or connections belonged to one person, how many organisations had anyone there, or how presence varied over the two days.

This is an evidential limit, not an accusation. The absence of those details from the published record does not prove that AFRINIC possessed no registration file, platform log or internal note. Private records might have existed. They are simply not established here and cannot be used to fill the public gap. Institutional analysis becomes more reliable when it keeps an unknown on the unknown side of the ledger instead of treating silence as proof either of perfect administration or of misconduct.

The access route is similarly incomplete in the surviving minutes. They do not identify the public PPM platform, the public joining link, the registration rule, the method of identity verification, a dial-in alternative or any bandwidth accommodation. They do not establish whether captioning or interpretation was available, how disability access was handled, whether technical failures occurred, how many attempted entries failed, or whether an unsuccessful participant could recover and rejoin. They do not identify excluded speakers or unsuccessful attempts to enter the speaking queue.

Virtual access can widen the possible floor while narrowing it in ways that are difficult to see. Removing a journey can help someone who could not afford travel. Requiring a stable connection can hinder someone whose bandwidth is unreliable. A meeting scheduled at a fixed UTC time can be convenient in one place and burdensome in another. A visual platform can aid some forms of engagement while creating difficulty for participants who need a dial-in route, captions or another accommodation. None of those outcomes may be inferred for AFRINIC 32 without evidence.

They instead explain why a virtual meeting’s public receipt should record aggregate access conditions and failures.

NRS’s account of AFRINIC as a private member-based registry helps place this problem correctly. NRS may research, document, convene and advocate for accountable coordination, but it did not run AFRINIC 32, certify the people who joined or determine consensus. Its substantive relevance lies in keeping the organisation’s private character visible. Openness can improve a private process without converting it into a public institution. Wider access is valuable because it can bring more facts, operational experience and objections into a technical discussion, not because every open link creates an electorate.

Heng Lu’s distinction between a stakeholder and a principal sharpens the point. A stakeholder may be affected, may hold useful expertise or may contribute evidence. A principal authorises another to act. Joining a meeting demonstrates participation under the meeting’s rules; it does not demonstrate that absent operators, organisations, end users, members or states delegated authority to the joiner. The word “community” can describe people who participate in a shared technical process. It cannot, by itself, identify a legal principal or certify a regional mandate.

The Day 2 queue rule illustrates how participation becomes an institutional outcome. First come, first served sounds neutral because the sequence does not ask a chair to prefer one viewpoint over another. The one-minute limit sounds even-handed because each admitted intervention receives the same nominal time. Yet the rule’s fairness and evidential value depend on facts the minutes do not preserve: who could see or enter the queue, when each entry was made, how ties or reconnects were handled, how many people waited, who withdrew or timed out, whether anyone was excluded, and whether the one-minute limit was applied consistently.

BTW’s analysis of rough consensus by microphone order supplies a useful comparator, not additional evidence about this particular meeting. Its relevance is structural. A queue does not merely organise speech; it shapes which objections become audible before time expires and which arguments the public record can later summarise. The first intervention can frame the issue that follows. Repeated points can consume scarce time. A late technical objection can look isolated because earlier speakers have already set the room’s apparent direction. None of this proves bias or capture at AFRINIC 32.

It shows why queue admission and order belong in the meeting receipt.

Equal minutes are not necessarily equal opportunities. A participant with a stable connection and immediate command of the meeting language may use sixty seconds differently from a participant whose connection drops, who relies on interpretation or who must compress a complicated operational consequence into a short intervention. Again, the record does not establish that such inequalities occurred. It establishes only a first-come-first-served rule and a one-minute limit. The institutional task is to preserve enough aggregate evidence for a later reader to distinguish a rule’s formal symmetry from its practical operation.

The minutes do preserve useful substance. They summarise concerns raised during the Simple PDP Update discussion about trust in mailing-list participation, the possible use of multiple accounts and whether inputs could be manipulated. Those are recorded concerns, not findings that manipulation happened. Treating them as proof would repeat the very mistake this analysis rejects: moving from a voiced possibility to an established institutional fact without the intervening evidence.

The minutes also explain the intended sequence after the meeting. Meeting consensus would lead to a last-call period of at least two weeks, followed by a co-chair recommendation to the Board, Board ratification and AFRINIC implementation. No consensus would return a proposal to mailing-list discussion. Each is a separate state. A meeting assessment is not last-call closure. Last call is not a Board decision. Ratification is not implementation. The separation gives the process opportunities for further notice, response and correction; it also means that no single view of the virtual room should be allowed to stand in for the whole chain.

AFRINIC 32 left evidence of that separation in practice. The minutes record no-consensus dispositions for the Simple PDP Update, the PDWG Guidelines and Procedures proposal, the Chairs Election Process proposal, the Policy Compliance Dashboard and the Day 2 policy proposals, sending them back to mailing-list discussion. No consensus should not be rewritten as a policy rejection on the merits. It describes a procedural disposition: the chairs did not determine that the proposal had reached the required state at that point, and discussion was to continue.

For the Board Prerogatives proposal, the Day 2 minutes record a conditional consensus assessment after the authors agreed to remove a specified section. This single account should neither be inflated into proof that every part of the meeting was fully documented nor dismissed because other proposals returned to the list. It demonstrates that chair judgement could connect a particular change to a stated disposition. The public-interest question is how consistently that relationship between proposal version, objection, response and assessment can be reconstructed across the entire docket.

The Day 2 open-microphone summary adds another valuable boundary. It said online webinars would continue for discussion and that consensus would not be determined in those webinars. That separation resists a common form of institutional drift. Additional discussion may broaden understanding, but a later webinar should not quietly inherit the decisional status of the recognised policy floor. Preserving the boundary makes continued engagement safer: people can exchange evidence without being told afterwards that mere attendance in an auxiliary session carried an unstated formal consequence.

The immediate significance of AFRINIC 32, then, lies neither in celebrating virtual participation as automatically inclusive nor in treating an incomplete minute as evidence of failure. The event shows a private coordination function doing necessary work under difficult conditions. It also shows why virtual procedure needs more than a meeting link and a narrative summary. When the room itself is software-mediated, the access path, attendance states, queue behaviour and record design become part of the process that rough consensus is meant to assess.