Summary

  • AFRINIC announced on 28 August that AFRINIC-38 will be a one-day hybrid Public Policy Meeting in Cape Town on Wednesday, 18 November 2026; the programme, venue, registration and participation details were still to follow.
  • Staff said the date was chosen to coincide with Africa Tech Festival and the South African Peering Forum in Cape Town, because many AFRINIC Resource Members would already be there and no alternative date and location fitted the second 2026 meeting.
  • The date is also inside IETF 127's official 14–20 November interval and ITU PP-26's 9–27 November interval. Those overlaps do not prove that a known person will be excluded, that an agreement was breached or that another date would perform better.
  • A versioned calendar-tradeoff receipt should preserve the participant groups, conflicts checked, co-location assumptions, rejected alternatives, hybrid programme and later unique-participation outcomes behind the claim of maximum participation.

One Wednesday sits inside four calendars

The easiest fact in this story is the date. AFRINIC Communication posted the meeting announcement to the RPD mailing list on 28 August. AFRINIC-38 will take place in Cape Town on Wednesday, 18 November. It is described as a one-day, hybrid Public Policy Meeting for discussing and reviewing Internet number-resource policies in the AFRINIC service region.

The date can be placed against three other public records without estimating anyone's itinerary. IETF's official page schedules IETF 127 in San Francisco from 14 through 20 November. The official Africa Tech Festival site schedules its Cape Town programme for 17–19 November at the Cape Town International Convention Centre. ITU schedules its Plenipotentiary Conference in Doha from 9 through 27 November. Eighteen November lies inside all three intervals.

That arithmetic establishes simultaneity. It establishes nothing about the size of the shared participant set. The fact that an AFRINIC policy contributor could also attend the IETF does not show how many would. The fact that governments and communications officials attend ITU conferences does not show which of them intended to join AFRINIC-38. The fact that Africa Tech Festival attracts telecom operators does not make its audience identical to the AFRINIC policy community.

This distinction is the centre of the article. A calendar overlap is observable. A participation effect requires a second record.

AFRINIC gave a rationale, not a result

The first objection on the RPD list focused on IETF 127 and asked AFRINIC to move the meeting. A participant also asserted that Internet organisations had an understanding intended to prevent such clashes. The frozen public sources do not establish a binding agreement, and the objection should not be promoted into an institutional finding.

AFRINIC staff member Ben Roberts supplied the other side of the trade. He said the week and city were chosen because Africa Tech Festival and the South African Peering Forum would take place there, encouraging maximum participation from people who were already travelling to Cape Town. In a follow-up, he emphasized the cost of travel within Africa and said many AFRINIC Resource Members in particular would already be present. He also said that no alternative date and location allowed AFRINIC to fit in the second meeting of 2026.

That is a coherent operational argument. Co-location can reduce the marginal cost of participating. A person who has already budgeted for one journey may be able to add a one-day policy meeting more cheaply than a separate trip. A registry serving a region with expensive and indirect air connections has reason to look for that economy.

But purpose is not outcome. The statement does not publish a candidate-date table, a count of the members expected to be in Cape Town, a list of policy proposers whose availability was checked, or a comparison between the participants gained through co-location and those placed in simultaneous programmes elsewhere. It does not need to prove the choice was perfect. It does need a record if “maximum participation” is later used as evidence that the policy process reached the people it was meant to reach.

Co-location saves one cost and creates another

The economic intuition behind co-location is strong because travel is lumpy. Airfare, visas, accommodation and time away from a network do not fall neatly by the hour. Once a participant has paid those costs for a Cape Town week, adding AFRINIC-38 may be relatively cheap.

Time, however, cannot be co-located in the same way. Africa Tech Festival's public dates cover the whole day of 18 November. AFRINIC had not yet published its programme at the research cutoff. Without session times, a participant cannot know whether a policy presentation, consensus call or response window competes with a commercial, technical or regulatory session already on the calendar.

The same applies across cities. A contributor in San Francisco or Doha might be able to open a remote AFRINIC room. That is access to a channel. It is not proof of attention, speaking opportunity or simultaneous-session capacity. A hybrid link does not add another hour to the day, move a maintenance window or make a policy author available during two live discussions.

This is not an argument that remote participation is inferior. For some contributors it is the only practical route and can be fully effective when the programme, moderation and recording design support it. The narrower point is that “hybrid” describes a delivery mode. It cannot by itself settle the scheduling tradeoff the institution made.

A registration total would answer the wrong question

After the meeting, AFRINIC may be able to report a large number of registrations. That would be useful operational data. It would not test the original rationale on its own.

Registration does not show who attended a policy session, remained connected, spoke, presented a proposal or participated in a consensus call. Counting browser connections can double-count people across devices and reconnects. Counting the whole week can obscure a one-day PPM. Counting organisations can hide whether the same small set supplied most of the interventions. Counting countries can hide whether the operators and proposers directly affected by a policy were present.

Nor should representation be reverse-engineered from a good turnout. An open meeting does not acquire a mandate over absent resource holders merely because many people entered the room. Participation supplies evidence, objections and technical discipline. It does not turn attendees into principals for everyone who could not attend.

The appropriate test is closer to the promise AFRINIC actually made. Did the co-location choice bring the participant groups it expected? Which groups encountered a competing live programme? Were proposal authors and directly affected operators able to take part at the decision-bearing moments? Did hybrid design recover some conflicts, and which remained? Those questions require modest, privacy-safe categories rather than a public attendee dossier.

Publish the calendar trade before the attendance story

A calendar-tradeoff receipt should begin before the programme is final. It should name the event, announcement time, decision owner and policy purpose. If the meeting is expected to carry proposal presentations, consensus calls or a Last Call trigger, those decision-bearing stages should be identified.

The next fields should state the intended participant groups. Resource Members, policy authors, operators, remote contributors, staff and people already attending co-located events may overlap, but they are not interchangeable. The receipt should say which group each scheduling assumption concerns.

Then comes the option set: candidate weeks and locations, minimum operational constraints, public calendars checked and the date of each check. An alternative can be rejected for cost, venue capacity, staff availability, notice time or collision with another event. A short reason is enough. Publishing the reason matters because “no alternative” is otherwise impossible to distinguish from “no preferable alternative under the constraints we selected.”

The co-location hypothesis should be equally concrete. Which event was expected to reduce marginal travel? What public or internal evidence supported that expectation? Which participant groups were at risk of a simultaneous session elsewhere? No names, fare records or travel plans need be exposed. A forecast can be expressed as privacy-safe groups and ranges, including an explicit unknown state.

When the programme is published, the receipt should add Cape Town time and UTC for proposal presentations, chair actions, consensus calls and breaks. It should record remote speaking, queue and recording capabilities, whether repeat sessions exist, and what happens if an author cannot present live.

Finally, the record should return after the meeting with unique participation measures. That means de-duplicated attendance rather than registrations, policy-session participation rather than festival-wide footfall, speaking and proposal coverage, and regional or affiliation breadth at a safe level of aggregation. The result should be compared with the stated objective, not with a target invented after the fact.

The absent ZAPF date is also a boundary

AFRINIC staff's same-week explanation names the South African Peering Forum. At the research cutoff, the forum's public site did not supply a usable date or populated programme; the frozen request returned an error. It is therefore proper to attribute the co-location claim to Roberts, but not to state a ZAPF date as independently verified.

That small limit demonstrates the value of a versioned receipt. Event calendars change. Pages fail. Provisional programmes become final. A planning decision may reasonably rely on information available to staff that is not yet public. The answer is not to pretend the information did not exist or to treat the missing page as proof against it. The answer is to preserve what was relied on, when it was checked and how it was later corrected.

The same discipline applies to the controversy. The mailing-list thread contains opposed views about travel cost, visas, safety, time zones, meeting frequency and the practices of other organisations. Those claims describe stakeholder experience and concern. They are not findings about motive, legality or the eventual attendance of AFRINIC-38.

A defensible choice should remain falsifiable

AFRINIC's decision can have a strong benign reading. It needed a second 2026 Public Policy Meeting, saw a rare concentration of African network and technology participants in Cape Town, and used a one-day hybrid format to reduce the additional travel burden. Nothing in the frozen record proves bad faith, deliberate exclusion or a defective process.

The very plausibility of that account makes a receipt valuable. If the strategy works, the record will show which participants were reached and allow the institution to reuse the method. If it works only for some groups, the next meeting can preserve the benefit while changing the programme clock. If a conflict proves immaterial, future critics will have evidence rather than suspicion. If important proposal authors or operators were missing at decision-bearing moments, AFRINIC can correct the design without having to concede that choosing Cape Town was irrational.

An institution should not be asked to avoid every overlap. In a crowded global calendar, that may be impossible. It should be asked to expose the trade it actually made. The announcement gives AFRINIC-38 a date. The public record should now give that date a participant model, an option history and an outcome that can be checked.

Sources