Summary

  • Version 2 replaced three seats reserved for peer-RIR or NRO-related officeholders with three seats drawn from AFRINIC’s participating community, retained two former-chair seats, and gave the Board a choice among eligible former-chair candidates rather than relying on Version 1’s random ordering.
  • It also made the appeal more intelligible: a three-quarters Board threshold constrained discretionary replacement; conflicts became case-specific; the review record, grounds, supporter requirements, remedies and duty to publish reasons became more explicit.
  • Those improvements did not make participation equivalent to authorization or turn the committee into a court. A durable design would keep local technical knowledge while publishing selection receipts, protecting terms, separating conflict substitutions, supporting an autonomous docket and confining remedies to AFRINIC’s private policy process.

L3 — Five seats were redrawn while the fourteen-day clock kept running

The smallest useful unit of analysis is the change between two charters. Version 1 was dated 18 July 2017. Version 2 was published with the date 26 November 2017, the day on which the Board unanimously approved the update. The Board resolution, however, described the updated Terms before it as dated 2 November. That difference should be left visible rather than tidied away. The available record establishes an updated text dated 2 November in the resolution and a published Version 2 dated 26 November; it does not explain the intervening document history.

Precision about those dates is part of the accountability problem, because a review body is credible only when readers can identify the instrument governing it without having to infer missing steps.

The resolution also recorded an earlier public-comment stage. It recited that draft Terms had been published for comment following an August 2017 resolution and said that the Board had considered the comments. That is evidence of notice and consideration at a general level. It is not evidence of who commented, what each person proposed, which objection prevailed or how a particular clause changed. A responsible account therefore credits the comment process without inventing a legislative history.

The same discipline applies to the vote: unanimous Board approval proves adoption of the revised charter, not agreement among every affected network operator, policy participant or resource holder.

Nor did approval mean that a standing committee sprang into operation that day. Resolution 201711.378 said the first appointment of the Standing Committee would occur during the 2018 Annual General Members’ Meeting. It contemplated a possible interim committee if one became necessary and said the Board would appoint it as close as possible to the Terms. The record does not establish that such an interim body was in fact appointed. The institutional act on 26 November was approval and publication of a design.

Conflating the charter with a seated body would erase the difference between assigning powers on paper and putting named reviewers in a position to use them.

Version 1’s five-seat arrangement provides the baseline. Two places belonged to people who had served as AFRINIC Policy Development Working Group chairs or co-chairs. Two more belonged to chairs, co-chairs or equivalent officeholders from policy groups at other Regional Internet Registries. The fifth was reserved for a chair or vice-chair of the Number Resource Organization Number Council or Address Supporting Organization Address Council from outside the AFRINIC service region. The design mixed experience from AFRINIC’s own procedure with office-based experience elsewhere in the number-resource coordination system.

That outward-facing architecture did not guarantee independence. A title in another registry could indicate familiarity with comparable processes, but it could also indicate membership in a close professional network with shared habits and assumptions. The fifth seat’s outside-region condition created distance, not an independent mandate. Still, Version 1 made institutional distance legible: three seats were categorically tied to offices outside AFRINIC, and the two internal former-chair categories sat beside them. The categories themselves disclosed where reviewers were expected to acquire their experience.

Version 1 also used random permutation after applying exclusions to order candidates in the relevant pools. Randomness is a modest device, not a moral quality. It cannot make an ineligible person eligible, cure a narrow candidate pool or prove that the selected reviewer understands the dispute. What it can do is reduce discretionary choice among already eligible candidates. When an appointing body cannot simply prefer one eligible name over another, the selection rule removes one opportunity for patronage, strategic anticipation or suspicion.

That benefit exists even if nobody ever acts improperly: an institution should be designed so that confidence does not depend on private assurances about individual virtue.

Version 2 retained the two former-AFRINIC-chair seats but altered both eligibility and selection. Its initial pool required qualifying service at at least two Public Policy Meetings during the preceding five years, and the Board would choose among eligible former chairs. The change combined an experience threshold with discretion. Requiring repeated service could filter out someone with only fleeting familiarity. Yet Board choice, unlike random ordering, made the appointing institution’s preference part of the result.

The text also allowed the Board to review candidates appointed under the former-chair provision and, where applicable, replace them with candidates satisfying that criterion. The significance is structural, not accusatory: the revised rule made the Board’s judgment more consequential at the point where the appeal body acquired its experienced members.

The larger composition change concerned seats three, four and five. Version 2 removed the categories for peer-RIR policy officers and the outside-region NRO NC or ASO AC officeholder. In their place it called for three people selected within the AFRINIC community through a route described as similar to the selection of AFRINIC community representatives to the NRO NC. Candidates were to be PDWG members, demonstrate participation on the Resource Policy Discussion mailing list or at an AFRINIC Public Policy Meeting, and have qualifying participation earlier than six months before the nomination call.

The last requirement is awkwardly worded in the instrument. It should be reported as written rather than converted into a more convenient rule.

This shift brought the majority of the committee into an AFRINIC-centred participant pool. It was not merely a geographic substitution. It changed the evidence of qualification from holding defined offices in peer institutions to having participated in AFRINIC’s own policy field, and it changed the implied source of legitimacy from external procedural experience to local involvement through a community-selection analogue. The description of that route does not reveal the actual nomination process, electorate, tally or decision sequence. One cannot infer who voted, how choices were counted or whether a particular selection ever occurred.

But the textual direction is clear: Version 2 preferred a majority with demonstrated exposure to AFRINIC discussion over a majority drawn from named offices beyond it.

Both versions kept ordinary membership exclusions for AFRINIC employees, directors and current PDWG chairs or co-chairs. Those exclusions matter because they prevent the most direct forms of role overlap. A sitting chair should not review the chair function; a director should not sit as an ordinary reviewer of a process administered under Board authority; an employee should not have to reconcile review obligations with organisational reporting lines. But categorical exclusions are only the first layer of independence. They say who may not sit routinely.

They do not answer who selects the remaining candidates, who supplies the committee’s working support, who fills vacancies or what happens when a permitted member has a conflict in a particular case.

On proposal authorship, Version 2 moved from a blunt rule to a tailored one. Version 1 made an author or co-author of the policy under appeal ineligible. Version 2 allowed that person to remain a committee member generally but required recusal from deliberation and abstention from all votes concerning the relevant appeal. If the committee or the Board believed a replacement necessary, the Board would appoint a temporary substitute. The revision preserved expertise between cases rather than treating one act of authorship as a permanent institutional disability.

That is a sensible administrative improvement, provided its two decisions are kept separate. The first decision is whether a conflict requires temporary substitution beyond recusal and abstention; Version 2 allowed the committee or Board to form that view. The second is who appoints the substitute; that remained with the Board. Recusal protects the immediate decision from a conflicted vote. It does not by itself protect the composition of the replacement panel from appointing influence.

The rule was more proportionate than blanket ineligibility, but its substitute mechanism still passed through the same institution holding the committee’s broader appointment authority.

Replacement of the committee as a whole, or of any member, shows a similar pattern of improvement without separation. Version 1 gave the Board a right to replace any or all members at any time and stated no vote threshold in the instrument. Version 2 retained the power but required at least three-quarters of the Board. That was a genuine constraint. It raised the cost of casual or factional replacement and made a narrow Board majority insufficient. Any assessment that treats the threshold as cosmetic misses a real alteration to the decision rule.

Yet a supermajority is not the same as tenure protection. Version 2 did not state a cause standard for discretionary replacement or require the Board to publish reasons for using the power. It retained Board interfaces across initial appointment, discretionary selection among former-chair candidates, insufficient-candidate filling, vacancies, temporary conflict substitutions and extension requests. Both versions contemplated replacement of the whole committee at least every three years or more often, with a qualified delay while an appeal was under way.

The three-quarters requirement constrained one lever, but did not detach the review body from the institution whose policy process it reviewed.

Version 2’s most substantial gains lay in what the committee was asked to examine and explain. The revised Terms required a report assessing the facts placed before the committee and ruling on the appellant’s grounds. They identified an existing record: the Resource Policy Discussion mailing-list material and the meeting video or transcript. They asked whether the Policy Development Process had been followed and whether, on that record, the chairs’ judgment of consensus or lack of consensus was correct. This was not an invitation to design the committee’s preferred policy from scratch.

It was a second look at procedural compliance and the declared state of consensus using material already generated by the policy process.

That distinction guards against two opposite errors. The first is excessive deference, under which reviewers merely ask whether chairs used the vocabulary of consensus. The second is substitution of policy taste, under which five reviewers decide what outcome they personally favour. Version 2 pointed towards an intermediate discipline: test the announced conclusion against the governing process and the recorded discussion. The committee’s own judgment still mattered, especially when deciding whether the record supported consensus. But a defined evidentiary base made that judgment contestable.

Readers could compare reasons with messages, video or transcript instead of being told simply to trust institutional intuition.

The available remedies also became clearer and more graduated. Version 2 allowed the committee to remand the chairs’ decision for reconsideration, reverse it or uphold it. Remand recognises that an error may be curable by returning the question to the original decision-makers with guidance. Reversal recognises that the record can sometimes support a definite conclusion against the chairs’ declaration. Affirmance confirms that a challenge can receive genuine review without succeeding.

A menu containing all three gives the committee a way to match response to defect, rather than treating every established problem as requiring the same institutional consequence.

Published rationale completed that architecture. Version 2 required findings reports to include reasons for decisions made. Reasons make an appeal more than a change of voice. They show whether reviewers addressed the appellant’s grounds, how they understood the record and why they selected remand, reversal or affirmance. They can also discipline the reviewers before publication: a conclusion that cannot be defended in a coherent account may need to be reconsidered. The duty did not guarantee excellent reasoning, and the record does not show how it operated in any particular case.

It nevertheless supplied a test that outsiders could apply to the committee’s work.

The filing side was correspondingly concrete. An appellant had to hold a good-faith belief that the chairs had erred, discuss the matter with the chairs and at least three other PDWG members, and file within fourteen calendar days. The three supporters had to be distinct from the complainant and send separate emails. The filing needed to identify the challenged announcement and provide evidence that prior discussion had failed. If a chair did not respond to or conclude those discussions within one week, that non-response was enough to satisfy the discussion conditions.

These rules gave an appellant a checklist and prevented silence from consuming the entire route. They also imposed labour at speed. Fourteen calendar days is short when a complainant must understand an announcement, organise three distinct supporters, secure separate messages, document attempted discussion and frame grounds capable of record-based review. The one-week rule limited the ability of an unresponsive chair to defeat filing through delay, but it also left little spare time.

Composition and access cannot be separated: the fairest panel is useless if a worthy appeal cannot reach it, while an accessible filing route is weak if the reviewers’ selection cannot command confidence.

The verified delta is therefore mixed but coherent. Version 2 moved the majority of seats inward, introduced Board preference among eligible former chairs, converted author conflict from blanket exclusion to case-specific recusal, raised the vote needed for discretionary replacement, specified a review record, expanded and clarified remedies, demanded reasons and detailed the path into review. It did not establish how community selection worked in practice, identify a first panel, show use of substitution or replacement, or prove that any control altered an outcome. Those absences are not footnotes to be filled by later events.

They define the boundary between what the 2017 texts can teach and what they cannot.