Summary

  • Resolution 202101.598 adopted new Election Processes and Guidelines “as amended”, but the surviving resolution register does not attach the exact file, hash, signed text, redline, roll call or reasons needed to reproduce the Board’s 20 January decision.
  • The published process imposed genuine ex ante controls across nomination, candidate screening, member standing, proxies, physical and virtual voting, trustees, counting, disputes and announcement, while concentrating several appointments and operational gates in the Board, CEO and NomCom.
  • The proportionate remedy is neither to dismiss the rules nor to inflate AFRINIC’s authority: it is to preserve a versioned election-control receipt that links private corporate authority, the adopted instrument, each administrative gate and the final corporate handoff.

L3 — The amended rulebook behind one sentence

A decision shorter than the system it authorised

Resolution 202101.598 is almost austere. In the January group of Board acts, the register identifies a meeting held on 20 January 2021 and records that the Board reviewed AFRINIC’s Election Processes and Guidelines. It then says the Board resolved to adopt the new Election Processes and Guidelines “as amended”. That is the numbered act. It is also the point at which a reader encounters the central documentary problem: “as amended” tells us that change mattered, but the surviving public register does not carry the amended instrument beside the resolution.

The absence is consequential without being scandalous. A resolution can be authentic as a corporate decision while its public evidentiary trail remains incomplete. The sealed public record does not supply the reviewed file name, a version identifier, a cryptographic hash, a signed copy, a redline, the amendment reasons, a roll call, abstentions or conflict declarations. It does not establish that none of those materials existed privately. A Board pack, legal advice, an internal copy or other records may have existed.

The defensible claim is smaller: a reader using the surviving record cannot reproduce exactly what words the Board reviewed and adopted on that date.

The available public process page is detailed, but timing prevents it from simply filling that gap. The archived page identifies 10 March 2021 as its last-modified time. The preserved capture dates from 17 April 2021. The resolution came on 20 January. There is therefore a public text that represents AFRINIC’s published mechanism for the cycle, but not proof that every byte of that March text was present in the Board’s January instrument. Some or all of it may have been. The record cannot distinguish the possibilities.

That distinction is easy to lose because the two live AFRINIC addresses now return 404. The resolution register survives in an archive capture dated 6 December 2022, and the process page in the April capture. The archive carries official AFRINIC pages after the live locations stopped serving them; it is not another institution corroborating AFRINIC’s account. Counting the original address and its preserved copy as two witnesses would manufacture confidence. One carried page proves one represented record, subject to the limits of that record.

The best reading therefore keeps four states apart. There may have been proposed text. The Board says it reviewed text. Resolution 202101.598 says it adopted a new text as amended. A later-visible public page presents a detailed process. None of those states should be silently substituted for another. This is not pedantry. In a system whose rules decide who may be nominated, who may vote, who holds a ballot and who announces a result, a version difference can change the practical distribution of control even if the broad architecture looks the same.

The strongest case for publishing rules in advance

The version weakness should not obscure what the published process did well. AFRINIC was a functioning private corporation organising selection for its own Board and Governance Committee. It had a legitimate reason to tell members, candidates, volunteers and staff how the cycle was meant to work before nominations and votes. Rules fixed in advance reduce the space for surprise. They let participants distinguish a disagreement with the design from an improvised decision taken after a result becomes foreseeable.

The page anchored the process in private corporate authority. It said the Board approved it under Bylaws article 13.2, applied it to Board seats and, mutatis mutandis, to elections following the same process, and allowed additional Board guidance to NomCom under article 9.2. These references did not transform the rules into public law. They did something more modest and appropriate: they located an internal selection mechanism within the corporation’s own constitutional framework.

The published text also recognised that the practical environment had changed. It expressly covered voting at either a face-to-face or a virtual Annual General Members’ Meeting. Its stated changes from the 2018 and 2020 arrangements included covering physical and virtual meetings, renaming the NomCom section, clarifying that the Board representative had a vote, using the main announcement list, consolidating voting procedures and adding responsibility to the timeline. The pandemic-era need for both meeting modes was not treated as a reason to abandon procedural definition.

The rulebook separated nomination from candidacy. A member’s act of putting a name forward did not itself place that person on the final ballot. The Board approved nomination inputs early under Bylaws section 12.14(i). NomCom prescribed qualifications, called for candidates, considered the resulting nominations and selected the slate. The Board could offer non-binding advice on a draft, while NomCom retained final responsibility for the candidate list. That sequence matters because a nomination is an input, screening is a gate and finalisation is an institutional decision.

Treating all three as “the candidate list” would erase where judgement entered.

It also specified the nomination channel. A member in good standing could make one nomination for each open position through an authorised contact. Self-nominations, and nominations made by people who were not such contacts, required support from authorised registered contacts of at least two members in good standing. The rule was not simply “members nominate”. It connected each type of submission to an identifiable source of corporate standing and support.

The process gave content to good standing for election purposes. A member had to have completed the relevant formalities, remain current and have paid membership and resource fees already due; invoices not yet due were excluded. That definition did not answer every possible eligibility question, but it reduced one area of ambiguity. A person assessing eligibility could at least ask whether the same stated test had been applied to the same temporal snapshot.

Candidate presentation received its own controls. With consent, candidate information was to be displayed in a public comment system. Comments could continue until the day before the election, while inappropriate material could be removed. Candidate withdrawal was also time-sensitive: withdrawal at least seven days before electronic voting could remove the name, whereas a later withdrawal left the candidate on the ballot and caused votes for that person to be treated as nil and ignored. Those provisions recognised that the information surface and the ballot surface had different operational deadlines.

The page distinguished ordinary representation from proxy authority. For Board and Governance Committee elections, it identified Resource Members in good standing and all Registered Members as eligible classes. A corporate Resource Member could be represented by a registered contact without resorting to a proxy. That is a legally and operationally important separation. A registered contact acting for the organisation is not necessarily borrowing another voter’s authority, and a proxy instrument is not merely a convenient label for any representative.

Face-to-face and virtual procedures then handled proxy mechanics differently while retaining common limits. In the face-to-face route, electronic voting could operate alongside stamped paper ballots, and proxies applied to paper voting. The instrument had to carry specified evidence of authority. No person could hold more than five proxies, and a Board candidate could not act as one. Under the virtual route, an eligible member could appoint a proxy by the stated five-day deadline. Issuing that proxy removed the member’s own ability to vote electronically.

The five-proxy ceiling remained, as did the prohibition on a Board candidate acting as proxy.

Ballot custody and counting were not left as an undifferentiated administrative task. The named electronic trustees were AFRINIC’s CEO, the NomCom chair and AFRINIC Legal Counsel. The face-to-face section described trustee signatures on the results; the virtual section said the trustees confirmed them. For paper ballots, a NomCom representative, Election Committee members and as many as four volunteers drawn from Resource Members or other regional Internet registries were to perform the count. Where both channels operated, electronic and paper totals were added. A tie produced repeat voting, and the highest total won.

Disputes, announcement and publication were similarly identified as separate stages. The process assigned election disputes to Bylaws article 10.2. The NomCom chair was to announce the result, including electronic and paper totals for each candidate. The CEO was then to email and publish the final result within forty-eight hours. A spoken announcement, an email and public publication may carry the same numbers, but they are different receipts at different moments. The rulebook’s decision to name them separately was sound.

Finally, the timetable did more than list dates. It allocated responsibilities among the Board, NomCom, Election Committee and AFRINIC staff from nomination preparation to a scheduled Board and Governance Committee election on 4 June 2021. It also recorded a special two-year term for Seat 6 after Robert Ford’s resignation, intended to preserve the sequential director-election cycle. These details show a mechanism adjusted to the circumstances of that particular cycle rather than a purely abstract set of principles.

None of this proves the timetable was executed. It proves that the published design contemplated the stages. That is still valuable. A clear design gives later observers something against which to compare appointments, eligibility decisions, ballot custody, counts and publication. Ex ante specification cannot substitute for execution evidence, but without it there is no stable baseline at all.

A minimum specification, not an expanding constitution

Heng Lu’s minimum initial specification approach helps explain why the benign case is strong. An institution should fix the smallest set of rules necessary to let participants coordinate, leaving later operational decisions bounded rather than pretending to foresee everything. In an election context, the minimum is not necessarily short. It can include nomination authority, candidate screening, voter eligibility, proxy rules, ballot channels, custody, counting, disputes and publication because each element prevents a foreseeable collision.

The test is functional. Does a provision tell the private corporation and its participants how to move from one institutional state to the next without arbitrary surprise? The definition of good standing does. The separation of a registered contact from a proxy does. The cap of five proxies does. Naming the trustees does. The late-withdrawal rule does. The forty-eight-hour publication obligation does. Each supplies enough advance information for someone to know what should happen at a gate.

The approach also warns against confusing a coordination artefact with reality. A timetable is not an event. A list of trustees is not a trustee ceremony. A proxy deadline is not a validated proxy register. A counting rule is not a tally certificate. A dispute clause is not a dispute log. The more detailed the text becomes, the easier it is to feel that the process has been proved merely because it has been described. In fact, specification and evidence of performance are complementary, not interchangeable.

BTW’s analysis of the January 2020 rulebook supplies an immediate comparator for this point. It shows that the 2021 page belongs to a continuing practice of pre-election specification rather than appearing without institutional ancestry. Its separate treatment of bylaw versions sharpens the other half of the lesson: constitutional authority should be connected to an identifiable text and adoption chain. Those comparisons illuminate what to inspect in the 2021 material. They do not independently establish what the January Board saw or what election administrators later did.

The resolution and the process page therefore deserve neither automatic reverence nor cynical dismissal. Together they show a serving Board performing an ordinary corporate task and a detailed public mechanism addressing genuine operational problems. Between them lies a missing reproducible link. Good governance analysis can hold all three propositions at once.