Summary

  • AFRINIC’s 2022 notice and virtual-election rule used a five-day proxy deadline, its timeline called 28 May “five clear days” before Election Day, a separate face-to-face rule used 24 hours, and the bylaws required proxy authority to be produced before the meeting. The public record does not identify which path governed any particular member.
  • Advance authentication has a strong legitimate purpose: it can verify corporate authority, enforce the five-proxy ceiling and candidate prohibition, prevent duplicate voting and establish a determinate electorate. Those purposes make precise receipts, reasons, cure, review and reversible access states more important, not less.
  • The disclosed material proves the architecture of the rule, not a private case. It does not establish a rejection, an excluded voter, a changed margin, an invalid election or a network consequence.
  • A privacy-safe proxy ledger could let members reconcile submissions, decisions, corrections and voting-access transitions without revealing identities or ballot choices.

A deadline with a long shadow

The calendar entry was spare. Electronic voting and proxy registration opened on 19 May 2022. Proxy registration and appointment closed on 28 May, described in the official election process as five clear days before the Board and Governance Committee Election Day. The Annual General Members’ Meeting was scheduled for Friday 3 June, beginning at 10:00 UTC during AFRINIC-35 in Mauritius. The agenda said the online election platform would close at 11:00 UTC.

Read only as dates, the sequence seems unremarkable. Read as an access system, it is much more consequential. A member entitled to vote could choose among personal participation, electronic voting and representation by proxy, subject to the applicable rules. But the abstract right to appoint a proxy did not itself put anyone inside the electorate. Someone had to obtain the right instrument, establish authority, complete and sign it, send it through the accepted channel, have it received, pass validation, become an effective appointment and obtain ballot access. Each verb marks a different state.

A deadline can attach to any one of them, and attaching it to the wrong one without saying so can transform a manageable administrative rule into an invisible exclusion point.

That distinction matters because the vote and the proxy appointment did not close together. The proxy route shut earlier. Once that route closed, electronic voting remained open until Election Day. Yet the official process also said that issuing a proxy removed the appointer’s ability to cast an electronic vote. There was therefore a potential transition between two mutually exclusive voting paths: the principal’s electronic access and the proxy’s delegated access. A well-built system makes the handover atomic.

A poorly explained system can produce a dead zone in which the principal has lost one route while the representative has not securely gained the other.

No public evidence supplied for this article shows that such a dead zone occurred. No disclosed member record shows a form, receipt, rejection, correction or ballot. That is an essential limit. The point is not to turn a systems risk into an allegation. It is to identify what a rule capable of changing access must record if members are to distinguish ordinary administration from arbitrary exclusion.

Four formulations, no licence to choose among them

The first difficulty is public and documentary. AFRINIC’s official pages did not present the relevant timing rule in one uniform formulation.

The dedicated 2022 AGMM notice said a member entitled to attend and vote could appoint a proxy, whether or not that person was an AFRINIC member, to attend and participate on the member’s behalf. It directed the member to a proxy instrument accessible from the member’s MyAFRINIC profile. To be effective, the notice said, the instrument had to be completed and submitted no later than five days before the election date.

The virtual-meeting proxy section in the official 2022 election process used the same basic architecture. An eligible member could appoint a proxy for the Board and Governance Committee elections, and the MyAFRINIC form had to be completed and submitted no later than five days before the election date. Form access opened when electronic voting opened. The timeline on that page, however, described the 28 May close as five clear days before Election Day.

Elsewhere on the same official election-process surface, the face-to-face proxy path used a different procedure. It said a proxy had to be duly approved by the chief executive at least 24 hours before Election Day and transmitted to the chair of the Elections Committee. It also described a downloaded and signed form emailed to the AFRINIC CEO no later than 24 hours before the election date.

The 2020 bylaws added another boundary. They allowed a member entitled to vote to exercise that right through e-voting, presence or proxy, required a signed written proxy appointment, and said a proxy was not effective unless the appointment notice and any supporting authority were produced before the meeting began.

Those formulations may have had a coherent internal explanation. Meeting mode may have selected the path. One text may have governed submission while another governed approval or ultimate effectiveness. “Five clear days” may have been intended to specify how days were counted. The bylaws may have established an outer corporate requirement while the election process imposed an earlier operational cutoff. But the public record sealed for this analysis does not authorise a writer to settle those relationships.

It also contains a meeting description that allowed physical or virtual attendance while another official passage described the pandemic-era meeting as online. The honest conclusion is that the version and mode boundary is visible, while its member-specific application is not.

This is not pedantry about drafting. A person standing at a deadline needs to know which act stops the clock. Is a form timely when the appointer presses submit, when AFRINIC’s system records receipt, when an administrator validates the signature, when the CEO approves it, or when the appointment becomes effective? Which timezone governs? Does “five days” mean five calendar periods counted backwards from the election date, or five clear days excluding one or both endpoints? If a member used the path presented inside MyAFRINIC, could a separate emailed form still matter?

If the meeting was hybrid in presentation but virtual for voting, which proxy procedure applied?

The published pages identify parts of this machinery but do not answer all of those questions. Nor do they reveal which undisclosed workflow controlled a particular submission. The divergence should therefore be preserved, not smoothed away by paraphrase. A clean retrospective sentence saying “the deadline was 28 May” captures the timeline but conceals the rule-selection problem that an actual appointer would have faced.

The strongest case for closing early

Before criticising the design, it is worth stating the administrative defence in its best form. A proxy is delegated corporate authority, not an informal request to pass along somebody else’s preference. The organisation running the election must be able to determine whether the appointer is entitled to vote, whether the signer can bind the member, whether the representative is eligible to serve, and whether the same authority has been asserted twice.

AFRINIC’s published constraints made those checks real rather than theoretical. No member entitled to vote could carry more than five proxies. A candidate standing for election as a director could not act as a proxy. The process distinguished a proxy from a registered contact representing a corporate Resource Member; a registered contact in that capacity did not need a proxy. It also distinguished the five-proxy ceiling from organisations for which the voter was already a registered contact. Those categories cannot be enforced reliably by glancing at a name at the last moment.

The electronic platform added another legitimate concern. It was available during the election period to members in good standing and could be used in person or by proxy. Issuing a proxy removed the appointer’s ability to cast an electronic vote. That rule aimed at an obvious integrity requirement: one underlying voting entitlement should not become two votes merely because it moved between a principal and a representative. Administrators needed a determinate list, and they needed time to test it before a finite election closed.

A fixed advance cutoff can serve all of those purposes. It can reduce last-minute confusion, give administrators time to verify corporate signing authority, identify prohibited candidates, enforce the carrier ceiling, prevent duplicate appointments and prepare the electronic platform. It may also reduce opportunities for rushed pressure or strategic reassignment immediately before the ballot. Private associations routinely set filing deadlines because human review and secure system changes take time.

That defence has force. It is stronger than a presumption that every deadline is suspect. It also tells us what evidence good administration should generate. If authentication is the reason for the lead time, the process should show exactly when authority entered the system, which rule was applied, what was checked, what failed, whether the failure could be corrected and who reviewed a contested decision. The more legitimate the operational need, the easier it should be to describe the states that satisfy it.

A deadline is therefore not the opposite of due process within a private association. It is one component of due process. The problem begins when the date is public but the state transition is hidden: when a member can see a closing day but cannot tell whether timeliness attaches to sending, receipt, validation or approval; when an error message does not distinguish a correctable signature defect from an untimely filing; or when electronic access disappears before delegated access is confirmed. These are design failures, not arguments against advance administration as such.

The cast of the transaction

Proxy systems become confusing when several actors are compressed into “the voter.” The 2022 materials require more discipline.

The member is the corporate participant that holds the relevant entitlement under the association’s rules. A registered contact is a person already recognised in connection with a corporate Resource Member. The official process said that such a contact could represent that organisation without a proxy. That existing representational status is not the same thing as a newly delegated proxy appointment.

The appointer is the member exercising the power to delegate. The signer is the natural person whose signature purports to express that member’s authority. They may be closely connected, but analytically they are not identical. The appointer can be eligible while the signer’s authority still requires evidence. Conversely, a technically valid signature cannot create an entitlement that the appointer does not hold.

The proxy is the appointment or delegated authority. The proxy carrier is the person who acts under it. Treating those terms as interchangeable obscures two different validation questions: whether the instrument is effective and whether the named representative may carry it. A representative’s existing role as a registered contact for other organisations also must not be miscounted as a proxy merely because the same human appears in both contexts.

A candidate is a person standing in the election and, under the published rules, could not serve as a proxy. That prohibition calls for a status check against the candidate list. The Elections Committee, or ECom, is part of the election-administration structure identified in the face-to-face path, whose text called for transmission to its chair. The CEO appeared in that path as the recipient of an emailed signed form and as an approving actor. Those descriptions do not justify collapsing ECom and CEO into a single decision-maker.

NomCom, the Nomination Committee, is another governance actor associated with the electoral setting, but it should not be casually substituted for ECom or the CEO. The sealed public facts used here do not assign NomCom the member-specific proxy receipt, approval or review decisions at issue. Naming it in an organisational chart would not prove that it touched an appointment.

MyAFRINIC is the member-facing account path from which the notice said the proxy instrument could be accessed. It is a channel and control surface, not an actor with judgment. The electronic-voting platform is the ballot-access system, not the proxy form itself. Opening a form in MyAFRINIC, completing it there and later obtaining a ballot route on the election platform are separate events even if they feel like one experience to a member.

Keeping that cast separate prevents two common mistakes. The first is to treat corporate eligibility as proof that every human action taken in its name was authorised. The second is to treat a procedural defect in one representative path as proof that the member lacked the underlying right. An authentication system exists precisely because entitlement, human authority, appointment and access are related but distinct.

Twelve states between permission and participation

The mechanism can be understood as a chain. Each link answers a different question, and none should be inferred merely because a later or earlier link exists.

Form access: Could the eligible member reach the designated instrument? The virtual-election rule said access opened when electronic voting opened, and the notice located the form in the member’s MyAFRINIC profile. Access depends on account status, contact status and platform availability, but the public record contains no member-level access event.

Completion: Were all required fields populated in the form presented by the controlling process? Completion is a content state. A blank or partial instrument can be accessed and saved without being complete.

Signature: Did a natural person execute the written instrument? The bylaws required a signed written appointment. A signature’s presence and the signer’s authority remain separate questions.

Submission: Did the user perform the act designated to send the instrument through the required channel? The notice and virtual rule spoke of completing and submitting through the MyAFRINIC path; the face-to-face passage described email to the CEO. Submission is normally the state visible to the sender.

Receipt: Did the organisation’s system or designated recipient actually receive the instrument, and at what time? Receipt is not always identical to submission. Network delay, email routing, portal failure or asynchronous processing can separate them. No such failure is proved here, but a robust rule states which side of that boundary bears the risk.

Validation: Did the instrument satisfy the applicable checks? Validation can include member eligibility, signer authority, completeness, meeting mode, channel, candidate status, the five-proxy ceiling and duplication. It should produce a reasoned state, not merely silence.

Approval: Did the actor designated by the governing path approve the proxy? The face-to-face language expressly referred to CEO approval. It is unsafe to read that approval requirement into every other path without evidence, just as it is unsafe to ignore it where that path controlled.

Effective appointment: Did the legal and procedural conditions converge so that the proxy could act? The bylaws tied effectiveness to production of the appointment notice and supporting authority before the meeting. An uploaded form may therefore be submitted but not yet effective.

Cure: If a timely instrument had a correctable defect, could the appointer supply missing evidence or correct an error while the administrative purpose of the cutoff could still be protected? The official public surfaces in the sealed record do not disclose a cure rule. That absence is not proof that cure was denied; it is a question the public rule should answer.

Review: Could a member obtain rapid reconsideration of a disputed state before voting became impossible? Review is distinct from cure. Cure changes the material; review tests whether the original material or decision was handled correctly.

Ballot access: Once the appointment became effective, did the proxy receive the correct ability to participate, while the appointer’s incompatible electronic-vote path changed state? Access is an entitlement presented by the platform. It is not proof that a vote was cast.

Ballot cast: Did the authorised participant actually complete the vote? A valid proxy may choose not to vote or fail to finish. The public rule pages prove no individual ballot and disclose no link between any appointment and a ballot choice.

These states form neither an accusation nor a hypothetical biography. There is no basis to invent a member who logged in on a particular afternoon, received a particular error or appealed to a particular employee. The value of the state model is institutional: it shows how a right can exist at the first link and still fail to become participation at the last. It also shows where a receipt, reason or review belongs.

The deadline needs an object

Saying “the deadline was five days” leaves the central operational question unfinished. A deadline must terminate a defined action. In a paper office, staff may infer whether the decisive event is delivery over the counter or approval by a clerk. In a digital election, software can give each event a timestamp, making ambiguity both less necessary and more consequential.

Suppose a portal records four moments: the appointer presses submit; the server accepts the payload; an administrator opens the case; and validation finishes. Only the first two are substantially controlled by the appointer. If timeliness depends on validation, a member could submit before the cutoff yet lose because the organisation processed the file after it. If timeliness depends on receipt, the organisation must define receipt and produce an acknowledgement. If it depends on the sender’s submission, the system needs a tamper-resistant record of that action. The policy choice may vary, but it must be made in advance.

The same is true of timezone. The scheduled AGMM and online-platform close were expressed in UTC, but the exact proxy cutoff timestamp and timezone are not disclosed in the sealed public record. A calendar date without a zone leaves a distributed membership to guess whether the day ends in Mauritius, at UTC midnight, at a server location or in each member’s locality. Five clear days can also be counted differently depending on whether the election day and submission day are excluded.

None of this proves anyone missed the cutoff because of ambiguity. It explains why a versioned election rule should contain a machine-readable deadline beside the human explanation: a date, exact time, timezone, counting convention, authoritative channel and decisive event. The public page should also identify the meeting mode that selects the process. If a rule changes, the old version should remain visible with an effective timestamp rather than being silently replaced.

One entitlement, two mutually exclusive paths

The electronic-vote disablement rule deserves special attention because it turns proxy authentication into a state-management problem. Preventing double voting is plainly legitimate. But the safe sequence is not simply “proxy issued, principal disabled.” It must define when issuance has occurred and ensure that the alternate route is live.

An appointer who merely opens a form should retain electronic access. So should an appointer who starts but abandons an incomplete draft. Submission may justify placing the case under review, but it need not mean that the appointment is valid. A validation failure should not strand the entitlement if the principal can still lawfully vote or correct the defect. Approval, effectiveness and ballot enablement must be coordinated.

The safest design resembles a transactional handover. The principal’s electronic access begins active. A submitted appointment enters a pending state without creating two usable ballots. Validation tests the delegated route. Only when the proxy’s access is ready does the system commit the switch that disables the principal. If validation fails, the principal’s path remains active or is promptly restored, subject to the rule. If a review reverses a rejection, the change is appended with timestamps and the handover is completed. If a proxy is revoked within a permitted period, the reverse transition is equally explicit.

The official record does not describe that technical implementation, and this analysis does not claim AFRINIC used or failed to use it. It identifies the evidence that would be needed to audit the published promise that issuing a proxy removed the appointer’s electronic vote. Without state-transition records, observers cannot reconcile the anti-duplication purpose with continuous access to one lawful voting route.

This is where a seemingly modest authentication deadline gains a long shadow. The cutoff is not only a date on the form. It can influence which identity the voting system recognises and when. The right to participate becomes operational only through the coupling of the membership record, the proxy record and the ballot-access record.

Correctable defects are not all the same

Election rules often treat lateness as different from a defect in a timely filing. That distinction is especially useful here. A form received after an unambiguous cutoff may be excluded to preserve finality. A form received on time with a missing attachment, an unclear signer title or a correctable mismatch raises a different question: can the defect be cured without frustrating the reason for the deadline?

A proportionate system classifies defects. Some go to underlying eligibility. Some go to authority. Some are formatting or documentary omissions. Some involve a prohibited proxy or a carrier already at the five-proxy ceiling. Some may be resolved by selecting another representative; others cannot be fixed once the ballot begins. The acknowledgement should identify the category and the factual basis without exposing unnecessary personal information.

Cure need not mean an unlimited extension. It can be a short, rule-based window available only when the original submission was timely and the defect is objectively correctable. The important point is that the deadline’s purpose should govern the response. If administrators need five days to authenticate authority, a member who submits on the first day should receive a defect notice early enough to respond. Holding every status until after the cure has lost practical value defeats the very lead time invoked to justify the cutoff.

Review also must fit the election clock. A right to complain after the election may produce a record but not participation. Rapid review could be assigned to a person or panel independent of the initial validation decision, with access to the rule version and immutable event history. The reviewer should be able to affirm, reverse or remand a decision and trigger the corresponding access state before the platform closes. Independence here is functional, not sovereign: it means a private association checks its own consequential administrative decisions through a process not controlled solely by the original decision-maker.

Again, the public pages do not establish a particular cure request, refusal or review. The absence of disclosed detail defines a watchpoint, not a finding of misconduct. Responsible analysis separates what a safeguard would look like from a claim that a specific person was denied it.

What the public record cannot carry

The official sources prove dates, channels, restrictions and public rule language within their version and meeting-mode boundaries. They do not prove private execution. That boundary should govern every inference.

There is no disclosed member-specific proxy submission. There is no receipt time, signer-authority file, defect code, correction, approval, rejection or review outcome. There is no evidence showing when an appointer’s electronic access changed, whether a proxy received ballot access or whether either person cast a vote. There are no published counts in the sealed record for eligible members, forms accessed, forms submitted, appointments accepted, forms rejected, defects corrected or proxies carried.

The record also does not establish causation. It does not show that the proxy cutoff changed turnout, altered a margin, selected a winner or affected the legal validity of the election. It does not show a network outage, a resource decision, a customer impact or a routing consequence. An article that moved from rule ambiguity to any of those claims would be manufacturing evidence.

Those limits do not make the subject trivial. Governance can be evaluated at the level of system design before a documented victim appears. A public rule can be tested for clarity, version control and reviewability even when private files remain private. Indeed, focusing on architecture rather than an invented anecdote produces a fairer critique. It lets the legitimate administrative objective remain visible and avoids accusing staff of motives the record does not reveal.

It also keeps separate several neighbouring debates. This is not a general article about how election calendars can be used as instruments of control. It is not an argument about one person accumulating large numbers of proxy mandates; the 2022 rule in fact stated a ceiling of five, and the evidence contains no proxy-carrier distribution. It is not the 2025 story about staff contacting voters. It is not a dispute about good standing, a signed Registration Service Agreement or debt age. It is not a full history of the 2022 election.

The narrow question is how a five-day authentication cutoff converted a formal proxy right into a usable route and how that conversion could be audited.

The private authority that AFRINIC actually has

NRS describes AFRINIC as the member-based regional registry serving Africa and explains why operators depend on the registry’s coordination functions. That context matters: an organisation maintaining important technical records and member services needs coherent corporate governance. NRS did not write, apply or audit the 2022 proxy process, however, and its institutional description cannot establish what happened to a form.

Heng Lu’s governing doctrine supplies a different boundary. A registry is a bookkeeper, service provider and coordinator. Its ledger is not a throne. AFRINIC can administer its own private meeting, authenticate a written delegation, limit proxy carrying, prohibit candidates from serving as representatives and choose a reasonable filing period. These are ordinary powers grounded in association rules and contract.

They are not public sovereignty. AFRINIC is not a legislature for a territory, a police force, a prosecutor, a confiscator or a court. Missing an internal proxy deadline may close one route to a corporate ballot if the rule is valid and properly applied. It does not prove civic unworthiness, regional disloyalty or wrongdoing. It cannot lawfully be inflated, as a matter of this doctrine, into a justification for punishment through number-resource administration, routing services or operational interference.

The distinction protects both sides of the analysis. It prevents critics from describing a private deadline as though it were state coercion. It also prevents the registry from borrowing the language of public mandate to insulate an internal decision from scrutiny. The proper question is narrower and more exacting: did the private coordinator administer the authority it actually possessed through a clear and reviewable process?

Heng Lu’s no-private-punishment principle is especially relevant when discussing remedies. A defective proxy appointment should be handled inside the election process. It should not contaminate unrelated registry services or be treated as evidence against an operator. The record here contains no evidence that AFRINIC imposed such an external consequence. The principle defines the limit, not an allegation that the limit was crossed.

Why operators still have reason to care

LARUS argues that predictable registry governance matters to operational continuity. Decisions inside a registry can shape confidence, planning and the environment in which networks obtain essential coordination services. That is a legitimate reason for operators to care about the quality of election administration even if they never file a proxy.

But relevance is not causation. LARUS does not prove that the May 2022 cutoff interrupted a network, changed a customer outcome or affected a number resource. The chain from a proxy rule to an operational event cannot be assumed. A cautious analysis treats predictable governance as part of institutional resilience while refusing to assign a technical consequence that the record does not contain.

That bounded concern points back to records. Operators accustomed to change logs, acknowledgements and reversible configuration states may reasonably expect comparable discipline when an election system changes who can exercise a member entitlement. The analogy has limits—a ballot must remain secret, and corporate authority contains personal information—but the underlying engineering principle survives: consequential transitions should be explicit, attributable, timestamped and auditable.

What the BTW comparisons illuminate—and what they do not

BTW’s analysis of AFRINIC election legitimacy emphasises ex ante clarity, independent review and auditable process. Those principles help evaluate the proxy cutoff because they ask whether the electorate could know the rules before acting and whether consequential decisions could be checked. The comparison does not prove that a particular appointment was mishandled in 2022.

BTW’s election-calendar analysis owns the broader claim that scheduling can become an instrument of control. This commission is narrower. It does not infer a general calendar strategy from one proxy date. The relevant interval is the space between 28 May and 3 June and the authentication work that was supposed to occur there.

The article about a proxy holder carrying fifty mandates examines concentration. It is useful as a warning that delegation rules affect representational power, but it cannot be imported into this record. AFRINIC’s disclosed 2022 rule capped one eligible member at five proxies, distinguished those proxies from registered-contact representation, and published no aggregate carrier distribution in the sealed sources. There is no evidence here of a fifty-mandate holder or any other concentration pattern.

The article about staff calling voters concerns a disputed 2025 authentication context. It may sharpen the general question of who contacts whom and how authority is verified, but it does not prove 2022 staff conduct. Combining the episodes would erase the evidence cutoff and turn analogy into testimony.

Used properly, the BTW sources are analytical comparators. They help identify clarity, review, audit, timing and authentication as governance concerns. Used improperly, they would supply facts absent from the 2022 record. The source boundary is part of the argument: institutional accountability improves when every claim carries a clear provenance and scope.

A receipt is a small constitutional object

For a private association, the most useful safeguard may be surprisingly mundane: a receipt. Not a generic “thank you” page, but a durable record that tells the appointer what the system believes has happened.

The receipt should identify the election and meeting mode, the rule version and its effective time, the authoritative submission channel, the decisive timeliness event, the received timestamp and timezone, and a privacy-preserving case reference. It should state the current status: received, awaiting authority evidence, under validation, accepted, rejected, cure requested, under review, superseded or withdrawn. The status should never be presented as a ballot choice.

If the case is not accepted, the receipt should give a reason code and a concise explanation. A missing signature is different from an ineligible appointer. An unauthorised signer is different from a prohibited candidate. A carrier-ceiling conflict is different from a duplicate appointment. A late submission is different from a timely form awaiting processing. These differences determine whether cure is possible and whether the principal’s electronic-vote access should remain active.

The receipt should also show the access transition without exposing the secret ballot: principal electronic access active, handover pending, proxy access enabled, principal access disabled, or principal access restored. That is operational information about entitlement, not information about whom anyone supported. Every change should append to an immutable history rather than overwriting the prior state.

Finally, the receipt should name a review path and its election-sensitive deadline. A member should know where to challenge a factual or rule-selection error, what material the reviewer can consider, and when a decision will be delivered. If the organisation cannot complete review before the vote closes, the system should make clear whether and how one lawful participation route is preserved. A remedy that arrives only after irreversibility should not be advertised as equivalent to timely access.

Calling such a receipt constitutional does not make AFRINIC a state. It recognises that, inside a private institution, small documents allocate practical power. The receipt tells the participant which rule the organisation has chosen, what evidence it recognises and whether a right has crossed into a usable state. It is the point where an otherwise opaque database becomes an accountable promise.

Aggregate reconciliation without ballot surveillance

Individual receipts solve only half the problem. Members also need to know whether the system reconciled as a whole. That can be done without publishing names, company identities, account identifiers, signatures or votes.

A privacy-safe report could begin with the eligible-member denominator for the relevant election. It could state how many eligible accounts were offered form access, how many forms were started if that metric is collected, how many were submitted, and how many arrived before or after the deadline. It could then report appointments accepted on first review, forms assigned reason-coded defects, defects cured, rejections maintained, reviews opened, reviews resolved and cases still unresolved when voting closed.

The report could separately describe the proxy-carrier checks: how many accepted representatives carried one, two, three, four or five proxies; how many proposed appointments conflicted with the ceiling; and how many involved a candidate prohibition. Small cells could be suppressed or combined where disclosure might allow re-identification. Registered-contact representation should be reported in a separate category so that it is not confused with proxy volume.

The crucial reconciliation would connect appointment states to access states. The number of effective appointments should match the number of proxy routes enabled, subject to clearly explained exceptions. Every principal route disabled because of an effective proxy should correspond to one enabled delegated route. Pending, failed, withdrawn or reversed appointments should reconcile with principal access that remained active or was restored. The totals should balance without exposing whether any ballot was cast or how anyone voted.

If the platform records ballot participation, only coarse aggregate counts should be reported, and those counts should not be joined to proxy-carrier groups where that could weaken secrecy. The objective is not to reconstruct votes. It is to demonstrate that the access-control machinery neither duplicated nor extinguished entitlements without an accounted state.

An external or functionally independent reviewer could verify the ledger logic and publish an assurance statement. The reviewer would need access to the rule versions, event logs and reason codes, not to political preferences. The system could commit to the event ledger cryptographically before publishing aggregates, but technical sophistication should not substitute for plain explanation. Members need to understand what was counted and what was not.

The counterfactual is straightforward. Had the 2022 process published such a reconciliation, observers could assess the operation of the five-day cutoff without demanding private files. They could see whether forms moved through the system in time, whether defects were handled consistently, whether reviews remained live before the platform closed and whether delegated access balanced disabled principal access. The sealed public record contains no such aggregate ledger. Saying what it could have shown is not claiming what its numbers would have been.

Versioning is part of fairness

The divergent public formulations also point to a document-control problem. Election rules are often spread across bylaws, notices, process pages, timelines, FAQs and portal instructions. Each document may be defensible in isolation while leaving a participant uncertain about precedence. A modern election should publish a small rule manifest that resolves the stack.

The manifest would identify the governing bylaws version, the election-process version, the meeting mode, the eligible member classes, the relevant channels and the exact deadline. It would state which text controls if the notice and process page differ. It would distinguish submission deadline, approval target and final effectiveness requirement. Every page would link back to the same manifest, and every portal receipt would store its version identifier.

If an administrator later corrects a typo or clarifies a timezone, the change should be appended with a timestamp and explanation. Material changes after proxy access opens should trigger direct notice to eligible members and a reasoned decision about whether the deadline must move. Historical versions should remain available so a reviewer can judge a filing against the text presented when it was made.

Versioning protects administrators too. Without it, staff may be accused of applying a rule that was never visible even when they followed an internal instruction in good faith. A durable manifest shows that the same decision criteria were available to participants and validators. It narrows disputes from competing screenshots to an identified instrument.

This kind of clarity would not answer every question. Corporate-authority evidence can be complex, and privacy rules may limit disclosure. But it would remove avoidable ambiguity over five days, five clear days, 24 hours and before-meeting production. It would say which phrases perform different jobs and which path applies to a particular meeting mode.

The discipline of not inventing a victim

Governance reporting often becomes vivid through a person: a member races against a clock, a form bounces, an official refuses an appeal. No such story is established here. Inventing one would make the article easier to narrate and less true.

The absence of a member-level record requires a different kind of human interest. The human question lies in the experience any participant should be able to have: knowing which form is authoritative, seeing a reliable timestamp, understanding a defect, preserving one voting route and receiving an answer while that answer still matters. These are not abstractions to the people acting for organisations. They are the interface between corporate rules and individual action.

Restraint also prevents the deadline from being used as evidence for a preferred verdict about the election. We cannot say the cutoff changed turnout. We cannot calculate a missing electorate without a denominator. We cannot infer a margin or winner. We cannot declare the election valid or invalid on this record. We cannot attribute strategic intent to staff, the CEO, ECom, NomCom, candidates or members.

What we can say is precise. The official notice offered a five-day MyAFRINIC path. The virtual rule repeated it. The timeline used “five clear days” and fixed 28 May as the close. The face-to-face rule described a 24-hour emailed and approved path. The bylaws required written authority produced before the meeting. The rule prohibited candidates from serving as proxies, capped proxy carrying at five and distinguished registered-contact representation. Issuance disabled the appointer’s electronic vote. Those facts create an authentication architecture with reviewable risk points.

That architecture is enough for a serious institutional analysis. It asks the organisation to publish a system that lets a member prove compliance and lets the membership audit aggregate treatment. It does not need a scandal to justify good design.