Summary
- AFRINIC's 2025 disclosure assigns USD 931,849 to the board election held on 23 June, USD 111,576 to the election held on 12 September and USD 1,043,425 to election costs for the year. The labels are an institutional accounting presentation, not a causal audit of every expense.
- The second ballot's invoice is only one part of rerun cost. Members repeat voter designation and scrutiny, candidates repeat campaigning, staff defer ordinary work, extraordinary administration continues and uncertainty delays restoration of accountable control.
- Cost cannot justify certifying a result that cannot be trusted. It should instead discipline remedy: officials must compare credential correction, quarantine, recount, partial rerun, delayed certification and total cancellation, then explain why the selected option is the least disruptive sufficient response.
- Future election budgets should include a contingency reserve, preservation duties, unit-cost disclosure and a public remedy-cost statement that identifies who bears direct expense, delay, participation burden and residual legitimacy risk.
The disclosed price of two election exercises
AFRINIC's additional disclosure for 2025 provides an unusually concrete starting point. It lists USD 931,849 as the subtotal associated with the board election held on 23 June and USD 111,576 for the election held on 12 September. The stated total associated with elections is USD 1,043,425. For an organisation emerging from an extended governance crisis, that figure is not a side note. It is part of the cost of restoring an elected board.
The June subtotal is dominated by professional and extraordinary-administration lines. The disclosure assigns USD 407,518 to Judgement Limited, USD 157,699 to 3 Hare Court, USD 55,484 to Queens Court Chambers, USD 44,605 to Georges Penny Chambers and USD 225,000 to the receiver's fee. It also lists Civica, a hotel, live streaming, connectivity, Nomination Committee airfares and accommodation, usher fees, election-committee allowance, taxis and stamps.
The September subtotal lists a receiver fee of USD 75,000, USD 19,600 for Voatz, USD 2,900 for candidate identity checks, USD 14,046 as appreciation for staff involved and a small incidental amount.
Those lines should be read carefully. The page calls them costs associated with each election. It does not demonstrate that every legal or professional charge arose solely from ballot mechanics, that no item served the wider receivership, or that the allocation has been independently tested for this specific analytical purpose. “Associated with” is broader than “caused only by.” A fair account preserves the figures and their stated labels without inventing a causal precision the disclosure does not claim.
Even with that caution, the difference between the two subtotals is informative. A rerun is not necessarily as expensive as the first attempt. Existing candidate knowledge, a shorter timetable, online voting, changed verification and fewer venue costs can reduce the incremental price. Nor does the smaller September subtotal mean cancellation was cheap. The relevant comparison is not only USD 111,576 against zero. It includes costs already sunk in June, costs of investigating and litigating the first process, continued extraordinary administration and the burdens imposed on every entity.
The costs that do not appear as election invoices
A member organisation pays in staff time. Someone must read revised rules, verify eligibility, designate a voter, check communications, assess candidates and vote again. Larger members may absorb those hours. A small network may rely on one executive who is also responsible for operations, customers and finance. Repetition therefore changes the effective electorate even if formal eligibility remains constant. The entities with the least spare capacity are most likely to drop out.
Candidates pay through renewed campaigning and uncertainty. They must decide whether old endorsements remain valid, whether positions need restating and whether participation implies acceptance of the cancellation. A candidate may incur travel, legal or advisory expense. More subtly, the second contest can change reputation: an allegation about the first process may attach to candidates even when no finding concerns them.
Staff pay through diversion. Election support competes with registry services, security work, member support, financial recovery and institutional rebuilding. The opportunity cost is not the entire payroll. It is the value of urgent tasks delayed because the same people must repeat accreditation, communications, testing and incident response.
Members collectively bear the cost of prolonged extraordinary authority. The 26 June 2025 court order extended the time for a new election and board constitution to 30 September. The order supplied the time needed for another attempt; its short public text does not price the extension. Every additional week before an elected board can act may affect oversight, strategic decisions and confidence, even where the receiver is acting lawfully and diligently.
The wider network community pays through uncertainty rather than a direct levy. Regional registry continuity affects confidence in governance, policy implementation and relationships with other Internet institutions. That does not mean a disputed ballot immediately disrupts routing. It means the organisation's ability to make durable decisions remains contested while the election is unresolved.
Finally, legitimacy itself has a replacement cost. If members believe a whole election was discarded without sufficient reasons, the second poll may attract suspicion. If officials preserve a questionable result to avoid expense, confidence may be worse. The cheapest immediate option can be the most expensive institutional option once future participation and legal challenge are considered.
Cost belongs inside proportionality, not above integrity
The wrong lesson from a seven-figure disclosure would be that officials should avoid reruns because elections are expensive. Sunk spending cannot convert an unreliable result into a reliable one. If defects are pervasive and valid votes cannot be separated, cancellation may be the only defensible remedy regardless of price.
The right lesson is that remedy must be granular. Before cancelling every contest, a decision-maker should identify the affected credential, voting channel, seat and possible margin. Can a member cure a document defect? Can an uncast credential be quarantined? Can count arithmetic be corrected by recount? Can certification pause while authenticity is checked? Can only one seat be rerun? Total cancellation sits at the top of a ladder, not at the first report of irregularity.
Each rung has costs and residual risks. A cure period is cheap but unsuitable for proven falsity. Quarantine protects unaffected votes but may be impossible once secret ballots are mixed. A recount addresses counting error but not unequal eligibility. A partial rerun preserves more voting effort but can produce strategic spillovers between seats. A total rerun offers a clean event while discarding valid participation and changing conditions.
A reasoned decision should therefore include a remedy-cost statement. It need not monetise trust or every hour. It should state direct expected expense, time to elected control, affected members, candidate burden, operational diversion, evidence preserved, valid votes discarded and residual challenge risk. The decision-maker can then explain why a more expensive remedy is necessary or why a narrower one is sufficient.
Cost analysis should never become a weapon against complainants. A member who reports a credible defect should not be blamed for the institution's contingency expense. The cost arises from the need to produce a reliable election and from the quality of controls chosen in advance. Charging challengers or threatening them with institutional loss would suppress evidence and make later crises more expensive.
Budget before the ballot breaks
Election budgets should separate ordinary conduct, assurance and contingency. Ordinary conduct covers voter communications, platform or venue, staffing and results. Assurance covers independent testing, observer access, identity verification, custody controls and certification. Contingency covers evidence preservation, a short certification pause, independent review and a possible partial rerun. A whole-election reserve may be impractical, but the board should know how it would fund one without compromising essential registry services.
Contracts should price failure states. A voting provider's fee should state what happens if polling pauses, logs must be exported, a recount occurs or a second event is required. Professional advisers should identify which work relates to the election and which relates to broader litigation or administration. That improves later disclosure without prejudging whether the spending was justified.
After the event, actual cost should be reported against budget with the same category definitions. The institution should explain whether a rerun reused systems, candidate checks or voter records, and which safeguards changed. Reuse can save money, but it should never carry forward a defect. New verification can improve confidence, but it also changes participation and must be justified.
The ultimate objective is not a cheap election. It is a reliable transfer of authority at a cost members can see and institutions can sustain. AFRINIC's 2025 disclosure makes that objective measurable. It also shows why the proportionality question cannot be postponed until after a cancellation has already made most costs unavoidable.
Annulment is not one act
The word “annulment” creates an impression of a single switch. An election exists, an authorised person flips the switch, and the election disappears. That image is convenient and usually wrong. Elections pass through stages: nominations are accepted; voters are accredited; voting opens; ballots are cast; voting closes; votes are reconciled; results are calculated; an official certifies them; winners are announced; appointments take effect; and challenges may follow. At each stage, a different intervention may be available. A temporary suspension is not the same as refusing to certify.
Withholding an announcement is not the same as declaring every valid ballot legally ineffective. Ordering a rerun is not the same as finding that no candidate was elected.
That distinction matters because AFRINIC’s 23 June 2025 election occurred in an exceptional institutional setting. The organisation had no functioning elected board. A receiver appointed under Mauritian law was charged with preserving the organisation and facilitating the reconstitution of its board. The receiver publicly stated that concerns had been raised about voter documentation, reported those concerns to authorities, and on 26 June announced a decision to annul the election. A later communication said suspicions related particularly to powers of attorney and that a police investigation was continuing.
Those statements establish that an extraordinary intervention occurred. They do not, by themselves, settle the source or limits of the power used.
The receiver’s April 2025 communiqué described a court mandate to “hold the ring,” preserve AFRINIC’s assets and value, organise board elections and enable appointment of a chief executive. It also located the receiver’s powers in the Eighth Schedule to the Insolvency Act and AFRINIC’s constitutional documents. This was a strong basis for administration and urgent protection. Yet a mandate to organise and complete an election does not answer every remedial question. A returning officer authorised to run a poll may correct a clerical error. A court may set an election aside. A membership meeting may resolve an unprovided-for issue. The legal consequences differ.
A serious inquiry must therefore disaggregate the action. Did the receiver pause certification while facts were checked? Did he determine that votes affected by suspect documents could not be separated from unaffected votes? Did he declare the entire election void from the beginning? Did he exercise a protective power pending court direction? Did the Supreme Court later ratify, merely acknowledge, or independently authorise the rerun? Public language moved quickly from concerns to annulment and then to a judicially extended deadline for another election. Each step needs its own authority.
The point is not semantic. If the original ballots remained legally capable of certification, a rerun could disenfranchise people who voted validly. If the original election was incurably compromised, certifying it could install a board without a reliable mandate. The institution needed an answer fast, but urgency did not eliminate the obligation to say which legal act was being taken and by whom.
The receiver’s exceptional but bounded office
A receiver is not simply a substitute board member. Receivership is a legal office created for a defined purpose and supervised by a court. The officeholder may control assets, operate an undertaking, make urgent decisions and preserve value. The precise authority depends on the appointment order, governing legislation and any later directions. Where an organisation lacks a board, the receiver may also perform functions that would otherwise be impossible. But the receiver’s legitimacy comes from those instruments, not from an electoral mandate.
AFRINIC’s 2025 communications repeatedly described the receiver as the actor empowered to organise the vote because no directors or chief executive were in office. Later election guidelines went further, stating that the Supreme Court conferred exclusive election authority on the receiver and that, for the election, the receiver would be deemed to exercise board powers. That assertion helps explain who could issue instructions to staff, appoint election bodies and approve arrangements.
It still leaves two questions: whether the court order expressly included a power to void a completed vote, and whether an implied protective power extended beyond a temporary pause.
Implied authority should be matched to necessity. If credible evidence suggested forged powers of attorney, the receiver could hardly be required to announce winners immediately and surrender control before an inquiry. Preserving the status quo could justify withholding certification, securing records and asking the court for directions. Those actions are reversible. They protect both the organisation and the possibility of giving effect to valid ballots later.
Final annulment is different. It destroys the legal effect of votes and changes the route by which a board will be constituted. The stronger the interference with member rights, the clearer the authority should be. An implied power to preserve assets does not naturally answer whether an election defect requires exclusion of particular votes, a recount, a rerun for one seat or cancellation of all eight contests. That is an adjudicative judgment. It demands a stated test of materiality and a reason why narrower remedies cannot restore confidence.
The receiver also faced an institutional conflict that no allegation of personal bad faith is required to identify. His court mandate depended on completing a valid election, while his tenure continued until that task could be achieved. Annulment both protected the integrity of the task and extended the period before elected directors could replace extraordinary administration. A responsible system treats this as a structural conflict, even when the receiver acts conscientiously. The answer is independent review, not speculation about motive.
The most defensible reading is therefore two-layered. The receiver possessed broad operational and protective authority, including the ability to stop immediate certification when credible irregularities threatened the organisation. A final decision that the entire membership vote had no effect should have rested on express language in the appointment order, a specific bylaw power validly assumed by the receiver, or prompt confirmation by the Supreme Court after affected parties had an opportunity to be heard. Public confidence required the receiver to identify which layer he was using.
What NomCom could decide
AFRINIC’s Nomination Committee had significant pre-ballot authority. Article 9 of the bylaws gives NomCom general responsibility for pre-election matters. Historical board-election guidance described duties that include calling for candidates, prescribing qualifications, reviewing nominations, interviewing where appropriate and finalising candidate lists. In the receivership election, NomCom also communicated election arrangements and, under published guidance, supervised aspects of the poll in coordination with the election body.
These functions can determine whether an election reaches the ballot in a lawful form. If a candidate is ineligible, NomCom may exclude the nomination under its mandate. If a nomination deadline or qualification rule was misapplied, NomCom may be able to correct the slate before voting. It may also report an incident discovered during voting and recommend a pause. None of this necessarily gives NomCom power to invalidate ballots after members have voted.
The distinction follows from institutional design. A nomination committee protects the gate to candidacy. It is not ordinarily an election court. Giving the same body authority to select candidates, supervise polling and conclusively judge disputes about the resulting vote would concentrate too much power. A committee whose own decision might be challenged cannot be the sole final reviewer of that challenge.
Article 10.2 of the AFRINIC bylaws is also relevant. The 2026 election guidelines paraphrase it as requiring members present at a general or special meeting to resolve collectively and by consensus an election issue not expressly covered by the bylaws. The guidelines say that if an uncovered issue is identified before the meeting, the election should not be interrupted; the matter should instead be raised for members to resolve. This provision points away from a free-standing committee power to invent a final remedy for every gap.
The receivership context complicates application. The June 2025 election was designed under court supervision, and the absence of a board made ordinary escalation difficult. Yet difficulty is not a transfer instrument. NomCom could document facts, preserve materials, advise the receiver and explain how a defect affected candidate eligibility or conduct. It could perhaps decline to finalise a result if a governing rule made its certification essential. A decision that the whole election was void required a separate foundation.
NomCom’s role in a sound authority matrix is therefore substantial but bounded: detect, investigate within mandate, correct nomination errors, recommend protective measures, and give a reasoned view on whether the announced rules were followed. It should not possess an unreviewable power to cancel a member election that has produced winners. If its findings trigger cancellation, another actor should adopt the remedy and accept responsibility for it.
ElecCom, trustees and the difference between custody and judgment
The Election Committee, often called ElecCom, occupies the operational centre. AFRINIC’s bylaws give it general responsibility for handling elections and require liaison with NomCom for the holding of polls. Published election materials have assigned it tasks involving voter verification, voting arrangements, supervision and adherence to electoral guidelines. The 2025 appointment announcement described ElecCom as responsible for strict compliance and accountability at every stage.
Operational custody creates immediate authority. If a voting credential appears compromised, ElecCom must be able to quarantine it. If the voting platform fails, the committee may need to pause access. If physical ballot materials do not reconcile, it should refuse to proceed blindly. A committee unable to halt an unsafe operation would be decorative.
But custody is not final adjudication. An election committee can record that eleven powers of attorney appear duplicated, that a credential was used from conflicting locations, or that the number of physical ballots exceeds checked-in voters. It can separate affected records and calculate whether they could change an outcome. It should not transform suspicion into a conclusive finding of fraud without an applicable standard and fair opportunity for response.
The same principle applies to election trustees or an independent voting provider. Technical officials may certify that a cryptographic tally was produced correctly or that sealed materials were opened in accordance with instructions. They can say whether the recorded votes match the inputs they received. They cannot determine whether a disputed corporate representative had legal authority under a membership agreement unless that question was assigned to them and supported by appropriate expertise.
This separation is essential in the AFRINIC case because “voter documentation” can describe different defects. A document may be forged. It may be genuine but signed by a person lacking corporate authority. It may comply with a member’s local law but not with an election instruction. It may have been accepted inconsistently by staff. Each possibility points to a different remedy and decision-maker. Police may investigate forgery; a court may determine legal validity; election officials may apply published accreditation criteria; the receiver may protect the organisation while those questions remain open.
ElecCom’s strongest legitimate power was therefore to preserve integrity at the point of operation: stop, segregate, record, report and, where authorised, decline to certify. A complete annulment required an additional judgment about legal consequence. Calling ElecCom “independent” did not enlarge its jurisdiction. Independence describes freedom from direction; authority describes what it may decide.
Members were principals, not emergency judges
AFRINIC’s resource members were the intended electors. Their voting rights gave the election democratic content and made the resulting board accountable to the membership. The bylaws also reserve significant corporate powers to members in general meetings. It is therefore tempting to say that members could simply decide whether their own election stood.
That proposition is only partly correct. Members can adopt resolutions, elect directors and resolve certain questions left open by the bylaws. Article 10.2 gives those present a consensus role over unprovided-for election issues. A general meeting may also receive reports, challenge officials and direct institutional reform. These are meaningful powers.
Yet an assembly is poorly suited to determining contested evidence about its own electorate in real time. Members may be candidates, supporters, disputed voters or competitors. Attendance may not match the accredited electorate. A consensus rule can allow one interested party to block a remedy. Sensitive documents cannot always be projected in a meeting. The people affected by an allegation need notice and an opportunity to respond.
Member sovereignty must therefore be made procedural. Before voting begins, members should approve or at least receive the rules identifying who may pause, certify or challenge the election. If a novel issue arises at the meeting, the chair should frame a precise question and record the consensus. If material facts emerge after the meeting, a special meeting can consider a recommendation, but adjudication of fraud or document validity may still require a court.
Members also possess a collective legitimacy claim against overbroad cancellation. If only a small, identifiable group of credentials is disputed, valid voters are entitled to ask why their ballots cannot be preserved. If a defect affects one regional seat, candidates in other seats can ask why the whole election must be rerun. The burden should be on the actor proposing annulment to explain indivisibility.
The membership’s role in the authority matrix is thus to establish rules, exercise reserved powers, resolve genuine gaps where feasible, and hold decision-makers accountable. It is not to conduct a show-of-hands trial on unresolved allegations. An election gains legitimacy from members choosing among candidates, but remedies gain legitimacy from law, evidence and fair procedure.
The Supreme Court’s supervisory position
The Supreme Court of Mauritius had a qualitatively different authority. AFRINIC was under receivership by court order. The receiver’s mandate, extensions and ability to continue until a board election were matters of judicial supervision. The court could interpret its own orders, direct the receiver, preserve the status quo, hear challenges and grant remedies affecting the election timetable.
AFRINIC’s public communications show that the court was approached repeatedly. A 19 June 2025 communiqué explained litigation and directions immediately before the vote. After the receiver announced annulment, communications said that the underlying concerns were reported to the court and that an extension was granted to hold new elections. This sequence demonstrates judicial involvement. It does not show, without the orders and reasons, whether the court decided that the first election was legally void.
An extension is not necessarily ratification. A judge may extend a receiver’s mandate because the practical task remains unfinished, while leaving disputes about the earlier decision to another hearing. Equally, an order may expressly approve a rerun and thereby resolve the immediate authority question. The operative text matters. Public summaries should not convert procedural accommodation into a merits judgment.
The court was the actor best placed to resolve a clash between the receiver’s protective duties and members’ electoral rights. It could receive confidential evidence, impose safeguards, distinguish suspect from valid votes, and decide whether a narrower remedy was workable. It could also address the receiver’s structural conflict by appointing an independent assessor or directing how reasons should be published.
Judicial supervision should not mean judicial administration of every ballot. Courts are slow and may lack specialised election expertise. ElecCom should still handle operations, and NomCom should still manage candidacy. The court’s role becomes essential when an intervention exceeds ordinary administration, materially changes member rights or rests on contested legal authority.
For future emergencies, the receiver or board should seek directions before final annulment whenever time permits. If immediate action is unavoidable, the decision should be framed as interim and brought before the court within a fixed period. That approach preserves safety without allowing emergency language to become permanent authority.
An authority matrix for each remedy
The central reform is a published authority matrix. It should not merely list institutions. It should pair each remedy with its source, trigger and review. For a temporary technical pause, ElecCom might act when continued voting risks loss of ballot integrity. The pause should be short, logged and reported to candidates. NomCom and the receiver should be notified, but neither should rewrite the record.
For suspension of certification, ElecCom or designated election trustees might act when reconciliation fails or a material incident remains unresolved. The suspension should preserve all ballots and expire unless a named reviewer extends it. A preliminary statement can explain the category of concern without accusing individuals.
For exclusion of specific votes, the decision-maker must apply a published eligibility rule and provide affected members an opportunity to establish authority. The standard should distinguish missing documentation from proven falsity. A record should show how excluded votes affect each contest. Review should be available to an independent officer or court.
For correction or recount, election officials should act under rules that preserve the original data and permit candidate observers. The result of a recount should be certified separately, with a comparison explaining the change. A recount is suitable for counting error, not for resolving the legal validity of representation documents.
For postponement before voting, the receiver or board may need authority when nominations, voter registration or essential systems are not ready. The decision should state a new timetable and protect candidacies already accepted. Courts should supervise postponement when a judicial deadline or receivership mandate is affected.
For partial reruns, the actor should identify the affected seat, electorate or voting channel and explain why unaffected contests can stand. This is often the least disruptive response to a localised defect. It requires a rule governing whether original candidates and voter lists remain fixed.
For total annulment, the threshold should be highest: proven or strongly evidenced defects that are material, outcome-relevant or so pervasive that no reliable separation is possible; a reasoned decision; representations from affected parties; preservation of evidence; and review by the court or an expressly authorised independent tribunal. The power should never be inferred solely from the ability to organise a new poll.
For final certification, an identified official should attest both the count and compliance with essential rules. Certification should not prevent later legal challenge, but it should mark the point when winners take office unless a court orders otherwise. Without such a matrix, every actor can claim responsibility when convenient and deny jurisdiction when challenged.
Materiality and the missing middle
Election disputes often jump from “there was an irregularity” to “the election is invalid.” That leap omits materiality. Not every defect affects voter eligibility, ballot secrecy, the count or the result. A governance system must define the missing middle: how evidence of an error becomes a remedy.
The first inquiry is authenticity. Was a document actually false, altered or duplicated? Suspicion based on formatting or late submission is not a finding. The second is authority. Did the person purporting to represent a member possess legal authority, regardless of whether the document looked unusual? The third is compliance. Did the submission meet the announced election requirements? A genuine authorisation can still be late or incomplete. The fourth is causation. Did acceptance of the disputed credential lead to a vote? The fifth is materiality. Could those votes change a seat, and can they be isolated without revealing ballot choices?
Ballot secrecy creates a hard case. If an invalid voter received a secret ballot, officials may know that the person voted but not for whom. Where the margin is smaller than the number of invalid ballots, the result may be uncertain. That can justify a rerun for the affected contest. If the margin is larger, the defect may not be outcome-determinative, though repeated unequal treatment can still damage legitimacy.
Pervasive procedural inequality may justify intervention even when arithmetic cannot show a changed winner. Suppose one faction obtained exceptions to documentation rules while another was rejected. The injury is not only the count; it is unequal access. Still, total annulment requires an explanation of scope. If the inequality affected only in-person voting, could verified electronic ballots stand? If it affected only one seat, why cancel seven others?
The receiver’s public statement referred to serious irregularities and an ongoing police investigation. A police inquiry is relevant but not a substitute for election findings. Criminal investigation asks whether offences may have occurred and whether individuals can be charged under a demanding standard. Election administration asks whether the declared result is reliable. An election decision may be required before a criminal case ends, but it must use its own stated evidentiary standard.
A reasoned annulment decision should therefore contain a redacted materiality analysis: number and type of disputed credentials; stage at which the problem arose; seats potentially affected; margins; inability to segregate votes; rules applied; alternatives considered; and reasons narrower remedies failed. Personal information and investigative details can be protected. The logic cannot.
Reasons, confidentiality and the right to answer
Election integrity can require confidentiality. Identification documents contain sensitive information. A forgery inquiry may be prejudiced by premature disclosure. Election vendors may hold security details that should not be made public. These constraints explain redaction; they do not justify a decision without reasons.
The public needs institutional facts rather than private accusations. A notice can say that a specified number of authorisations were challenged, that independent verification could not confirm them before certification, that the number exceeded a margin, and that ballots could not be separated. It need not name the people or reveal passport numbers. It can state the legal provision and standard without disclosing police evidence.
Affected members need more. Before their votes are excluded, they should be told the defect and allowed to respond through a secure channel. A company may explain who had signing authority, correct a mistaken registry record or show that a document was authentic. Candidates should be permitted submissions on remedy, especially where officials propose cancelling an uncontested or unaffected seat.
The decision-maker must also disclose conflicts. NomCom and ElecCom members may have professional ties to candidates or members. The receiver’s tenure may be affected by timing. Lawyers may have acted in related litigation. A recusal record can protect both the institution and the individuals involved.
Reasons discipline authority. They force the decision-maker to connect fact, rule and remedy. They permit a court to review the decision without rerunning the entire election. They allow members to distinguish an unavoidable protection measure from convenience or overreach. They also create precedent for the next incident.
Silence has a compounding cost. When members cannot inspect even a redacted account, competing narratives fill the gap. One side treats any irregularity as proof of capture; another treats any cancellation as proof of manipulation. The organisation then reruns an election in an environment where the legitimacy of the referee is already contested. Security cannot repair that alone.
AFRINIC should publish an incident decision after any legal restriction lifts, even if later elections make the immediate result moot. The document should state who decided, under what authority, on what date, using what standard, and with what review. Historical accountability is part of operational recovery.
Why a fresh election does not cure an unlawful cancellation
A rerun can produce a clear result and a functioning board. That practical success does not retroactively answer whether the first vote was lawfully cancelled. If institutions treat the second election as a complete cure, they create a dangerous precedent: any official capable of arranging another ballot can erase the first and rely on later participation to legitimise the act.
Members may vote again for many reasons. They may want AFRINIC to recover. They may fear that boycott will hand power to opponents. They may accept the court’s timetable without conceding the receiver’s legal theory. Participation is not necessarily waiver. Candidates may also face changed conditions: reputational damage, campaign fatigue, altered voter registration and new rules on proxies or powers of attorney.
The second election can nevertheless reduce practical remedies. A court may be reluctant to displace a board already installed after a later, better-protected vote. Damages may be unavailable or inappropriate. The remaining remedy may be declaratory: clarify that the earlier power was absent or exercised incorrectly, and define future limits. Such a declaration still matters because AFRINIC’s governance will face other disputes.
The legitimacy of the replacement election depends partly on the explanation of the first. If voter-registration rules are tightened without identifying the prior weakness, members cannot assess proportionality. If proxy voting is removed because of allegedly forged authorisations, the institution should explain why less restrictive verification was inadequate. Otherwise, an emergency response can permanently narrow participation.
The 2026 election guidelines prohibit proxy voting in virtual elections in light of alleged forged powers of attorney and use a prior register as a baseline. Those choices may be reasonable safeguards. They also show how one disputed episode can reshape future rights. That is precisely why the original authority and evidence should be recorded.
A lawful rerun should therefore be accompanied by a preservation clause: holding the new election does not destroy records or preclude review of the cancelled election. It should also state whether candidates, voters and accepted nominations carry over; which rules changed; and which actor approved each change. Recovery should not depend on institutional amnesia.
A defensible emergency sequence
AFRINIC can convert the lessons into a sequence suitable for future crises. First, an election official receiving a credible incident report should open a numbered case, preserve relevant records and notify a small group defined in advance. The notice should avoid conclusions and identify whether voting can safely continue.
Second, the authorised operational officer may impose the least disruptive temporary measure: quarantine a credential, extend voting for affected members, pause one channel, or suspend certification. The action should have a short expiry and should not alter ballots already cast.
Third, an investigator independent of candidates and the original decision should establish facts. Technical, corporate-authority and criminal questions may require different expertise. The investigator should report what is confirmed, unresolved and immaterial. Evidence should be retained under clear access controls.
Fourth, affected voters and candidates should receive the allegations relevant to them and a practical period to respond. Urgency may shorten the period, but it should not eliminate it. Responses and recusals should form part of the record.
Fifth, the designated remedial authority should apply a published ladder: correction, exclusion, recount, partial rerun, then total annulment. It should explain why every less intrusive option is limited public evidence. Where the organisation is in receivership, a total annulment should be submitted to the supervising court for confirmation.
Sixth, a public redacted decision should follow. It should identify authority, evidence categories, materiality, remedy, timetable and review rights. It should distinguish allegations referred to police from findings made for electoral purposes.
Seventh, certification of the replacement result should include an independent compliance statement. The board taking office should not rewrite the incident account, but it should commission a governance review and propose bylaw amendments to close gaps.
This sequence is slower than a one-line cancellation. It is faster than years of legitimacy litigation. More importantly, it prevents a receiver, committee, vendor or chair from being forced to invent jurisdiction under pressure.
Certification should carry an authority statement
The final safeguard is deceptively simple: every election certificate should identify the authority behind it. A tally certificate can establish that a voting provider counted the recorded ballots correctly. An eligibility certificate can establish that the accepted voters met published requirements. A compliance certificate can establish that election officials followed the timetable and custody rules. A legal certificate can state that the person declaring the result possessed the power to do so. Combining all four into an unsigned announcement makes it impossible to see which proposition has actually been verified.
AFRINIC should require a short authority schedule beside every future result. It should name the official responsible for voter accreditation, the custodian of ballots, the official who calculated the result, the actor who certified compliance and the legal instrument authorising appointment of the winners. If a court order temporarily changes the normal allocation of functions, the schedule should cite the order and explain the change. If any official acts subject to a reservation or unresolved challenge, that limitation should also appear.
This would not turn an election notice into a judgment. It would make institutional responsibility legible. Members could distinguish a technical assurance from a legal conclusion, and a reviewing court could identify the exact decision under challenge. Election officials would be less likely to exceed their mandate because the certificate would require them to state which proposition they were competent to attest.
The same schedule should accompany an interruption. A suspension notice should identify its expiry, evidentiary trigger and route to review. A refusal to certify should say whether the problem concerns the tally, eligibility, procedure or legal authority. An annulment decision should state who converted that defect into a remedy and why a narrower correction was unavailable. These labels prevent the broad word “integrity” from doing the work of several unproved conclusions.
Institutional memory also improves. Future boards, members and election committees should not have to reconstruct authority from scattered communiqués years later. A durable certificate allows later reviewers to compare the action taken with the bylaws and court directions in force on that date. That record is particularly important when extraordinary administration ends, because the people who inherit the organisation may not have participated in the emergency decisions.
Authority statements cannot cure an unlawful act. They make unlawful or uncertain acts visible soon enough to be reviewed. In a crisis, that is not paperwork at the edge of legitimacy; it is part of the mechanism by which legitimacy is preserved.
The answer is a hierarchy, not a name
Who had power to annul AFRINIC’s election? The most accurate answer is not a single office. Different actors possessed different powers at different stages. ElecCom could protect the conduct and custody of the vote. NomCom could manage candidacies and report non-compliance within its mandate. Members could exercise reserved corporate powers and resolve certain gaps. The receiver could take urgent protective action and, under court authority, perform functions needed to restore governance. The Supreme Court could interpret the mandate, supervise the receiver and conclusively authorise a remedy that displaced member voting rights.
The receiver was the practical centre because AFRINIC had no board and the court had charged him with reconstitution. That made an immediate suspension defensible when serious documentary concerns emerged. It did not make every form of annulment self-validating. The final destruction of an election’s effect required a clear legal source, material evidence, reasons, proportionality and independent review. If the appointment order expressly supplied that authority, it should have been cited. If it did not, court confirmation was the safer constitutional route.
This conclusion does not decide whether suspect powers of attorney were genuine or whether the June result was reliable. It states the prior question that institutional recovery cannot skip. Good ends do not merge offices. A receiver preserving value, a committee handling ballots, members choosing directors and a court supervising insolvency each derive legitimacy from different sources.
AFRINIC’s crisis exposed the danger of drafting election rules for ordinary times while relying on improvisation in extraordinary ones. The next version should define not only how votes are cast but how authority moves when the poll breaks. It should say who can pause, who can investigate, who can exclude, who can void, who can review and when a decision expires.
An election is not legitimate merely because ballots were counted. Nor is its cancellation legitimate merely because concerns were serious. Legitimacy rests on a chain: authorised actor, proven trigger, fair hearing, proportionate remedy, recorded reasons and available review. AFRINIC needed that chain more than it needed a confident verb.