Summary

  • AFRINIC's September 2025 board election ended a period without ordinary elected directors, but personnel replacement did not automatically reconcile historical records, complete delayed audits, settle authority disputes or redesign the powers attached to registry services.
  • A clean-restart story is attractive because it assigns failure to former officeholders and treats a new election as institutional renewal. The harder evidence points to a systems problem involving board quorum, executive delegation, member records, financial oversight, legal strategy, data custody and weak separation between ordinary administration and high-consequence registry action.
  • Effective reform should map every decisive power to a named legal source, evidence threshold, second approver, audit trail, continuity safeguard and independent remedy. A bylaw amendment that only changes titles, committees or election mechanics will not be enough.
  • The restored board should preserve useful staff knowledge without allowing oral history, inherited access or executive convenience to substitute for reproducible records. Reform must make the institution governable by people who were not present during the crisis.
  • AFRINIC will have escaped the crisis only when an adverse decision, disputed election, management vacancy, banking interruption or court order can be contained without turning network-resource records and dependent operators into leverage.

The new board inherited an old machine

The most seductive sentence in institutional recovery is also the least testable: new people will do better. It may be true. It may even be necessary. AFRINIC could not return to ordinary corporate government without directors, and the September 2025 election was therefore a material change. A board could meet, appoint committees, supervise management, address delayed accounts and begin a bylaw review. Those are real gains after years in which ordinary authority had fractured.

But the new directors did not enter an empty company. They inherited contracts, accounts, litigation positions, access credentials, member files, registration records, policies, technical systems, staff relationships, pending requests and institutional memories created under earlier authority. They also inherited the same legal person incorporated in Mauritius, the same 2020 constitution, the same dependence of operators on recognised registry data and much of the same external RIR architecture. An election changed the occupants of the boardroom. It did not reset the machine they were asked to govern.

That distinction is the clean-restart problem. Reform language often treats a crisis as a contaminated chapter that can be closed by electing credible directors, recruiting a chief executive and resuming meetings. The image is reassuring because it reduces structural uncertainty to character. If the wrong people caused the failure, the right people can reverse it. No difficult redesign of powers, incentives or exit rights is required.

AFRINIC's own public updates show why this is incomplete. In October 2025, the board and receiver said financial audits for 2022, 2023 and 2024 were still being completed, that work paused since 2022 had to be reviewed, that legal cases remained, and that a comprehensive bylaw review was needed. In March 2026, the board described continuing legal challenges, collaboration with a receiver awaiting formal discharge, recruitment of a substantive chief executive and the need to address governance gaps. Those statements do not prove present failure.

They prove that restoration involved inherited obligations which an election result could not extinguish.

The relevant question is therefore not whether the directors are new. It is whether the controls are new enough that good conduct is supported, error is visible and bad conduct cannot easily become operational power.

Elections restore an organ, not an institution

A company needs a board, but a board is one organ among several. It does not personally authenticate every member, reconcile every payment, validate every resource-holder request, preserve every backup, administer every certificate or inspect every legal instruction. Directors govern through delegation. The quality of the institution depends on what happens below the resolutions.

This matters especially for AFRINIC because the crisis exposed several kinds of authority at once. Corporate authority determined who could bind the company. Electoral authority determined who could become a director. Registry authority determined who could update number-resource records. Technical authority determined who could operate databases, reverse DNS and routing-security services. Legal authority determined who could litigate, settle, comply with a court order or preserve disputed assets. Financial authority determined who could pay staff, suppliers and counsel. These authorities interacted, but they were not interchangeable.

A newly elected director may have a sound mandate to vote at board level and still lack direct knowledge of how a historical account was approved. A technically competent employee may understand the database and still lack authority to decide a disputed membership status. A receiver may have power to preserve the company and organise an election without becoming the permanent source of number-resource policy. A court may control a corporate remedy without intending to administer routing-security entities. A clean-restart narrative obscures these boundaries because it compresses them into one idea of leadership.

Institutional repair begins by separating them. Each important act should answer six questions. Who may initiate it? Which evidence establishes the relevant fact? Who approves it? What record makes the decision replayable? What continuity interest must be protected while it is disputed? Who can review or reverse it? A board resolution can approve this architecture. It cannot substitute for it.

The September 2025 board was therefore a gateway, not a cure. It supplied an elected body capable of authorising repair. Its legitimacy should be judged partly by whether it reduces future dependence on exceptional people, temporary officeholders and judicial improvisation.

The failure was not confined to one office

AFRINIC's public record since 2019 has included an allocation controversy, internal review, litigation, board-quorum failure, receivership, disputed election arrangements, an annulled June 2025 vote, a replacement election and continuing legal proceedings. These events involved different actors and legal questions. They should not be collapsed into one accusation or one causal story.

They do, however, reveal recurring control surfaces. Historical allocation files raised questions about access, approval, record reconciliation and audit escalation. The board crisis raised questions about quorum, vacancy, delegation and who could act when ordinary directors could not. Litigation raised questions about remedies, frozen funds, authority to represent the company and protection of live services. Elections raised questions about membership status, powers of attorney, voter designation, independent verification and who could annul a result. Receivership raised questions about temporary operational authority and handback.

Changing directors does not automatically change any of those surfaces. A new board can receive the same incomplete dashboard. A new chief executive can inherit the same concentration of credentials. A new committee can rely on the same undocumented staff explanation. A new election can reuse the same baseline register. A revised public statement can leave the same remedy path. Unless the underlying controls change, personnel renewal may only place new names above old uncertainty.

The point is not that every inherited practice was defective. Mature institutions depend on continuity. Experienced staff know systems, customers and historical exceptions that cannot be rediscovered cheaply. Existing contracts and technical arrangements may be sound. A reckless purge would destroy evidence and competence. Clean restart is a fallacy not because continuity is bad, but because it pretends continuity does not exist.

The correct reform posture is selective inheritance. Preserve what can be verified. Quarantine what cannot. Document tacit knowledge. Reauthorise high-impact access. Reconcile open exceptions. Record why a rule or delegation survives. Institutional memory should become evidence, not privilege.

Bylaws can allocate power without controlling its use

AFRINIC's 2020 constitution defines membership, member powers, director appointment, board powers, the chief executive, meetings, committees and other corporate machinery. It is indispensable. It is also necessarily general. A constitution may say the board manages the business or may delegate powers. It rarely specifies the exact evidence needed to change a resource-holder name after a merger, revoke a credential during a dispute or reconcile an election register.

This creates a familiar reform trap. After crisis, institutions rewrite bylaws because the document is visible, can be voted on and carries constitutional symbolism. Clauses are added on term limits, vacancies, committees, eligibility, conflicts or meeting procedure. The resulting text may improve governance. It may still leave decisive administrative power unchanged.

The 2026 bylaw review is therefore important but should be judged against operational questions. Does it create a lawful continuity authority when the board loses quorum? Does it identify which powers cannot be delegated to management? Does it require notice and reasons before high-consequence registry action? Does it preserve status while an independent challenge is heard? Does it distinguish a member vote from operational resource control? Does it require a verifiable handover when a receiver, interim committee or chief executive leaves? Does it prevent one temporary arrangement from quietly becoming precedent?

The constitution should also define what belongs outside constitutional discretion. Accurate records, separation of duties, immutable logs, backup testing and credential inventories are control obligations. They should not depend on whether a particular board values them. Likewise, no bylaw can transform broad mission language into unlimited authority over the economic use of number resources. The legal source and scope of each coercive power still need to be shown.

A useful bylaw review narrows ambiguity. A performative review expands rhetoric while leaving staff and members to discover the real rule during conflict. AFRINIC should prefer boring precision over grand renewal.

Committees are not controls unless information reaches them

The post-election board established finance, audit, remuneration and legal committees. That is a conventional and sensible response. Specialised committees can give directors time and expertise to examine subjects that a full board cannot treat deeply at every meeting.

Yet a committee name does not create oversight. It receives whatever management chooses or is required to report. If an anomaly never enters the audit packet, the audit committee cannot escalate it. If legal spending is reported only as a total, the legal committee cannot distinguish continuity work from strategic litigation. If financial statements are delayed, the finance committee supervises through partial visibility. If access changes are treated as technical administration, no committee may see their governance significance.

Each committee therefore needs an information covenant. The covenant should identify mandatory reports, direct access to responsible staff, protected access to auditors and counsel, exception thresholds and matters that cannot be closed without committee acknowledgement. It should also preserve the full board's responsibility. Directors should not be able to say that a difficult question belonged to a committee and therefore ceased to be theirs.

The audit committee's most important report may not be the annual financial opinion. It may be a monthly control-exception register: privileged access added or removed, high-impact record changes, unresolved reconciliations, overdue corrective actions, policy exceptions, failed backup tests, unusual credential recovery, complaints involving staff authority and any event that could affect member rights or registry continuity.

The legal committee should receive a matter map, not only legal advice. For each case, the board needs the claim, forum, remedy sought, operational dependency, cost to date, settlement authority, record-preservation duty and continuity safeguard. Legal privilege can protect advice without making aggregate control invisible.

The finance committee should separate ordinary operations, litigation, crisis recovery and restricted continuity funds. The remuneration committee should test whether incentives reward closure speed or accurate, reviewable outcomes. Committee architecture becomes reform only when the evidence architecture changes with it.

The chief executive problem is a delegation problem

Recruiting a substantive chief executive is an obvious part of normalisation. A stable executive can coordinate staff, execute board decisions and carry responsibility that should not remain with an interim group. But the choice of individual is less important than the design of delegated authority.

During calm periods, broad delegation appears efficient. Registry work contains thousands of routine acts, and directors should not micromanage them. The danger is that high-consequence acts can be described as routine administration. Account suspension, resource deregistration, certificate action, disputed contact changes, litigation instructions and exceptional payments may all pass through executive channels before the board sees their combined effect.

The new chief executive should receive a written authority schedule. It should distinguish routine service, reversible protection, material commercial commitment, contested rights decision, policy interpretation and emergency action. Each class should carry financial and operational thresholds. Some acts require a second executive. Some require committee notice. Some require prior board approval. Some require independent review before becoming final. Emergency authority should expire and generate an automatic retrospective record.

This is not distrust of the incoming executive. It protects that person from inheriting impossible ambiguity. A chief executive should not have to decide whether an old practice was lawful based on corridor memory. Staff should not have to guess whether a judicial direction overrides a board delegation. Members should not learn the boundary only after an adverse action.

Interim management also needs a clean handback. Every credential, contract, pending decision, exception, legal instruction and unresolved risk held by an interim committee should be inventoried and accepted by the substantive executive under board oversight. Otherwise the title changes while shadow authority persists.

Staff continuity should become documented capacity

The people who kept AFRINIC services operating during governance uncertainty hold valuable knowledge. Reform that treats all continuity as contamination would be reckless. Registry databases, member histories and technical services contain context that cannot be reconstructed from public pages.

The risk is key-person government. An employee may be the only person who remembers why an organisation name differs from a contract, which court document affects an account, how a legacy allocation was inherited or why a technical exception was granted. This knowledge allows service to continue. It also makes the institution dependent on unrecorded judgment.

AFRINIC should convert tacit knowledge into institutional evidence through paired review, written case histories, role-based access and succession tests. A second qualified person should be able to reproduce a material decision from the record. Sensitive evidence can remain restricted, but the existence, custodian, validation and governing rule should be visible to authorised reviewers.

Privileged access should be rebuilt from role, not inherited by convenience. The restored board should require a complete access census covering production systems, member data, billing, document stores, source repositories, domain and certificate controls, cloud accounts, bank portals and legal archives. Every access right should identify owner, purpose, approver, last use and review date. Shared accounts and dormant credentials should be eliminated or tightly contained.

Change control matters most at the boundary between data and rights. An edit to a contact field may be routine; an edit that changes who can authenticate a transfer or create a route authorisation is not. The system should classify the consequence rather than rely on the screen used to make the change.

The goal is not bureaucracy for its own sake. It is replaceability. A resilient registry can survive the departure, conflict or incapacity of any one competent person without losing the reasons behind its state.

The member register remains a constitutional dependency

The September 2025 election restored a board through member voting, but the electorate depended on AFRINIC records. Membership class, agreement status, fee standing, corporate authority, designated voter data and platform access had to converge. The June election had been annulled after reported suspicions of irregularities, particularly around powers of attorney, while investigations had not reached final conclusions. The replacement design strengthened direct designation and verification.

Those changes were important. They did not make the membership register self-authenticating. The 2026 election guidelines used the September 2025 voter register as a baseline while inviting members who had not participated to register. Baseline reuse can reduce repeated burdens and preserve a settled electorate. It can also carry forward an unexplained omission or exception.

A clean restart would treat the successful election as proof that the register problem had ended. A control-based reform treats the election as the first audited state in a continuing chain. It preserves the provisional and final populations, every correction, the rule version, the cutoff, member notices, challenge outcomes, platform activation and tally reconciliation. Later reuse should cite changes from that state, not merely copy names.

The board also has an incentive conflict. Its authority came from the register it now supervises. This does not invalidate the election. It makes independent assurance important. An outside reviewer should be able to begin with the full legal member population and reproduce eligibility without relying on the preferences of current directors or candidates.

Member accountability requires more than a ballot. A member should know its status, who may speak for it, which fees or documents affect voting, how to correct an error and what record proves the correction. Aggregate denominators should show how many members were eligible, designated, activated, challenged and able to vote. A board elected through a reproducible chain is stronger than one protected by celebratory language.

The registry ledger must not inherit political victory

The central operational danger in any reform is to let control of the company become control of the truth. An elected board has corporate authority within the constitution. It does not acquire freedom to make number-resource history conform to its political interpretation.

The registry should therefore preserve a distinction between governance state and resource state. Governance state records directors, delegations, committees and authorised executives. Resource state records holder identity, status, contacts, transfer history, disputes, security assertions and operational delegations. One can affect the other through lawful decisions, but no governance transition should silently rewrite resource facts.

This requires a last-verified-state rule. When authority is disputed, the registry preserves the last state that passed defined verification. Protective locks may prevent irreversible change. A lock should identify scope, reason, decision-maker, review route and expiry. It should not become an unreasoned punishment that disables unaffected functions.

Board members should not have direct power to edit production records. Executive staff should not be able to convert political instruction into a high-impact record change without recorded legal and evidentiary review. Auditors should be able to trace every privileged action. Resource holders should receive notice when their authority, security or public record changes, subject to narrowly defined security exceptions.

The design also protects the board. Directors cannot credibly be blamed for every data action if they establish lawful delegation, monitor exceptions and refuse private intervention. A clean separation lets directors govern the institution without becoming informal registry operators.

Reform succeeds when a losing candidate, adverse litigant or unpopular member can trust that political defeat will not alter the ledger without a separate, reviewable basis.

Authority should be mapped act by act

AFRINIC's recovery needs an authority register more precise than an organisation chart. The chart shows reporting lines. The register shows power.

For every high-consequence act, it should identify the legal source, authorised role, necessary evidence, approvals, notice, technical execution, audit record, review body and continuity treatment. The list should include member admission and termination; voter eligibility; executive appointment; litigation and settlement; bank mandates; resource-holder changes; transfers; deregistration; account suspension; RPKI certificate issuance and revocation; reverse-DNS changes; emergency credential recovery; release of backups; public statements about disputed status; and activation or termination of external support.

The register should expose combinations that create excessive power. A person who can validate evidence, approve a decision and execute the technical change controls the whole chain. A person who can alter a member record and voter designation can affect the principal that supervises the institution. A person who can instruct counsel and approve payment can sustain litigation without an independent budget check. Separation is meaningful only at the level of these combinations.

Some powers should require positive concurrence. Others should be subject to veto by a control function. A security officer may stop a dangerous change without deciding the merits. An independent reviewer may preserve status without becoming a permanent administrator. A board committee may require escalation without directing the technical outcome.

The authority register should also show absences. If no lawful actor can decide a contested request during board incapacity, the answer should be a published preservation rule and a route to temporary independent authority. It should not be an improvised assumption that whoever remains available inherits all powers.

This document would do more for reform than a promise of better leadership because it can be tested against real acts.

Remedies determine whether incentives changed

An institution behaves differently when error can be corrected before it becomes irreversible. AFRINIC's reform will remain cosmetic if members and resource holders must still choose between accepting an adverse administrative act and funding emergency litigation.

The first remedy should be a reasoned internal reconsideration by someone who did not make the initial decision. It should have access to the complete decisional record and power to correct ordinary error. High-consequence cases need an independent route with authority to preserve continuity, order focused evidence and issue reasons. Courts remain necessary for coercive relief and legal questions, but they should receive a narrowed record rather than an operational emergency created by absent internal review.

The remedy should follow the affected function. A billing dispute should not automatically threaten resource registration. A contested transfer should not disable unrelated reverse DNS. A suspicious credential should trigger protective authentication without deciding holder entitlement. A membership-election dispute should not rewrite resource control. This is claim isolation: attach the hold to the proven risk rather than to the whole relationship.

Time is part of the remedy. A decision delivered after a transaction fails, a certificate expires or customers migrate may be formally correct and practically useless. AFRINIC should publish response and review targets by consequence, not only average ticket closure. Emergency continuity decisions need hours or days; complex merits can take longer if the last verified state remains protected.

Correction should leave a trace. The old state, new state, reason, approver and downstream effects should be preserved. Reversal data should feed governance. A high reversal rate in one decision class is evidence about training, rules or incentives, not merely individual mistakes.

When remedies are real, leadership matters less. That is a feature. Good directors should want an institution capable of correcting them.

Financial repair must distinguish resilience from institutional defence

Delayed audits and substantial legal exposure make finance central to AFRINIC's recovery. A clean-restart story can turn the first approved accounts into a symbol of restored normality. The more important question is whether the accounts show how money supported continuity, litigation, reform and ordinary service.

Members need a functional expenditure map. Staff, systems, security, registry operations, audit, elections, legal cases and recovery projects should be distinguishable. Legal costs should identify matter categories and decision authority without exposing privileged strategy. Election costs should be separated from ordinary meetings. One-time recovery costs should not disappear into a permanent operating baseline.

Reserves should have purposes and draw conditions. Funds needed to keep records, authentication, reverse DNS, RPKI, communications and essential staff operating during a banking or governance shock should be protected from ordinary institutional ambition. Legal contingency should be visible and bounded. A board should not be able to describe every expenditure defending its preferred authority as registry continuity.

Procurement and payment rights also need separation. The same person should not select counsel, approve an open-ended mandate and release every invoice without independent review. Emergency payments may need rapid treatment, but they should expire into normal oversight. Vendor continuity plans should identify who can pay when bank access or signatory authority is disputed.

The delayed financial years are an opportunity to reconstruct decision trails, not only produce opinions. Exceptions, related parties, legal commitments, unreconciled balances and post-balance-sheet risks should be explained at the level needed for member judgment. An audit opinion supports confidence in the statements. It does not prove that every governance choice was wise.

Financial normality arrives when future directors cannot finance an institutional crisis invisibly, even if members trust them today.

External support needs limits as well as competence

ICANN, the Number Resource Organization, other RIRs, courts, government bodies, advisers and service providers have all had reasons to care about AFRINIC continuity. Their expertise and assistance can protect systems that one distressed company cannot safely maintain alone.

External support also creates authority risk. A peer registry may be able to operate a service without having the affected operator's contract. ICANN may express concern about compliance and backups without becoming AFRINIC's board. A government may seek continuity without acquiring power to choose directors or resource holders. A court may preserve assets without deciding technical policy. The new board may welcome assistance without being able to transfer powers it does not possess.

Every support arrangement should therefore identify trigger, function, data access, decision limits, funding, liability, duration and exit. Technical access should be no broader than necessary. Advice should not be described as command. Mutual aid should not become permanent policy control. Temporary services should return a complete change history.

This discipline also prevents the restored board from using international recognition as a substitute for member accountability. Support from established Internet institutions can indicate confidence or practical cooperation. It does not prove that every internal reform is legitimate. Conversely, member voting cannot make an unsafe technical arrangement sound. Different sources of authority answer different questions.

The best external support makes AFRINIC more independently governable and its records more portable. Support that deepens dependence merely changes the identity of the gatekeeper.

Reform should be tested against adverse scenarios

Institutions often validate reform with ordinary cases. A cooperative member updates a contact. A routine invoice is paid. A board meeting reaches quorum. A backup is restored in a technical exercise. These tests are useful and too easy.

AFRINIC should test the redesigned controls against the cases most likely to divide its leaders. A large resource holder disputes an adverse decision while litigation is pending. A director has a relationship with a party to a transaction. A chief executive leaves without notice. A bank delays access. A court order is urgent but ambiguous about technical consequences. A privileged credential appears compromised. A member claims exclusion days before voting. A backup differs from the live record. An external partner offers emergency operation under a mandate that does not cover contested changes.

For each scenario, observers should be able to identify the authoritative state, decision-maker, evidence, permitted protective act, forbidden act, review path, communication duty and restoration condition. The exercise should include disagreement. A design that works only when everyone interprets the facts alike is not a crisis design.

Results should be reported without exposing security secrets. Members need to know whether the institution can preserve services, isolate disputes, recover credentials, reconcile records and hand authority back. Failed tests should produce dated corrective actions and accountable owners.

The decisive test is reversibility. Can a mistaken protective act be undone? Can the complete prior state be proved? Can affected operators continue while merits are decided? Can a temporary administrator leave without retaining hidden access? Reform that survives these questions changes incentives because every actor knows unilateral action will be visible and containable.

A staged repair is safer than ceremonial rebirth

AFRINIC does not need to suspend ordinary service until every reform is complete. That would punish operators for institutional weakness. It needs a staged repair whose early steps reduce the most consequential risks.

The first stage is preservation: complete access census, verified backups, change-log protection, legal-hold discipline, inventory of pending high-impact cases, bank-signatory continuity and publication of the current authority schedule. This creates a stable base without deciding every historical controversy.

The second stage is reconciliation: member and voter records, resource-holder authority, delayed audits, contracts, litigation matters, privileged accounts and unresolved exceptions. Reconciliation should classify certainty rather than force false finality. Verified, disputed, pending evidence and preserved by lawful order are different states.

The third stage is remedy: reasons, reconsideration, independent review, continuity holds, deadlines and a record of reversal. This changes the experience of present members before constitutional reform is complete.

The fourth stage is constitutional adjustment: bylaws, committee mandates, vacancy rules, non-delegable powers, emergency authority and handback. Drafting should be informed by the earlier control map rather than abstract governance fashion.

The fifth stage is external assurance: independent tests of backups, access, election registers, high-impact decisions, financial controls and service substitution. Assurance reports should explain scope and limitations.

The sixth stage is portability and failover: operators receive verifiable records, emergency service can be substituted under a lawful mandate, and return or succession can occur without rebuilding history. This stage makes institutional failure survivable.

The order matters. Ceremonial rebirth announces trust first and asks for evidence later. Staged repair produces evidence from the beginning.

What the clean-restart instinct gets right

The clean-restart instinct is not wholly irrational. Institutions need moments of closure. Staff cannot work indefinitely under suspicion. Directors need room to decide. Members may reasonably want to stop litigating the past. Constantly relitigating every historical act can paralyse service and turn reform into factional punishment.

New leadership can create practical discontinuity. Directors without responsibility for old positions may ask questions incumbents avoided. A new chief executive may reassign duties, release information and end personal conflicts. Committees can reopen delayed work. A new electoral mandate can give the company a legitimate corporate organ that temporary administration could not supply.

The mistake is turning this possibility into a control. Good intentions are not auditable. Independence is not proved by novelty. A new person can be captured by the same information asymmetry, incentives and powers. They may make different decisions, but the institution remains fragile if outcomes still depend on personal restraint.

Reform should therefore protect the new leaders from impossible expectations. They should not be asked to certify every historical fact by taking office. They should be required to classify uncertainty, preserve evidence and commission review. They should not be expected to guarantee that no dispute will occur. They should be expected to ensure disputes cannot disable unrelated operations. They should not promise a perfect past. They should publish a controllable present.

Closure is legitimate when it follows verification, remedy and explicit residual uncertainty. It is dangerous when it means forgetting which powers failed.

The measure of recovery is ordinary dependence without institutional fear

AFRINIC serves networks that should not need to understand every Mauritian corporate case or election rule to maintain accurate number-resource records. The ultimate purpose of reform is to make ordinary dependence safe and unremarkable.

That condition has practical signs. A member can identify its recorded rights and representative. A resource holder can obtain a verified history. A buyer or lender can understand whether a record is disputed. A staff member can act from a written authority schedule. A director can see material exceptions. An auditor can reproduce privileged changes. A court can issue a narrow order with a clear technical translation. A temporary provider can preserve service without acquiring policy power. A chief executive can leave without taking institutional memory. A losing candidate can challenge the election without threatening the ledger.

AFRINIC's public steps since the September 2025 election - committees, audits, executive recruitment, engagement and bylaw review - are capable of contributing to that future. They should be assessed as inputs, not conclusions. The board's most credible message would not be that the crisis is over because the board exists. It would be that no future board, including this one, can exercise the old combination of opaque power and weak remedy.

The clean restart says the institution has changed because its leaders changed. Durable reform says the institution has changed because authority is narrower, evidence travels with decisions, operators retain continuity and mistakes can be corrected before they become network events.

AFRINIC does not need institutional amnesia. It needs institutional memory strong enough that no faction owns it and institutional controls strong enough that no faction needs to be trusted without verification.

Sources and analytical limits

AFRINIC's 2020 constitution is used for the company's member, board, chief-executive, delegation, meeting and committee structure. It establishes formal powers and does not prove how every power was exercised in a particular case.

The Supreme Court of Mauritius judgment in Benjamin Adzenyamebeye Eshun v African Network Information Centre (AfriNIC) Ltd, 2023 SCJ 63 supports the board-quorum and member-rights context. The Court of Civil Appeal judgment in African Network Information Centre (AfriNIC) Ltd v Cloud Innovation Ltd and another, 2024 SCJ 473 supports the later receivership and authority context. The article does not treat either judgment as a general verdict on all AFRINIC governance.

The receiver's April 2025 election communication, 15 July 2025 notice, replacement election guidelines and voter-register publication establish the attributed mandate, June annulment, absence of final investigative conclusions at that date, replacement design and published electorate mechanics. They do not independently establish misconduct, outcome effect or the accuracy of every private record.

The October 2025 joint board-receiver communique is used for the board committees, delayed 2022-2024 audits, pending legal work and bylaw-review commitment. AFRINIC's March 2026 member update and 2026 election guidelines are used for the board's account of current priorities and reuse of the September 2025 voter-register baseline. These are institutional statements, not independent assurance that every reform objective was achieved.

No complete privileged-access inventory, board packet, staff delegation schedule, legal matter ledger, member database, resource-change history, audit work paper, bank mandate, election evidence file or sealed court record was available for this analysis. The proposed authority register, committee covenants, staged repair and adverse-scenario tests are governance recommendations. They are not findings that a named current director or employee has acted improperly, and the article does not infer individual responsibility from office, access or association.