Summary

  • Serge Kabwika Ilunga died on 23 December 2021 while serving as an AFRINIC director and as one of three named members of the Audit Committee.
  • AFRINIC declared his Board seat vacant, directed that the company record be updated, and in January 2022 used Resolution 202201.669 to reconstitute the committee.
  • The strongest reading of that act is straightforward: it replaced a stale public committee state with a current one while the Board could still act.
  • The deeper weakness is temporal. AFRINIC’s bylaws required a majority of directors, and never fewer than five, for a Board-meeting quorum; a Board-dependent succession route can therefore become unusable when vacancies make succession most urgent.
  • A durable rule would trigger early, preserve records and access through a narrow temporary custodian, require a quorate appointment and recorded handover, expire quickly, and return ordinary authority through a member-controlled route.

On 23 December 2021, a governance diagram became a human absence. Serge Kabwika Ilunga died while serving on AFRINIC’s Board. He also appeared on the then-current three-person Audit Committee roster alongside Benjamin Eshun and Prof Habib Youssef. The company recorded the death, declared Board Seat 4 — Central Africa vacant, and directed that its record at the Registrar of Companies be updated. In January, Resolution 202201.669 reconstituted the Audit Committee.

That sequence matters more than the administrative sparseness of the published record. A named committee cannot sensibly continue to present a deceased director as one of its current members. Papers still need a destination. Questions still need an accountable recipient. Existing work, if any, must not become attached indefinitely to an officeholder who can no longer receive it. A numbered public act is therefore better than silence or the fiction that the old roster somehow continued unchanged.

The useful question begins where that benign account ends. Reconstitution depended on the Board. AFRINIC’s bylaws put management and supervision of the company’s affairs under that Board, and article 19.6 required a quorum consisting of a majority of directors and, in every case, at least five. The available evidence does not show that the Board lacked quorum in January 2022. Nor does it establish the exact day, incoming committee members, vote, chair or handover. What it does expose is a structural countdown: if a Board must repair a committee vacancy, the repair must occur while enough directors remain able to make it.

The succession clock, in other words, has to beat the quorum line.

The humane case for a short public resolution

Governance commentary can become bloodless when it treats death as a defect in an organisation chart. That would be especially misplaced here. The first fact was a person’s death, with consequences for his family, colleagues and community that are not reducible to a vacancy. Resolution 202112.661 separately recorded that Ilunga died while serving as a director and authorised a funeral contribution. That contribution says nothing about committee succession and should not be recruited as evidence about it. It belongs to the company’s response to a human loss, not to an argument about committee mechanics.

The corporate consequences nevertheless had to be addressed. Resolution 202112.660 declared his seat vacant and directed an update to company records. That was one state change. The Audit Committee roster created another problem because it still included him. The June 2021 roster had named Serge Ilunga, Benjamin Eshun and Mark Elkins. After one member resigned, Resolution 202108.628 reconstituted the committee in August as Benjamin Eshun, Serge Ilunga and Prof Habib Youssef. The December death therefore did not merely reduce the Board. It also made the August committee roster stale.

Swift reconstitution has an obvious and defensible purpose. Staff who hold papers need to know where they may send them. Directors need to know who is expected to receive committee matters. External advisers need a current point of contact. Members have a legitimate interest in knowing that a public roster no longer describes an impossible present. A short resolution can update that public state without disclosing confidential deliberations or the substance of any matter before the committee.

There is also value in the form. A numbered act gives the change a place in the company’s record. It marks that the change was not an informal assumption made by staff or an indefinite improvisation by an executive. It creates a reference point against which later records can be compared. Even where the public notice is concise, the act says that the corporate state was affirmatively changed.

That is the strongest case for Resolution 202201.669, and it should be stated without qualification before asking more of it. The resolution was a useful private corporate repair. It prevented a known stale committee roster from being left as the last public word. It showed the practical virtue of using ordinary authority while that authority remained available. Nothing in that conclusion requires claiming that the committee performed well, that any particular work was completed, or that the change affected networks, customers or markets.

Six states that should not be collapsed into one

The phrase “reconstituted the Audit Committee” sounds like a single action. Operationally, it sits in the middle of several different states. Separating them is essential because a public appointment record can prove one transition while leaving others unknown.

The first state is the death of a serving director. It is an irreversible fact about a person and an immediate fact about officeholding. The second is the formal vacancy of the Board seat and the instruction to update company records. Those two states are connected, but they are not identical. A death occurs; the company must then recognise its corporate consequences in the forms that its own rules require.

The third state is the staleness of a committee roster that named that director. Staleness is not a moral judgment and does not imply misconduct. It simply means that the recorded membership no longer matches reality. Nor does staleness tell us whether the remaining members continued to hold their positions, whether committee rules allowed them to meet, or whether any matter was pending. It identifies a mismatch, not its full operational consequences.

The fourth state is a new decision by a capable Board. Resolution 202201.669 occupies this point in the sequence. It changed the committee’s private corporate state. Yet “capable” is doing important work in that sentence. The relevant organ needs whatever authority, notice, eligibility and quorum its rules demand. A succession clause that names the Board as appointing actor assumes that the Board can assemble and decide.

The fifth state concerns custody: papers, access, open work and escalation channels need controlled destinations. These do not move merely because a public list changes. A document repository may have access controls. A mailbox may still route to an old account. An external adviser may hold an outdated contact list. A pending question may have no recorded owner. These are examples of the kinds of transitions a serious succession process must address; they are not claims about what happened at AFRINIC.

The sixth state is restoration. Temporary arrangements should end when ordinary authority returns or when the regular appointment is effective. Without an expiry point, a narrow continuity measure can harden into an alternative chain of power. Restoration is therefore not an afterthought. It is part of the original design.

Keeping these states separate prevents two opposite errors. The first is to treat a public resolution as proof of every hidden transfer: acceptance, conflicts review, access provisioning, custody, open-work handover, committee quorum and effective operation. The second is to treat the absence of those details from the public record as proof that no private record or handover existed. Neither inference is justified. Publication is evidence of what was published. Silence sets a limit on public knowledge; it is not a finding of failure.

Appointment is not acceptance, and a roster is not custody

Committee succession is often described through nouns: vacancy, member, chair, resolution. The risk lives in the verbs. Who detected the vacancy? Who disabled old credentials? Who preserved the record? Who accepted the appointment? Who disclosed a conflict? Who received open matters? When did the change take effect? When would an interim arrangement expire?

The published record establishes the act at month precision: Resolution 202201.669 reconstituted the Audit Committee in January 2022 after the death of a serving director. The available material does not preserve the resolution’s verbatim entry. It does not establish the exact January day, the incoming roster, a proposer or seconder, a vote, a chair, an effective time or a term. Filling those blanks would create false precision.

The omissions also show why institutional analysis must distinguish layers of proof. Appointment is the authorised selection of a person or set of people. Acceptance is evidence that the appointee agreed to serve. A conflicts record identifies interests that may constrain participation. Provisioning gives the person the access needed to perform a role. Custody identifies who is responsible for the integrity and availability of records. Work transfer identifies open matters and their new owners. Committee quorum determines when that group can act. Performance asks what it subsequently did and how well.

Those are related, but none is interchangeable with another. A person may be appointed but never accept. An appointee may accept but lack access. Access may exist while custody is undocumented. A roster may be current while open matters remain assigned to a departed office. Conversely, an organisation may have completed each private transition while publishing only the roster change. Responsible analysis should demand clarity about the distinction without pretending to possess absent facts.

This is particularly important for a committee whose work may involve sensitive company material. Public accountability does not require dumping confidential papers into view. It can be served by a compact transition receipt: the fact of appointment, recorded acceptance, a conflicts check, the effective time, confirmation that controlled custody transferred, a count or classified inventory of open matters, and an expiry or review point. Such a receipt can make the chain legible without revealing the substance of protected work.

The point is not bureaucratic abundance. It is the alignment of authority with liability. If a new officeholder receives access, recommendations or decision rights, that person should carry a duty proportionate to the control received. If an officeholder departs or dies, credentials and responsibilities should not remain invisibly anchored to that office. A title without the means to act is empty responsibility; control without a recorded duty is unaccountable power.

Why five matters even when the committee had three names

The decisive number in this story is not the size of the committee. It is the threshold for the organ that could reconstitute it. AFRINIC’s bylaw article 19.6 required a Board-meeting quorum of a majority of directors and, in any event, not fewer than five. Articles 19.7 and 19.8 provided one vote per director and a majority of votes cast by eligible Board members.

That framework reveals a dependency between two separate bodies. A three-person committee might lose one member, yet the remaining two do not automatically gain authority to choose the replacement. The Board’s appointment power does not move to them merely because they are closest to the work. Nor does it necessarily move to an executive or to the last available director. The committee’s need for continuity and the Board’s capacity to supply it are separate variables.

The evidence does not establish that AFRINIC’s Board was below quorum when it adopted Resolution 202201.669. The responsible inference is narrower and more useful: because the Board was the necessary actor, the succession path could operate only while a Board meeting could validly act under the company’s rules. Every vacancy that reduces the available Board can narrow that window. A mechanism that waits for incapacity before it triggers may be perfectly worded and completely unusable.

This is the timing paradox. The more severe the vacancy problem becomes, the greater the need for an orderly appointment; but the same vacancies can remove the appointing organ’s ability to make it. A rule that says “the Board shall appoint” describes authority in ordinary conditions. It does not, by itself, solve the case in which the Board can no longer reach its own threshold.

The lesson should not be inflated into a claim about what caused AFRINIC’s later company-wide difficulties. BTW’s account in The Board Without Quorum owns that later collision and its chronology. Here it serves only as a bounded stress comparator: an institutional power that depends on a functioning Board deserves to be designed before a Board becomes unable to exercise it. Resolution 202201.669 did not prove that the later crisis was inevitable, avertable, or caused by this death. It shows that ordinary authority was useful when it was used.

A registry is a bookkeeper, not a throne

The institutional stakes require precision because regional internet registry debates easily slip from technical importance into imagined public authority. NRS describes AFRINIC as a member-based registry. That character matters. The company coordinates technical records and services for its members and service region; it is not the government of that region.

Heng Lu’s ledger-not-throne analysis supplies the controlling distinction. The defensible functions are narrow and important: protect the uniqueness and integrity of number-resource records, support accurate registration and contactability, preserve security-relevant coordination, record recognised changes and keep necessary technical services available. Those functions may be consequential for operators. Their importance does not crown the institution that performs them.

AFRINIC possesses no sovereign, legislative, regulatory, police, prosecutorial, punitive, confiscatory or public-law adjudicative power. Its service region is not territory. A Board resolution is a private corporate act, not legislation. Its Audit Committee is a private corporate work surface. The committee cannot regulate a continent, punish an operator, confiscate property or adjudicate public rights merely because “Audit” appears in its name.

This boundary improves rather than diminishes the succession analysis. If the committee were treated as a miniature public authority, an emergency might be invoked to justify indefinite residual power. Once it is understood as a bounded company mechanism, the correct design becomes clearer: preserve necessary records and work, transfer limited access, document responsibility, and end temporary authority promptly. Continuity protects a function. It does not grant immortality to a roster, committee, Board or gatekeeper.

The distinction also keeps operator-facing consequences honest. LARUS has explained why predictable registry governance can matter to infrastructure operators. Unclear decision paths, uncertain records or discontinuous services may complicate planning even when routers keep forwarding packets. But that general risk channel does not prove that Resolution 202201.669 caused an outage, a financial loss, a customer effect or a market result. No such causal claim is supported here.

Predictability is valuable at a more modest level. An operator benefits when registry records have known custodians, state changes follow legible processes, and essential contacts do not vanish into organisational ambiguity. Members benefit when authority can be traced to a current officeholder. Staff benefit when they are not forced to improvise ownership after a vacancy. None of this requires attributing public power to the registry or technical effects to a committee resolution.

The strongest counterargument: perhaps the short act was enough

A fair reading must confront the simplest objection. Corporate resolutions are often brief because the detailed work happens elsewhere. The Board may record an appointment while staff complete access changes, counsel maintains corporate records, appointees give private acceptance, and the committee’s terms govern its operation. Publishing every step might expose confidential information or burden a public register with operational detail. Why demand more from a concise roster change?

This objection is strong. The public absence of a handover record is not evidence that no handover occurred. A short resolution can be the visible tip of a complete private process. Indeed, unnecessary publication can create its own risks by exposing identities, security practices or sensitive matters. Good governance is not measured by the number of pages released.

The answer is not maximal disclosure. It is minimal verifiability. A succession system should be able to demonstrate that certain transitions occurred, even if it does not disclose their protected substance. The public record might remain short while a versioned private receipt establishes acceptance, conflicts review, credential changes, record custody, open-work ownership and effective time. Members or an authorised reviewer could later verify that the process was completed without reading the underlying material.

More importantly, even a flawless private handover would not answer the timing dependency. Detail cannot cure incapacity after the fact. If the rule requires a quorate Board, the system must detect and act on vacancies before the Board loses quorum. The relevant criticism is therefore not that Resolution 202201.669 was too short. It is that any succession design resting exclusively on a future Board decision needs an earlier trigger and a bounded bridge for the case in which ordinary appointment is temporarily unavailable.

That distinction matters because it avoids converting transparency into theatre. The goal is not a longer resolution. The goal is a sequence that remains usable under stress and leaves enough evidence to assign responsibility.

A counterfactual built around restraint

Consider first the ordinary case. A serving director dies or otherwise leaves office. The vacancy trigger is recorded immediately. Credentials specific to the departed office are suspended automatically, without waiting for a discretionary finding. Records remain in immutable custody, meaning that preservation does not depend on a personal account or device. A response clock starts.

The clock should be short enough to prevent a stale roster from persisting, but long enough for the authorised Board to receive notice and act properly. Its purpose is not to force a predetermined choice. It is to make delay visible before delay becomes incapacity. The Board, while quorate, appoints the new committee state. Appointees record acceptance and conflicts. An open-work inventory identifies matters by protected reference and status rather than publishing their substance. The transition receives an effective time.

Now add a narrow temporary custodian. The role should be defined by function, not by a person’s prestige. Its authority would be limited to preserving records, maintaining access boundaries, receiving new papers, logging deadlines and routing urgent procedural notices. It could not issue new substantive findings, broaden the committee’s remit, dispose of matters, confer permanent access or appoint its own successor. The role would exist to prevent loss and confusion, not to govern through an emergency.

The custodian’s access should be least-privileged and logged. Where an existing credential was tied to a departed officeholder, suspension should be automatic. Necessary records should sit in a repository controlled by role and retention policy, not in a personal mailbox. Every transfer should produce a versioned receipt: what category of material moved, who released it, who received custody, when the transfer took effect, what access was disabled, and what remains unresolved.

Appointment would remain with a quorate Board under ordinary conditions. The temporary role would not become an alternative appointment channel simply because it holds the papers. This keeps custody distinct from substantive authority. The person able to preserve a file does not thereby acquire the power to decide the matter contained in it.

The hard case begins if the Board cannot act before the clock ends. Here, design must resist the seductive one-person escape hatch. Giving an executive, one remaining director or the custodian indefinite authority to constitute the committee would solve continuity by creating an unchecked power. That cure would be worse than the dependency it addresses.

A proportionate bridge would preserve only the minimum state and shift restoration to a member-controlled route already specified in advance. Members could be called through a defined process to restore an appointing body or confirm a narrowly framed interim arrangement. The precise legal form would need to conform to the company’s governing instruments and applicable law; this article offers no conclusive company-law opinion. The institutional principle is that incapacity should activate collective restoration, not crown a lone residual officeholder.

Every temporary authority should have a short expiry. It should end automatically upon a valid Board appointment, member-controlled restoration or a fixed time limit, whichever arrives first. Extension should require a new, reviewable act rather than silent rollover. After ordinary authority returns, the temporary custodian should deliver the receipt, surrender exceptional access and have that closure recorded.

The counterfactual is deliberately modest. It does not promise uninterrupted substantive committee work in every circumstance. It protects custody, makes delay visible, preserves the possibility of later action and prevents temporary control from becoming permanent. That is what continuity should mean for a private technical coordinator.

What the January act can and cannot teach

Resolution 202201.669 teaches that stale institutional states require affirmative correction. The progression from the June roster, to the August reconstitution after a resignation, to the December death record, and then to the January reconstitution shows that committee membership did not simply update itself. A Board decision was required to establish a new state.

It also illustrates the value of acting before institutional capacity is in doubt. This is an inference from the sequence and the quorum rule, not language attributed to the resolution. The company had a mechanism, and the mechanism was used. The same mechanism would lose practical value if its required actor later became unable to assemble the minimum quorum.

The act does not disclose the incoming roster in the material available for this analysis. It does not prove acceptance, a conflicts process, committee quorum, document custody, credential changes, transfer of open work, contact with external advisers or later performance. It does not show that any matter was caught, missed, caused or prevented by the committee. BTW’s separate analysis of internal-audit effectiveness owns questions of scope, data access, escalation and warning performance; importing those questions here would confuse succession with performance.

Nor does the act prove any operational outcome for the internet. It did not legislate, regulate, police, prosecute, punish, confiscate or adjudicate. It changed a private company committee. The responsible significance lies in corporate continuity and accountable custody, not in invented claims of continental authority or network causation.

Finally, the absence of public detail cannot bear more weight than it can carry. It is proper to say that public proof stops at the published acts and rules. It is improper to convert that limit into an accusation. The most useful response to an unknown is a better evidence design for the future: a compact, protected transition receipt and a clear line between appointment, access, custody and performance.

Continuity belongs to the work, not the officeholder

The temptation after a vacancy is to preserve the institution’s appearance. Keep the committee name, fill the empty chair, update the webpage, move on. Appearance matters because stale information can mislead. But the deeper continuity object is neither the chair nor the roster. It is the bounded work that should survive a change in personnel: preserved records, known custody, controlled access, traceable decisions and an available escalation path.

Heng Lu’s function-versus-gatekeeper distinction is especially useful here. Accurate registry records, contact services, reverse DNS, RPKI, transfer history and operationally safe state changes may deserve continuity. The particular company, Board, committee or officeholder touching those functions does not become immortal because the function matters. Inside the company, the same logic applies. Committee work may need an orderly transition; no particular committee membership has a right to endure.

This approach avoids two forms of institutional mythology. One imagines that replacing names on a roster guarantees functional continuity. The other imagines that preserving a function requires preserving the existing gatekeeper and all of its powers. Neither is true. A function can pass to a new, bounded and accountable arrangement. A gatekeeper can change while records and services remain protected.

Authority-liability alignment provides the test. The person who can open the repository, receive the papers or make a recommendation must be identifiable and responsible. The person whose appointment has ended should not retain unexplained access. The body that appoints should have valid authority to do so. Temporary control should end visibly. When these conditions are met, succession is more than nominal. When they are absent, a fresh roster can conceal a broken chain.

The narrow lesson of a narrow resolution

It is easy to ask Resolution 202201.669 to carry a grand story about AFRINIC. It should not. The act neither proves institutional success nor supplies a diagnosis of later failure. It records a specific repair after a specific death. Its value lies precisely in that narrowness.

The Board had previously reconstituted the Audit Committee after a resignation in August 2021. When Ilunga died in December, the resulting vacancy affected both a Board seat and the current committee roster. The company separately recognised those states, then reconstituted the committee in January. That sequence replaced assumption with an authorised act.

Yet the bylaw threshold exposes the edge of the mechanism. A Board-dependent repair is available only while the Board can act. Waiting until the appointing body crosses the quorum line turns a succession rule into an instruction addressed to an actor that cannot answer. The protection must therefore begin earlier: detect the vacancy, suspend office-bound credentials, secure records, start the clock, convene the authorised decision, record acceptance and transfer, and set an expiry.

If ordinary authority cannot be restored in time, the bridge should remain narrow and collective. A custodian can protect papers without deciding their substance. Members can have a predefined route to restore the appointing structure without pretending that an executive or residual director has inherited public power. Review and automatic expiry can prevent an emergency arrangement from becoming a permanent constitutional shortcut.

This is not a call for more authority. It is a call for authority to arrive earlier, carry matching responsibility and end on schedule. That is the difference between continuity and opportunism.

The January resolution was useful because it occurred on the usable side of the dependency revealed by AFRINIC’s own quorum rule. The public evidence does not establish the exact mechanics of its adoption or implementation, and it does not need to. The enduring lesson is temporal: succession is not a clause waiting patiently in a rulebook. It is a sequence that must begin while someone with legitimate private authority can still complete it.