Summary

  • Resolution 202105.610 separated three acts that are easy to blur: Board approval of the submitted 2020 statement set, signature of that set by two authorised officeholders, and signature of the associated representation letter by the same two people.
  • Naming Eddy M. Kayihura as CEO and Subramanian Moonesamy as Chairman made accountability concrete and avoided an anonymous or open-ended delegation; the strongest reading is that this was a sensible authorisation for an immediate corporate task.
  • The public record reviewed here does not identify the exact approved file, signature dates, completion state, expiry, substitute, or successor mechanism. Those absences justify a continuity question, not an allegation that the documents were unsigned or that no internal records existed.
  • A durable design would preserve named responsibility while adding a privacy-safe completion receipt and a narrow rule requiring fresh recorded confirmation if vacancy, incapacity, or turnover intervened.

L3 — Two names below one Board vote

The decisive feature of Resolution 202105.610 is not a figure in the accounts. It is the movement from a Board’s collective approval to action by two identified people. In May 2021, AFRINIC’s Board approved, as submitted, the financial statements for the year ended 31 December 2020. It then named Eddy M. Kayihura, in his capacity as Chief Executive Officer, and Subramanian Moonesamy, in his capacity as Chairman of the Board, and authorised them to sign the statements on behalf of the Board. The resolution also recorded a second signature authority: the same two officeholders could sign the associated letter of representation.

Those sentences describe three institutional acts, not one. First came approval: the Board accepted the submitted statement set. Second came execution of the statements: the named CEO and Chairman were authorised to place the Board’s signatures on that approved set. Third came execution of the representation letter: the same people were authorised to attest through a separate document associated with the statements. A reader who compresses all three into “the accounts were approved” loses the very feature that makes the resolution useful. Approval answers what the collective body decided. Signature answers who could execute that decision.

The representation letter records another channel of responsibility, even though its contents are not established by the evidence reviewed for this article.

This separation matters because corporate action is rarely self-executing. A Board can resolve that something should be done, but a company still needs human agents to authenticate documents, communicate decisions, and complete formal steps. Names identify the agents. Offices explain why those people, rather than any other employees or directors, were entitled to act. The CEO stood on the management side of the attestation chain; the Chairman stood on the governance side.

By pairing them, the resolution made it possible for members, auditors, and later officeholders to locate responsibility at the point where a Board decision became an executed document.

The strongest benign case is therefore strong indeed. A short resolution naming the incumbent CEO and Chairman may have been exactly the right tool for an immediate, bounded task. It did not say that any future CEO, any future chairman, or any unnamed delegate could sign at will. It did not create an indefinite class of authorised actors. It placed the job with two recognisable people who occupied the offices most closely connected to management and Board responsibility. Compared with an authorisation addressed vaguely to “management” or “the company,” that precision reduces ambiguity.

Compared with a permanent delegation to whoever might later claim an office, it can also reduce the risk of authority drifting beyond the Board’s immediate intention.

Nothing in the available record establishes that the resolution was illegal, suspicious, or defective. Nor does the absence of a public completion record establish that the statements or representation letter went unsigned. The immediate task may have been completed without delay and in the conventional way. Internal minutes, an auditor’s file, a signed copy, or a corporate register may contain details that were not visible in the official resolution register. Responsible analysis must distinguish what the public record does not say from what the institution did not do. They are not the same proposition.

That distinction begins with the date. The official register places Resolution 202105.610 in May 2021. The archived register links meeting records for 11 May and 26 May, but the excerpt reviewed here does not determine which meeting adopted this resolution. There is no sound basis to invent a day, proposer, seconder, tally, attendance list, or sequence of debate. “May 2021” is the warranted precision. More detail would look exact while being less truthful.

The same discipline applies to the documents. The resolution says that the financial statements were approved “as submitted,” but the available public record does not identify an approved-file hash or another immutable version marker. The official annual-report package identifies the reporting material associated with the year ended 31 December 2020, yet this analysis does not infer from that association a particular signature timestamp or a cryptographic identity between every relevant file.

There is also no established basis here to quote or reconstruct the representation letter, to describe its assurances, or to say when or whether each named person signed it.

These are not merely clerical niceties. A resolution can authorise a signature only with respect to some document. If the record identifies the document solely by a general description, a later reader may have to compare dates, filenames, publication histories, and copies to determine whether the signed item was the version that the Board actually approved. Most of the time, that reconstruction may be easy. When officeholders change, records move, or a dispute arises years later, the cost grows. A simple version identifier at the moment of approval would turn a probable match into a verifiable one.

Completion has the same structure. Authority and execution are different states. Resolution 202105.610 proves that the Board authorised Kayihura and Moonesamy to sign. It does not, standing alone, prove that either or both signatures were completed, that both were mandatory, or that the associated representation letter reached its intended recipient. It also does not prove non-completion. The correct public conclusion is deliberately modest: the resolution records a clear route to execution, while the reviewed register does not close the loop with execution evidence.

For AFRINIC’s members, that loop matters because the organisation is a private membership registry, not an abstraction. NRS’s institutional account makes the member-facing dimension clear: accountability should be legible to the constituency whose registry depends on the company’s continuity and record-keeping. Members need not receive confidential management representations or replicas of private signatures to know that an annual attestation step was completed.

They can reasonably be told which approved statement version was involved, which offices were authorised, whether execution occurred, and whether any subsequent office change required renewed authority.

Heng Lu’s agency analysis helps explain why this is more than a document-management preference. A collective body delegates action to human agents, while members and network operators can bear the institutional downside if authority becomes hard to verify. Alignment requires a traceable answer to six practical questions: who authorised, who acted, what document was involved, through which office, during what period, and with what evidence of completion. Resolution 202105.610 answers several of those questions well. It identifies the authorising body, the intended actors, their offices, and the two classes of document.

The remaining questions concern version, time, completion, and what happens if the human actor and the office cease to coincide.

The later institutional history makes that last question vivid without changing the meaning of the 2021 act. AFRINIC experienced a Board-quorum crisis in 2022–23, a history examined separately by BTW. Other separately examined events concern a different delegation tied to a named CEO and a later judicial treatment of turnover. Those episodes do not show that Resolution 202105.610 remained open, was used after turnover, failed, or contributed to the later crisis.

They establish a narrower and more useful proposition: at a private institution, authority that depends on named, functioning offices should be designed to remain intelligible when those offices cease to function normally.

LARUS brings the operational reason into view. Network operators rely on registry continuity, and unclear governance records can increase planning, verification, and diligence costs even when no outage or financial loss is proved. The relevant harm mechanism here is not a claim that a signature issue interrupted infrastructure. No such claim is supported. It is the quieter cost of uncertainty: staff and advisers may have to reconstruct whether a corporate act was completed, whether an incumbent’s departure mattered, and whether a successor could lawfully finish a pending step.

Better records lower that cost before a stressful transition makes it urgent.

BTW’s earlier analysis of another year’s accounts also supplies a useful boundary. Board responsibility, statement signature, management representation, external audit, and member-facing action can be separate assurance layers. This article does not repeat that broader account-assurance framework or evaluate audit scope. Its concern is narrower: within the 2021 resolution, the Board itself distinguished approval from the two signature authorisations. The fact that the same two people were named for both documents does not collapse the documents into one, and it does not turn authorisation into proof of execution.

There is an equally important boundary on the institution’s power. AFRINIC’s Board had private corporate authority to approve the company’s statements and appoint agents to sign them. That is real authority within a company. It is not sovereign, legislative, regulatory, police, punitive, or confiscatory public power. The six absences matter because registry administration is sometimes narrated in language that inflates coordination into government. Resolution 202105.610 did nothing of the kind. It made two officers responsible for two private attestation tasks.

It neither legislated for network operators nor conferred coercive state power on the CEO or Chairman.

Heng Lu’s bookkeeper frame is especially apt. A regional Internet registry performs valuable technical bookkeeping and coordination: it maintains records, supports uniqueness, and helps operational systems remain orderly. Good bookkeeping requires faithful corporate records as much as accurate registry records. The modesty of that description is a strength. One need not call AFRINIC a regulator or sovereign to insist that its own approval and signature trail be durable. Indeed, the case for a completion receipt is stronger when framed as disciplined private administration rather than as a claim of public-law supremacy.

At the visible level, then, Resolution 202105.610 did several things well. It recorded an approval, bounded the reporting period, named both people, stated both offices, identified the statement-signing function, and separately mentioned the representation letter. It created a bridge from a Board vote to executable responsibility. What it does not publicly reveal is the final state of that bridge: the exact file that crossed it, the dates of execution, whether both signatures were required and supplied, any auditor dependency, or the event that closed the authority. That is the point from which the deeper continuity question begins.