Summary

  • AFRINIC's annual general meeting in Dakar on 24 May 2004 was reported to have endorsed and adopted the company's whole proposed constitution. The surviving consent formula is only “all members present”: it supplies neither a count nor the denominator of eligible voters.
  • The certified final text, draft-to-final redline, eligible-member register, notice pack, exact resolution, quorum record, proxies, abstentions, voting method, chair's declaration, authenticated minutes and Registrar receipt are not in the available record.
  • A separate decision in which five full members present approved a temporary election-rights measure, against a reported total of seven registered full members, is not the constitution vote and cannot supply its tally.
  • If validly adopted, the constitution ordered private rights and duties within a Mauritius company. It did not make AFRINIC a sovereign, regulator, police force, prosecutor or confiscatory authority, and it did not authenticate a mandate from Africa's states, networks, operators or end users.
  • The useful remedy is evidence reconstruction: publish the exact instrument, its version lineage, the eligible electorate, the member act and the filing record while preserving registry continuity and using corporate and judicial remedies for corporate disputes.

The act that can be named

The afternoon of 24 May 2004 in Dakar matters because it is the point at which AFRINIC says its members moved from an earlier, tentative corporate document to a final constitution. The setting was AFRINIC's first public policy meeting, held over 23 and 24 May, but the legally relevant setting was narrower: an annual general meeting of a company incorporated in Mauritius. Public-meeting atmosphere and corporate decision-making occupied the same place. They were not the same constituency.

The contemporaneous narrative reports that proposed changes were considered, endorsed and adopted by all members present. Later AFRINIC material describes what emerged as its final constitution. Those two propositions are the firm centre of the record. They support the occurrence of a member-company act and the institution's subsequent reliance on it. They do not reveal the complete words placed before members, the identities or voting powers of those members, or the formal record by which the act could be independently reconstructed.

Precision about the object of adoption is therefore essential. This was presented as a complete replacement of tentative bylaws, not merely a vote on four detached motions. For identification only, the bounded map of reported amendment clusters is: the legal-form change to a company limited by guarantee; replacement of Trustees with Directors; an appointed seventh board seat; and board-seat numbering and election sequencing. That one map does not disclose clause wording, legal consequences clause by clause, or the final arrangement. It is an inventory of subjects discussed around a whole-instrument proposal, nothing more.

The distinction sounds fussy until one asks a practical question: which words governed the company the next morning? A slide that names subjects is not a constitution. A meeting narrative that says changes passed is not the signed resolution. A later reference to “final” is not a file with an authenticated date and lineage. Corporate authority attaches to a text and an authorised act, not to a general recollection that something constitutional occurred.

A version without a stable identity

AFRINIC later reported that it had been incorporated in Mauritius in February 2004 under an earlier bylaws document. It also said supporting governance documents were published on its website in April. At Dakar, an archived presentation described the existing bylaws as tentative, invoked conformity with Mauritius law and contemplated complete replacement. The meeting narrative then said the proposed changes were adopted and that the organisation would thereafter be governed by the constitution linked on its site.

Subsequent AFRINIC material referred to a final constitution, while a later-2004 RIR-system newsletter spoke of newly adopted bylaws.

This sequence is evidence of institutional continuity. It is not version control. The earlier incorporation document, the April web material, the archived slide deck, the now-vanished link that the meeting report pointed towards, and later bylaws cannot safely be collapsed into one file. There is no certified final constitution in the surviving set. There is no hash, Registrar stamp or stable file identity that fixes the exact bytes adopted on 24 May. Nor is there a redline showing how the February text became the instrument supposedly approved in Dakar.

The slide deck is especially tempting because it is close in time and describes the proposal. Yet its nine pages are not a complete constitution. Its mixed use of old and proposed terminology reinforces the warning: it may illuminate intention, but it cannot stand in for the final legal text. Even attribution is unsettled. The meeting narrative names Alan Barrett in connection with the bylaws presentation, while the deck names Adiel Akplogan. That conflict may reflect authorship, delivery or file attribution; the evidence does not decide among them.

It is a modest uncertainty, but it demonstrates why a convenient document should not be promoted beyond what it proves.

Later repetition cannot repair the missing identity. When an institution consistently calls a document final, it establishes how the institution described and relied upon its history. It does not establish that the file now at hand is the file adopted then. Recognition by ICANN or repetition within the RIR system may corroborate AFRINIC's account of its institutional development. It cannot generate a lost redline, certify an absent resolution or make the exact text appear by retrospective consensus.

The electorate hidden inside the word “all”

The strongest surviving consent statement is also the easiest to overread: all members present endorsed and adopted all proposed changes. “All” conveys unanimity among the people counted by the narrator. It does not tell us how many people were counted, whether each held a vote, which membership class they occupied, whether proxies were recognised, or what proportion they represented of the entire eligible electorate.

Four populations must remain separate. Event attendees were people at the wider meeting. Corporate members were persons or organisations holding a status under the company's rules. Voting-eligible members were the subset entitled to decide the defined constitutional business. A continental constituency would comprise the states, networks, operators, resource holders and end users across a service region. Presence in a conference room proves none of the transitions from one population to the next.

Contemporaneous and later institutional accounts give conflicting estimates for the wider event: approximately 80, more than 85, approximately 120 or 125 attendees, with around 30 countries also mentioned. None is an adoption denominator. Country representation is not a share register. A crowded hall is not an electoral roll. The language of regional participation may show interest and breadth of attendance, but it cannot authenticate corporate voting power, still less political consent on behalf of a continent.

One nearby number creates a particular trap. A separately described resolution temporarily expanded election eligibility for that meeting. Five full members present approved it, and the narrative referred to seven registered full members. That was a decision about election rights, not the vote on the constitution. The five approvals, the five present and the seven registered belong exclusively to that separate item. They cannot be converted into a constitution tally, denominator or quorum. The constitution record remains denominator-free beyond reported unanimity among members present.

That separation is not pedantry. If a number from one motion migrates to another, a later reader may calculate an apparent majority, infer a threshold and declare compliance. Each step would rest on the same category error. The reliable statement is smaller but stronger: the narrative attributes unanimous assent to the members it describes as present; it supplies no constitution-specific headcount or percentage.

Ten questions the record leaves open

The missing evidence can be stated as a finite ledger rather than a cloud of suspicion. Each gap has a distinct evidential consequence, and none should be filled by institutional custom.

  1. G01 — Exact final text and stable version identity. No certified or otherwise authenticated file fixes the constitution adopted on 24 May. The slide deck, the vanished contemporaneous link, the earlier incorporation bylaws and a later bylaws document are not interchangeable substitutes.

  2. G02 — Draft-to-final redline and amendment motions. No complete draft chain shows which words changed between incorporation and Dakar, or whether amendments from the floor altered the proposal. The subject inventory is not a clause-level redline.

  3. G03 — Eligible-member register, class and voting entitlement. The record does not identify the constitutional electorate or show each participant's authority. Attendees, members present, full registered members in the other decision and countries represented remain different categories.

  4. G04 — Exact adoption tally and denominator. Beyond the narrative's unanimity among members present, there is no proved vote count, percentage or headcount for the constitutional act. No number from the separate election-rights decision may be borrowed.

  5. G05 — Notice date, recipients and attached resolution. AFRINIC's later account of publishing documents online in April does not prove that compliant corporate notice reached the proper recipients or included the exact proposed instrument and special-resolution wording.

  6. G06 — Quorum, proxies, abstentions, method and declaration. There is no preserved quorum calculation, proxy set, abstention record, voting method, poll or chair's declaration. Narrative unanimity does not establish any of these facts.

  7. G07 — Authenticated minutes, signed member act and filing receipt. The public board-resolution index contains acts near the meeting but no member special resolution adopting the constitution. Board decisions cannot substitute for the missing member record, and later recognition cannot substitute for a Registrar receipt.

  8. G08 — Point-in-time Mauritius law. The available official Companies Act compilation is revised and visibly incorporates amendments, including a 2004 change effective after the Dakar meeting. It is useful for understanding the legal framework, but it is not proof of every word in force on 24 May or of AFRINIC's compliance with that historical text.

  9. G09 — Presenter identity. The narrative names Alan Barrett; the archived deck names Adiel Akplogan. The discrepancy is real, while its explanation is unknown. It should not be resolved by guesswork or used to discredit what either document can independently establish.

  10. G10 — Overall attendance. The surviving estimates conflict. No single figure should be chosen for neatness, and no attendance estimate should be used as a corporate voting denominator.

This ledger sets the limits of the conclusion. Missing evidence does not prove that notice was defective, quorum absent or the vote invalid. Silence is not proof of compliance either. The proper status is unresolved. Anyone asserting a more definite procedural conclusion bears the burden of producing the instrument-specific record.

What Mauritius company law helps explain—and what it cannot prove

The available official compilation of the Mauritius Companies Act explains why the missing material matters. It describes a company constitution as capable, within statutory limits, of modifying internal rights, powers, duties and obligations. Provisions inconsistent with the Act are void to the extent of inconsistency. It also gives the constitution a contract-like effect between company and members and among members.

The same compilation describes a special-resolution route for adopting, altering or revoking a constitution, together with notice to the Registrar. Its definition of a special resolution refers to at least 75 per cent of votes entitled and voting unless a higher constitutional threshold applies. It also contains notice, quorum, proxy, poll and chair-declaration provisions, and benchmarks that include 14 days in relevant contexts.

Those rules furnish audit questions, not answers about Dakar. Without the point-in-time statutory text, the then-operative constitution and the meeting papers, one cannot confidently map each current provision onto the 24 May act. The compilation itself displays later amendment annotations. A provision changed after the meeting cannot simply be projected backwards. Nor does the existence of a statutory duty prove that AFRINIC discharged it.

This caveat cuts both ways. It prevents a critic from declaring non-compliance merely because no document appears in the current set, and it prevents AFRINIC from treating the general statute as evidence that the required steps must have occurred. Legal framework and event proof are different things. The framework identifies the records that should settle a dispute; the event record determines whether those records exist and what they show.

If the adoption was valid, its effect was important but bounded. A constitution can allocate internal powers, organise the relationship between members and officers, establish decision rights and create enforceable expectations within company law. Members, directors, the company, the Registrar and courts may all have roles in that private-law order. The constitution's reach comes from statute, corporate membership and judicial enforceability—not from a public constituent act by Africa.

Internal authority is not continental authority

The word “constitution” carries public-law grandeur. In this setting it names the governing document of a Mauritius-incorporated company. That document may be central to the company's internal legitimacy. It is not a constitution for Senegal, Mauritius, Africa, Internet operators or address holders as a political class.

The distinction rests on the origin of authority. Corporate members can consent to rules that bind the company and themselves, subject to law. They cannot, merely by voting, acquire authority from people who were not members, did not appoint them and were not part of the electorate. An organisation's service geography does not become a constituent body by metaphor. Serving a region is not the same as speaking for it.

Official publication also has a bounded function. A first-party meeting report proves that AFRINIC made a report and supports the occurrence it describes. A government-hosted statute establishes the content of that compilation. An ICANN archive shows what AFRINIC represented in a recognition setting. An RIR-system newsletter shows institutional understanding. None proves legitimacy merely because it is official. Official acts establish acts, documents and claims; legitimacy requires separate evidence of lawful authority and consent.

Recognition is similarly limited. Coordination institutions may rely on AFRINIC as the regional registry coordinator. Such reliance can matter operationally and may become deeply embedded. It does not cure an absent corporate record, confer ownership of address space or delegate sovereign powers. Administrative convention is not a title deed to the Internet, and dependency is not consent.

AFRINIC's proper description is therefore a private institutional bookkeeper and coordinator. It records and administers number-resource registrations and supports predictable coordination. It does not own Internet address space. It is not Africa's government. Its constitution granted no sovereign, punitive, police, prosecutorial or confiscatory authority. A registry cannot turn a corporate document into permission to investigate as police, accuse as prosecutor, punish as sovereign or seize as owner.

Why the missing file matters in daily operations

The adoption question is not antiquarian. A constitution is a root reference for later acts. Directors rely on it to understand powers. Members rely on it to understand rights. Courts and counterparties may need it to trace authority. If the original version is unstable, later amendments and practices become harder to place on a reliable lineage.

The first operational risk is mistaken continuity. A later consolidated document may quietly become treated as though it were the original Dakar text. Provisions added later can then be projected backwards, and the history of authority becomes circular: the current institution authenticates the current document by citing its own uninterrupted use. The longer that assumption persists, the more expensive correction becomes.

The second risk is electorate confusion. If attendees become “the community”, and “the community” becomes the source of corporate authority, the institution can move between legal categories without showing who authorised what. Open participation is valuable, but it is not a substitute for a member register or defined voting rights. Consultation may inform a decision. It does not authenticate the decision-maker.

The third risk is mandate inflation. A registry that sees its internal constitution as evidence of continental sovereignty may treat administrative choices as commands over networks and resource holders. That is precisely where recordkeeping can mutate into private enforcement. Operators need accurate registrations, reliable services, predictable rules and appealable decisions. They do not need a corporate coordinator to invent coercive powers because the infrastructure depends on its records.

The fourth risk is false certainty in disputes. One side may declare the whole institution illegitimate because an archival chain is incomplete; another may declare every historic act unassailable because operations continued. Both responses exceed the evidence. The prudent position preserves service while reconstructing authority. Corporate validity questions belong in contractual, company-law and judicial channels. They do not justify registry self-help punishment, abrupt service disruption or historical erasure.

The conclusion the evidence can bear

AFRINIC's members were reported to have adopted the company's proposed whole constitution at its Dakar annual general meeting on 24 May 2004. The report says all members present agreed. Later institutional material calls the outcome final or newly adopted. This is credible evidence that a corporate act occurred and that the organisation relied upon it.

The record does not identify the exact authenticated instrument, the complete path from earlier draft to final text, the eligible electorate, the constitution-specific tally, the notice, the resolution wording, quorum, proxies, abstentions, voting method, chair's declaration, signed minutes or Registrar receipt. It also does not settle the precise historical statutory text for every procedural issue. These are defined gaps, not permission to improvise.

The legal implication is bounded. A valid company constitution can organise internal authority and have contract effect within Mauritius law. It cannot confer a public mandate over a continent. The institutional implication is equally bounded: AFRINIC may coordinate number-resource registrations as a private bookkeeper, but it has no sovereign or coercive jurisdiction. The practical implication is constructive: find, authenticate and publish the documentary chain without interrupting the service that operators rely upon.