Summary

  • AFRINIC’s published 2004 architecture assigned Seat 1 to Northern, Seat 2 to Western, Seat 3 to Southern, Seat 4 to Eastern, Seat 5 to Central and Seat 6 to Indian Ocean. Seats 1 and 2 began with three-year terms, Seats 3 and 4 with two-year terms, and Seats 5 and 6 with one-year terms. Later terms were to be three years, allowing one pair to come up each year and avoiding simultaneous turnover of all six elected lines.
  • The numbers classified candidates, labelled ballot objects, attached primaries and alternates, and set the retirement sequence. They did not create six subregional electorates or six territorial mandates. Selection was made through a meeting-wide, organisation-based private corporate electorate, and the winner entered company office rather than becoming the political delegate of a geography.
  • The process said the seat order followed decreasing subregional LIR or ISP membership totals, but the readable evidence does not contain the underlying counts, cut-off time, source definition, deduplication method, tie rule, signed calculation, verifier or exact adopted clause.
  • One archived nomination notice literally assigns Central and Indian Ocean to “Seat 4 and 6”, even though the election-process document and contemporaneous meeting report place them in Seats 5 and 6. That unresolved documentary conflict must remain visible; it is not evidence that Central actually occupied Seat 4.
  • AFRINIC was and remains, for purposes of this analysis, a private number-resource bookkeeper and coordinator. Seat numbering did not confer sovereign, regulatory, punitive, police, prosecutorial, judicial or confiscatory authority. Official records establish reported acts and texts; their official status does not establish legitimacy or public mandate.

The small machine hidden inside six large labels

The most revealing feature of AFRINIC’s six numbered regional seats is not geography. It is time.

The map attached a stable numeral to each of six attributed subregions: Northern to 1, Western to 2, Southern to 3, Eastern to 4, Central to 5 and Indian Ocean to 6. The inaugural terms then descended in pairs: three years for 1 and 2, two years for 3 and 4, and one year for 5 and 6. Once that opening phase had done its work, each renewing pair was to receive the regular three-year term. The stated purpose was continuity through annual renewal by pair.

This was a compact succession machine. Without unequal opening terms, six directors beginning together on equal three-year clocks could all have reached expiry together. The three-two-one ladder displaced that cliff. At the first annual interval, the one-year pair could mature; at the next, the two-year pair; at the third, the three-year pair. A regular three-year term for each renewed pair would then preserve the annual rhythm. The inequality was therefore transitional rather than permanent in the published design.

The numbering made the rhythm legible. A notice could refer to Seat 5 without re-describing Central, its candidate qualification and its position in the opening cohort every time. A ballot could carry the same identifier. A retirement notice, vacancy record or later election could point back to a stable line. An alternate could be attached to that line without being mistaken, at least in a well-kept ledger, for a separate numbered office. Numbering reduced the coordination cost of keeping eligibility, selection, tenure and succession aligned.

That administrative usefulness is precisely why the labels require discipline. “Northern seat” or “Central seat” sounds more politically substantial than “corporate directorship filled through a geography-limited candidate line.” Repetition can turn shorthand into a claim. A stable geographic noun can invite readers to imagine a constituency, voters living within it, instructions flowing from it and a director answerable to it. The reviewed record proves none of those additions.

There were six attributed operating geographies in a private Board design, not six governments. The number ordered a corporate mechanism; it did not draw a public electoral district. The geography constrained who could stand for a labelled line. It did not establish that the people, states, operators, number-resource holders or end users within that geography had selected the officeholder, could instruct the officeholder, were entitled to reports from the officeholder, or could remove the officeholder. The office remained a director’s office in a private company.

This distinction does not trivialise geography. Different parts of Africa and the Indian Ocean present different infrastructure, market, language, access and operator conditions. Requiring geographic variety among candidates can improve the Board’s information. But information is not authorization. A person can bring experience of a place without possessing the place’s political voice. The design can be defended as a way to see more of the operating environment while still being described accurately as corporate selection machinery.

A ranking rule whose arithmetic is missing

The archived election-process document states that actual LIR or ISP membership in the six subregions was to determine the seat numbers in decreasing order. On that account, the subregion with the largest total received the lowest-numbered position and the ordering continued down through the six lines. The published mapping was not presented merely as alphabetical geography; it purported to translate a measured membership distribution into seat order.

The result mattered because number and inaugural duration were linked. Once Seats 1 and 2 were assigned three years, Seats 3 and 4 two years, and Seats 5 and 6 one year, the ranking placed subregions into different opening-term cohorts. A membership snapshot therefore did more than arrange a list. It helped determine which attributed geography received the longer opening runway and which reached its first renewal sooner.

Yet the evidentiary chain stops before the calculation. The reviewed material contains no readable table of subregional totals. It does not reveal the precise cut-off timestamp, the source from which organisations were counted, the treatment of duplicates, the definition of membership used in the computation, or the handling of organisations whose geographic classification might have been contested. It gives no tie rule. It supplies no signed worksheet, identified verifier, approval record or audit showing how the rule generated the final six-line map.

Those omissions do not prove that no calculation was made. Nor do they license an invented reconstruction. The proper statement is narrower: the process published a decreasing-membership rule, and the event record published the resulting map, but the surviving readable record examined here does not allow an outsider to reproduce the arithmetic.

That difference between a stated rule and a reproducible application is central to accountability. A rule can be neutral in form while leaving material discretion in its inputs. Which date counts? What qualifies an organisation for a subregion? How are affiliates treated? What happens if two totals tie? Who signs the result? Each answer could alter position in the opening ladder without changing the rule’s one-sentence description.

The map should therefore be regarded as a reported institutional act with high evidentiary support, while the claimed measurement behind its ordering remains incompletely auditable. The distinction protects both sides of the inquiry. It does not dismiss the official account simply because the worksheet is absent. It also refuses to treat publication by the institution that designed the process as independent proof that every input and judgment was correct.

There is a further reason not to overread the ranking. A count of member organisations is not a count of people, states, networks, users, economic value or public consent. Even a perfectly reproduced table would establish only the metric the process selected. It would not show that a subregion with more organisations deserved greater political standing, nor that a lower seat number was a constitutional honour. The numbers became operationally important because they were connected to timing, not because they measured the democratic weight of a territory.

Four layers that the label must not collapse

The architecture becomes clearer when four layers are kept separate: candidate geography, selector identity, corporate office and affected population.

First, candidate geography. The nomination notice required candidates to reside in Africa and in the region associated with the line for which they stood. That is a candidate-side gate. It seeks regional variety by restricting the eligible pool. It can make it more likely that a director brings experience of conditions associated with a particular operating geography. It does not reveal who voted.

Second, selector identity. The design used a meeting-wide, organisation-based private corporate electorate. The record reviewed here does not establish six separate subregional rolls choosing only their own labelled lines. It therefore cannot support the picture of Northern electors choosing a Northern political representative, Western electors choosing a Western one, and so forth. The selecting body and the candidate’s attributed geography were different dimensions of the process.

Third, corporate office. The person selected entered a Board position governed by the company’s internal instruments and applicable company law. The number could define the election line and its clock, but it did not replace the corporate character of the office. A geographic qualification is not, by itself, an instruction instrument. No reviewed document establishes that a director owed a separate territorial duty to a subregion or could be commanded by every affected organisation located there.

Fourth, affected population. Operators, states, resource holders and end users could all be affected by the quality and stability of the registry’s administration. Their exposure did not make them members of the selecting electorate. Their physical presence inside a label did not automatically give them appointment, instruction, reporting or removal rights. Operational dependence is real, but it is not the same thing as delegated political agency.

Collapsing these layers creates two opposite errors. The first is romantic: treating a geographically qualified director as the voice of a coherent regional public. The second is cynical: assuming that because the electorate was private, regional knowledge had no value. The evidence supports neither. The mechanism could widen the Board’s field of vision without turning the Board into a parliament.

This is also why institutional terms such as “representative” require care. In election material, such language may describe a regional line or the expected informational perspective of its officeholder. It should not be silently upgraded into sovereign representation. A service geography is not a people capable of issuing one political mandate. Each attributed region contains plural states, markets, operators and end users with interests that can conflict. No single corporate selector can infer authorization from that diversity merely by giving a seat a geographic name.

One numbered line, a primary and an alternate

The election architecture attached primary and alternate ordering to each numbered regional line. That continuity feature must not be counted as additional numbered seats. Six lines remained six lines. A primary and an alternate associated with Seat 3 did not create Seat 3A and Seat 3B as two simultaneous numbered directorships.

This distinction matters because alternates can solve one problem while creating another. They can reduce the disruption caused by a vacancy or unavailability. If the next person is already identified through the same selection exercise, continuity need not depend on an improvised appointment. But the legal and operational effect of the alternate depends on rules that the readable record does not supply.

The missing instrument would need to answer basic questions. What event activated the alternate? Did the alternate hold any voting status before activation? Was fresh consent required? How was a vacancy declared? Did activation last for a meeting, for an absence or for the remainder of a term? Could the alternate cease to qualify geographically? Which record would show activation and deactivation? Without those provisions, it would be wrong to describe every alternate as a full director at all times, and equally wrong to ignore the continuity position altogether.

The gap reinforces the value of seat numbers as ledger keys. A properly maintained record could show the numbered line, primary, alternate, qualification evidence, term start, expected expiry, retirement, vacancy event, activation instrument and present status. That makes the architecture observable without inflating the alternate into another regional office. The identifier is useful because it can connect records; it is dangerous when the institution supplies the label but not the connected records.

The “Seat 4 and 6” conflict cannot be edited away

One archival inconsistency is unusually concrete. The nomination page grouped Northern and Western as Seats 1 and 2, and Southern and Eastern as Seats 3 and 4. For Central and Indian Ocean, however, its third line literally printed “Seat 4 and 6”, while also assigning that tail pair the one-year opening terms. The election-process document places the one-year pair in Seats 5 and 6. The contemporaneous meeting report likewise maps Central and Indian Ocean to Seats 5 and 6.

The combined record strongly supports the intended six-line sequence—especially because Seat 4 had already been assigned to Eastern on the same nomination page, and because the one-year cohort elsewhere is 5 and 6. But a strong inference about an apparent publishing error is not the same as documentary resolution. No corrected notice or contemporaneous explanation in the reviewed record establishes why the page said 4 rather than 5.

The conflict must therefore be carried literally: “Seat 4 and 6” in the nomination notice versus Seats 5 and 6 in the process and meeting report. It must not be silently harmonised. It must not be cited as proof that Central occupied Seat 4. And it must not be dramatised into evidence that the entire map was invalid. It is a bounded conflict that exposes a larger records problem: the smallest typographical discrepancy can become consequential when seat numbers connect eligibility, ballots and term clocks.

A corrected notice would do more than improve historical neatness. It would show which document the institution treated as controlling at the time and whether participants received consistent information. An account of the correction could identify when the error was noticed, who authorised the fix and which downstream records were checked. In the absence of that chain, the prudent approach is to disclose the contradiction and rely on the mutually consistent process and meeting records for the reported map, while retaining uncertainty about the faulty page.

What the act legally changed—and what it could not change

The contemporaneous report describes numbered-seat machinery as one amendment adopted at the May 2004 AGM. That establishes the institution’s report of an internal act. It does not provide the exact clause, clause number, filed instrument or redline. The later recognition application describes the May AGM and a final constitution, but it is a retrospective applicant account. It does not fill the missing text or independently prove that every legal step was completed.

The narrow legal effect supported by the seat materials is procedural. A stable number connected an attributed candidate geography to a corporate Board line and an inaugural renewal cohort. The design could govern nominations, ballot labels, primary-and-alternate ordering, retirement sequencing and later renewal. It could make internal office succession more predictable.

Nothing in those materials creates a public office. No territorial principal is identified. No subregional instruction right, reporting compact or removal mechanism is established. The labels do not transfer authority from states, operators or end users to the director. Nor do they transfer ownership of number resources to the company. The numbering act cannot bootstrap a private coordination institution into a sovereign for the territory named on a ballot.

The current official Mauritius Companies Act consolidation, dated 15 May 2026, helps illustrate the present corporate accountability architecture. In that current frame, Board powers are subject to the Act and the company’s constitution; directors are to act honestly, in good faith and in the company’s best interests, and the relevant duties run to the company. That is consistent with understanding the numbered office as a company office, not a territorial delegation.

But the date boundary is essential. The 2026 consolidation is not a certified statement of the statute exactly as it stood on 24 May 2004. It cannot establish the event-date wording, validate the missing seat clause or prove compliance with historical notice, quorum, voting or filing requirements. A point-in-time statutory text and the filed constitution effective on the event date would be required for stronger historical legal conclusions.

The private-company boundary also limits substantive authority. AFRINIC can maintain accurate number-resource records, coordinate uniqueness, run its lawful internal affairs and organise corporate elections. It is a private bookkeeper and coordinator. It has no sovereign, regulatory, punitive, police, prosecutorial, judicial or confiscatory authority. A well-run election can make its Board internally accountable; it cannot create powers the institution never possessed.

That point is practical, not semantic. A Board can influence a service on which operators rely. Its decisions can create continuity or uncertainty. But influence over an essential coordination function does not make the Board a regulator, and operational dependence does not make registry records a form of public title that the institution may punish, prosecute, adjudicate or confiscate at will. The numbered-seat design should be judged as governance of a narrow corporate function, not as a vote on who governs the Internet in Africa.

The strongest defence of the design

The strongest case for AFRINIC’s 2004 arrangement is straightforward. A new continent-wide coordination company needed both geographic range and continuity. Six candidate gates reduced the risk that all elected officeholders came from one market cluster. Stable numbers made the lines easy to administer. Alternates created a pre-identified continuity position. Unequal opening terms phased the Board into predictable annual renewal. Equal three-year terms after the opening phase prevented the temporary ranking from becoming permanent tenure inequality.

The meeting-wide corporate electorate can also be defended. Separate local electorates might have tied selector and candidate geography more tightly, but they could have increased the risk of capture by a small local network. A common private electorate could assess all six lines as parts of one Board. That design encouraged an institution-wide view while geographic eligibility kept diverse operating environments visible.

Membership totals offered an apparently neutral ordering device. A start-up institution had to decide which pair received one, two or three opening years. Using an observed participation measure could appear less arbitrary than private negotiation. The later equality of terms meant the ranking’s formal tenure consequence should diminish after the first cycle.

These are real virtues. Staggering is a rational continuity tool. Regional qualification can improve information. Stable identifiers can improve records. An alternate can reduce vacancy risk. None of those benefits requires a claim that the seats were political constituencies. Indeed, the defence becomes stronger when the claim is modest: the system was useful because it organised Board composition and time, not because it reproduced a continent’s democratic will.

The strongest criticism without mandate inflation

The strongest criticism is not that the regional seats lacked sovereign legitimacy, because they never needed sovereign status to perform a private corporate function. The stronger objection is that the design allowed a transient and unreproducible membership snapshot to acquire durable institutional salience.

Seat numbers are sticky. Even after the opening term inequality disappears, a low number can feel senior, foundational or more important. If numbering was tied to 2004 membership distribution, the labels may freeze an early market structure into the institution’s vocabulary. Later audiences can forget that the ranking served a one-time phase-setting function. They may interpret a durable number as a durable hierarchy.

The missing calculation compounds the risk. Without source totals, a cut-off, a tie rule and a signed result, later observers cannot determine whether the published map followed the stated formula. Without the exact adopted clause, they cannot see whether the legal text matched the process page. Without term and alternate ledgers, they cannot reconstruct the clock. When evidence thins, institutional memory supplies the missing links—and memory tends to convert familiar labels into assumed authority.

Regional language can then obscure the selector. If public communications say a director “represents” Central or Northern without identifying the limited meaning, readers may assume a regional public chose and controls that person. The label lends a democratic tone to a private selection structure. This is legitimacy inflation: the governance mechanism may be valid for the company, yet its wording suggests consent by people and organisations who were not shown to be selectors.

The remedy is not to erase geography. Doing so would discard its informational value. Nor is the remedy to invent a public mandate after the fact. The proportionate answer is calibrated language and auditable records: regional expertise rather than territorial voice; company office rather than public office; a corporate electorate rather than a regional people; and an explicit ledger showing how each seat, term and alternate was administered.

Eleven limits that should remain attached to the history

A responsible account of the six seats has to preserve what is not known, because each missing item corresponds to a different accountability claim.

First, the exact adopted clause and clause number are missing. Meeting and process narratives should not be turned into invented constitutional wording.

Second, the underlying LIR or ISP totals by subregion are not readable in the reviewed record. No numerical margin should be supplied.

Third, the calculation chain is missing: there is no available signed ranking, tie rule, named verifier, approval record or independent audit. The published decreasing-order rule proves the stated method, not every application decision.

Fourth, the nomination notice conflicts with the process and meeting report by printing “Seat 4 and 6” for the final pair. The reason is unknown.

Fifth, the precise term commencement, expiry and retirement dates are absent. One, two and three years are durations, not a complete calendar.

Sixth, the alternate instrument is absent. Activation, voting status, vacancy succession and deactivation cannot be reconstructed.

Seventh, whether later renewals occurred exactly on the published annual pair schedule is outside this inquiry and unproved. Architecture is not evidence of later compliance.

Eighth, the amendment-specific notice, quorum, resolution, voting and filing record is missing beyond the report’s statement of adoption by members present. Figures from other acts or elections must not be imported into this one.

Ninth, a certified version of Mauritius company law in force on 24 May 2004 is not present. The 15 May 2026 consolidation is current framing only.

Tenth, no subregional compact establishing consultation, instruction, reporting or removal rights has been shown. Candidate residence and a regional label do not supply one.

Eleventh, independent corroboration is limited. AFRINIC publications and the later application preserved in the ICANN archive are useful evidence of what was reported, proposed and described. Official documents prove acts, not legitimacy; custodial location and institutional authorship do not certify representation, mandate or legal completeness.

Together, these limits produce a precise conclusion. The six-seat map, inaugural ladder, later three-year terms and continuity purpose are well supported as published institutional arrangements. The calculation behind the ranking, the exact legal text and the detailed operation of tenure and alternates are not fully reproducible. The record supports corporate machinery. It does not support territorial political authority.