Summary

  • The checked public record does not establish that candidates for the two AFRINIC-selected NRO Number Council seats had to reside or work in the AFRINIC service region. The preserved election procedure described the election as open to anybody; the geographic language it published concerned voters.
  • Before the vote, AFRINIC said anybody within its service region could vote. Afterward, the minutes, result page and meeting report described eligible voters as people “residing and working” in the region. The final voter roll, ballot design, verification method, constituency denominator and reason for the change in wording remain unavailable, so 104 candidate-vote units cannot be turned into 104 voters, turnout or a continental mandate.

On 26 April 2005, at AFRINIC-2, five names appeared in a contest for two seats on the NRO Number Council. Two decades later, the most revealing part of the election is not the winning margin. It is the placement of a few words in the surviving documents.

The archived procedure says the election was open to anybody. In its separate voter section, it says anybody within the AFRINIC service region could vote. The records published after the event use a different and more specific description: eligible voters were people residing and working in that service region. Those are voter words. They are not a candidate-qualification clause, and the conjunction is “and,” not “or.”

That precision matters because the premise under examination—a rule under which residence in or employment within the region qualified a person to stand—does not survive contact with the checked public record. A plausible regional rationale is not the same thing as a published rule. A country field is not a residency decision. A biography is not an employment audit. An official result is evidence of what the institution recorded and did, but it does not independently verify every eligibility fact underneath the result.

The better account of AFRINIC’s first public Number Council selections is therefore not a story about a candidate restriction. It is a record audit. What did the governing arrangements authorize? Which people were said to be eligible for what? How was a regional connection expressed before and after the ballot? What can the published numbers establish? Which constituencies actually authorized the selected members? And where does administrative coordination end before it is mistakenly recast as sovereignty?

The institutional act was a selection, not a transfer of public power

The NRO Memorandum of Understanding created a Number Council with a symmetrical regional formula. Each of five regional Internet registries supplied three members: one appointed by the registry’s board and two selected through that region’s policy forum. On that formula, the full council had 15 members. AFRINIC’s three places amounted to one fifth of the council, while its two forum-selected places amounted to two thirds of its own allocation. Neither percentage was population-weighted, and neither purported to measure Africa’s inhabitants, states, networks or resource holders.

The procedure for the two forum selections was meant to be open, accessible, documented and transparent. That language matters as a procedural commitment, but it does not answer every eligibility question. “Open” does not itself define the electorate. “Documented” does not guarantee that a voter ledger is public. “Transparent” does not identify the evidence used to test residence or work. The adjectives create an expectation against which the surviving record can be assessed; they are not substitutes for the missing operational rules.

The third AFRINIC place followed a different route. AFRINIC’s Board appointed Alan Barrett. That appointment must be kept separate from the two places contested at the meeting. It tells us nothing about whether candidates for the public selections had a geographical qualification. Collapsing the appointed seat and the elected seats would erase the composition rule that framed the event.

The Number Council also carried a bounded role. Its members advised within the NRO structure and served as the Address Council in the ICANN Address Supporting Organization arrangement. The relevant instruments described advisory, coordination and global-policy-process functions. They did not legislate for a continent, create a territorial jurisdiction or grant police, taxation, judicial or confiscatory powers. AFRINIC’s power in this episode was the power to coordinate a regional selection under an institutional arrangement.

That is the correct scale of the act. AFRINIC was a ledger coordinator participating in a contractual governance mechanism, not a sovereign licensing a political electorate. The selection could influence advice and policy coordination concerning Internet number resources. It could not turn a service-region line into a border of citizenship or make a meeting ballot the voice of a continental people.

The candidate rule and the voter rule occupy different parts of the record

The preserved election page is the critical document because it set out the process before the event. It supplied a nomination timetable and requested details about nominees, including affiliation, physical address and country. Yet it did not state that a nominee had to reside in the service region. It did not say that employment there was sufficient or necessary. It did not request nationality or citizenship, define either concept, or describe an adjudication process for candidate geography. Instead, it said the election was open to anybody.

Country, address and affiliation fields can help identify and describe a nominee. They might also have been useful to administrators. But a form field does not become an eligibility condition merely because someone could use it for screening. To establish such a condition, the record would need to connect the information to a rule: who was covered, what the required nexus was, which documents counted, who decided difficult cases and whether a rejection could be challenged. None of that candidate machinery appears in the checked sources.

The nominees page later displayed five profiles and country labels. Those publications establish that AFRINIC presented the candidates in that way. They do not establish anyone’s citizenship, residence or employment. A country label can refer to location, affiliation, origin or a field entered on a form; without an express definition it cannot bear more evidentiary weight than the process gives it. The source set therefore does not support a nationality controversy, and creating one would distract from the real documentary issue.

The geographic boundary appears elsewhere. Before the event, the voter section said anybody within the AFRINIC service region could vote. That phrase still leaves questions—what “within” meant, how it was tested and which version of the region list controlled—but its grammatical object is the voter. It cannot be moved to candidacy without changing the rule.

After the event, AFRINIC’s minutes, results and meeting report offered the formulation “residing and working” in the service region. The two elements are conjunctive in the text. They must not be rewritten as alternative paths of residence or employment. Perhaps the post-event wording was loose. Perhaps it reflected a final instruction that is no longer public. Perhaps an administrator treated one form of connection as enough despite the published “and.” Each possibility is conceivable; none is proven.

The chronological mismatch is consequential. A person “within” a region on election day is not necessarily both residing and working there. Conversely, a person residing and working in the region could be temporarily outside it. The two descriptions may have been intended as equivalent shorthand, but they are not textually identical. Without a final certified rule or adjudication guide, it is impossible to say which test was applied or whether the conjunction described actual ballot access.

This is why the audited question must stay on residence and employment, not citizenship. The documents offer regional presence, residence and work as the relevant concepts. They do not offer nationality as a gateway. Turning “country” into citizenship would invent both a legal category and a dispute that the record never records.

What happened on 26 April

The event mechanics are clearer than the eligibility mechanics. AFRINIC held the election at AFRINIC-2 on 26 April 2005. Five candidates received published totals. Jean Robert Hountomey received 33 candidate-vote units and Sylvia Geha Kezengwa received 27; they were declared elected. Ali Drissa Badiel received 23, Yaovi Atohoun 16 and Wycliffe Ochieng 5. The difference between the second-place total and the next total was four candidate-vote units.

The five numbers sum to 104. That arithmetic is sound, but its label must remain exact: 104 is the aggregate number of candidate-vote units. The checked record does not disclose how many choices each voter could make, whether every voter used the same number of choices, how many ballots were cast, or how invalid or blank ballots were treated. It follows that 104 is not a count of voters and not a turnout figure.

Shares calculated from 104 have the same limitation. Hountomey’s total was about 31.73 per cent of candidate-vote units; Kezengwa’s was about 25.96 per cent; Badiel’s about 22.12 per cent; Atohoun’s about 15.38 per cent; and Ochieng’s about 4.81 per cent. The winning pair together received about 57.69 per cent of candidate-vote units. These figures describe distribution among the published candidate totals. They are not electorate shares, approval rates or measures of a mandate.

The distinction is more than statistical housekeeping. If voters could select two candidates, a single person might account for two units in the sum. If choice rules differed or voters could cast fewer than the maximum, the relationship between totals and people would be different again. Because the ballot design is missing, converting candidate totals into human voters would supply a denominator that the record withholds.

Other numbers cannot repair that absence. A public participant snapshot dated 13 April displayed 36 rows and included people from within and outside Africa. The meeting report later said that more than 100 people participated in AFRINIC-2. The 36-row list was a dated attendance snapshot, not a voter roll. The broader attendance statement was also not a certified electorate count. Neither number can be used to calculate turnout, and the apparent proximity between “more than 100” participants and 104 candidate-vote units is not evidence that each participant voted once.

The calendar contains a smaller unresolved detail. The process identified 15 March at 20:00 as the nomination deadline, 42 elapsed days before the 26 April election. Its generic schedule described nomination closure as election day minus 45. The record does not explain the three-day discrepancy. It may be an innocuous scheduling adjustment, but it is another reason to avoid treating a high-level process outline as a complete execution log.

AFRINIC joined the NRO Memorandum of Understanding on 27 April, the day after the ballot, and the joinder confirmed its entitlement within the NRO arrangement to select three individuals. An ASO announcement on 28 April described AFRINIC members as having voted for the two representatives. That is narrower than AFRINIC’s own pre-event description of anybody within the service region and its post-event description of people residing and working there. The source set does not reconcile “members” with those broader person-based formulations.

The appropriate conclusion is not to choose whichever wording produces the neatest story. It is to preserve the discrepancy. The election happened; five candidate totals were published; two winners were recorded. But the surviving texts do not disclose a single, stable and fully specified electorate definition across the process, event record and subsequent announcement.

Registration collected information; the record does not show verification

The AFRINIC-2 registration form asked for job, organisation, street address, city and country, among other meeting-related details. Those fields could have furnished clues about whether someone lived or worked in the service region. They do not prove that the information was checked, that it controlled access to a ballot, or that registration and voting were identical statuses.

Verification is a chain, not a collection box. An auditable system would identify the rule, evidence required, time at which eligibility was measured, person or body responsible for decisions, treatment of ambiguous cases and final certified voter list. If both residence and work were required, it would explain how each was demonstrated. If “within the region” was the operative test, it would explain whether that meant ordinary residence, employment, physical presence, organisational affiliation or something else. The public record checked for this article supplies none of those links.

This does not prove that AFRINIC performed no checks. An internal practice could have existed and later been lost, never published or kept in records outside the accessible set. Meeting staff may have known participants personally or relied on self-reported details in an early, lightly formalised process. But absence and nonexistence are different propositions. The defensible finding is narrower: the method cannot be audited from the checked record.

The same boundary applies to official documents. Minutes are strong evidence that AFRINIC recorded the election and its description of eligible voters. A results page is strong evidence of published totals and declared winners. A meeting report is strong evidence of the institution’s contemporaneous account. Official status, however, does not transform a statement about eligibility into independent proof that each person met an undisclosed test. It proves the statement and act, not every fact presumed beneath them.

The missing voter roll has several effects. It prevents a count of electors. It prevents analysis of how many were members, operators, resource holders, employees, residents or visitors. It prevents comparison between eligible people and participants. It prevents checking whether both residence and work were uniformly required. And because individual adjudications are absent, it prevents examination of how borderline or inconsistent cases were treated.

The service-region boundary itself is not fixed conclusively for election day by the available regional list. An archived AFRINIC page from August 2005 displayed 55 economy or territory entries. That is useful context for the institution’s service area later that year. Because the capture postdates the April election, it cannot conclusively establish the precise boundary administrators used four months earlier. Treating it as contemporaneous would erase a meaningful timing qualification.

Thus the open questions are specific. Was there an unpublished candidate practice despite the open-to-anybody language? Did “residing and working” state the actual voter rule or merely summarise it imprecisely? Were residence and work both assessed? Did meeting registration feed a separate election list? Was member status required, as the ASO announcement might suggest, or was regional connection enough? How many individual voters and ballots were involved? The public sources close none of these questions.

A regional electorate is not automatically a continental constituency

The election documents identify a selection mechanism inside a service region. They do not establish an authorization chain from every party affected by Internet number-resource governance. Meeting participants, eligible voters, AFRINIC members, network operators, resource holders, end users, residents, governments and the inhabitants of Africa are different sets. The record cannot make them interchangeable simply by using the word “regional.”

This distinction is central to the agency question. Operators and resource holders may bear technical, financial and continuity consequences from policy coordination. They may have incentives and exposure unlike those of observers or occasional meeting participants. Yet the source set provides no denominator showing how many such principals were eligible, registered, present or voting. It supplies no principal-by-principal authorization under which the winners could be said to represent all operators or all resource holders.

Participation has value without becoming universal authorization. A regional forum can convene expertise, expose decisions to challenge and choose people for defined institutional roles. Those functions do not require the fiction that everyone affected has delegated a political mandate. The ballot authorized two people to occupy two council seats under the NRO arrangement. It did not authorize them to speak for every African, every state, every network or every end user.

The terminology used after the vote makes the denominator problem harder, not easier. If the electorate was people residing and working in the service region, then it was a person-based set with two asserted connections. If it was AFRINIC members, as the later ASO announcement said, it was an organisational or membership-based set. If it was anybody within the region, as the pre-event page said, its scope might have been broader still. No checked document certifies which population ultimately controlled.

A service region is an administrative geography designed to divide coordination responsibilities. It is not a country and does not create citizenship. It is not a sovereign people and does not erase the diverse legal, commercial and operational relationships inside it. The fact that AFRINIC coordinated number-resource functions for a region did not grant it ownership of that region’s networks or public jurisdiction over its inhabitants.

This authority boundary protects analysis from exaggerating both legitimacy and risk. On one side, it prevents a meeting election from being marketed as the democratic voice of a continent. On the other, it prevents an internal eligibility rule from being described as territorial punishment. The event was institutionally consequential but legally bounded: an administrative selection for an advisory and policy-coordination body.

The practical stakes remain real. Selection quality can affect the advice and policy-process work performed through the Number Council and Address Council. Geographic access rules can influence who pays the cost of travel, registration, information gathering and participation. Operators can face downstream infrastructure consequences from governance decisions. But the package does not quantify an election-specific loss, identify a 2005 participation-cost ledger or show remote-voting arrangements. Mechanisms should not be converted into invented amounts.

The strongest case for an unwritten regional-connection rule

The most persuasive contrary reading begins with institutional purpose. A seat arising from a regional policy forum was meant to have a regional character. The election page asked nominees for country, affiliation and physical address. The meeting form requested job, organisation and location. The published candidate profiles appeared regionally grounded. In a new and lightly formalised system, administrators might reasonably have expected a genuine connection to the region without drafting a complete eligibility code.

On this view, “open to anybody” could have meant that anyone could put forward a name and that no nominations committee would narrow the field, not that a person with no regional connection could ultimately take a seat. The “residing and working” language might have captured the intended regional nexus but appeared only in post-event summaries. Or the conjunction might have been casual prose, with residence or employment treated administratively as alternatives.

This steelman is institutionally intelligible. A regional forum has a legitimate reason to avoid a ballot detached from the region it serves. Early governance systems often rely on personal knowledge and self-reporting. The absence of detailed public machinery does not prove that nobody applied a practical boundary.

But that remains inference. The explicit openness language sits in the election procedure, while the explicit geographic wording sits in the voter section. The post-event sources use the conjunction in “residing and working”; no checked text authorises an alternative reading. No candidate clause, rationale memorandum, screening decision, voter ledger or verification protocol bridges the divide. A good-faith administrative explanation cannot be promoted into a historical rule merely because it would make sense.

The counterfactual record clarifies the problem. If candidates expressly needed either residence or employment, one would expect a clause defining those alternatives and stating how a nominee proved one of them. If country meant citizenship, one would expect nationality language, passport evidence or rules for dual nationals and noncitizen residents. If registration data were a verification system, one would expect the election rules to connect those fields to ballot access. If the participant list were the electorate, one would expect the result to identify it as such and reconcile people, ballots and candidate totals.

None of those expected links appears.

The surviving point from the contrary case should be retained rather than dismissed: regional connection was a sensible concern, and internal practice may have been thinner and less formal than modern expectations. That surviving point justifies further archival inquiry. It does not cure the current evidentiary gap.

What the audit establishes—and what it leaves open

The checked record establishes a bounded set of facts. AFRINIC conducted two regional-forum selections for the NRO Number Council at AFRINIC-2, alongside a separate Board appointment. The public procedure described the election as open to anybody and did not publish a candidate residence, employment, citizenship or nationality condition. Its geography appeared in the voter section. Post-event records described eligible voters as people residing and working in the service region. Five candidates received 104 aggregate candidate-vote units, and the two highest published totals won.

The same record leaves the implementation layer open. It does not show whether an unpublished candidate practice existed. It does not reconcile the pre-event “within” formulation, the post-event conjunctive formulation and the later “members” description. It does not identify the evidence used to establish residence or employment, the decision-maker, the appeal path, the voter list, ballot rules, number of voters or constituency denominator. It does not let attendance stand in for voting or candidate totals stand in for turnout.

The doctrinal conclusion is equally bounded. AFRINIC coordinated a ledger and a selection mechanism; it did not become a government. A regional procedure can be legitimate within its own institutional scope without acquiring sovereign force. Conversely, calling a process regional does not prove authorization from all principals who bear its consequences. The right question is not whether the meeting somehow embodied Africa. It is whether the selection followed its defined rules and whether the record allows those rules and their application to be checked.

For this event, the documentary gap is the story. The supposed candidate residence-or-employment rule is not established. The published geographic rule is voter-side, and its post-event language is conjunctive. The official record proves the recorded words, totals and acts; it does not certify invisible verification. That conclusion is narrower than a sweeping judgment on the election, but it is stronger because it does not ask the sources to say what they do not say.