Summary

  • Chairs Elections Process Draft 1, dated 2 November 2019 and posted to the RPD list on 5 November, tried to replace a thin chair-selection practice with explicit qualifications, an election calendar, electronic ballots, objections, ratification and vacancy rules.
  • Its voters would have been people subscribed to the RPD mailing list for at least six months. That was a visible eligibility boundary, but the text did not establish that one account represented one verified person, an authorised organisation or the region’s affected operators and users.
  • The AFRINIC Board sat at several gates: publication of acceptable candidates, possible disqualification, judgment of objections, ratification, interim appointment and highest appeal. Draft 1 allowed delegation but did not supply a complete separation-of-functions design.
  • The AFRINIC-31 record says the proposal had no consensus and returned to the list. The available record does not show Draft 1 adopted, implemented or binding, and it does not prove fraud, manipulation or a corrupted election.
  • The contemporaneous attachment identifies itself as AFPUB-2019-GEN-003-DRAFT01; AFRINIC’s later and current archive trail calls the same 2 November version AFPUB-2019-GEN-007-DRAFT01. The reason for the discrepancy is not established.
  • A verifiable election could give PDWG chairs a stronger mandate to organise discussion and assess rough consensus inside a private process. It could not give AFRINIC or its chairs sovereign, legislative, regulatory, police, prosecutorial, punitive, confiscatory or public-law adjudicative power.

A narrow office with consequential hands

The four-page proposal began with a practical complaint. The AFRINIC Consolidated Policy Manual described the selection of Policy Development Working Group chairs too briefly. The authors, Jordi Palet Martinez and Souad Abidi, wanted more than an improvised choice around a meeting. Their Draft 1 supplied qualifications, dates, terms, a ballot, a fraud response, an objection period and a route through vacancies. In institutional terms, that was a serious improvement in visibility. A rule that can be read in advance is easier to contest, follow and repair than a custom reconstructed after a dispute.

But visibility is only the first test of an election. The harder test is the authority record behind it: who was entitled to vote, which human being controlled each credential, what mandate that person carried, which institution administered the ballot, and where a voter could challenge an error. Draft 1 moved the chair election from an underspecified practice toward a recognisable procedure, yet it left those questions partly unanswered. It named an electorate without fully identifying its members. It described a one-vote aspiration without specifying the system that would make the aspiration auditable.

It created several review gates, but gathered many of them around one institutional centre.

That matters because a PDWG chair is neither decorative nor sovereign. The office is procedural. Chairs organise discussion, identify whether objections have been addressed, assess rough consensus and help determine whether a proposal advances, returns for revision or fails to move. Those decisions can alter the timing and practical status of work affecting IPv4 transfers, routing security, registry information, fees or other matters on which operators plan. A chair does not legislate, confiscate or regulate by public law. Even so, the chair’s procedural judgment can consume or save months of engineering, legal and management attention.

The correct response is not to inflate the office into government. Nor is it to say that internal legitimacy is irrelevant because AFRINIC is private. A private technical coordinator still needs procedures proportionate to the consequences of its own decisions. The chair requires a mandate that is real enough to discipline the office and narrow enough not to manufacture powers the electorate itself does not possess.

Draft 1 is therefore best read as an attempted authority circuit. At one end stood a mailing-list subscriber. At the other stood an office able to declare whether rough consensus existed. Between them were candidate eligibility, screening, a seven-day electronic vote, a tie formula, objections, Board ratification and appeal. The design question is whether that circuit reliably connected identifiable human voters to a bounded procedural office. On the evidence published in 2019, it did not yet do so.

What Draft 1 actually built

The proposal’s merits are easiest to see by reconstructing it rather than reducing it to the controversy that followed. Draft 1 retained the working group’s broad character: anyone participating online or in person could be part of the PDWG. For the election, however, it drew a narrower line. Anyone who had been on the RPD mailing list for at least six months before the election process could vote. Candidates also needed at least six months on the list. This was an effort to connect both candidacy and voting to sustained exposure to the forum rather than to whoever happened to appear at a public policy meeting.

The candidate channel included a second connection to AFRINIC. A person had to be an AFRINIC member’s membership contact or be nominated by two membership contacts of an AFRINIC member. Candidates were to provide biographical information and describe their contributions, participation, experience and intended achievements. AFRINIC staff could not serve as a chair. The chair position was voluntary and unpaid. A chair role was also meant to be incompatible with other roles directly involved in phases of the policy process, unless the nomination explained how a conflict would be resolved if the candidate won.

The proposal also addressed geographic concentration. The two chairs generally could not come from the same country. If no other acceptable candidate existed, Draft 1 allowed an exception, then supplied a way to exit that situation at the following election. This was not a complete theory of regional representation, and nationality alone cannot prove whom a chair represents. Still, the clause showed that the authors understood the symbolic and practical concern raised by two-person leadership drawn from one national setting.

Terms were equally explicit. Chairs would serve staggered two-year periods. A chair could win one consecutive re-election, then would need to wait at least a year before running again. Staggering promised continuity; the consecutive-term limit promised some renewal. Neither feature established electoral legitimacy by itself, but both made succession more predictable than open-ended incumbency or simultaneous replacement.

The calendar began on the first day of the month three full months before an incumbent’s term ended. Nominations would remain open for that natural month—between 28 and 31 days depending on the calendar. AFRINIC would then publish the acceptable candidates. A ten-day information period allowed confirmed information about candidates to emerge, and the Board could disqualify a candidate. Voting would last seven calendar days. Results were to be announced immediately.

Draft 1 anticipated close outcomes. A difference of less than 10 per cent would be treated as a tie. If more than two candidates had stood, a seven-day runoff would take place between the top two. The text, however, did not say what the 10 per cent measured: total eligible voters, votes cast, valid ballots, the leading candidate’s tally, or the difference between the top two as a proportion of some other base. That is not a minor drafting curiosity. Tie mathematics can decide whether an election ends or reopens, and the denominator determines the answer.

It also anticipated misconduct. The proposal warned against electoral use of the RPD list in circumstances showing collusion and contemplated candidate disqualification. Sufficient indications of fraud could disqualify affected candidates or invalidate and repeat the process. These provisions demonstrated that the authors were not naïve about strategic behaviour. Yet terms such as evidence, sufficient indications, collusion and fraud require standards, a finder of fact, a reasoned decision and a remedy. Naming a risk is not the same as designing a trustworthy procedure for proving it.

After results, the process still was not complete. Objections from the community had to reach the Board within seven days. The Board would decide whether an objection was significant and proven. If there were no objections, or if the Board did not accept them, the process moved to ratification. If ratification failed, the Board would appoint an interim chair until the following public policy meeting. A later election would fill only the balance of the original term. A similar replacement mechanism applied when a chair resigned or failed to meet the role’s responsibilities.

Finally, the Board was the highest appeal instance, although functions could be delegated to Election and Nomination Committees.

Taken together, these clauses were much more than a preference for online voting. They formed a complete architecture extending from who could stand and vote to how a vacancy would be bridged. That completeness is precisely why Draft 1 deserves close attention. Its value lies not only in what it fixed, but in how clearly it revealed the unresolved junctions between participation, identity, institutional control and mandate.

Six months on a list: a boundary, not an identity

The proposed electorate was Draft 1’s central innovation. Six months of RPD subscription offered three plausible advantages. It discouraged an election-day influx organised solely to affect the result. It treated online participants as more than remote spectators. And it replaced the physical happenstance of being in the meeting room with an eligibility rule visible in advance. In a region where travel expense, visas, time and employer budgets affect who can attend a meeting, removing the room as the only practical ballot box could broaden access.

Yet a mailing list records accounts, not necessarily unique people. The distinction is decisive. One person may control more than one address. A shared address may represent several people. An address may remain subscribed while its human user is inactive, changes role or leaves an organisation. An automated or archival account may receive mail without exercising judgment. Draft 1 said mechanisms would try, as much as possible, to ensure one vote per voter, but it did not describe how eligible accounts would be resolved into unique natural persons.

Nor did six months of subscription establish participation. Tenure shows that an address remained on a list for a period. It does not show that the holder read the discussions, contributed, understood the proposals or was accountable to anyone affected by them. Draft 1 wisely avoided a subjective test of whether a contribution was good enough, but it did not replace that subjectivity with an objective activity rule. List tenure and participation remained different concepts.

A third distinction separates a person from a principal. Someone may be a stakeholder because a registry decision affects the person’s work, organisation, customers or network. That does not automatically make the person an authorised representative of all those interests. Draft 1 connected candidates to AFRINIC membership contacts, but its voter clause did not say that a voter had to be a membership contact, carry an organisation’s delegation or vote in an individual capacity. It therefore left open whether the electorate was a collection of individuals, representatives of member organisations, or a mixture of both.

Those models are not interchangeable. An individual-voter system can honour personal expertise and participation, but it should say that institutional size or membership does not multiply voting power. An organisational system can connect ballots to accountable principals, but it must identify which organisations qualify and who is authorised to speak for each. A mixed system needs rules preventing a person from voting once personally and again through one or several organisations. Draft 1 did not settle among these choices.

Most importantly, a list electorate could not represent Africa by assertion. AFRINIC’s service region is vast, but it is not a polity that has delegated government to a mailing list. Many operators, users and affected businesses may never subscribe. Some subscribers may live or work outside the region. Some participants may have deep technical knowledge without any mandate from an African network. Others may carry organisational authority but rarely post. The word “community” can describe a forum. It cannot, on its own, prove a constituency, a denominator or consent.

Lu Heng’s analysis of multistakeholder governance gives this distinction its sharpest form: stakeholders are affected parties, not automatically principals; attendance is not mandate; a service region is not a polity; coordination is not sovereignty. Applied here, that doctrine does not make the RPD list worthless. The list is a valuable discussion and evidence channel. It allows proposals to be tested, objections to remain visible and implementation risks to be recorded. But participation in that channel does not silently transfer public authority to the group or to the chair it elects.

An auditable electorate would have begun with a declared purpose. Was the election choosing a facilitator trusted by recurring participants, a representative of member organisations, or an officeholder accountable to a defined set of affected principals? Once that purpose was stated, the process could identify the relevant human voters, publish eligibility counts, correct errors and distinguish personal participation from organisational authority. Draft 1 began with the tenure rule instead. It made the boundary more definite, but not yet authoritative.

Candidate legitimacy before the ballot

Voter identity was only half the entrance problem. The candidate path mixed objective requirements with discretionary judgment. Six months on the list, a membership-contact connection, a ban on AFRINIC staff, a two-term sequence and specified disclosures could all be checked. The conflict rule also addressed a real problem: a chair who occupied another role directly involved in a policy phase could be asked to manage a process in which the chair had a separate responsibility or interest.

But several words did substantial work. AFRINIC would publish “acceptable” candidates. During the information period, confirmed information could lead the Board to disqualify someone. The same-country restriction could be relaxed if there were no other acceptable candidates. Each of those steps required an answer to the same questions: acceptable by which criteria, confirmed through what evidentiary process, decided by whom, explained in what form, and corrected through which independent route?

Objective rules do not eliminate judgment. A biography can be incomplete. A claimed contribution can be disputed. A membership contact may lose the role between nomination and voting. A conflict-mitigation promise may be too vague. A candidate may challenge information supplied by an opponent. Good administration makes those judgments reviewable by stating standards in advance, giving notice, offering a reply period, recording recusals and publishing reason codes. Draft 1 supplied a ten-day window but not that full decision structure.

Contemporaneous RPD discussion identified this seam. Participants questioned the Board’s involvement in appointment, disqualification and appeal. They raised the possibility of multiple accounts, the problem of voter identification, the meaning of collusion, candidate diversity and whether objection outcomes would be published. These were not proof that any actor intended to abuse the system. They were design tests. A process can be vulnerable to arbitrary outcomes without anyone acting arbitrarily; the vulnerability arises when the rules leave too much of the path implicit.

The country-diversity rule illustrates both the benefit and the limit of such drafting. Avoiding two chairs from one country can reduce a visible concentration and broaden the experience around the table. Yet country is an imperfect proxy for employer, commercial interest, technical school, language, gender or the communities affected by registry policy. AFRINIC-31 participants also raised gender equality. Draft 1 did not offer a general representational formula, and it should not be credited with one. Its nationality clause was a bounded eligibility feature, not evidence that the resulting pair would speak for the region.

The unpaid nature of the office adds another constraint. Voluntary service can protect independence from a salary line, but it can also narrow the candidate pool to people with supportive employers, personal resources or unusually flexible time. The proposal did not claim to solve that participation cost. Candidate disclosures could help voters assess experience, but they could not equalise the capacity to campaign, attend meetings or absorb the work. This is one reason a credible election must be described as a mandate from its actual electorate, not as a universal choice by everyone affected.

An electronic ballot without a visible authority record

Draft 1’s electronic vote had a strong democratic intuition. If chair selection occurred only at a physical meeting, the result would depend heavily on who could reach the venue. A seven-day remote ballot could separate voting from travel and let eligible list participants act across time zones. It could also give voters more time than a show of hands to consider candidate information. For a technical community already organised substantially through email, electronic voting was not an eccentric mechanism.

The weakness was not that the ballot was electronic. It was that the proposal did not specify the authority record around the ballot. The text did not identify the platform, the credential-issuance method, the privacy model, the deduplication method, the frozen electoral roll, the audit trail or the correction route. It did not say how a voter would learn that someone else had obtained a credential in the voter’s name, whether a lost ballot could be reissued, how a duplicate would be resolved, or which counts would be published after the election.

“One voter, one vote” depends on the noun as much as the number. If “voter” means a subscribed address, multiple addresses can distort the human count. If it means a natural person, the administrator needs a defensible way to bind accounts to identity while protecting privacy. If it means an organisation, the process must record authorisation and prevent parallel credentials. Draft 1 expressed an intention to ensure one vote as far as possible; it did not establish which of these meanings would govern.

The present-day election-integrity work published by NRS is useful here as an analytical principle, not as evidence about what happened in 2019. NRS asks whether a person can test if a vote was registered in that person’s name, supports lawful reporting of irregularities, and expressly does not claim an AFRINIC mandate. The important principle is a voter-verifiable authority record: public confidence improves when there is a privacy-safe register, a way for the voter to verify registration, and a lawful challenge path.

That principle avoids two opposite errors. Full publication of identities and vote choices would expose voters and could chill participation. Total institutional opacity would force everyone to trust the administrator’s assertion that the list and count were correct. A privacy-preserving commitment can sit between them: the electorate and eligibility count are frozen; each eligible human can verify inclusion; each issued credential is unique; each voter can confirm that a ballot was recorded without revealing the choice; aggregate issued, returned, invalid, duplicate and corrected counts are published.

Draft 1’s immediate-results rule also needed that supporting record. Speed is valuable, especially when a term is ending, but an instant number without reconciliation is not transparency. The seven-day objection window started after the result. A voter who could not verify inclusion or ballot receipt would have little evidence with which to object. The Board, in turn, would have to assess significance and proof using a record the proposal did not define.

The undefined tie denominator compounds the issue. Imagine a reported margin below 10 per cent. Whether that is a tie cannot be independently tested until the base is known. The correct base is a policy choice, not a mathematical inevitability. It might be valid votes, all ballots returned or some specified comparison of candidate totals. Draft 1 chose the threshold but left the base unstated. A better text would publish the formula and apply it to reconciled counts before any election began.

Nothing in the published record establishes that an election was actually conducted under Draft 1, let alone that electronic voting caused manipulation. AFRINIC-31 participants raised concerns about exposure to e-voting fraud, voter verification and multiple accounts. Those concerns show why a design needed more detail. They do not prove fraud, a bad count or an invalid election. Institutional criticism is strongest when it identifies the missing control without inventing the failure the control was meant to prevent.

The Board at every gate

Draft 1 identified the AFRINIC Board as a responsible institutional actor, which is not in itself a flaw. A private association cannot run a durable process through nobody. Someone must authorise resources, appoint administrators, decide whether criteria are met and keep the office functioning when a vacancy occurs. Naming responsibility can be better than hiding it behind an undefined community.

The concern was accumulation. In Draft 1, acceptable candidates would be published through AFRINIC. Confirmed information could reach the Board and produce disqualification. Post-result objections went to the Board, which assessed whether they were significant and proven. The election then proceeded to Board ratification. A failure to ratify led to a Board-appointed interim chair. The Board was also the highest appeal instance. It could delegate functions to Nomination or Election Committees, but the text did not provide a complete map separating screening, administration, first decision and review.

That concentration matters even when every decision-maker acts in good faith. If the body connected to candidate screening also judges complaints, ratifies the winner, supplies an interim replacement and decides the final appeal, a candidate or voter may struggle to identify an independent correction point. Delegation does not automatically create independence. A committee may be functionally separate, or it may act under the same instructions and return contested decisions to the same ultimate authority. Draft 1 did not say.

The Board’s candidate role needed objective standards and reasons. Its objection role needed an evidence threshold, notice to affected parties and a disposition that explained the result. Ratification needed bounded grounds; otherwise an electoral winner could face a second, undefined political choice. The interim-appointment power needed a strict clock and a statement that it was for continuity rather than a substitute for voters. Final appeal needed a reviewer institutionally distinct from the first contested decision.

These are not demands for a public court. AFRINIC is a private Mauritius-incorporated technical bookkeeper and uniqueness coordinator. It can administer its association and create review within its own procedures. The proportionate design is a private remedy for a private office: published criteria, separate functions, conflicts and recusals, a correction receipt, a reasoned result and a bounded appeal. The remedy should be capable of correcting the election without pretending to exercise public-law adjudication.

The Board chain also sharpens the distinction between legal form and rhetorical community. A vote might be described as the choice of the community, yet ratification and appeal could still place decisive authority with the corporate institution. Conversely, the Board might say it was merely giving effect to community choice while retaining several opportunities to determine which choices reached or survived the ballot. Neither description should substitute for a power map. Draft 1’s clauses show who could act, and they show why the design needed more separation.

LARUS’s accountability framing is useful on this point. Network operators need continuity, predictable administration and a route to challenge decisions that affect planning. But continuity is not a reason to collapse all roles into one centre. A temporary appointment may keep a process running, while strict term limits, published reasons and independent review keep the temporary power from becoming self-justifying. Accountability is built from traceable actions and correction, not simply from naming the organisation that remains responsible.

Objections, vacancies and the missing recall

Draft 1’s objection window was a genuine procedural advance. It recognised that announcing a result should not foreclose challenge and that complaints should arrive within a known period. Seven days gave the process a clock. The fraud provisions also acknowledged that some defects could require candidate disqualification or a repeated process rather than a rhetorical rebuke.

But the text left the objector’s practical burden uncertain. What counted as a significant objection? What proof was required? Could a voter see the registration and count records needed to support a claim? Would the candidate receive notice and an opportunity to answer? Would accepted and rejected objections be published with reasons? Would a Board member connected to a candidate or disputed decision recuse? The available Draft 1 text did not complete those rules.

Its vacancy clauses were clearer about continuity. If ratification failed, the Board appointed an interim chair until the next public policy meeting, after which an election filled the remainder of the original term. Resignation or failure to meet responsibilities triggered the same mechanism. This limited the replacement term and preserved the staggered sequence rather than resetting the calendar around a new full term.

That was not an electorate-driven recall procedure. Draft 1 did not state that voters could petition for, trigger or decide a recall ballot. “Failure to meet responsibilities” required someone to determine both the responsibility and the failure, while the Board’s interim appointment carried the office to the next meeting. Describing these clauses as recall would overstate the electorate’s power and obscure the actual allocation. The draft handled resignation, failure and vacancy; it did not give voters a full removal vote.

Again, the problem is not that every private office must mimic a constitutional system. It is that labels should match mechanisms. A vacancy rule preserves service. A removal rule identifies grounds and a decision-maker. A recall gives the electorate a way to withdraw its mandate. Draft 1 contained the first and part of the second, not the third. That boundary belongs in any accurate account of its election architecture.

The Luanda test and the no-consensus result

The proposal went from the RPD list to AFRINIC-31 in Luanda, where a 30-minute session on 5 December 2019 put its design under direct scrutiny. The meeting minutes record floor and remote concerns about electronic voting, voter verification, multiple accounts, the adequacy of six-month list tenure, the loss of face-to-face voting, candidate presentation, repeat candidacy and gender equality. The RPD thread in the preceding weeks had also tested Board involvement, disqualification, collusion, objections and independent appeal.

This mix of concerns is revealing. Some pulled in opposite directions. Remote voting could reduce the privilege of travel, while the loss of face-to-face voting worried participants who valued deliberation around the meeting. A subscription threshold could deter sudden mobilisation, while list tenure alone could admit dormant or duplicate accounts. A term limit could promote renewal, while repeat candidacy might preserve experience. A country rule could resist one form of concentration, while leaving gender and other forms of representation unresolved.

The co-chairs, Moses Serugo and Abdulkarim Oleyede, later reported no consensus. On 20 December the proposal returned to the list. The record names unresolved concerns about electronic voting and candidate criteria. The minutes also state that nobody at the public policy meeting supported the policy. That sentence should be preserved with its evidentiary limit: the published material supplies no support headcount, attendance denominator or roll-call record from which a reader can independently reconstruct the claim.

The absence of a denominator is not a reason to reverse the result and declare hidden support. It is a reason to distinguish an official procedural record from an auditable measurement. AFRINIC’s official minutes prove what the meeting record and co-chairs reported. They do not, without more, prove the identity or authority of every person counted, the total relevant population, or a continental mandate. The correct conclusion is the narrow one: Draft 1 was judged not to have consensus and was sent back for work.

AFRINIC’s 2019 annual report recorded the proposal as “Under Discussion” at year end. The available evidence does not show Draft 1 adopted, implemented or binding. It does not show an election conducted under the draft. It does not show fraud, manipulation, capture, illegality, discrimination, bad faith or concrete loss caused by the proposal. A design can fail to command consensus because its safeguards are incomplete without being a vehicle for proven wrongdoing.

This outcome also illustrates the office’s own mandate problem. Co-chairs had to interpret a discussion, identify unresolved objections and report a consensus disposition. Their procedural power was real inside the forum, even though it was not public law. The more traceable their appointment and reasoning, the easier it is for participants to accept a no-consensus decision as a disciplined assessment rather than a personal preference. The Draft 1 debate was thus self-referential in a productive way: it tested the procedure for selecting the people who would judge whether procedures had achieved rough consensus.

One draft, two identifiers

The record carries a discrepancy that should not be cleaned away. The contemporaneous four-page attachment posted to the RPD list identifies itself twice, including in its revision-history line, as AFPUB-2019-GEN-003-DRAFT01. AFRINIC’s later and current archive trail identifies the same version dated 2 November 2019 as AFPUB-2019-GEN-007-DRAFT01. The GEN-007 form appears in the AFRINIC-31 minutes, the 2019 annual-report trail and the revision history associated with the later proposal page.

The evidence available here does not establish when the number changed, why it changed, or which administrative step produced the difference. Both identifiers therefore belong in the public account. GEN-003 describes what the attached contemporaneous PDF says about itself; GEN-007 describes how AFRINIC’s later and current records catalogue the 2 November version. Silently normalising one to the other would make the archive look cleaner than the evidence permits.

The discrepancy does not prove that the text was altered, hidden or manipulated. The relevant sources connect both numbers to Draft 1, but they do not explain the numbering history. Its importance is more basic: an election architecture concerned with traceability should itself be cited with documentary precision. When the identifiers differ, accuracy requires the uncertainty to remain visible.

Draft 2 is later context, not a repair kit

A second draft was submitted on 19 August 2020. The proposal page and the staff assessment dated 30 August 2020 are later evidence. They cannot be poured backward into the November 2019 text. In particular, later clauses must not be presented as safeguards that Draft 1 already contained, and later staff questions must not be described as a contemporaneous legal review of the first draft.

The later staff assessment did ask questions that confirm the institutional seam was real: whether chair selection belonged outside the policy process under the Bylaws, and whether making the Board the final election appeal displaced the appeal mechanism in section 3.5 of the Consolidated Policy Manual. These questions are useful as later institutional recognition of the boundary between chair elections, Board authority and policy-process appeal. They are not proof of what Draft 1’s authors intended beyond their text.

AFRINIC’s 2021 annual report later recorded Draft 2 as expired. That status belongs to Draft 2. It does not transform Draft 1 into an adopted policy, nor does it establish that Draft 1 remained operative until 2021. Keeping the instruments separate protects both criticism and credit. Draft 1 should be judged for the election architecture it actually proposed in 2019, not for provisions or assessments that arrived the following year.

Why a procedural mandate has economic weight

The chair office remains narrow, but narrow does not mean inconsequential. Network policy proceeds through time. A proposal is drafted, discussed, revised, tested for objections, assessed for rough consensus and possibly moved toward last call. A chair’s choices about how objections are classified and whether they have been addressed can accelerate, pause or redirect that sequence. The decision does not become legislation, yet organisations still plan around its procedural effect.

The costs are unevenly distributed. A large operator or recurring participant may maintain staff who follow a mailing list, attend meetings, brief executives and revise network or legal plans as proposals change. A smaller operator may rely on one engineer or executive whose attention is already divided among operations, customers and compliance. Repeated uncertainty consumes travel budgets, legal review, engineering time and monitoring effort. Those with greater resources can remain in the room—or on the list—longer than those with less.

Where a proposal concerns scarce IPv4 resources, transfers, RPKI, registry data, fees or compliance, delay can shape expectations about network expansion and continuity. That does not mean a chair owns the resources or commands the networks. It means private procedural decisions can alter the environment in which operators decide when to invest, whom to contract with, how to manage risk and whether a policy change is likely to proceed.

An auditable chair election lowers one specific risk: the risk that participants accept or reject a procedural judgment only because they trust or distrust the individual chair. If voters can test eligibility, ballot registration and the count, and if objections receive independent reasons, the office rests less on personality. The election does not guarantee a correct consensus assessment. It provides a defensible explanation for why these particular people have the temporary responsibility to make that assessment under known rules.

The limit is equally important for markets. A better internal mandate cannot turn an AFRINIC procedure into title over Internet number resources. It cannot create a police power to punish a network, a legislative power to bind non-consenting operators, a prosecutorial power to pursue alleged wrongdoing, or a judicial power to settle public rights. Treating procedural legitimacy as regulatory legitimacy would amplify the economic stakes beyond what the institution can lawfully and practically justify.

The discipline runs in both directions. Operators should not pretend that a private chair decision is irrelevant merely because it is not public law; it can affect the schedule and uncertainty surrounding technical coordination. AFRINIC should not treat the effects of coordination as evidence of sovereignty. A credible system acknowledges operational consequence while maintaining the line around institutional authority.