Summary

  • A ballot from each eligible legal member can be formally valid even when several members share one ultimate controller. Formal validity does not answer whether the institution intended to represent legal entities, independently directed operators, resource-linked responsibility or a mixture of those principles.
  • Corporate affiliation and proxy representation are different relationships. An affiliate may cast its own ballot under a group instruction; a proxy may carry another independent member's mandate. Election reports should disclose both without treating a common voting contact as proof of common control.
  • The relevant fact is election-specific voting authority on a fixed date. Ownership, joint-control rights, board appointments, management agreements and reserved matters can change between the record date and the ballot. A current corporate-family list applied backwards is not a reliable audit.
  • Every result should publish at least four denominators: eligible legal members, eligible voting weight, represented control groups and actual ballots or voting weight by control group. In weighted systems, both organizations and entitled votes remain necessary.
  • A reproducible control map needs source-linked edges, effective dates, confidence, a reason code and a correction path. Brand names, shared domains, common directors, BGP adjacency and one human voting for several organizations are leads, not proof.
  • Candidate affiliation and board recusal belong in the same system. Group influence can enter through nominations and post-election decisions even where ballot shares look dispersed; recusal records must identify the matter, conflict category, participation limit and effect on quorum.
  • The institution must choose openly among legitimate models: preserve one vote per legal member with group disclosure, cap votes per control group, use one-group-one-vote for specified constitutional matters, or create a mixed structure. Data cannot make that constitutional choice, but it can stop the choice from remaining hidden.
  • Number Resource Society can research published rules, represent member concerns and advocate comparable disclosure. It cannot certify group control, administer elections, adjudicate affiliation disputes or alter ballots; implementation belongs to each RIR and review belongs to an empowered independent process or competent court.

Twenty ballots can be valid without representing twenty independent decisions

Consider a close registry election in which twenty eligible organizations cast ballots. Each organization has its own incorporation record, membership agreement, account and named voting contact. Nothing in the ballot file is duplicated. No proxy limit is breached. The election service accepts every credential and produces a correct cryptographic tally.

Now add one fact: all twenty organizations are subsidiaries of the same corporate group, and a group public-policy office approved their voting instruction. The votes have not become fake. The companies may be real operating entities with different licences, customers, networks and local directors. The staff who submitted the ballots may have followed each company's internal authority perfectly. Yet it would be misleading to describe the result as twenty independent corporate judgments without disclosing the shared control.

Change the fact again. Suppose the parent owns all twenty companies but local boards retained an express right to decide registry elections independently. Some subsidiaries support one candidate, others another. A rule that automatically collapses the group to one vote would now erase documented autonomy. Common ownership is strong evidence of potential direction, not an infallible substitute for election-specific authority.

This is the mechanism the title tests. It does not assert that a named telecommunications group holds twenty memberships in any RIR or that any published election was improper. Twenty is a stress-test number. It is large enough to reveal the difference between legal-member validity and independent control, while remaining small enough to matter in low-turnout elections or closely divided constitutional votes.

The first governance duty is therefore descriptive. A registry should be able to say how many legal members voted, how much voting weight they carried, how many independently directed control groups were represented, which classifications were uncertain and what rules converted those populations into the certified result. Until those counts exist, arguments about fairness begin with different electorates and talk past one another.

The rulebook must identify the unit it means to represent

One-member-one-vote sounds complete, but the word member can describe only the legal unit named in the governing document. It does not explain why that unit deserves separate political weight after control has been centralized. Nor does it say that affiliates should be combined. Corporate law ordinarily treats subsidiaries as separate persons even when accounts are consolidated and strategy is directed from a parent.

RIR voting systems already embody different representative theories. RIPE NCC's published Articles provide one vote for each unsuspended member and limit the number of proxies one person may carry by reference to all possible votes. ARIN assigns voting rights to General Members in good standing and has the vote exercised through designated Voting Contacts. APNIC links voting entitlement to membership tiers, from one vote at the lower end to 64 for the Extra Large tier. LACNIC also uses resource-linked voting bands, from one to 11 votes under its published bylaws.

These designs may represent legal equality, resource stewardship, fee responsibility, operational scale or historic institutional compromise. None automatically answers the corporate-group question. A one-member-one-vote system can multiply a group's voice through separately incorporated affiliates. A weighted system can concentrate influence within one large member or distribute it among several affiliates. A proxy rule can limit one human's delegated mandates while leaving coordinated affiliate voting untouched.

The institution should state its constitutional objective in plain language. If every separately liable legal member receives a vote regardless of shared control, that is a defensible choice. If voting power is meant to measure independent operating judgment, common control is directly relevant. If resource scale is intentionally represented, group consolidation may be part of the design rather than an anomaly. A mixed purpose needs a mixed explanation.

The worst outcome is an unstated switch. An institution should not celebrate legal-member plurality when that produces a large denominator, then describe the same ballots as independent operator support when common control has not been measured. The certified count and the interpretation of that count must use named units.

Corporate affiliation is not the same relationship as a proxy

Election systems often record proxies more carefully than affiliates because a proxy is visible at the ballot boundary. Member A authorizes a person to vote on its behalf. The election administrator can check the authorization, enforce a cap and record that the mandate came from A. The relationship is explicit even though the ballot may remain secret.

Affiliate direction can be invisible at the same boundary. Subsidiaries A, B and C each use their own authorized contact. No formal proxy exists. A parent company's governance committee has nevertheless decided the group's position. Three credentials enter the system and three ballots are valid, while the shared instruction remains outside the election record.

The reverse mistake also matters. One professional administrator, lawyer or industry representative may be the voting contact for several unrelated organizations. ARIN's 2025 election results reported 772 General Members casting ballots through 607 Voting Contacts. The difference does not prove that the organizations shared ownership or a common ballot instruction. It proves only that organizations, contacts and ballots are distinct counts. A control audit that treats every shared contact as one corporate group would manufacture concentration.

The data model needs separate edges. AFFILIATE_OF describes corporate control. PROXY_FOR describes a delegated electoral mandate. VOTING_CONTACT_FOR describes the person authorized to submit a ballot. INSTRUCTION_APPROVED_BY describes the internal authority that determined the position, where disclosure is required. REPRESENTS_CONSTITUENCY describes an association or national registry mandate that is not ownership.

Each edge should have its own proof and effect. A proxy may be capped even when the members are independent. Affiliate votes may remain fully valid but be group-deduplicated for reporting. A shared contact may require no adjustment at all. A national representative may aggregate hundreds of local interests under a published mandate without becoming their corporate parent. Precision prevents a transparency reform from turning every form of coordination into suspicion.

The decisive entity is a dated control-and-instruction graph

A corporate-family list is not enough. The audit needs a graph built for the election. Nodes should include legal members, service accounts where relevant, parent companies, joint ventures, controlling public bodies, voting contacts, proxy holders, candidates and the ballot entitlements certified for the event.

Edges should describe an actual relationship: majority ownership, joint control, power to appoint a board majority, contractual management authority, reserved-matter veto, proxy mandate, voting-contact appointment, candidate employment or declared independence of instruction. Every edge needs a start date, and where applicable an end date. A graph without dates applies today's ownership to yesterday's vote and can reverse history.

Formal sources deserve the greatest weight. Competition decisions can identify transaction parties and whether control is sole or joint. Securities filings and audited accounts disclose subsidiaries and material investments. Company registers establish legal existence, directors and some ownership changes. Court orders can transfer or restrain control. Registry-recognized merger or acquisition records connect a legal change to membership or number-resource administration. A signed member declaration can disclose who authorized the election instruction when public corporate records cannot.

Weak signals can locate questions but should not decide them. Shared email domains, postal addresses, directors, nameservers, upstream providers, autonomous systems or brands may reflect common control, outsourcing, a franchise, a service relationship or coincidence. A group website may lag behind a sale. An acquired brand may persist after authority has moved. A route can remain technically separate under common ownership.

Each group assignment should therefore store the source, observed fact, inference, confidence, effective period and reviewer. The public report can show a concise reason code such as majority-owned, jointly-controlled, parent-directed ballot, independent mandate documented or unresolved. Sensitive supporting material can be reviewed confidentially, but the classification cannot be an unexplained private label.

Control for an election is narrower than beneficial ownership

Beneficial-ownership reporting is a useful starting point, but it answers a different legal question. It seeks the natural person or ultimate entity that enjoys ownership or control. A registry election needs to know who could instruct or approve a member's ballot on the record date.

A publicly listed parent may have thousands of shareholders. Treating each fund as a controller of an operating subsidiary would be nonsensical. Corporate authority normally runs through the parent board and management structure. A state-owned operator may answer to a ministry, statutory board or holding company, but not every public network in the country necessarily shares one election instruction.

Joint ventures are harder. Two parents can share decisive rights while neither has unilateral control. The venture may have a board with reserved matters and an independent management team. For reporting, it can be assigned to a joint-control cluster without pretending it belongs wholly to either parent. If both parents also have member affiliates, the graph should show overlapping influence rather than forcing a tree where the legal structure is a network.

Minority protection is not the same as direction. An investor may veto a sale of all assets while having no authority over routine governance votes. A creditor may temporarily approve major expenditure during restructuring. An insolvency practitioner may displace ordinary management for a defined period. A trust or regulatory separation may prevent a parent from instructing a subsidiary on specified matters.

The practical question should appear on the declaration: “Which person or body had authority to determine or approve this member's vote for this election?” A member can answer with its own board, a parent committee, a joint body or another documented authority. The answer does not expose the ballot choice. It identifies the centre of judgment the institution is counting.

A minimum affiliate declaration can be short and consequential

The registry does not need a complete commercial-intelligence dossier. A minimum election declaration can collect five facts.

First, the legal member's current name and identifier. Second, the ultimate corporate control group or a statement that no higher controlling entity exists. Third, any other eligible member known to share that control. Fourth, the body authorized to determine this election's vote. Fifth, changes expected or completed between the eligibility record date and the close of voting.

The form should distinguish sole control, joint control and declared independent instruction within a group. It should also ask whether the same person is acting as a formal proxy for other members, because that relationship is governed separately. A checkbox saying “no affiliates” is not enough where public records show a parent; the member should be able to explain a regulatory or governance separation.

The declaration should be signed by an officer or other person authorized to speak for the member, not necessarily the voting contact. False statements should have proportionate electoral consequences, such as correction of the disclosure, loss of eligibility for a future period or referral under the bylaws. A disputed classification should not threaten the member's number resources, routing entities or registration continuity. Governance sanctions and registry operations address different risks.

For large public groups, the registry can pre-populate known affiliates and ask members to confirm. For private companies, confidential documents may be reviewed by an independent election assessor. The public output can name a confirmed group where authoritative records already do so, or publish an anonymized group identifier where disclosure would be disproportionate.

The point is not to collect ownership for its own sake. It is to produce a verifiable denominator and identify conflicts before ballots are counted.

Four denominators are the minimum, not an optional appendix

An election result can be mathematically correct and rhetorically misleading if it reports only the denominator most favourable to a legitimacy claim. Group-level disclosure should make four populations standard.

The first is eligible legal members. This is the formal population under the bylaws and the denominator against which legal-member turnout is calculated. The second is eligible voting weight. It matters wherever members have more than one vote and remains useful when one legal member carries a tiered entitlement.

The third is eligible control groups. Every legal member is assigned to a confirmed group, an independent-member node or an unresolved cluster. The fourth is participating control groups and their associated ballots or voting weight. This shows whether turnout that appears broad at member level depends heavily on a small number of group instructions.

Contacts and proxies should be reported alongside, not substituted for, these four. ARIN's published separation of eligible organizations, voting members and Voting Contacts demonstrates why. RIPE NCC meeting reports distinguish the total membership, members registered to vote and ballots cast. LACNIC reports organizations participating as well as votes cast in a weighted system. APNIC's tier table makes clear that organizational and vote counts answer different questions.

Suppose 1,000 legal members are eligible, 300 vote, and those 300 belong to 250 control groups. Legal-member turnout is 30%. Group turnout depends on the total number of eligible groups, perhaps 900. If one group supplied twenty of the 300 member ballots, its member-ballot share is 6.7%, while its control-group share is one of 250. Neither number says how the group voted. Together they show how formal plurality and independent direction differ.

Every percentage should name its numerator and denominator. “Participation reached 30%” is incomplete. “Thirty per cent of eligible legal members cast ballots, representing 27.8% of eligible control groups and 32% of entitled voting weight” can be audited.

The twenty-vote stress test should be replayable before certification

A registry can test its rules without alleging that a real group coordinated. Create a synthetic election with twenty affiliate members and vary one fact at a time.

In Scenario A, one parent has sole control and issues a common voting instruction. All twenty affiliates are eligible and use separate contacts. Under a pure legal-member constitution, all twenty ballots count. The result report nevertheless shows one control group contributing twenty ballots. Under a group cap of five, fifteen ballots or portions of voting weight require a published treatment. Under one-group-one-vote for constitutional amendments, the group contributes one such vote.

In Scenario B, the same parent owns the affiliates, but each local board has documented independent authority over registry elections. Ten vote, six abstain and four do not participate. The control graph retains the ownership edge while the election-instruction graph records independent nodes. A system that mechanically deduplicates by parent would fail this scenario.

In Scenario C, twenty unrelated members appoint the same proxy. A proxy cap may invalidate or reassign some mandates, but no corporate-group adjustment follows. In Scenario D, ten affiliates and ten unrelated members share an outsourced administrator who merely submits their independently approved ballots. Contact deduplication would again be wrong.

In Scenario E, a transaction closes after the eligibility record date but before voting. The old parent had authority at record date; the new parent controls instruction by ballot date. The rules must state which date governs affiliation and whether a material change triggers re-certification.

The election administrator should publish expected outcomes for these scenarios before a live dispute. Test fixtures can run against the software that prepares entitlements and disclosure reports. A constitutional rule that cannot produce a stable answer to a synthetic twenty-member group is not ready for a close real election.

Control dates need the same discipline as ballot dates

Corporate control does not move at one universal moment. A transaction is announced, reviewed, approved, closed, registered and operationally integrated on different dates. Each may matter for a different purpose.

The election should define at least three dates. The eligibility record date establishes which legal members and voting weights may participate. The affiliation snapshot date establishes the control graph used for disclosure or any group rule. The instruction date identifies the authority that approved the ballot. These dates can be the same, but the rules should not assume they are.

Official merger decisions provide reliable evidence of control structure and closing conditions. The Virgin Media and O2 combination, the Orange and MasMovil joint venture, and the Vodafone and Three joint venture illustrate why joint control, remedies and effective dates require specific records. They are not evidence that any of the companies held or coordinated registry votes. They demonstrate the kind of public source an assessor can use to timestamp a group edge.

Registry procedures add another clock. RIPE NCC asks organizations to report mergers, acquisitions and other changes in business structure. ARIN's transfer process requests evidence that relevant assets or a registrant were acquired. APNIC's policy distinguishes a subsidiary whose infrastructure remains fully independent from one whose infrastructure has effectively merged. Those processes protect registration accuracy; an election audit can use confirmed facts without treating resource-transfer recognition as the sole test of ballot authority.

A material change notice should be mandatory from the record date through certification. If a parent sells a member, a court appoints an administrator or a joint venture becomes solely controlled, the assessor updates the edge and reruns the published metrics. Historical reports retain the version used at certification rather than silently adopting the new family tree.

Uncertainty belongs in the result, not in a hidden analyst note

Some group relationships will be unresolved. Private ownership records can be incomplete. A joint-control agreement may be confidential. A member may contest that its parent can direct the ballot. The institution should not convert every ambiguity into either independence or affiliation merely to produce a clean total.

The report can publish a lower and upper concentration range. Confirmed affiliations form the lower-bound grouping. Plausible relationships supported by incomplete evidence form an upper-bound grouping. Documented independent mandates can be shown separately even where ownership is common. Cases with no adequate evidence remain unresolved rather than being assigned by brand recognition.

Confidence should attach to each edge, not to the company as a whole. Majority ownership established in audited accounts may be high confidence. A shared director can be low confidence. A signed election declaration can be high confidence about internal authority but limited to that election. A press report can corroborate but should not normally be the sole basis for restricting a vote.

The certified legal tally need not wait for every public-source question if the rules preserve legal-member voting. If a group cap affects ballot validity, however, unresolved classification becomes consequential and requires an empowered review before certification. That is one reason disclosure-first reform is easier to implement than an immediate cap.

Visible uncertainty protects both sides. It prevents a group from disappearing behind formal entities, and it prevents an assessor from presenting an allegation as established control. Year-to-year improvements in coverage become measurable rather than hidden behind an implausible claim of completeness.

Voting secrecy and group transparency can coexist

A group-level report does not need to reveal how a member voted. It needs to disclose the structure through which voting authority entered the election.

Public results can show that a named or anonymized control group had twenty eligible affiliates and that seventeen participated, without showing candidate selections. Where even participation would reveal a choice—for example, a one-question election with a published voter list and a very small group—the report can use ranges, delayed publication or cell suppression.

The independent assessor may need access to ballots only to compute group-level support ranges or test whether a cap was applied. If so, access should be separated from the registry board and staff with political interests in the outcome. The assessor publishes aggregate verification and destroys or seals event-specific linkage according to a retention policy.

The public control graph should minimize personal data. Corporate relationships established in public records can be named. Voting contacts do not need home addresses or private email details in the group report. Confidential shareholder agreements can be summarized through a reason code and assessor attestation. The registry should state the purpose, access rules, retention period and correction process before collecting the information.

Secrecy cannot become an excuse for withholding every denominator. Election administrators already report turnout without publishing ballots. The same separation works here: disclose who was eligible, the structural relationships relevant to weight, the number participating and the certified totals, while protecting the content of each secret vote.

Candidate affiliation is the other half of the same graph

Group influence does not begin when a ballot is submitted. It can enter through candidate recruitment, nominations, campaign endorsements and the supply of people able to serve on a demanding board.

Candidates should disclose their current employer, controlling group, material directorships, significant consulting relationships and recent executive roles for a defined lookback period. The purpose is not to stigmatize industry experience. Network institutions need directors who understand operations. The purpose is to let members see whether several nominally separate candidates share one current economic sponsor or whether a candidate may face recurring company-specific conflicts.

The control graph can connect candidate nodes to the same group identifiers used for member disclosure. Election reports can then state the distribution of current affiliations among nominees and elected directors without inferring that affiliated candidates will act identically. Historical employment should be dated and should not produce a permanent label.

Campaign support also merits a declaration. A group may lawfully endorse candidates or provide staff time. Disclosure becomes material when support is coordinated across affiliates or when a candidate's apparent independent campaign is financed or organized by one group. Rules should distinguish ordinary public endorsement from resources that create an obligation or conflict.

This data makes post-election conflict management possible. The institution can anticipate matters involving a director's group, identify whether several directors might need to recuse and plan for quorum. Without the common graph, affiliations are rediscovered ad hoc during a crisis and handled inconsistently.

Recusal must be a reproducible event, not a private gesture

A director saying “I did not take part” is not enough for an auditable conflict system. Recusal needs a record that can be reviewed without publishing confidential board deliberations.

The record should identify the meeting and agenda item, the director's relevant group affiliation, the conflict category, whether the director received papers, whether the director joined discussion, whether the director voted, who decided the treatment and whether quorum changed. Options can include full recusal, participation for technical information without decision, disclosure with no restriction, or appointment of an independent committee.

Group mapping matters because conflicts can be indirect. A decision may concern a subsidiary, joint venture or competitor of the director's employer. Not every group connection requires withdrawal. The policy should define materiality and allow a reasoned determination. Automatic recusal for every sector-wide issue would deprive the board of expertise and could make it impossible to act.

Repeated recusals create an institutional signal. If directors linked to one or two groups must withdraw from many matters, the problem is broader than individual ethics. Nomination and board-composition rules may need adjustment. A skills matrix can track operator experience, region, institution type and independence without imposing a simplistic ban on affiliates.

Recusal also does not cure concentrated voting. A group may help elect several directors who later comply perfectly with conflict rules. The legitimacy question remains whether the electorate understood the group's structural voting share. Ballot disclosure and board conflict management are complementary controls, not substitutes.

The audit should be reproducible from inputs to published denominator

An election audit earns trust when another qualified reviewer can reproduce the structural counts from the same evidence and rules. A spreadsheet of unexplained parent names is not enough.

The public methodology should define the node types, edge types, authoritative sources, snapshot dates, confidence thresholds, treatment of joint control, treatment of documented independent mandates, handling of unresolved cases and calculation of group-level voting weight. Version the methodology and identify every change from the prior election.

The machine-readable dataset can use stable pseudonymous member identifiers if the membership list is not public. Each edge carries a source reference or confidential-evidence attestation, effective dates, reason code, confidence and reviewer. The transformation from member entitlements to control-group totals should be published as code or precise pseudocode.

Input totals must reconcile. The sum of eligible legal members across group, independent and unresolved classifications must equal the certified eligible-member total. In a weighted system, the sum of member weights must equal authorized voting weight. Every ballot must map to one legal member and one event-specific group classification. Proxy mandates and voting contacts remain separate tables.

Change logs should show additions, removals, group splits, group mergers, corrections and expired evidence. A hash of the final input snapshot and the audit code can be published with the report. Confidential evidence need not be exposed for the computation to remain reproducible; the independent reviewer can attest that each restricted edge met the declared standard.

This discipline also catches ordinary errors. Duplicate member identifiers, a parent applied before closing, a stale group assignment after divestment or a weighted entitlement attached to the wrong affiliate will cause reconciliation or temporal tests to fail before they become a political dispute.

The audit must test its own false positives

Control mapping is powerful enough to suppress legitimate plurality if it is careless. The quality programme should deliberately sample cases likely to be misclassified.

One sample should cover shared voting contacts among unrelated organizations. Another should cover subsidiaries under common ownership that claim independent election authority. A third should cover franchises or brands without common corporate control. A fourth should cover state-related entities with different statutory mandates. A fifth should cover joint ventures where neither parent has sole direction. A sixth should cover recently divested members whose websites still display the former brand.

For every sampled case, the assessor records what the automated or public-source method inferred, what primary evidence established and whether the group assignment changed. False-positive and false-negative rates should be reported against the reviewed sample, with the limits of extrapolation.

Members need a practical correction route. They should see their proposed group classification before the election, receive the cited basis and submit contrary evidence. A decision-maker independent of the initial analyst should issue a reason. Urgent review targets should align with the election calendar. A correction after certification may improve history but cannot restore a lost governance opportunity.

The reviewer must also be able to reject strategic claims of independence. A parent resolution authorizing one group instruction outweighs a subsidiary's bare assertion that it “operates separately.” Conversely, common branding cannot override a legal agreement that reserves the vote to an independent board. The standard is evidence of direction, not the outcome a party prefers.

Publishing error analysis makes the system contestable. It prevents group deduplication from becoming a new, opaque registry power.

A disclosure-only model preserves legal votes and changes interpretation

The least disruptive institutional choice is to preserve every ballot valid under existing bylaws and add group-level reporting. In the twenty-affiliate scenario, all twenty ballots count. The result states that one control group supplied twenty ballots or the associated voting weight.

This model respects corporate personality and avoids giving an assessor power to invalidate votes through a disputed classification. It can be introduced without changing the mathematical result, though collecting and publishing affiliation may still require a rule amendment and privacy assessment. It creates a longitudinal evidence base before stronger remedies are considered.

Disclosure changes political interpretation. A candidate who received 100 member ballots may have support from 85 control groups rather than 100. A constitutional amendment backed by 60% of voting weight may depend on one group's affiliate cluster. Those facts do not nullify the result. They tell members whether the current constitution produces the plurality they intended.

Disclosure can also deter structural gaming. A group may retain legitimate affiliates, but it cannot present the affiliates as unrelated voices. Members can evaluate candidates and rule changes with the full denominator. Boards can see whether nomination or recusal reforms are needed.

The limitation is equally clear. If one group can determine outcomes under the legal-member rule, disclosure only describes the power. It does not constrain it. The institution must then decide whether concentration is acceptable, an intended consequence of its membership model or a reason for constitutional change.

A group cap limits multiplication but creates boundary power

A second model caps ballots or voting weight associated with one control group. The cap might be a fixed number, a percentage of eligible voting weight or a maximum only for specified decisions.

In the twenty-affiliate scenario, a five-vote cap forces a treatment of fifteen otherwise valid member entitlements. They might be disregarded, proportionally scaled, allocated by a group instruction or selected by the affiliates. Each method changes rights and incentives. A group could reorganize ownership, invoke independent mandates or use joint ventures to avoid the cap. The classification process becomes part of the election itself.

A cap can protect against multiplication of legal entities in a one-member-one-vote system. It may also penalize genuine subsidiaries with separate liabilities, minority owners and local obligations. In a resource-weighted system, it can contradict the deliberate decision to give greater stewardship weight to larger holders.

The rule therefore needs a clear harm threshold, a stable control definition, advance notice, independent review and a transition period. It should specify how joint control counts, how newly acquired affiliates are treated, how divestments restore weight and whether documented independent instruction creates an exception.

No registry should present a cap as a neutral data-cleaning exercise. It is a redistribution of constitutional power. The control audit supplies evidence for that debate; it does not make the decision automatically.

One-group-one-vote is coherent only if the group is the constitutional member

The strongest deduplication model gives one vote to each independently controlled group. It directly aligns the denominator with centres of corporate direction.

The appeal is obvious. Incorporating another subsidiary does not create another vote. A merger combines electoral voice when it combines control. Divestment creates a new voice when independence becomes effective. The legal form no longer determines political multiplication.

The difficulties are equally serious. Membership contracts remain with legal entities. Fees, liabilities and resource records may be distributed among subsidiaries. Joint ventures produce overlapping groups. State control is not always unitary. A parent can own a company without directing its election position. The registry or its assessor acquires substantial authority to decide who is one political person.

If one-group-one-vote is chosen, the institution should consider making the control group a recognized constitutional category rather than an audit overlay. Affiliated legal members could designate which entity exercises the group vote, document the mandate and retain their operational memberships. Independently mandated affiliates could seek a defined exception. Joint-control clusters would need explicit representation rules.

The change would require more than an election procedure. It would alter who the institution says its constituents are. That choice deserves member approval under the amendment rules, a published impact analysis and safeguards against retroactive reclassification.

Different matters can justify different aggregation rules

An institution does not have to use one group rule for every decision. Ordinary board elections, fee votes, service resolutions and constitutional amendments affect members differently.

One option is to preserve legal-member voting for ordinary governance while applying a group cap or parallel group majority to amendments that alter core rights. Another is a double-majority rule: a proposal must receive the required legal-member or voting-weight majority and support from a required share of participating control groups. This prevents a small number of large groups from acting alone without erasing their formal entitlements.

A mixed system can also protect minority operating contexts. The electorate might retain existing votes while board-composition rules limit the number of directors currently affiliated with one group. Nomination thresholds can be group-deduplicated even if ballots are not. Proxy caps and affiliate disclosure can operate together.

Every hybrid adds complexity. The report must show both calculations and identify what happens when they disagree. Members should be able to reproduce the result from the published rules. A double majority announced only after a controversial vote would be illegitimate; it must be part of the constitution before the election.

The advantage of multiple institutional levers is proportionality. A registry can address the precise risk—affiliate ballot multiplication, concentrated nominations, proxy aggregation or post-election conflicts—without pretending they are one problem.

Weighted voting makes the group calculation indispensable

In a tiered system, counting affiliate organizations without their weights can hide more than it reveals. One group may have three members carrying 64 votes each; another may have twenty members carrying one vote each. Legal-member concentration and voting-weight concentration point in opposite directions.

APNIC's published tiers and LACNIC's one-to-11 bands demonstrate why the audit needs both. The group table should sum eligible weight, weight cast and any proxy-carried weight without confusing those values with the number of organizations. It should also state whether resource holdings are measured separately from voting entitlements, because tier formulas and caps can make them diverge.

Corporate changes can affect weights on different dates. A merger may leave memberships separate until a registry process is completed. Resource holdings may move before or after the legal closing. A group may cross a tier threshold. The snapshot rules should identify the entitlement actually certified, not reconstruct a hypothetical combined tier after the fact.

The constitutional question remains explicit. If resource-linked responsibility is the intended basis of voting, group-level aggregation might confirm rather than reduce a large group's weight. If tiers were designed per legal member, multiple affiliates may produce a result the original formula did not anticipate. The audit cannot infer purpose from arithmetic. Governing documents and member debate must supply it.

Reporting both dimensions prevents a familiar rhetorical switch. An institution cannot call the system organization-based when defending multiple affiliate votes and resource-based when defending their combined weight without acknowledging the interaction.

Operational registration must remain insulated from electoral disputes

Affiliation classification is a governance function. It should not become a route to suspend number resources, remove reverse DNS, revoke certificates, alter registry records or interrupt member services.

A member can be late or wrong in an electoral disclosure while remaining the accurate registered holder of resources. The appropriate remedies concern voting eligibility, public correction, candidate conflicts or future governance rights. Technical continuity should be preserved unless a separate, properly authorized registration or legal process establishes a reason for operational action.

This separation protects the audit's credibility. Members are more likely to disclose complex ownership and joint-control arrangements if a classification dispute cannot endanger their networks. It also prevents staff from using the control graph as a general-purpose enforcement database beyond its stated purpose.

The registry may reuse verified corporate facts where lawful, but each use needs its own mandate and decision standard. A merger record used to update a resource holder is not automatically proof of common ballot instruction. An election declaration does not authorize a transfer. A candidate-affiliation edge does not establish operational control of an ASN.

Data access should reflect those boundaries. Election assessors receive what they need for the snapshot. Registration staff do not gain unrestricted access to secret ballot linkage. The board cannot rewrite a disputed affiliation during certification without the independent process. Clear separation makes the system more accurate and less threatening.

The annual cycle should make correction ordinary

Group disclosure will fail if it is attempted only in the final week of voting. A workable cycle begins with continuous member responsibility to report material control changes, followed by a scheduled pre-election snapshot.

Six months before the election, the registry can refresh public corporate sources and notify members whose known affiliations changed. Four months before, members confirm or correct the proposed group, identify independent mandates and declare voting-authority arrangements. Two months before, the independent assessor resolves ordinary disputes and publishes aggregate preliminary denominators.

The eligibility record date freezes legal entitlements under the bylaws. A later material-change window captures acquisitions, divestments, court appointments and control transfers. Before voting opens, members see the final structural classification that will be used for disclosure or any constitutional group rule. Emergency review remains available through certification.

After the election, the report publishes legal and group denominators, proxy and contact counts, uncertainty ranges, methodology version, corrections and audit hash. Members can challenge factual errors for the historical record. The next cycle starts from the corrected graph rather than rebuilding every relationship from scratch.

This timetable makes disclosure a routine institutional control, not an accusation activated when one candidate loses. It also creates trend data: group count, largest group share, unresolved classifications, concentration by voting weight, candidate affiliations and recusals can be compared across elections.

Number Resource Society has a bounded advocacy role

Number Resource Society can help make the issue legible to members who see separate legal entities on a registry roll but cannot tell which denominators an election report uses. It can research published bylaws, compare affiliate and proxy rules, explain control-map methods and represent members seeking clearer disclosure or a fair correction process.

NRS can advocate a minimum reporting template: eligible members, eligible weight, control groups, participating groups, voting contacts, proxies, uncertainty ranges, candidate affiliations and aggregate recusals. It can commission independent research using public corporate and election records, provided the sources, inference limits and correction route are explicit. It can help a member assemble evidence for the responsible RIR or an empowered reviewer.

Its boundary must remain categorical. NRS does not certify ultimate control, administer an RIR election, validate ballots, assign voting weight, decide an affiliation appeal, direct a recusal or alter a registry record. Those powers belong to the RIR under its governing documents, an independent election assessor or reviewer with a defined mandate, and a competent court where legal rights require adjudication.

NRS also should not present group deduplication as an operational service that substitutes for registry governance. Its role is advocacy, research, convening and member representation. A proposed model can be tested in public analysis; implementation requires the institution whose constitution and election are affected.

That division strengthens the reform. Members gain an advocate without creating another authority that can silently decide who counts.

The 2027 test is whether the same result can be counted four ways

By 2027, a credible election report should let a reader reproduce four views of the same event.

The legal-member view shows every eligible entity and valid ballot under the bylaws. The voting-weight view shows entitlements and votes cast under any tier system. The control-group view deduplicates members according to dated, source-linked and reviewable authority. The human-mandate view shows voting contacts and proxies without mistaking either for ownership.

The views will not always tell the same story. A result can have high member turnout and lower group turnout. A small number of organizations can carry a large share of weight. Many groups can use a few professional contacts. Twenty affiliates can cast twenty valid ballots under a constitution that deliberately protects legal entities.

The divergence is the evidence. It allows members to ask whether the institutional design still matches its purpose. It supports precise reform instead of allegations about brands or size. It gives candidates and directors a common conflict vocabulary. It makes control changes visible on the dates they matter.

The title's question therefore has no universal numerical answer. When one corporate group casts twenty registry votes, a legal-member constitution may count twenty, a group constitution may count one, a capped system may count five, and a double-majority system may record twenty legal votes plus one group. What matters is that the institution chose the rule before the election, disclosed the shared control, protected independent mandates, and made the calculation reproducible.

Conclusion: the hidden constitutional choice must become visible

Corporate personality is a legal fact. Corporate control is another fact. Voting instruction is a third. Registry elections become hard to interpret when they treat the first as conclusive evidence of the other two.

The solution begins with structure rather than suspicion. Separate affiliates from proxies. Map control and instruction with effective dates. Publish legal-member, voting-weight, control-group and contact denominators. Give members a correction route. Connect candidate affiliations to proportionate recusal records. Reconcile every input to the certified total and preserve a versioned audit.

Only then should the institution decide whether legal entities keep separate votes, groups face a cap, constitutional matters require a group majority or the system remains unchanged. Each choice can be defended if its purpose and tradeoffs are explicit. None should be smuggled into a data-cleaning rule or inferred after a close result.

One corporate group casting twenty ballots is not automatically evidence of capture. It is evidence that the registry must say what it counts as representation. A transparent constitution can answer twenty, one or something between them. An opaque constitution answers only after power has already been exercised.

Sources and scope

The article uses a hypothetical twenty-affiliate group to test electoral mechanisms. It does not assert that the named telecommunications groups hold multiple memberships, coordinate ballots or exercise improper influence in any RIR. The merger sources establish examples of sole or joint corporate control and effective-date evidence; an election-specific conclusion would require member, ownership and voting-authority evidence under the published audit method.