APNIC's 2023 Proxy Storm: When Corporate Votes Crossed Borders

Summary

• APNIC’s 2023 procedure let a member’s Corporate Contact transfer its electoral authority to a proxy who could be outside the organisation. MyAPNIC authentication, confirmation to the parties and loss of the member’s own voting ability formed an identifiable chain of authority.
• Each appointment carried the member’s tier-based entitlement, which ranged from one to 64 votes. The important measure was therefore not merely the number of proxies, but the total voting weight gathered through them.
• Crossing an organisational or national boundary did not make a proxy defective. The unresolved governance issue was whether numerous separately valid mandates accumulated under common direction without adequate aggregate disclosure.
• The campaign generated nearly 40 complaints about nominee conduct. An external chair from a law firm reached nine breach findings, but threats of injunction proceedings meant that eight were not made public before voting ended.
• The challenged candidates did not win, and the Election Chair reported no complaint about the operation of the voting machinery. Those facts narrow what can responsibly be claimed, but they do not answer how concentrated delegated authority became.
• September changes addressed candidate eligibility, corporate-group representation, residence, participation, economy-level outcomes and electoral oversight. Their wording shows a response to identified vulnerabilities; it does not measure whether those measures later worked or prove that a particular 2023 appointment was invalid.
• A useful public account would preserve secret candidate selections while reporting mandate bands, represented organisations, weighted concentration, geographic spread, corporate-group review and the handling of conduct complaints.

Follow the authority before judging the geography

The most revealing event in APNIC’s 2023 election was not that a vote could travel across a border. It was that an organisation’s power could move through a defined series of steps and then disappear into an undisclosed total. A Corporate Contact could designate another person to vote for the member. That person did not have to be employed by the appointing organisation. Authentication took place through MyAPNIC, the appointment was confirmed, and the member could no longer exercise the same vote directly. What began as an organisational entitlement ended as authority held by a named proxy.

That sequence matters because it separates documentary validity from political effect. On the available account, the procedure was designed to establish who could make an appointment, connect the appointment with a member account and prevent the original holder from also voting. Those are substantial controls. They make it inappropriate to describe every external appointment as an evasion, much less to infer fabrication solely from the proxy’s location or employer. Delegation was part of the permitted architecture of the election.

Yet verification of each link did not reveal the shape produced when many links met. A system can authenticate ten, twenty or more mandates one at a time while giving the public no indication that they lead to the same representative, associated campaign or commercial interest. Individual records may be orderly even when the combined distribution of decision power is sharply uneven. The controversy arose in that gap between a series of accepted transactions and the aggregate influence those transactions might create.

The distinction can be stated without assuming wrongdoing. Appointment proves that a member authorised a person under the operative process. It does not establish why that member delegated, how actively it considered the decision or whether the proxy solicited many similar mandates. Nor does a collection of mandates by itself establish coercion or coordinated voting. Those are separate propositions requiring evidence that the public reporting did not provide.

Geography can still be relevant evidence. A pattern involving multiple economies may help describe the reach of a campaign, reveal that authority is being centralised far from appointing organisations or prompt questions about how requests were made. But a map cannot decide legitimacy. APNIC is a regional institution, corporate structures cross jurisdictions, and members may have sound reasons to rely on experienced people elsewhere. Treating national alignment as the test would confuse the location of an intermediary with the quality of the authorisation.

The stronger inquiry begins with authority and then adds scale. Who was entitled to appoint? Was the instruction deliberate and authenticated? How many organisational mandates arrived at the same destination? How much tier-weighted power did they contain? Were the organisations related, independently controlled or merely approached by the same campaign? Which checks examined those relationships? The 2023 record answers parts of the first two questions. It leaves the decisive aggregation questions largely unilluminated.

A proxy appointment was a transfer, not an extra ballot

The 2023 arrangement did not simply give another person permission to assist a voter. It reassigned the member’s capacity to cast its ballot. The Corporate Contact occupied the crucial starting position: that contact could choose the proxy through the member-facing system. Confirmation then connected the appointing organisation and the recipient of the mandate. Once the designation took effect, the organisation gave up its ability to submit the vote itself. That last condition prevented the same entitlement from being exercised twice.

This makes the authority chain more important than the ordinary language of representation might suggest. The member remained the source of the entitlement, but the proxy became the person able to deploy it during the applicable voting window. The proxy’s act was attributed to the member for electoral purposes even though the representative could stand outside the member’s workforce. Authority, identity and organisational affiliation were therefore related but non-identical features of the transaction.

APNIC’s online voting terms and conditions provide the relevant institutional setting for member voting through MyAPNIC, authorised contacts and two-factor authentication. Those controls support confidence that a submitted action was attached to an eligible organisational account. They are principally controls over access, attribution and auditability. They do not, on their own, demonstrate that several eligible accounts were acting independently of one another.

A useful audit would consequently test four different propositions. First, the member organisation actually held an entitlement. Second, the person initiating the delegation possessed the required corporate-contact authority. Third, the system bound the designated person to the proper number of votes and disabled a competing direct ballot. Fourth, the collection of appointments was examined for concentrations that could make nominally separate instructions function as a single bloc. The public description gives meaningful assurance on the earlier stages and much less on the fourth.

This is where the language of participation can mislead. A member has participated in a formal sense if its authority is validly exercised through a proxy. That tells readers nothing about how much independent decision-making occurred across the participating organisations. Ten organisations may each reach their own judgment and select the same trusted representative. Alternatively, an organiser may do most of the political work while the organisations merely approve a transfer. Both patterns can produce ten authorised mandates, yet they represent different distributions of deliberation and control.

The available evidence does not let an observer choose between those descriptions for the election as a whole. It would be unfair to label inactive every member that appointed someone else, since delegation may be the practical route to involvement. It would be equally unwarranted to treat the number of appointing accounts as proof of ten independently developed preferences. The proper conclusion is narrower: APNIC could verify an appointment without publicly demonstrating the independence of the wider set in which it sat.

Weighted mandates changed the scale of delegation

Counting appointment forms would not have measured the power being transferred. APNIC members did not enter the election with equal-sized ballots. Under the published membership tiers and voting rights, entitlements extended from one vote at the Associate level to 64 at Extra Large. Between those points, the allocations increased through the tier structure. A proxy therefore received the appointing organisation’s bundle, not a standard unit shared by every member.

Consider what follows without inventing any 2023 totals. Five proxy appointments could carry less weight than one appointment from a larger-tier member. Twenty organisations represented by a single person might sound more consequential than ten, but the smaller group could command the greater electoral force. Neither the number of individuals participating nor the number of organisations authorising them is an adequate substitute for the weighted sum. Every account of concentration needs at least those three measures kept apart.

This also complicates statements about turnout. A report can say that many members took part while a comparatively small set supplied much of the voting power. It can likewise show a large proportion of weighted votes used even if relatively few organisations acted. Both descriptions may be correct. Participation counts answer how broadly the formal electorate appeared; weighted totals address how much decision capacity entered the result. The former should not be presented as if it settles the latter.

Delegation adds another dimension. The public needs to know not only the overall distribution of member entitlements but also how those entitlements were rearranged among the people who actually controlled ballots. Before appointments, voting weight is distributed across member organisations by tier. After appointments, several bundles may be held by one proxy. Without a post-delegation view, a published tier table describes the starting allocation while concealing the operational allocation at the moment choices are made.

That concealment need not be deliberate. Election reporting often focuses on the winning candidates, vote totals and proof that the machinery worked. A proxy system creates a less familiar reporting obligation because the person exercising power may not match the organisation to which it was originally assigned. The more flexible the appointment rule, the more useful it becomes to show the resulting distribution in a form that protects individual choices.

A sensible account would disclose concentration bands rather than identities. It could state how many proxies held one mandate, how many held several, and how many occupied higher bands. Alongside organisation counts, it could give the amount of tier-weighted authority inside each band and the number of economies represented. None of that would expose which candidate received a particular ballot. It would tell members whether the election remained broadly distributed after delegation transformed the initial entitlement map.

Cross-border representation was neither proof nor irrelevance

The phrase “crossed borders” captures a visible feature of the dispute, but it should not become a verdict. A member can have operations, owners, staff, advisers or trusted professional relationships in more than one economy. A regional registry necessarily brings together organisations whose practical connections do not fit neatly within national lines. Permitting an external person to serve as proxy can help an organisation participate when its own authorised personnel cannot or do not wish to manage the election directly.

That practical benefit should be taken seriously. A prohibition based only on residence, nationality or employment would exclude legitimate delegation and could narrow participation. It might favour organisations with larger internal governance teams while burdening smaller members that depend on outside expertise. Nothing in the 2023 account establishes that crossing an economy boundary invalidated a mandate. The applicable procedure explicitly allowed a proxy who was not an employee of the member.

Still, geography can reveal how an influence network operates. If a campaign obtains mandates from organisations spread across several economies and places them under a common representative, the pattern demonstrates organisational reach. It may also show that the relevant centre of electoral decision-making differs from the locations attached to the member accounts. Those observations justify scrutiny of concentration; they do not prove improper pressure, false authority or a common beneficial owner.

The same discipline applies to corporate affiliation. Several organisations may belong to a shared group and therefore have a reason to coordinate. Unrelated members can also support the same candidates or rely on the same adviser. Common representation is an indicator of aggregated control at the ballot-casting stage, not conclusive evidence about the legal or economic relationships among the appointing members. A review should examine those relationships rather than assume them from a proxy’s identity.

Members and affected customers must also remain distinct. The election authority belonged to eligible member organisations under the voting arrangements. Customers served by those organisations may experience consequences from APNIC governance, but being affected does not itself supply the member’s ballot or a right to instruct its proxy. A campaign that invokes broad customer interests is making a representational claim. It is not demonstrating that those customers participated in the authorisation.

This is why the best response to cross-border appointments is a visible, reviewable chain rather than suspicion of borders. The record should establish the organisational source of each mandate, the authority of the person granting it, the weight transferred and the checks applied when mandates converge. Aggregate reporting can then show geographic spread without exposing individual preferences. Borders become one descriptive dimension among several, while the central question remains where effective decision power accumulated.

The absent appointment graph was the central evidence gap

The publicly available account contains several important pieces of the 2023 story: the authorised proxy route, unequal voting weights, an unusually contentious campaign, nearly 40 conduct complaints, nine adverse findings by an outside chair, delayed publication of most of those findings, warnings about impersonation and threatening calls, the defeat of the challenged candidates and later institutional changes. What it does not contain is a comprehensive map connecting appointing members to proxies and the weighted authority each proxy held.

Such a map need not be published with names. Internally, APNIC could see the individual relationships needed to administer the election. Public assurance could have been produced from that information through aggregation. The report could group proxies by the number of member organisations represented, calculate the voting weight controlled in each group, state how many economies supplied the mandates and describe the review of apparent corporate affiliations. Readers would gain a picture of concentration while candidate choices remained secret.

The missing graph prevents two competing accounts from being tested. One is that a coordinated interest accumulated enough authority to create a serious capture risk. The other is that the visible campaigning reflected energetic but dispersed member support, with lawful appointments that never produced dominant control. Candidate results alone cannot decide between them. A bloc can fail to elect its preferred slate and still reveal a structural vulnerability; a noisy campaign can also lose without ever possessing the power attributed to it.

Absence of aggregate evidence should not be converted into evidence of misconduct. It establishes uncertainty about scale. The distinction is important because governance analysis can become self-sealing: secrecy is treated as proof that something improper was hidden, while any denial is dismissed because the data are unavailable. A more rigorous position identifies the specific information that is missing, explains what competing hypotheses it could distinguish and stops before assigning guilt.

The same standard protects APNIC from overly broad criticism. Publication of appointment concentration might have shown that the disputed network controlled only a modest portion of weighted authority. If so, the data would have narrowed allegations and strengthened confidence in the outcome. If the figures showed a much larger concentration, members could debate whether additional review or limits were warranted. Non-publication deprived the institution of both possibilities.

The graph also matters over time. Later reform can change eligibility rules or oversight bodies, but without a 2023 baseline there is no reliable measurement of improvement in proxy concentration. One can compare text before and after September. One cannot compare the actual distribution of mandates unless the same indicators are calculated for both periods. Reform language demonstrates an institutional response; a longitudinal concentration series would demonstrate whether power became more dispersed.

A complaint ledger should separate allegation, finding and notice

Complaint statistics become useful only when the institutional state of each matter is visible. Receipt means that APNIC has recorded an allegation; it does not say whether the allegation was complete, credible or ultimately supported. A matter under review occupies another category because evidence and responses may still be developing. An adjudicated finding is different again: it records the conclusion reached by the responsible process, while leaving open the separate question of what consequence, if any, was available. A calm report would keep these stages apart instead of placing every item beneath one headline total.

Timing requires its own field because a conclusion can exist without being known to the electorate. The record should show whether disclosure occurred before voting closed, afterward, or remained pending at the relevant point. That sequence matters without assuming that earlier publication was always possible or that later publication was worthless. Information released during the choice period can inform judgment; information released after it can support accountability, explanation and future safeguards, but cannot become part of the knowledge voters actually possessed when they acted.

Reporting the date and procedural status would expose that difference without speculating about anyone’s hypothetical ballot.

Voter knowledge should therefore be described from the materials available at the time, not reconstructed from the completed record. In the 2023 contest, the pertinent distinction is between complaints that had arrived, conclusions that had been reached, and findings that members could examine before the window ended. The existence of later-public information cannot establish that voters knew it earlier. Equally, lack of pre-close disclosure does not erase a finding subsequently acknowledged. A chronology can hold both propositions together and show where the informational setting diverged from the eventual public account.

Remedy belongs in a further column rather than being inferred from the finding itself. A report can identify any stated consequence, unresolved challenge or absence of a recorded disposition without claiming powers that the published materials do not establish. This matters especially where threatened proceedings influence when information is released. The threat, the publication decision, the underlying conclusion and the available institutional response are separate events. Combining them can make caution look like exoneration or make a finding appear to carry a sanction that has not been shown.

Separation gives every stage its proper evidentiary weight.

The resulting public ledger could remain concise. For each grouped class of matters, it could present receipt status, review status, finding status, disclosure phase and recorded disposition, with explanatory notes where timing was affected by a dispute. Aggregate presentation would avoid turning untested allegations into personal labels. It would also let members see how much of the complaint environment was allegation, how much had become an institutional conclusion, what portion was knowable during voting and what response followed.

That is a more faithful account of electoral information than either a raw complaint count or a final list stripped of timing.

Conduct complaints changed the informational setting

The election’s political environment cannot be reduced to proxy administration. APNIC received nearly 40 complaints concerning nominee conduct. An external chair drawn from a law firm concluded that nine matters involved breaches. Only one of those findings became public before voting finished. The remaining eight were withheld during the electoral period after candidates threatened injunction proceedings that could have delayed the process.

That chronology created an information problem independent of ballot mechanics. Members were choosing candidates while most of the eventual adverse findings were not available to them. The decision to avoid a threatened interruption may have been understandable in procedural terms, but its electoral consequence is clear: the voting audience did not possess the same body of acknowledged findings that existed after the event. Whether the missing information would have changed a ballot is unknowable from the record.

A complaint, a breach finding and an enforceable remedy are three different things. The large complaint count shows the intensity of contestation, not that every allegation was substantiated. The nine findings record conclusions reached through the external-chair process. They do not, merely by existing, identify what coercive consequence could be imposed, whether a finding disqualified anyone or how a court would have treated a dispute. Threatened proceedings affected publication timing without themselves deciding the merits.

Those distinctions prevent the numbers from being used rhetorically. “Nearly 40” should not be treated as nearly 40 proven violations. “Nine” should not be understated as mere campaign noise, because an independent external process reached those findings. “Eight unpublished” does not mean the evidence was fabricated or suppressed for a partisan purpose; it describes the practical result of a publication decision made under threats of injunction litigation. Each number answers a separate question.

Warnings about impersonation and threatening calls add to the reason for examining the campaign environment. They suggest concerns about how people were approached and represented during the contest. Yet a warning is not a finding that every disputed proxy arose from impersonation or intimidation. The authority records, complaint files and campaign communications would perform different evidentiary functions. Conflating them would turn a serious environment into an unsupported conclusion about particular mandates.

Future handling should aim to give voters timely, carefully framed information whenever fairness permits. A pre-election account could state which complaints were received, which had reached findings, what remained contested and what practical consequence followed. Where publication is delayed, the institution can later explain the timing and status without presenting allegations as adjudicated facts. That approach recognises both procedural caution and the electorate’s interest in knowing material conduct conclusions before it decides.

The result ruled out some claims, not the governance concern

The candidates most closely associated in public controversy with the challenged commercial interest were not elected. That outcome is a necessary part of any honest account. It means the 2023 contest did not finish with the disputed slate taking the available positions. Describing the episode as a completed capture would therefore go beyond the documented result. The concern was about attempted or possible concentration within the electoral process, not proof that the alleged project secured office.

The Election Chair also reported no complaint about the mechanics of the election. This supports a distinction between system operation and campaign conditions. There was no reported challenge of the kind described to the machinery by which ballots were received and counted. At the same time, disputes about solicitation, representation, candidate behaviour or disclosure can exist even when the voting platform performs as designed. Mechanical integrity is important, but it does not certify the whole political environment.

Nor does defeat show that proxy concentration was harmless. A concentrated bloc may be insufficient to determine the result, may choose candidates who attract little support elsewhere or may be outweighed by greater participation from other members. The fact that a strategy failed does not reveal how much voting authority it assembled. That is precisely why the missing aggregate appointment data remain relevant after the winners are known.

The reverse inference is equally unsafe. A contentious campaign, widespread concern and subsequent reform do not establish that the losing candidates possessed a decisive bloc. Institutions frequently revise rules after discovering ambiguity or risk, including risks that did not determine an outcome. A reform can be prudent even when the feared event did not occur. Using the later rules as retrospective proof against individuals would confuse prevention with adjudication.

This is also where participation and decision power must be separated. An election may attract broader attention, more campaigning and more formal member involvement while still allowing authority to accumulate unevenly. Conversely, one organised faction can appear prominent in public discussion yet control too little weighted power to prevail. Public visibility, organisational count and electoral capacity are related only when data demonstrate the relationship.

The most supportable judgment is therefore bounded. The documented result resists the claim that the disputed candidates captured the council through the 2023 vote. The absence of a mechanics complaint resists an allegation that the counting system was shown to have malfunctioned. Neither fact supplies the unpublished distribution of proxy mandates. Confidence in the declared winners can coexist with a sceptical assessment of whether the reporting exposed enough about concentrated delegated power.

September’s reforms responded on several fronts

APNIC’s September changes addressed features that had become salient during the year. They restricted representation connected with a common corporate group, introduced regional residence and participation qualifications, limited how many successful candidates could come from one economy and established an Electoral Committee with responsibility for eligibility and conduct decisions. The package reached beyond proxy appointments and altered the framework around who could stand, how concentration among winners would be constrained and who would oversee disputes.

The corporate-group element recognised that formal separation among organisations may not reflect independent interests. Residence and participation requirements connected candidacy more closely to the region and to demonstrated engagement. The economy limit sought geographic spread in the final composition rather than treating a purely numerical result as sufficient. Creating an Electoral Committee concentrated responsibility for determinations that had proved difficult during the campaign.

These changes are evidence of institutional recognition that the earlier arrangements required strengthening. They are not a formal admission that every criticism of the 2023 contest was correct. Rulemakers can respond to uncertainty, appearance and future exposure as well as established misconduct. The amendments should therefore be read prospectively: they show which risks APNIC chose to address, not a retrospective judgment that named appointments or candidates violated rules that previously applied.

The reforms also focus heavily on candidates and composition. A common-group restriction can prevent related organisations from gaining multiple positions. A residence rule can shape who is eligible to stand. A one-economy limit can diversify the set of winners. An oversight committee can make eligibility and conduct decisions. None of those measures necessarily reveals how many member mandates a proxy collects or how much tier-weighted authority moves through a shared representative.

That remaining distinction matters because candidate-side and voter-side concentration are not interchangeable. Diverse candidates can compete before an electorate whose voting control is highly aggregated. A geographically varied council can be selected through concentrated mandates. Conversely, dispersed voting does not guarantee that the winners satisfy a desired regional balance. Each side requires its own indicators and safeguards.

Nor can the text of reform establish measured success. To evaluate the September package, APNIC would need to publish outcomes tied to its purposes: how eligibility reviews were handled, how corporate-group questions were resolved, how economy and residence provisions affected the field, how complaints were disposed of and whether proxy weight became less concentrated. Without those observations, one can say that rules changed and identify their design. One cannot responsibly claim the vulnerabilities have disappeared.

A concentration report can reveal structure without revealing choices

An aggregate concentration report should begin with separate units of observation. Member organisations are the legal sources of electoral authority. Authorised contacts are the people through whom member voting can be activated; the Corporate Contact performed the appointment step described for 2023. Proxies are the people who may ultimately exercise delegated bundles. Weighted votes express the size of those bundles under the tier structure.

Treating any one of these as a substitute for the others would obscure whether breadth existed at the organisational level, whether operational control was concentrated among a few contacts, or whether a modest number of appointments carried unusually large voting weight.

The geography table should be equally disciplined. It can count the economies associated with appointing organisations and, in a separate field, the economies associated with proxies or authorised contacts. Those distributions describe reach and centralisation, not validity or motive. Where a cell would make an organisation identifiable, neighbouring values can be combined into a broader band. The report should explain that an economy count is neither a count of independent interests nor evidence of improper cross-border activity. Its purpose is to show how widely authority originated and how narrowly, or broadly, its exercise was located.

Common control requires a distinct analytical layer because formal account separation may not equal independent ownership or direction. APNIC can review relevant relationships confidentially and publish only grouped results: organisations examined, associations identified, matters resolved as independent and matters that remained uncertain. The method should state what information was considered and where the assessment could not support a firm classification, without exposing confidential ownership details. The report must also avoid equating a shared proxy with common control.

Shared representation is a concentration fact; an affiliation conclusion depends on the separate review.

Threshold review should operate as an assurance step, not as an unpublished shortcut to invalidation. The report can describe how concentration bands triggered closer examination and then state, in aggregate, whether authority records, contact roles or possible affiliations received additional scrutiny. Because the supplied record does not establish a new numerical boundary, the design need not announce one here. What matters is that a trigger be defined in advance, applied consistently and linked to a reasoned disposition.

Review identifies a condition requiring attention; it does not by itself prove coordination, coercion or a defective mandate.

The published output can preserve choice secrecy through bands, suppression of identifying cells and separation of authority data from candidate selections. A reader should be able to compare the number of organisations, the number of authorised people, delegated weighted votes, economies represented, common-control review and threshold dispositions without learning how any member voted. Denominators should accompany every share so that eligible authority, exercised authority and delegated authority are not confused. Releasing the methodology alongside the tables would make annual comparisons possible and expose changes in classification.

This design makes concentration observable while keeping the protected act—the selection among candidates—outside the report.

Transparency can protect both secrecy and legitimacy

Ballot secrecy protects the link between an eligible voter and the candidates selected. It does not require the institution to hide every structural fact about how voting authority was exercised. APNIC could preserve the separation between identity and choice while reporting the aggregate characteristics of proxy delegation. Indeed, secrecy becomes more credible when readers can see that it is being used to protect preferences rather than to withhold the overall distribution of power.

The first useful disclosure would be a mandate histogram. Proxies could be grouped into bands according to the number of member organisations they represented: a single mandate, a small cluster and progressively larger clusters. The exact band boundaries could be chosen to avoid identification. For every band, APNIC could provide the corresponding sum of weighted votes. This would immediately show whether numerous small appointments or a few high-tier transfers drove concentration.

A second view would describe geographic spread. Reporting the number of economies represented within each sufficiently large band would illuminate cross-border reach without naming members. Such data should be presented neutrally. A broad footprint might reflect successful regional mobilisation, common professional support or central coordination. The table would establish the pattern, leaving claims about motive to evidence from the appointment and conduct reviews.

Third, an internal corporate-group assessment could be summarised. APNIC need not expose confidential ownership information. It could state how many appointments were reviewed for possible related control, how many were resolved as independent, how many were associated and what rule or safeguard applied. Where uncertainty remained, the methodology could say so. The purpose would be to demonstrate scrutiny, not publish a commercial registry of members.

Complaint reporting requires a separate ledger. Aggregate proxy statistics cannot substitute for information about conduct, and a breach finding does not quantify voting control. A report should distinguish allegations received, matters reviewed, conclusions reached, findings published before the ballot closed, findings released later and any available consequence. The 2023 sequence shows why timing belongs in that account: information delivered after voting has a different democratic function from information available beforehand.

Finally, reporting should identify its denominators. The number of member organisations entitled to vote, the amount of weighted authority available, the organisations that participated directly, the organisations represented through proxies and the weight actually exercised are different totals. Publishing them together would allow members to see both breadth and control. It would also prevent affected customers, campaign participants and formal voters from being merged into a single, inflated notion of community participation.

The durable test is concentrated control, not foreignness

The 2023 episode is best understood as a stress test of delegated, weighted governance. APNIC had a method for transferring a member’s authority to an authenticated proxy, including someone outside the organisation. It also had a tier system under which the transferred amount could vary from one vote to 64. What the public account lacked was the bridge between those two systems: an aggregate explanation of where the weighted mandates ended up.

That absence left the debate vulnerable to two simplifications. One treated external or cross-border proxies as inherently suspect, overlooking the regional character of APNIC and the legitimate value of delegation. The other pointed to individually authorised appointments as if validity at the transaction level answered every concern about concentrated power. Neither approach is adequate. Geography does not establish abuse, while formal authorisation does not reveal the political structure created by many authorisations together.

The campaign evidence calls for the same restraint. Nearly 40 complaints demonstrate an exceptionally contested environment. Nine breach findings by the external chair give part of that environment institutional weight. The non-publication of eight findings until after the voting period limited what members knew at the relevant time. Impersonation and threatening-call warnings sharpen concern. None of those facts, separately or collectively, supplies the missing proxy graph or proves that a specified mandate was invalid.

The declared outcome supplies further boundaries. The criticised candidates did not obtain the positions, and no complaint about election mechanics was reported by the Election Chair. Those facts matter to confidence in what happened, but they cannot show how close, dispersed or concentrated the underlying authority was. An unsuccessful concentration remains a governance issue if the rules allow it to become decisive in another election. An alleged concentration that never existed should not be accepted merely because the campaign was alarming.

September’s response improved the institutional toolkit by addressing corporate-group representation, candidate qualifications, economy-level composition and electoral oversight. Its significance lies in the vulnerabilities it recognises and the procedures it creates. Its success must be assessed through later evidence rather than inferred from the presence of new language. Proxy concentration remains a distinct measurement problem unless subsequent reporting makes it visible.

A balanced standard would welcome participation through lawful delegation while refusing to confuse the number of participating accounts with independent decision power. It would respect secret ballots while publishing non-identifying concentration data. It would treat findings as meaningful without inventing remedies they did not supply. It would recognise that members possess the election authority even though customers and others may experience the consequences of council decisions. Above all, it would ask whether the aggregate structure can be examined before a controversy becomes a result.

The lesson of the proxy storm is therefore not that authority should stop at an economy’s edge. It is that transferable authority needs an observable destination. When every mandate is checked alone but their combination remains hidden, the institution can establish procedural permission without demonstrating political dispersion. APNIC’s strongest answer would be a recurring, privacy-preserving account of delegated weight, affiliation review, geography and complaint timing. That evidence would allow cross-border representation to remain legitimate without requiring members to accept concentrated control on trust.