Summary

  • During the Board election segment of AFRINIC-10 in Cairo on 21 May 2009, the meeting report records “E-networks (multiple proxies),” a call of seventeen proxy votes plus one vote associated with Mark Elkins, and an immediate recount that remained seventeen plus one. Proxy voting was therefore an implemented part of the physical ballot, not a theoretical accommodation.
  • The election committee’s stated condition was that people carrying proxies that had been agreed and duly signed could vote. The public report preserves that rule and an aggregate account of its use, but it does not publish the instruments, principals, signatories, authority evidence, receipt records, acceptance decisions or ballot links needed to reproduce the checking instrument by instrument.
  • The proper conclusion is deliberately limited. A valid proxy can transmit a corporate member’s private voting right, and the absence of public forms does not prove fraud or invalidity. Yet a signature alone cannot establish the principal, the signatory’s authority, the instrument’s scope or the one-authorisation-to-one-ballot path. In any event, this internal election could fill offices in a private corporate coordinator; it could not confer sovereign, regulatory, police, prosecutorial, punitive, confiscatory or adjudicative power.

The moment the aggregate became ballot weight

The revealing moment in Cairo was not the announcement that proxies might be allowed. It was a count. In the Board election segment on 21 May 2009, the surviving AFRINIC meeting report records an item as “E-networks (multiple proxies).” It then records seventeen proxy votes plus one vote associated with Mark Elkins. The total was challenged or checked in the room, and the recount remained seventeen plus one. Whatever uncertainty surrounds the underlying instruments, those numbers establish one central fact: delegated member voting power entered the operational ballot process.

That distinction matters. Rules can remain unused. Permission can exist on paper without affecting a contest. Here, the report describes proxies being treated as countable units in the room. A bundle large enough to require an explicit call and recount was not incidental administrative paperwork. It was part of the weight passing through the ballot box. The subject is therefore neither proxy voting in general nor the philosophical merits of participation from a distance. It is the evidentiary quality of an actual transmission from a member legal entity, through a human authorisation, to a cast ballot.

The setting can be stated precisely without enlarging it. AFRINIC-10 was held in Cairo in May 2009. The rule and the election discussed here occurred on 21 May during the Board election segment. The report identifies the election committee as Ken Lohento, Mohamed Elghamry, Sunday Folayan and Adiel Akplogan. It does not assign a separate signature check, acceptance decision or ballot-handling act to any one of those four people. Naming the committee is supported; inventing an individual division of labour is not.

The document is also best described for what it is. It is an AFRINIC meeting report produced after the event, with PDF metadata dated 17 December 2009. It should not be promoted into verbatim minutes or a contemporaneous transcript. Its value is nonetheless substantial. It preserves the institution’s account of the rule, the physical voting method, the aggregate proxy call, the recount and the final ballot totals. Its limits are equally important. The report is narrative evidence about what AFRINIC recorded itself as doing. It is not the instrument file by which every delegated vote can now be independently verified.

This produces a useful discipline for reading the seventeen-plus-one entry. The number should be neither dismissed nor mythologised. It is strong evidence that proxies were actually used. It is not a list of seventeen identified principals. It is not proof that one person legally held every mandate. It is not proof that the instruments were defective. It is not proof that the instruments were all valid. And it is not a basis for reconstructing voting choices. The surviving typography simply does not create a clean table linking each member principal to a signatory, a proxy holder and a marked ballot.

The committee stated a condition, then applied a process

The report’s proxy rule was compact. It said that people carrying proxies that had been agreed and duly signed could vote. Each part of that sentence has evidentiary significance, but none should be expanded beyond the words preserved. “Duly signed” points to an instrument and a signature. “Agreed” points to some form of acceptance or assent. The report does not identify who performed that agreement, what standard governed it, when it occurred, whether deficiencies could be cured or what record the committee retained.

The phrase therefore supplies a threshold, not a completed audit. It tells a reader that the committee did not describe any scrap of paper as sufficient. The stated condition required agreement and a due signature. But it does not show the evidence by which a signature was treated as due. It does not say whether the committee checked a corporate register, a designated voting contact, a board resolution, a delegation letter or some other source of authority. It does not say whether an original was required, whether a copy could be accepted or whether a late instrument was rejected. The actor and method behind “agreed” remain unknown.

The physical process is clearer at an aggregate level. The returning officer described marking the name of the member, having the member or proxy approach the ballot box, and reconciling the number found in the box against the number expected. This is meaningful operational evidence. It indicates that a proxy was not merely added later to a spreadsheet or treated as an invisible adjustment. A person approached a physical ballot process on behalf of a member whose name was to be marked, and the box count was to be reconciled.

Yet those steps establish only part of the chain. Marking a member name can help prevent a second appearance under the same recorded identity. Having a proxy approach the box can connect an identified person to a physical act. Reconciling the box can expose a mismatch between expected and actual ballots. None of those controls, standing alone, proves that the member legal entity was eligible, that the human signatory could bind it, that the instrument covered this election, that it was still in force or that it had not been revoked or duplicated.

The rule, its application and its auditability must therefore be kept separate. The rule was the stated condition of agreement and due signature. Application was the committee’s treatment of proxies as participants in the physical ballot process, including the called and recounted aggregate. Auditability would be the ability to move backwards from each counted proxy ballot through issuance and acceptance to the instrument, authorised signatory and eligible corporate principal. The public report supports the first two propositions. It does not provide the exhibits needed for the third.

This separation protects the analysis from two equal and opposite errors. One error treats the absence of published instruments as proof that the proxies never existed or were fraudulent. That conclusion is unavailable. A confidential file may have existed and may have been checked. The other error treats the report’s assurance and aggregate count as if they fully demonstrate valid authority. That conclusion is also unavailable. A process can be described faithfully while its decisive instrument-level evidence remains outside the published record.

What seventeen plus one means, and what it cannot mean

The call is unusually important because of its scale within the aggregate ballot. The report records forty-one ballots cast and forty valid. If every one of the seventeen called proxies produced one cast ballot, those proxies represented 17 out of 41 ballots cast, approximately 41.46 percent. Under the same bounded assumption, the seventeen proxies plus the one vote associated with Mark Elkins represented 18 out of 41 ballots cast, approximately 43.90 percent. These figures describe only a possible share of all ballots cast under the stated assumption.

They cannot be converted into a share of valid ballots. The report says one of the forty-one cast ballots was invalid, but it does not identify that ballot. It does not state whether the invalid ballot was cast directly or by proxy. It follows that the public record does not establish that all seventeen proxy ballots were valid. It also does not support the tempting but unsound assertion that the proxies were 42.5 percent of the forty valid ballots. That calculation silently assumes the very fact the report leaves unknown.

The eighteen-item grouping is similarly bounded. The report’s typography associates one vote with Mark Elkins while separately describing seventeen proxy votes. It does not establish that all eighteen items were controlled by one legal person, marked in the same way, valid, or derived from the same kind of authority. Calling the group a bundle is a description of the recorded call and recount, not a finding that the ballots travelled as a single legal block.

The final results do not cure the missing map. The report records one regional ordering at 25 to 15 and another at 32 to 8. Those totals show the outcomes the report announced. They do not show how any proxy ballot was marked, whether the seventeen proxy ballots divided between choices, whether the associated vote aligned with them, or whether any proxy changed either result. Outcome causation remains unknowable from the record. The margins cannot be used to reverse-engineer choices that were never disclosed.

Nor can the names appearing later in the report be used to manufacture seventeen principals. The sequence could contain principals, voting contacts, holders or simply the order in which people were called. The available evidence does not provide a reliable one-to-one mapping. Treating the list as a hidden proxy register would replace an admitted gap with speculation. The disciplined reading is narrower: the report preserves an aggregate, not the underlying legal topology.

The immediate recount adds some confidence about arithmetic in the room. Seventeen plus one remained seventeen plus one when counted again. That consistency is useful. It indicates that the aggregate was not a fleeting transcription that immediately collapsed under checking. But arithmetic stability is different from mandate validity. Recounting the same physical or called units cannot show who authorised them. It confirms the number being handled, not the legal competence behind each authorisation.

The aggregate validity rate, forty valid ballots out of forty-one cast, was approximately 97.56 percent. Even that high overall rate says nothing specific about proxy validity. A direct ballot may have been invalid; a proxy ballot may have been invalid; the report does not say. Aggregate cleanliness cannot be allocated to an undisclosed subgroup. This is exactly why instrument-level records matter when a meaningful share of ballot weight travels by delegation.

The minimum principal-to-ballot ledger

A reproducible proxy process begins with the corporate principal, not the person standing near the box. AFRINIC membership rights belonged to legal entities acting through people. The first ledger entry should therefore identify the member’s legal identity and connect it to the operative corporate and membership records. A trading name, familiar attendee or database contact is not enough if it cannot be tied to the legal person holding the voting right.

The second entry is eligibility on the relevant date. It must answer whether that member was entitled to cast the internal corporate vote at the time. A proxy can transmit only a right the principal possessed. If eligibility cannot be shown, a perfectly executed proxy form has nothing valid to transmit.

The third entry identifies the human signatory and the basis on which that person could bind the member. A signature is a mark made by a human being; authority is a legal relationship between that human being and the corporate principal. They are not the same fact. The file might establish authority through a registered role, board approval, corporate delegation or another competent record. Whatever the basis, it must be dated and connected to the relevant member.

The fourth entry is the signed instrument itself. It should identify the principal and proxy holder without ambiguity. It should state the election to which it applies, the scope of the delegated act and the expiry or termination condition. A broad letter written for one purpose cannot automatically become authority for another. An undated or open-ended instruction invites uncertainty about whether the mandate remained current on 21 May 2009.

The fifth entry concerns the proxy holder. The person who receives authority should be identified and should accept the role under any applicable conflict restrictions. Identity matters because the physical participant must match the person named in the instrument. Acceptance matters because a mandate cannot be safely inferred merely from possession of paper. If one person carries more than one instrument, the ledger must preserve each principal as a separate chain rather than allowing the holder’s identity to swallow the underlying members.

The sixth entry is receipt. The record should show when the committee received the instrument, by what channel, in what form and under which deadline. Timing can determine whether the delegation was effective for the meeting. A timestamp also distinguishes a valid early submission from a document added after a dispute began. If a cure period existed, the record should show what deficiency was identified, when correction was requested and when a compliant version arrived.

The seventh entry is the committee’s decision. “Agreed” needs an operational meaning. The ledger should show the acceptance standard, the reviewer or controlled process that applied it, the checks performed, and the resulting status. Any rejection should carry a reason. Any cure should preserve both the original deficiency and the corrected record. Any challenge should remain attached rather than disappearing behind a final accepted label.

The eighth entry covers duplicates, revocations and conflicts. A member might send a replacement instrument, revoke an earlier one or submit inconsistent directions. A proxy holder might appear with overlapping claims. The system must distinguish a superseding mandate from a second mandate and must prevent two papers from creating two ballots for one underlying right. A current authorisation is not established merely because an earlier document once existed.

The ninth entry connects accepted authority to ballot issuance. This is the point at which a valid legal instruction becomes a permitted physical voting unit. The accepted instrument should generate or be linked to exactly the ballot entitlement allowed under the rules. The link need not expose ballot secrecy. It does need to prove that one eligible principal produced no more than one permitted ballot for the relevant contest.

The tenth entry records casting and reconciliation. It should establish that the issued ballot entered the process, that the member name or entitlement was marked consistently, and that the total in the box reconciled with the expected number. The AFRINIC report gives a public description of this aggregate control. A complete ledger would connect that aggregate to the issuance records without revealing how a secret ballot was marked.

The eleventh entry preserves count, retention, challenge, correction and the final record. Forms and acceptance evidence must remain securely available long enough for a proportionate review. Access should be controlled; alterations should be logged; challenges should identify the affected instrument; corrections should never erase the earlier state. The final record should allow an authorised auditor to prove the chain while maintaining ballot secrecy and protecting personal information.

This ledger is not bureaucratic decoration. Each field answers a different failure mode. Legal identity prevents a familiar label from substituting for the actual member. Eligibility prevents delegation of a right that did not exist. Signatory authority prevents an employee or contact from binding a company without competence. Scope prevents one instruction from drifting into another election. Receipt time prevents retrospective addition. Duplicate and revocation checks prevent multiplication of one right. Ballot linkage prevents an accepted form from producing uncontrolled weight. Retention makes later review possible.

The 2009 meeting report does not publish this chain. That does not establish that AFRINIC had no internal records. It establishes that the public evidence available here cannot reproduce the chain. The correct diagnosis is an auditability gap, not a finding of fraud. That distinction should remain visible in every conclusion drawn from the seventeen-plus-one call.

Why the signature is necessary but not self-authenticating

The strongest limitation on the committee’s stated rule lies in the phrase “duly signed.” A signature may be necessary, but it is not self-authenticating. It can show that someone marked a document. It cannot, without surrounding evidence, identify the legal principal, establish the signer’s competence, prove the document’s scope, demonstrate timely delivery or rule out revocation.

Corporate members make this especially important. Companies act through authorised humans, but many humans can be associated with a company without possessing the same powers. A network contact may understand operations. A billing contact may manage invoices. An employee may attend a meeting. A director or delegated officer may hold corporate authority. Those roles cannot be collapsed into one another simply because a name is familiar to the registry.

Paper formality can therefore clarify or launder a mandate. It clarifies when the form is tied to an eligible principal, a competent signatory, a bounded instruction and a controlled ballot. It launders when the visible signature becomes a substitute for asking whether the signer could do what the paper purported to do. The risk is not proof that laundering occurred in Cairo. It is the reason the missing evidence matters.

The same applies to “agreed.” Agreement could mean the principal and holder agreed, the committee accepted the form, or some other process occurred. The report does not settle the meaning. An analyst should not insert a preferred actor. Instead, the uncertainty should shape the evidence request: identify who treated each instrument as agreed, under what standard and on what date.

A clean chain also preserves the difference between authority and preference. A member may prefer a candidate but fail to issue a valid proxy. A supporter may endorse a proxy holder but lack power to bind the member. An official may believe participation is desirable but lack power to waive the instrument requirements. Preference can explain motive; it cannot complete a missing delegation.

This is why a called ballot is only the last visible portion of a longer structure. The person approaching the box is observable. The member principal behind that person may not be. The signature is visible. The corporate competence supporting it may lie in another record. The counted unit is visible. The one-to-one path preventing duplication may remain inside an acceptance ledger. A defensible system connects these layers rather than treating the last visible act as proof of every earlier one.

The strongest defence of the Cairo practice

The best defence of the proxy practice begins with corporate reality. A company cannot walk into a meeting. It acts through human representatives. Requiring a particular officer to be physically present in Cairo would make travel money, schedule, visa access and personal availability part of the practical distribution of internal voting power. A written proxy can allow an absent member to use a private membership right through an identified person.

That defence has real force. Proxy delegation can broaden effective participation without pretending that every attendee represents a population. It can preserve the member as principal while separating authorisation from physical presence. It can also create better evidence than an informal oral instruction because a dated written instrument can state who delegated what to whom.

The public does not need to see private signatures or corporate records for this defence to work. Confidentiality and data protection can justify restricting access to original forms and personal documents. Publication of every signature could expose individuals to misuse without adding proportionate public value. A committee can review sensitive material privately while publishing enough control information to make the process credible.

The report’s physical description also provides some assurance. The member name was to be marked; the member or proxy approached the box; the number in the box was to be reconciled. The seventeen-plus-one total was recounted and remained stable. These facts show an effort to convert delegated rights into countable units rather than an unexplained post-election adjustment.

Most importantly, the absence of exhibits is not evidence that checking did not occur. Records may have existed outside the report. The committee may have checked them. The forms may have been valid. The election may have been entirely proper in its handling of proxies. A fair analysis must leave those possibilities open because the surviving public record does not eliminate them.

The defence fails only if it is stretched too far. Privacy can explain why a signature is not public; it cannot explain why no aggregate controls, timestamps, hashes, rejection counts, conflict rules or independent attestation are available. A recount can confirm the number of units; it cannot confirm signatory authority. A meeting report can state a rule; it cannot replace the evidence showing that the rule was applied consistently to each instrument.

The right conclusion is consequently two-sided. Proxy delegation is a legitimate mechanism for transmitting a private membership vote when the chain is sound. The report provides meaningful evidence of application. But the report alone cannot establish the soundness of every chain. Admitting both propositions is stronger than choosing between blind institutional trust and unsupported accusation.

Privacy-compatible proof is possible

Auditability does not require publishing a folder of signatures. It requires designing public and confidential records so that absence of disclosure does not become absence of proof. The sensitive layer can hold the original instrument, corporate authority evidence, personal identifiers and the secret-ballot controls. The public layer can report the process and provide non-identifying evidence of its integrity.

For each accepted instrument, the institution could publish a stable anonymous identifier or cryptographic hash, a receipt time within the permitted window, an acceptance status and a confirmation that eligibility, signatory authority, scope, duplicate and revocation checks were completed. Aggregate reporting could show how many instruments were received, accepted, rejected, cured, replaced or revoked. It could state the conflict rules and retention period. None of this requires exposing the way a ballot was marked.

An independent attestation can bridge the two layers. A qualified reviewer with controlled access could confirm that each counted proxy ballot linked to one eligible member, one authorised signatory, one effective instrument and one issued ballot. The attestation could identify exceptions without publishing personal data. If a challenge arose, a court or authorised auditor could examine the underlying material under appropriate confidentiality.

Hashes and timestamps are not magic. A hash proves only that particular bytes have not changed after the relevant reference point; it does not prove that their contents were legally sufficient. A timestamp proves receipt or commitment at a time; it does not prove authority. Their value comes from placement in the larger chain. Together with documented acceptance standards, identity checks and ballot reconciliation, they make silent substitution or retrospective invention harder.

Rejection reporting is equally important. An institution that publishes only accepted totals gives no view of how the rule behaved at its boundary. A count of rejected and cured instruments, together with reason categories, shows whether the committee actually distinguished defective submissions from compliant ones. It also allows members to see whether standards were applied consistently without exposing private documents.

Retention closes the final gap. An audit trail that disappears after the meeting cannot answer a later, focused challenge. Secure preservation should maintain originals or authenticated copies, access logs, acceptance decisions and ballot-entitlement records for a defined period. Destruction policies should be disclosed in advance. If a record is corrected, the system should preserve the earlier state and the reason for change.

The Cairo report contains none of these public instrument-level details. Again, that absence is not proof that there was no private file. It means a reader today cannot move from the reported seventeen proxy votes to seventeen reproducible authorisation chains. Privacy can justify a protected evidence room. It cannot justify an evidence vacuum.

AFRINIC’s office remained private and narrow

Even a perfectly evidenced proxy chain would answer only an internal corporate question. It could show that an eligible AFRINIC member validly delegated its private vote and that the resulting ballot was handled correctly. It could support the selection of people to private company offices within the powers supplied by applicable corporate law and the operative corporate instruments. It could not transform the election into an act of public sovereignty.

AFRINIC is a private bookkeeper and coordinator. Its legitimate common function is thin: preserve uniqueness, maintain accurate records, support proof of control and contactability, record transfers and security metadata, and keep auditability, portability, failover and replacement paths available. The ledger records a network reality it did not create. It does not own the networks, assets or operator rights reflected in its entries.

That boundary matters precisely because elections carry the language of legitimacy. A ballot can validly select a corporate director. It cannot vote a private association into being a state. A member meeting is not a legislature. An election committee is not an electoral commission for a continent. A service region is not a sovereign people. A proclaimed community is not automatically a principal capable of delegating public power.

No result at AFRINIC-10, however cleanly authorised, could create regulatory, police, prosecutorial, punitive, confiscatory or adjudicative power over operators, networks or number resources. AFRINIC cannot acquire those powers by ceremony, repetition, institutional support or internal confidence. Its influence comes from useful coordination and shared recognition. Influence should not be misdescribed as jurisdiction.

This does not make the internal election meaningless. Private governance matters. Directors can make decisions within the corporate body, appoint or supervise officers where law permits, oversee records, approve expenditures and direct the institution’s legitimate service functions. Those acts can affect members and deserve accurate authorisation. The point is scope: the cleaner the election, the stronger the internal corporate mandate; the election still cannot enlarge the lawful office that the voters were entitled to fill.

The distinction between stakeholder and principal is useful here. Meeting participants, technical specialists, supporters and affected operators may contribute evidence, preference and warning. Only a competent principal can authorise a delegate, and only within the principal’s own power. An AFRINIC member can transmit its own membership vote through a valid proxy. It cannot, through that vote, grant AFRINIC sovereign authority that the member itself does not possess.

Operators remain the central actors in the running Internet. They finance infrastructure, configure routers, enter customer agreements, maintain services and bear continuity risk. The registry serves that reality by keeping a dependable common record. When the recordkeeper mistakes internal ritual for a throne, it reverses the relationship. The ledger begins to appear as the source of the network rather than evidence about it.

The 2009 proxy question is valuable because it forces attention back to the real direction of authority: from an eligible principal, through a bounded delegation, to a private ballot—not outward from a ballot toward invented public power.

The practical impact of an incomplete chain

The immediate impact of missing instrument-level proof is epistemic. Members, auditors and later decision-makers cannot distinguish a fully controlled process from one that merely reached the right aggregate by habit. They can see that seventeen proxies were counted and recounted. They cannot independently test whether seventeen eligible principals each generated one authorised ballot.

The next impact is institutional. If the public record relies on assurance without showing control design, every later dispute becomes a contest of trust. Defenders point to the report and the committee. Critics point to the missing forms. Neither side can resolve the disagreement from published evidence. A narrow technical recordkeeping problem becomes a legitimacy argument because the evidence architecture was not designed to carry the weight later placed upon it.

The operational impact is not that the 2009 result must be discarded. The record does not justify that remedy. It is that future reliance on the event should remain bounded. The report can support statements that proxies were permitted under the agreed-and-duly-signed condition, that they were used, and that a seventeen-plus-one aggregate was recounted. It cannot support statements about the validity or choices of each proxy ballot.

The governance impact is a warning against retrospective inflation. An institution may be tempted to treat a long-ago election as a source of continuous, enlarged legitimacy. But the event did only what a private election could do at that time. It selected corporate officeholders within the available authority. It did not create an enduring claim to govern operators, and incomplete evidence should make later institutional storytelling more modest, not more ambitious.

There is also a design impact. A registry is already a recordkeeping institution. Maintaining a proxy-authorisation ledger is not alien to its function; it is exactly the sort of narrow, factual coordination it should perform well. Accurate identities, dated authority, stable records, controlled access and reproducible reconciliation are bookkeeping virtues. They do not require a sovereign posture. They require disciplined administration.

The seventeen-plus-one moment therefore offers a compact test of institutional maturity. Can AFRINIC distinguish a number from the authority behind the number? Can it preserve confidentiality without sacrificing reproducibility? Can it accept a useful delegation method without treating signatures as talismans? Can it recognise that a sound internal vote remains internal? Those questions stay focused on the act in Cairo while reaching the core of registry accountability.

What the surviving record supports

The public evidence supports a short, firm set of conclusions. The election committee stated that people carrying proxies that had been agreed and duly signed could vote. The physical process contemplated a member or proxy approaching the box after the member name was marked, followed by reconciliation. The report records E-networks with multiple proxies, seventeen proxy votes plus one vote associated with Mark Elkins, and a recount that remained seventeen plus one. It records forty-one ballots cast and forty valid.

The same evidence requires an equally firm set of reservations. It does not publish the seventeen instruments or identify every principal. It does not explain “agreed,” show who checked a signature, establish each signatory’s corporate authority, state receipt times, publish acceptance or rejection records, resolve duplicates or revocations, connect each instrument to one issued ballot, identify the invalid ballot, or reveal how any proxy voted. It does not prove an effect on either contest.

Those reservations do not amount to an accusation. The forms may have been genuine and effective. The committee may have performed careful checks. The private file may once have been complete. The public report simply does not permit an independent finding on those questions. Unknown means unknown, not suspicious by default and not valid by default.

The strongest defensible judgment is thus narrower than either pole of the debate. AFRINIC-10 implemented proxy voting in a consequential part of its physical ballot process. A legitimate mechanism was available to transmit private member rights. But public reproducibility stopped at the aggregate. The evidence chain visible today runs from the committee’s rule to the call, recount and box totals; it does not run backwards through every instrument to an eligible corporate principal.

That is enough to shape future practice. An institution does not need to expose private signatures to show that its controls worked. It does need to preserve and attest to the principal-to-ballot path. Where a substantial share of ballot weight is delegated, that path is not secondary paperwork. It is the election’s authority ledger.

And the scope of what the ledger could authorise remains fixed. A complete chain could substantiate a private vote. It could not substantiate a claim of sovereignty. The most reliable reading of the Cairo record therefore combines procedural seriousness with institutional modesty: prove each delegation, count only what the proof permits, retain enough evidence for review, and never confuse a well-run corporate election with public authority.