Summary

  • LACNIC reported 13,178 organisations entitled to vote and 978 organisations participating in its May 2026 extraordinary Board election. The exact organisation turnout was 7.421%, which LACNIC displayed as 7%.
  • Those 978 organisations produced 2,145 weighted votes because LACNIC assigns one to eleven votes according to membership category and address space. The public result does not disclose the total eligible weighted vote power, so a weighted-turnout rate cannot be reconstructed.
  • Luis Fernando Arancibia Medina won with 889 weighted votes, or 41.45% of votes cast. Jorge Mario Valencia received 670, abstention received 303 and Juan Pablo Barrera Guzman received 283. The 219-vote margin was clear under LACNIC’s highest-votes rule, even though the winner did not receive an absolute majority of votes cast.
  • The Electoral Commission certified that it audited 100% of votes received. LACNIC also published a 313-page electoral roll, a roll-challenge period, provisional results and a voter-audit window. Those are meaningful controls. They do not reveal how eligible voting power was distributed or how many organisations stood behind each weighted total.
  • Nothing in the reviewed record establishes fraud, manipulation, a wrong tally, candidate misconduct, illegality or an invalid election. The narrower finding is that LACNIC published one participation denominator but not the other, then seated a director in an institution whose private corporate decisions can have consequences for number-resource holders across a large service region.

Four numbers, four different units

Election reports are most trustworthy when every number answers a clearly named question. LACNIC’s official result for the extraordinary Board election offers four headline numbers: 13,178, 978, 2,145 and 889. They cannot be compared safely until their units are separated.

The first number is the count of organisations entitled to vote. It is not a count of individual people and it is not the total number of weighted votes available. Each eligible organisation acts through a registered membership contact. That contact exercises an institutional right on behalf of the member.

The second number is the count of organisations that participated. LACNIC reported 978. Dividing 978 by 13,178 produces 7.42146%. LACNIC rounded that to seven percent. This is a valid organisation-level participation calculation.

The third number is the total weighted votes cast. LACNIC reported 2,145 across the three candidates and abstention. This total exceeds the number of participating organisations because members do not all have equal voting weight.

The fourth number is the winner’s weighted tally. Arancibia received 889 votes. Those votes cannot be translated into 889 organisations. A large member may carry more votes than a small member, and the public result does not identify the number of organisations behind a candidate’s total.

This distinction prevents an attractive but false calculation. Dividing 889 by 13,178 would compare weighted votes in the numerator with organisations in the denominator. It would not show that 6.75% of eligible organisations supported the winner. The record does not contain the organisation-level support count needed for that statement.

The correct arithmetic is more limited. The organisation turnout was 7.421%. The winner’s share of weighted votes cast was 41.445%. Those percentages describe different populations and answer different questions.

What happened in the extraordinary election

LACNIC’s Board issued the call on 15 January 2026. It said a permanent vacancy had arisen and invoked Article 22 of the bylaws. The elected director would begin on 1 July 2026 and complete the vacant term through 31 December 2028.

The Board had discussed the process in November 2025. Its approved minutes record debate about the timetable, the need for multiple candidates and the nationality restrictions produced by the composition of the existing Board. The Board unanimously approved the calendar and targeted a January announcement.

Candidate nominations opened on 24 March and closed on 7 April. LACNIC published the voter registry on 8 April. Members could challenge or correct that registry until 6 May. Candidate information appeared on 5 May, followed by a short period for questions and complaints.

Online voting opened on 21 May and closed on 28 May. LACNIC published provisional results immediately after the ballot. It then scheduled 2 and 3 June for voters to audit their votes and submit complaints about the voting process.

The calendar contained two certification paths. If no complaints were filed, official results and an Electoral Commission letter would appear on 3 June. If complaints required resolution, the Commission would respond between 4 and 8 June and publish the official result on 9 June.

The certificate is dated 3 June. It states that the Commission conducted the general count, audited 100% of votes received and certified Arancibia as the winner with 889 votes. That timing is consistent with the no-complaint publication path, although the bounded public record does not establish whether informal questions or non-dispositive complaints were received.

Arancibia now appears on LACNIC’s Board page as a director whose term ends in December 2028. The election was therefore not advisory. It filled a real corporate office on the body that manages and administers the institution.

The tally is reproducible up to a point

LACNIC’s candidate table can be checked. Arancibia’s 889 votes equal 41.445% of the 2,145 weighted votes cast. Valencia’s 670 equal 31.235%. The 303 abstentions equal 14.126%. Barrera Guzman’s 283 equal 13.193%. The entries sum exactly to the reported total.

Arancibia’s margin over Valencia was 219 weighted votes, or 10.210 percentage points of the votes cast. That is a decisive plurality under Article 25, which says that the candidates receiving the highest number of votes fill the available positions.

The winner did not receive an absolute majority of the weighted vote. That observation does not invalidate the result. LACNIC’s published election rule does not require a majority for a one-seat contest; it requires the highest number of votes. Nor does the reviewed text state a minimum turnout threshold for this election.

Low participation and rule validity must therefore be kept separate. A seven-percent turnout may be sufficient to produce a winner under the bylaws. It may still be inadequate evidence for a rhetorical claim that a broad regional community authorized the result.

The same separation applies to abstention. The 303 abstention votes were active ballot choices and belong in the 2,145-vote total. The 12,200 eligible organisations that did not participate made no recorded ballot choice. Their absence cannot be reclassified as abstention, approval or opposition.

Why the vote-weight denominator matters

LACNIC does not use one organisation, one vote. Its public table assigns one vote to End Users. ISP members receive between one and eleven votes according to category and the amount of address space assigned. The table also explains how fee discounts or waivers can affect the voting entitlement of organisations in special membership situations.

Weighted voting is not automatically illegitimate. A membership association may decide that members with different economic or operational relationships to the institution should carry different voting power. The rule is visible, is embedded in the bylaws and can be debated by members.

But a weighted system creates two participation questions. How many eligible organisations voted? How much eligible voting power was exercised? LACNIC answered the first for the Board election. It did not provide the public inputs needed to answer the second.

The 978 participating organisations cast 2,145 weighted votes, an average of 2.193 votes per participating organisation. That average says something about the participating pool. It does not show whether larger members participated at higher rates, whether smaller members dominated numerically or whether non-participation was concentrated in any tier.

To compute weighted turnout, a reader needs the aggregate voting entitlement of all 13,178 eligible organisations. The public voter roll lists organisations and countries but does not attach membership categories or vote weights. The result page gives votes cast but not votes available.

This omission matters because organisation turnout and voting-power turnout may diverge. If high-weight members participate more often, a small organisation turnout could still exercise a much larger share of eligible voting power. If low-weight members participate more often, weighted turnout could be closer to or below the organisation rate. The existing numbers do not choose between those possibilities.

The omission also obscures concentration. A candidate’s 889 votes might come from many low-weight members, fewer high-weight members or a mixture. Ballot secrecy properly prevents publication of the organisation-level choices. Aggregate tier data could reveal the structure without identifying any voter.

A long roll is useful, but not a reconciliation

LACNIC published a 313-page electoral roll for the extraordinary election. This is not a symbolic gesture. Members can check whether their organisation appears. Candidates can inspect the institutional electorate. Researchers can see the geographic breadth and the names of eligible organisations.

The roll also demonstrates the scale behind the 13,178 figure. It is a list of organisations, not a list of 13,178 individual electors. The designated membership contact is the person who exercises each organisation’s right.

Yet the roll is not a complete election reconciliation. It does not state each organisation’s voting weight. It does not show the count of eligible organisations by tier. It does not identify changes between the draft and final roll or aggregate the reasons for addition, removal and correction.

Those limits do not make the roll defective. Publishing organization names while withholding personal contact data is a sensible privacy choice. The point is narrower: the roll supports one denominator and cannot support the other.

A machine-readable summary could close the gap without attaching sensitive attributes to named members. LACNIC could publish the number of eligible organisations and total eligible votes for each tier, plus the number of participating organisations and votes cast for each tier. Country and language aggregates could be added where cell sizes are large enough to prevent inference.

What “100% audited” proves

The signed certificate deserves precise treatment. Five Electoral Commission members placed their names on a document stating that 100% of votes received were audited. It identifies the election, closing date, winner and winning tally.

That statement is stronger than an unattributed webpage. It creates a named institutional responsibility and says the entire received-vote population, rather than a sample, was within the audit.

The certificate does not describe the audit method. It does not state whether the Commission reproduced the tally from raw records, checked only a system report, reviewed credential issuance, inspected configuration changes or reconciled invalid and unused vote units. It does not identify an external auditor.

None of those absences proves that the Commission performed weak work. The certificate may rest on detailed records and competent controls that are not public. The proper conclusion is about the scope of public evidence, not the quality of unseen work.

The certificate also addresses votes received. It does not certify the unreported aggregate weight of eligible organisations, the representativeness of participating members or the political scope of the resulting mandate. A complete tally audit and a complete participation account are different controls.

The strongest defence of LACNIC’s process

LACNIC can make a serious defence. The Board did not fill the vacancy privately. It issued a public call, published dates, accepted multiple candidates and applied disclosed nationality restrictions. Members had time to inspect the electoral roll and challenge it.

The ballot was online. Provisional results were published before certification. Voters received a dedicated interval to audit their own votes and raise complaints. The Electoral Commission then signed a certificate and LACNIC published candidate and abstention totals.

The voter roll is unusually substantial public evidence. The vote-weight table is also public. The bylaws state how winners are selected. Readers can reproduce the candidate percentages and the organisation turnout.

The institutional defence is therefore not that people should simply trust “the community”. It is that a private membership organisation followed a disclosed process, enabled member review and obtained certification from the elected body assigned that function.

That defence has force inside the association. It is also bounded. Procedural compliance tells us that an office was filled under LACNIC’s rules. It does not tell us that the 978 participating organisations represented the 12,200 non-participants, non-member resource users, network customers, governments or populations within the service region.

The legal form sets the mandate boundary

LACNIC’s own bylaws are direct about legal form. They establish a non-government organisation headquartered in Montevideo and governed by the bylaws and applicable rules. They define members as holding institutional power, the Assembly as the governing body and the Board as the body that manages and administers the organisation.

This clarity is useful. It provides a lawful place for the election result. Active and Founding Members exercise defined rights. Their weighted ballots fill a Board vacancy. The Electoral Commission certifies the result. The director then holds powers given by the bylaws.

The distortion begins only when that corporate act is enlarged into a public mandate for Latin America and the Caribbean. A service region is a territory in which a registry performs functions. It is not a constituted electorate. A list of member organisations is not a census of operators, users or citizens.

Heng Lu’s doctrine separates stakeholders from principals. A stakeholder is affected by a decision. A principal supplies authority. LACNIC’s members are principals for defined association acts. Many other actors are stakeholders because registry records, transfer rules, RPKI and operational continuity matter to their networks.

The existence of those effects does not retroactively enroll every affected party into the 2026 ballot. Nor does it make the ballot meaningless. It means the mandate must stay where the legal and evidentiary chain places it: inside LACNIC’s corporate governance.

A second 2026 election confirms that the issue is structural

The extraordinary Board election was not LACNIC’s only low-participation statutory ballot in 2026. The Fiscal Commission election reported 12,710 eligible organisations, 1,002 participants and 2,170 weighted votes. The exact organisation turnout was 7.884%, displayed as eight percent.

That comparator should be used cautiously. The voter-roll date differed. The office was different. Candidate interest and member attention may differ between a Board vacancy and a Fiscal Commission seat.

Still, two one-seat statutory elections in the same year produced similar organisation participation and weighted-vote totals. That weakens an easy explanation that the Board figure was merely an extraordinary-election anomaly.

It also creates an opportunity. LACNIC can publish one common result schema across all statutory elections. The schema could standardise eligible organisations, eligible weighted power, participating organisations, votes cast, abstentions, invalid ballots, audit scope and complaints.

The Electoral Commission page shows why a common schema is needed

Another 2026 result page contains a facial inconsistency. The Electoral Commission election page reports 12,658 authorised organisations, 2,961 organisations voting, 2,961 total votes and seven-percent participation.

Those numbers cannot all describe the units printed beside them. If 2,961 organisations participated, organisation turnout would be about 23.4%, not seven percent. Because 2,961 is also the total weighted vote count, the most plausible public-reading explanation is a copied value or label.

That is an inference, not a finding about the underlying election. LACNIC may hold a correct participation count elsewhere. The page should be corrected or annotated so readers can tell which value is wrong.

The inconsistency does not prove a bad tally. It proves that prose labels and manual result tables are fragile. A structured election dataset with unit validation would reject an organisation count that does not reconcile to its stated percentage.

What the record cannot establish

The bounded evidence does not reveal the total eligible weighted vote power. It does not show participation by tier, country, organisation size or language. It does not show how many organisations supported each candidate.

It also does not establish whether every participating organisation used its full voting entitlement. The 2,145 total may equal the sum of all entitlements exercised, but the public result does not expressly provide that reconciliation.

The package does not establish the electronic system’s version, configuration, security assessment, event log or independent technical review. LACNIC describes the platform as open source and auditable, but that self-description is not a substitute for a result-specific audit record.

The reason for the permanent vacancy is outside this article’s bounded source set. That fact is not needed to calculate participation or evaluate disclosure. It should not be guessed.

Most importantly, there is no evidence here of manipulation, corruption, coercion, an incorrect winner or illegality. A missing denominator is not a hidden ballot. It is a missing aggregate disclosure.