A Board Roster Is Not an Election Record: What NANOG’s Public Gov

Summary

NANOG’s public material exposes meaningful but uneven parts of an election chain: nomination windows, candidate fields, committee descriptions, voting windows, selected denominators, aggregate votes and elected names. A Board roster is therefore connected to a visible electoral process, but it is not itself a complete public election record.
The disclosed numbers describe a defined membership electorate in particular years. They do not establish the views of non-voters, meeting attendees, sponsors, the whole network-operator sector or the wider public.
A small, dated aggregate record could make future elections easier to inspect without disclosing individual ballots: eligibility basis and denominator, nominations, candidate field, voting window and method, aggregate results, certification reference, material challenge category and an explicit secrecy boundary.

A list of names answers the smallest question

A public Board page is useful. It tells readers who has been elected to serve, who may speak or decide in an institutional role, and where a reader can begin to understand governance. But a roster answers only the smallest possible question: who is on the Board now. It does not by itself answer how the names reached the page, who could vote, when nominations opened, whether a candidate withdrew, what rule governed the ballot, how many valid votes were cast, whether a result was certified, or how a material complaint would be handled.

That distinction is not an argument that a Board needs to publish every administrative detail. Nor is it an allegation that an incomplete public file conceals a defect. Elections normally contain information that should remain private: individual ballots, contact details, support statements, personal disputes, and sometimes legally protected membership information. A volunteer body also has finite time. It may make more sense to spend that time running a sound election than building a permanent public archive with the density of a state election commission.

Still, there is an important middle ground between a bare roster and a raw-ballot release. A reader can reasonably ask for enough aggregate information to distinguish an appointment, a committee selection, a membership election and a public-relations announcement. NANOG’s own materials show why. Across 2019, 2020, 2021, 2023 and 2024, the organisation made different parts of its election process visible. Taken carefully, those pages do not prove a single uniform standard. They do show the outline of a public record that can be made more legible without turning a professional association’s vote into a spectacle.

The right analytical question is therefore modest: what can the published record actually establish, and what cannot it establish? The answer is more useful than either extreme. It allows readers to recognise published electoral information where it exists, to avoid turning missing fields into accusations, and to see what a proportionate future record might contain.

The chain that an election has to traverse

An election is not one event. It is a chain of decisions and facts. Someone has to define who is eligible. A nomination window has to open and close. A candidate field has to be formed, including any withdrawals. A voting period and ballot method have to be communicated. Votes have to be counted under a stated rule. A result has to be certified or otherwise accepted under the organisation’s procedures. In some cases, a governing-document question travels alongside the Board contest. If an issue is raised, there has to be a way to classify and resolve it, even when the details remain confidential.

Readers do not need all of those steps in the same level of detail. A private membership list should not become public just to establish that a denominator exists. An individual ranking should not be exposed to establish that ranked-choice voting was used. A challenge may contain personal material that should not be published. Yet an aggregate public account can still show that a stage existed and what its result meant.

That is particularly important in an organisation whose public language invokes membership. A member election is a bounded mechanism. It can create authority inside the organisation under the rules that define that membership. It does not turn everyone affected by NANOG, every person who attends a meeting, every sponsor, or every network operator in North America into a voter. The boundary is not a weakness. It is the condition that makes a denominator intelligible. Once the electorate is blurred into a larger public, even a perfectly counted ballot begins to carry claims it cannot sustain.

NANOG’s past-elections index states that elections take place each year in accordance with its bylaws and directs readers to annual summaries and results. That is an important starting point because it identifies continuity: elections are not described as a one-off exercise. But an index is a map, not the territory it maps. A reader still has to open the individual years to see whether the same fields are visible, whether the same voting method was in use, and whether the available information supports the same inference.

The rest of the public file makes one fact clear: disclosure is not all-or-nothing. Some years reveal nominations and committee size. One year supplies a denominator, aggregate votes and candidate-level figures. Later years identify a ranked-choice method and provide aggregate counts. These are building blocks. The mistake would be to force them into a claim that every election had identical public documentation, or that one visible field fills every unseen gap.

2019 and 2021 show the administrative skeleton

NANOG’s 2019 annual report describes a Board election with a four-week nomination window. Five candidates ran for two positions. The report says that a seven-member Election Committee oversaw a process that was open for 52 hours, and it names Benson Schliesser and Patrick Gilmore as the newly elected Board members. It also records a separate bylaw outcome: the organisation’s name changed from NewNOG, Inc. to NANOG, Inc.

That is a meaningful public description. It identifies a time-bounded nomination stage, a candidate field, the number of open positions, an Election Committee, a voting duration, elected names and a governing-document issue. A reader does not have to guess whether the names on the subsequent roster arrived through a random selection or a membership contest. The report makes clear that a Board election took place.

But the same document does not publish every field that might matter to a reader trying to reconstruct the election in full. It does not state the number of eligible voters. It does not give a total vote count. It does not state the ballot method. It does not display a candidate-level tally or a certification record. It does not describe a challenge process. None of those omissions establishes that the relevant controls did not exist. The report is a particular public account, not a complete claim about the organisation’s private administrative files.

The 2021 annual report offers a similar but not identical outline. It describes a November Board election, a one-month nomination window, five candidates for two Board positions, a five-member Election Committee and a process open for 52 hours. It names Dave Siegel and Steven Feldman as the elected members. Again, the public can see a nomination period, a candidate count, the number of seats, committee involvement, a voting duration and a result.

The differences matter. The reported committee size differs from the 2019 account. The nomination window is described differently. A public reader should not silently convert either description into an unchanging constitutional rule. The useful conclusion is narrower: both reports put an administrative skeleton in public view. They do not give a complete, year-by-year measurement of the electorate or ballot.

Why is that distinction worth retaining? Because a process can be transparent in one dimension and opaque in another without being either wholly open or wholly closed. A four-week nomination period is a real disclosure. So is the number of candidates. It tells a prospective candidate and a reader something about access to the contest. But it does not tell a reader how many eligible members saw the call, how many considered standing, how eligibility was assessed in each case, or how votes distributed among the field. Those are separate questions, and they need separate evidence.

The same logic applies to the 52-hour figure. A voting window is not a turnout rate. It says how long the ballot was open, not how many people could vote, how many did vote, whether reminders reached everyone, or what the result means beyond the stated electorate. It can be an important operational fact without carrying an implied verdict on participation.

2020 is a useful disclosure benchmark, not a universal template

The 2020 materials make the public record more granular. NANOG’s candidate page set out requirements that Board candidates must be members in good standing and must attend at least two of three yearly NANOG meetings while in office. It published a timetable for nominations, statements of support and voting. That kind of information matters because it shows that candidacy is not merely a public declaration. It is linked to a stated membership condition and an attendance expectation.

The candidate page does not, however, establish that every person who might have wished to stand was eligible, that every eligible person chose to stand, or that every stated requirement was independently audited in public. A requirements page tells readers the rule being presented; it is not a complete audit file. The distinction protects against a common overreach in governance reporting: treating an eligibility criterion as proof that the underlying population has been measured and verified in every relevant detail.

The 2020 results page adds another layer. It reports 553 eligible voters and 176 total votes cast for the Board election. It names Tina Morris and Vincent Celindro as elected and provides a candidate-level table. It also separately reports a bylaw election with its own eligible-voter and total-vote figures. That separation is valuable. It avoids merging a Board contest and a governing-document vote into one undifferentiated expression of support.

These numbers can support a precise statement: for the Board election described on that page, NANOG reported an eligible electorate of 553 and 176 votes cast. They can also support a historical statement about the listed winners and the published candidate-level results. They should not be made to do more. The figures do not reveal the identity of non-voters, their reasons for not voting, or the preferences of people outside the stated electorate. They do not establish a general participation pattern for later years. They do not establish that the public has a complete account of every ballot-control or certification step.

It is tempting to calculate a percentage immediately. That temptation should be resisted unless a reader has a genuine reason to compare like with like. A quotient is a measurement within a defined reported electorate. It is not a certificate of organisational legitimacy. It does not tell us whether a member received a reminder, whether the denominator was measured under the same rule in another year, whether people were travelling, or whether a particular election faced unusual circumstances. More importantly, it does not permit a leap from member participation to a mandate for people who were never part of the election.

The practical lesson of the 2020 page is not that every future public record must reproduce a full candidate-level table. Candidate-level figures may be useful in some voting systems and less meaningful in others. The lesson is that an aggregate denominator, aggregate total, dated timetable and clearly identified result can turn a result announcement into something readers can inspect. It is a benchmark for minimum clarity, not a demand that all elections use the same rule or display the same granularity.

2023 makes the electorate visible, while keeping its boundary intact

NANOG’s 2023 results page records general election voting from 17 to 25 October. It lists five active Board candidates—John Jason Brzozowski, Vincent Celindro, Ron da Silva, Cat Gurinsky and Marlin Martes—and identifies Les Williams as having withdrawn his candidacy on 6 October. It reports 762 eligible voters, 230 total votes and one abstention. It names Cat Gurinsky and Vincent Celindro as the elected candidates.

Those details make several points visible at once. A reader can see that a candidacy can change after the field first forms. A withdrawal is not the same as an unreported absence from a final list. A date range allows the voting period to be located in time. The eligible-voter number creates a defined denominator for the reported total. The abstention figure makes clear that a ballot need not be read only as a candidate preference. The winner names complete the public outcome.

The annual report adds that the Election Committee moved NANOG to ranked-choice voting in 2023 and reports the same 762 eligible voters and 230 votes. It names the same elected members and records no bylaw amendments that year. Taken together, the page and report establish that the 2023 election was described as using a ranked-choice system and that aggregate electorate and vote numbers were published.

They do not establish every possible claim about ranked choice. A label for a voting method is not a complete tally audit. The cited material does not reproduce all ranking rounds or every ballot. It does not tell readers whether every voter ranked the same number of candidates, whether each voter understood the system in the same way, or whether another method would have produced a better result. Those are different questions. A public report should not manufacture an answer simply because the phrase “ranked choice” sounds technically complete.

The same caution applies to the number 762. It is a reported eligible electorate for the election described; it is not a count of all people with an interest in NANOG’s work. It does not tell us how many members read a candidate statement, how many were present at a meeting, or how many people in the broader networking field would have wanted a vote. A membership system necessarily has a perimeter. The best public account identifies that perimeter rather than pretending it disappears.

This is where language about mandate often goes wrong. An election can confer a legitimate internal role on the candidates elected under the organisation’s rules. It does not follow that an electorate has conferred a general mandate to speak for every operator, every customer, every vendor or every technical community. Keeping those propositions separate does not diminish an election. It preserves its actual meaning.

The 2024 comparison should remain a comparison, not a verdict

The 2024 annual report says that NANOG continued to use a ranked-choice voting system for its annual election. It reports 698 eligible voters and 165 total votes, names Steve Feldman and Elizabeth Culley as elected, identifies an Election Committee and says that there were no bylaw amendments.

Readers naturally compare numbers across adjacent years. A comparison can be worthwhile: the source puts a denominator and aggregate total beside a stated voting method in both 2023 and 2024. But interpretation has to remain proportionate. The figures alone do not say why the reported eligible electorate changed from 762 to 698. They do not say why the total votes changed from 230 to 165. They do not show whether membership terms changed, whether the timing or communications differed, whether the candidate field had a different shape, whether a member chose not to vote for personal reasons, or whether every other condition was comparable.

The right conclusion is simple. The public record contains different reported aggregate figures for two different annual elections. It is not enough to diagnose a cause. A decline in a count is not self-interpreting. It should not be treated as evidence of disaffection, endorsement, defective procedure, or a measure of the elected Board’s authority.

That restraint is useful to NANOG as well as to its readers. A governance system should not be forced to defend every variation in a number as a moral verdict. Elections exist within a calendar, a membership cycle, professional demands, travel schedules and personal decisions. A compact public record can make the denominator and method clear without pretending to know the motives of everyone who did not cast a ballot.

At the same time, a reader need not ignore the difference. It is a legitimate monitoring point. Future annual material could make comparisons easier by recording the eligibility basis, the key dates, the candidate field, the voting method and the aggregate results in a consistent dated form. Consistency does not create a broader mandate. It makes the scope of a disclosed result easier to understand.

Ballot access and committee remit describe functions, not motives

NANOG’s member-voting help explains that a member can access a ballot during an election cycle through the member profile and that the member should confirm good-standing status. Its Election Committee description says the committee assists the Board by conducting elections for Board positions, proposed amendments to governing documents and other election-related matters directed by the Board. It also says that the committee reviews the bylaws periodically.

These pages help separate functions. A member profile is an access surface. Good standing is a stated membership condition. The Election Committee has a stated remit. The Board has a different role. A candidate page provides another view of candidacy conditions. A results page records an outcome. A useful public account does not merge those functions into a single anonymous actor called “NANOG.”

Role separation is valuable precisely because it lowers the temperature of analysis. If a reader sees an Election Committee named in an annual report, the safe claim is that the report identifies a committee with a stated election role. It is not safe to infer that the committee unilaterally controls eligibility, that it favoured a candidate, or that the Board’s relationship to the committee is improper. Likewise, a Board-directed election-related matter is not evidence that every Board action controls a ballot outcome. The public descriptions tell us what the pages say. They do not reveal private motives.

The distinction also clarifies what a better record should show. It should identify the function responsible for each stage without publishing a private deliberation. For example, a public result page could say that eligibility was assessed under the membership rule, nominations were handled during a stated window, the Election Committee administered the voting process, and a specified certification reference was issued. That is not a transfer of decision-making power. It is a map of the route through which a result became public.

A membership election is neither a meeting survey nor an industry referendum

NANOG’s election record needs to be read through the scope of its electorate. Members who meet the organisation’s rules are not the same population as everyone who has attended a NANOG event. Attendees are not necessarily a subset that maps cleanly onto active voters. Sponsors may have commercial relationships with events, but sponsorship is not itself a voting category in the cited materials. Network operators outside NANOG, end users, governments, vendors and members of the public may all be affected by discussions in the broader ecosystem without being parties to a NANOG membership ballot.

This does not make the election trivial. Organisations need defined procedures for choosing Board members. A membership vote can be an appropriate way to do that. But it does mean that claims should use the language the evidence supports. “Elected by the reported membership electorate” is a meaningful statement when the record supplies the relevant context. “Mandated by the industry” is a different and much larger claim. It needs evidence that these sources do not provide.

The error runs in both directions. One can wrongly inflate a bounded vote into a universal mandate. One can also wrongly dismiss a bounded election because it is not universal. Neither is useful. The first erases the people who were not eligible to vote. The second ignores the fact that a membership organisation may have a real internal decision rule. A good governance file helps readers hold both truths at once: internal authority can be real, and its scope can remain limited.

The 2023 and 2024 denominators are therefore best read as identifiers of scope. They tell a reader that the result was attached to a reported eligible electorate of a particular size in a particular year. They do not tell a reader what every non-voter wanted. They do not tell a reader whether people outside the electorate accept the result. And they do not have to. A public election record becomes stronger when it says exactly what it measures instead of reaching for a grander word than its evidence can carry.

The smallest useful public election record

The practical remedy is not a demand for exhaustive disclosure. It is a compact, repeatable public election record with a clear privacy boundary. Its purpose would be to let readers reconstruct the official path without exposing individual ballots or turning routine administration into a public performance.

Election year and record version. The page should say which election it covers and when the record was published or revised. A version date matters because candidate fields can change, corrections can be needed and an annual archive should not rely on readers guessing which page was current at a particular time.

Eligibility basis and aggregate denominator. The record should identify the rule or membership status used to define eligibility and provide the aggregate count of people eligible to vote when safe to do so. It need not publish a member list. The point is to distinguish a total vote count from a larger unknown population.

Nomination and voting windows. The 2019, 2020, 2021 and 2023 materials show why dates are useful. A reader can understand that a field was open for nominations and that a ballot had a bounded period. A window is not proof of equal awareness or equal access, but it is an essential piece of the process chronology.

Candidate field and withdrawals. A public final list should make clear who stood and whether an announced candidacy was withdrawn. The 2023 page illustrates the value of stating a withdrawal directly. This does not require publishing personal reasons, which may be private. It prevents a later reader from mistaking a changed list for a missing record.

Voting method. If the organisation uses ranked choice, a page should state that. If it uses another method, it should say that instead. A one-line method label cannot substitute for a technical audit, but it tells readers what sort of result is being presented and avoids retrospective confusion.

Aggregate votes, abstentions where reported and result. The 2020 and 2023 pages show the public value of aggregate totals. The result should name elected candidates, and an abstention total can be useful when it is part of the record. The purpose is not to force publication of every ballot or every intermediate tally. It is to allow the declared outcome to be connected to a stated aggregate.

Certification reference and material challenge category. A public record can point to a certified-results document or state the applicable result-certification reference. If a material issue is formally raised, the page can state a category and disposition—resolved, withdrawn, pending, or not published for a stated privacy or legal reason—without exposing a person’s private information. The absence of a public challenge detail should never be treated as evidence that a challenge existed. The field is about making a route available when an issue does arise.

An explicit secrecy boundary. The page should say that individual ballots, personal data and small cells are not published. This matters because readers need to know that aggregate disclosure is not a partial promise of personal exposure. The record is a public account of an election chain, not a voter register exposed to everyone.

None of these fields tells a committee whom to elect. None turns a member vote into an industry referendum. None assumes that every change in an aggregate figure has one cause. Together, however, they let a reader see where a result begins and ends. That is the practical value of a record.

Certification, review and public explanation are different safeguards

It is useful to separate three safeguards that public discussion often compresses into one. Certification is the organisation’s formal act of accepting or attesting to a result under its own rules. Review is the work of checking whether a question about eligibility, procedure or counting needs to be resolved. Public explanation is the reader-facing account that lets someone outside the election administration understand the scope of the result. They overlap, but none automatically supplies the other two.

A certified result can be real even when its public explanation is short. An internal review can be careful even when the underlying evidence cannot be published because it includes personal membership information. Conversely, a polished public summary can be easy to read while leaving unclear whether a result has reached its formal certification point. The remedy is not to make every internal document public. It is to identify the public status accurately.

A link or reference to the applicable certified-result record, a date, and a high-level disposition where a material matter is formally raised can tell readers much more than a winner list alone.

This distinction also protects candidates. A candidate who is not elected should not be made the subject of public speculation simply because a reader cannot see a complete administrative file. A nominee who withdraws should not have to disclose a private reason for the record to acknowledge that the final field changed. A voter should not have to give up ballot secrecy for an aggregate total to be credible. The point of an election record is to make institutional claims more exact, not to convert every person involved in an election into public evidence.

For NANOG, the useful public question is therefore not whether the cited pages reproduce every safeguard that may have existed. It is whether a reader can tell which safeguard the page is describing. A nomination timetable tells us about access to the field. A voter denominator tells us about the stated scope of an aggregate total. A method label tells us how to interpret the form of the result. A certification reference tells us where formal status rests. Keeping those distinctions visible is a more reliable form of accountability than treating any one disclosure as a proxy for the entire process.

The case for restraint is not a case for opacity

There is a strong counterargument to any proposal for more public election fields. Elections can become less trustworthy if people fear that their choices, affiliations or private disputes will be exposed. Volunteer committees can be overloaded by reporting duties. Candidate-level tables can create false precision under some methods or invite commentary that ignores the ballot rule. A formal public challenge category may tempt people to litigate ordinary disagreement in public. A membership organisation may also have contractual and legal obligations that limit what it can say about eligibility.

Those are real constraints. They are reasons to design a narrow record, not to write a maximalist transparency script. An organisation should not publish individual rankings. It should not publish a member roster to satisfy an outsider’s curiosity. It should not disclose small groups that could identify a voter. It should not expose private correspondence simply because it relates to an election. And it should not promise a public field that it cannot safely maintain.

The proposed record is deliberately compatible with those constraints. It relies on aggregate counts, dates, labels and high-level dispositions. It gives an organisation room to say that a field is withheld for a stated privacy, security, legal or proportionality reason. It preserves a distinction between a publicly inspectable process and a public performance of every private administrative act.

Restraint also protects governance judgement. A Board election is not a consumer rating. Members may have different reasons to vote, to abstain or to sit out a cycle. Candidates may withdraw for personal reasons. A committee may have to apply a rule to an unusual case. The point of public reporting is not to turn every variation into a verdict. It is to give readers enough context that a winner list does not have to carry the entire explanatory burden.

What would change this assessment

This assessment is intentionally conditional. It would narrow if NANOG already maintains a stable public record that consistently supplies the full aggregate chain described here: an eligibility basis and denominator, nomination and voting windows, final candidate field, voting method, aggregate results, certification reference and a documented privacy boundary. If such a page exists, the relevant task is not to invent a new mechanism. It is to direct readers to that record and assess whether its scope matches the claims made about an election.

It would also narrow if a proposed field proves unsuitable. A small electorate segment, a sensitive legal issue or a security concern may mean that aggregation should be coarser, delayed or omitted. Good governance is not the mechanical publication of every possible datum. It is the ability to say what a measure covers, what it excludes and why.

The assessment would become more urgent only if an organisation repeatedly presented a result as a broad mandate while declining to identify even the limited electorate, time frame or voting method behind it. The cited NANOG materials do not require that conclusion. They instead show an organisation that has, in different years, placed substantial pieces of election information in public view. The remaining opportunity is to make those pieces easier to read as one bounded chain.

A Board roster remains important. It identifies the people who hold an institutional role. But a roster is the end of a sentence, not the full account of how that sentence was written. A proportionate election record can preserve ballot secrecy, respect volunteer capacity and still give the public a clearer answer to the question that matters most: what, exactly, does this result show?

A comparison should preserve the boundary before it preserves the number

The practical value of a dated record is sometimes misunderstood as a demand for a league table. It is not. A year-to-year comparison becomes useful only after it preserves the terms of comparison. A denominator from one election cannot simply be laid beside a total from another if the eligibility basis, calendar, method, field or public description changed. A difference may be worth noticing, but noticing is not the same act as assigning a cause.

That is why a record should keep its definitions beside its numbers. If the eligibility rule is stated, readers know that a total is bounded. If the voting window and method are stated, readers can locate what sort of choice the total refers to. If an abstention is included, it can be read as an item in the published result rather than as a story about a person. If a candidate field changed, a dated withdrawal can prevent a later archive reader from treating an ordinary change as an unexplained inconsistency.

The same discipline matters when information is deliberately withheld. A page that says a small cell is not published for privacy does not turn the field into zero. A page that says a personal record is confidential does not invite a reader to invent a competing account. A page that offers only an aggregate result should be read as offering only an aggregate result. Public accountability is strongest when a reader can distinguish a measurement, an unknown, a protected detail and an inference rather than seeing all four collapsed into a single blank space.

For future readers, the modest ambition is continuity of meaning. They should not have to know the private history of a committee to see which election a page covers, who was eligible in aggregate, how the field and time window were described, what method was stated and where the declared result rests. They should also be able to see the limit: the record is a bounded account of an organisational election, not a census of an industry and not a demand that individual voters become public evidence.

A Board Roster Is Not an Election Record: What NANOG’s Public Governance File Can Show