Summary

  • A visible March-to-April 2025 sequence shows a motion to approve February minutes being deferred after discussion, an action to present edited minutes at the next meeting, and later approval of both February and March minutes.
  • That sequence closes the public approval state but does not reveal the edits, identify who made them or prove that a particular change contemplated in March was executed exactly as imagined.
  • Across seven public minutes, NANOG already records many useful fields: attendance, time boundaries, exact motions, votes, a bounded abstention reason, action labels, named actors, presented documents and executive-session presence or absence.
  • A lightweight closure index for material, non-confidential items could preserve exact state verbs, accountable roles, review dates, public evidence and explicit closure without turning minutes into transcripts or exposing protected information.

Six weeks, four public states

On 7 March 2025, NANOG's Board discussed a motion to approve its February minutes. The public record does not say that the motion was rejected; it says the motion was “deferred.” It then records a next step: “Present edited minutes at the next meeting.” On 18 April, the next public minute records a 7-0-0 vote approving the minutes of both 2 February and 7 March as presented.

That six-week sequence is unusually instructive because it resists two easy stories. Deferral did not mean rejection: the February minutes returned and received approval. But later approval does not reveal what edits were requested, who performed them, whether every contemplated edit was made or how one version differed from another. The March source does not disclose the discussion, the editor, an edit log or the exact February text. The April source closes one public state—the approval of those minutes. It does not expose a line-by-line revision history or independently verify the editing step imagined by a reader.

The distinction is not pedantry. It is the difference between tracing a governance decision and filling gaps with assumptions. The March entry supports “discussed,” “deferred” and “assigned for return.” The April entry supports “approved.” Neither document, alone or together, supports a detailed claim about the editing work between them. A reader can observe a state transition without pretending to see the work that produced it.

This is the central reading problem of public minutes. They preserve decisions in compressed form rather than reproduce every operational step. A motion can state exactly what the Board authorised. An action line can name a person, staff group or team. A later minute can report another vote or status. Yet “authorised,” “assigned,” “completed” and “verified” are not synonyms. Collapsing them makes public accountability both overconfident and unfair: overconfident because it claims more than the record shows, and unfair because an absent public update becomes a story that nobody acted.

NANOG is a voluntary network-operator forum and nonprofit organisation. Its Board minutes document acts within that organisation; they do not speak for every NANOG member, conference attendee, network operator or member of the public. A 7-0-0 Board vote can be credited as the specific Board act recorded that day without being enlarged into member-wide or industry-wide consensus. The public question is narrower: which fields allow a reader to identify the state of a material decision, and what small navigation layer would make later states easier to find?

Eight verbs, eight evidentiary boundaries

The most reliable way to read minutes is to begin with verbs. The relevant states are discussed, deferred, assigned, authorised or adopted or accepted, executed, verified, corrected and closed. They may look like a single progression, but the seven sources do not present a standard workflow, and an item need not pass through every state publicly. The vocabulary is a discipline for interpretation, not a claim about NANOG's internal procedures.

Public state What the state can establish What it cannot establish by itself
Discussed The subject came before the Board Agreement, decision or assignment
Deferred The immediate decision was postponed Rejection or abandonment
Assigned A next step was given to a person, role or team Performance or completion
Authorised, adopted or accepted The recorded Board act occurred in the motion's terms Execution, implementation or independent verification
Executed The assigned or authorised step is reported as carried out That the result met an external test
Verified The result was checked against a stated basis That every related question is resolved
Corrected A prior record or implementation changed A new Board act unless the record says so
Closed No further action is expected, or the item is explicitly superseded A licence to expose protected evidence

“Discussed” means a topic appeared before the Board. It does not establish agreement, a decision or an assignment. “Deferred” means the immediate decision was put off; it does not mean the proposition was rejected. “Assigned” means the minute identifies a next step and perhaps an actor or team. It does not prove that the work happened. “Authorised,” “adopted” or “accepted” records a Board act in the terms of the motion. It does not prove implementation, payment, configuration, delivery or external acceptance.

“Executed” requires evidence that the authorised or assigned step occurred. That evidence might be a later status statement, a published document or another non-confidential record, depending on the item. “Verified” is stronger and different: someone or some process checked the executed result against a criterion. A management statement that steps were taken is evidence of that statement; it is not automatically a published independent test. “Corrected” identifies a revision to a record, decision or implementation and should preserve what changed.

“Closed” should be explicit: the responsible body says no further action is expected, the item is superseded, or a final evidence link completes the public chain.

The danger lies in verb inflation. A carried motion becomes “implemented.” A named action becomes “done.” An accepted document becomes “final and verified.” A committee status becomes proof of outcome. Silence becomes “abandoned.” Each upgrade may sound natural in summary prose, but each changes the evidentiary state.

The opposite error is also common. Readers can treat any incomplete public chain as failure. Minutes may omit an operational update because the matter is routine, confidential, reported elsewhere or not scheduled for Board return. NANOG may maintain complete internal action tracking that the public sources do not show. The absence of a public owner, deadline, evidence link or closure field can be described as a navigation limit. It cannot be used to infer non-execution, poor performance, conflict, concealment, non-compliance or missing internal records.

This state grammar makes the March-to-April example useful without overstating it. The motion was discussed, deferred and given a next-meeting action in March. The minutes were approved in April. The public approval state advanced; the editing history remains unspecified. Precision allows both conclusions to stand without turning the gap between them into either proof of execution or an allegation of failure.

What the minutes already preserve

The seven documents are not sparse records. In different combinations, they identify attendees and regrets, call-to-order and adjournment times, workshops, presenters, motion identifiers, exact motion text, vote totals, action labels, named people or teams, documents presented, next-meeting dates, and the presence or absence of an executive session. One vote includes a bounded explanation for an abstention. Several actions identify a future review point or time horizon.

Those fields deserve credit because they make many Board acts independently readable. A reader does not have to infer that an agenda was approved when a motion and vote are printed. A conditional financial authority can be quoted in its own terms. A document list can be distinguished from the later wording of a motion. An executive-session marker can show where the public account appropriately stops.

Format varies over time. The 2019 minute is narrative. The November 2024 and March 2025 documents are visibly labelled draft. Later 2025 minutes use increasingly structured motion identifiers, presenters, vote and status fields, while action lines range from a bare staff assignment to named individuals and a one-year monitoring horizon. Variation may reflect changing secretarial practice, the subject matter and proportional effort. It is not evidence of concealment or institutional failure.

Nor can the seven minutes support an action-closure rate. They are selected around public decision states, not established as the complete population of Board decisions or tasks. The documents differ in period, format and subject. Turning the visible action lines into a percentage would create a denominator that the sources do not supply. The useful comparison is qualitative: which state fields appear, which links cross meetings, and where the public record stops.

The strongest existing feature is specificity at the decision point. The design opportunity is continuity after that point. A motion identifier can anchor a later update; an action line can carry a review date; a status can link to non-confidential evidence. The organisation need not publish more sensitive substance to make the already public pieces easier to navigate.

2019: disclosure and memory in the same sentence

The 3 May 2019 public minute identifies Board and staff participants and records a conference call beginning at 2:04 p.m. and ending at 3:02 p.m. EDT. The participant list shows who is named for the meeting. It is not a roll-call vote and does not prove that every listed person was present for every agenda item.

In a short section headed “Minutes,” the record says the Board discussed ensuring that confidential information would not be inadvertently published in public Board minutes. It adds that the Board secretary would edit the minutes and archive the details for future boards. This is a dated discussion and an identified intended practice. It is not, by itself, a permanent disclosure policy, a legal rule or evidence about the contents of every later public or private minute. The archived details mentioned there are not part of the seven-source public record.

Even within that boundary, the passage captures the right institutional tension. Public minutes serve present readers, while retained detail serves future boards. Confidentiality is not an obstacle accidentally encountered after publication; it is one of the design constraints the Board was explicitly considering. A closure layer must therefore allow a meaningful public status without demanding the underlying protected material.

For some items, “closed - confidential basis retained internally” may be more accurate than a public evidence link. For others, a source document or short status may be safely published. The key is not universal disclosure. It is a consistent way to tell readers whether the absence of evidence is an intentional boundary, an item awaiting review, or simply a state that the public record does not specify.

The 2019 minute also warns against turning a public archive into a substitute for institutional memory. Future boards may need detail that should not appear publicly. The public-facing improvement proposed here is therefore an index of material states, not a repository of every discussion. Internal retention and public navigation can be complementary rather than competing functions.

November 2024: a draft, an abstention and two motion fours

The 15 November 2024 document demonstrates how much meaning can sit in labels. It is visibly marked DRAFT. It lists attendees and states that there were no regrets. It records a 7-0-0 vote approving the meeting agenda. The draft label must stay attached when the document is described. It does not prove that the meeting lacked a later approved minute or that any Board act was legally defective.

The same document records approval of the 20 October minutes by six votes in favour, none opposed and one abstention. It explains the abstention as coming from a member who was not present at the October meeting. That is a proportionate, bounded reason. It should not be reframed as evidence of a conflict, a recusal or an undisclosed dispute. The source offers a simpler explanation, and there is no basis to go beyond it.

Another motion authorises withdrawal of up to $150,000 from the reserve account by the end of 2024, if needed to cover expenses, through a 7-0-0 roll-call vote. Every limiting phrase matters. “Up to” establishes a ceiling. “By the end of 2024” establishes the stated window. “If needed” makes the authority conditional. The vote proves the recorded authorisation; it does not prove that any money was withdrawn, when a transaction occurred, how much moved or whether the reserve was otherwise adequate.

The document then uses MOTION #4 for approval of proposed Code-of-Conduct changes and again for adjournment. The number collision creates a traceability risk: a later reference to “Motion #4” would be ambiguous without the subject or page. It does not show that either vote was invalid, that the numbering was intentional, or that anyone acted in bad faith. A stable identifier matters precisely because mundane duplication can make a record harder to connect later.

This single minute supplies four different lessons. Preserve document status. Accept a stated abstention reason within its limits. Read conditional authority literally. Treat identifier collisions as document control rather than scandal. Together they show why a closure index should carry the exact source link, meeting date and motion subject, not only a number.

The minute also records committee status, but a status update is not automatically a completed action. The public record works best when it lets its own language stand. If a later document uses a stronger verb, the state can advance. If it does not, a reader should not manufacture the transition.

March and April 2025: a visible return without an edit history

The 7 March public document is also visibly labelled DRAFT. It separates an informal workshop from the formal meeting: the workshop ran from 1:00 to 2:19 p.m., and the formal call to order came at 2:19 p.m. That boundary matters because the source does not disclose what, if anything, was decided in the workshop. The formal minute begins where it says it begins.

Within the formal record, the February-minutes motion is deferred after discussion and the action is to present edited minutes at the next meeting. The same minute separately records the immediate appointment of an additional Hackathon Committee member for a two-year term through a 7-0-0 vote. These items occupy different states. The minutes approval did not occur that day. The committee appointment was authorised as effective immediately. Even then, the appointment motion is evidence of the Board act, not evidence about the appointee's later service or performance.

The 18 April minute records a 7-0-0 motion approving both the 2 February and 7 March minutes as presented. This is the public return point for the earlier deferred approval. It supports a bounded conclusion: both sets of minutes were later approved. It does not reveal the March discussion, show a before-and-after February text, name an editor or establish that a specific edit was implemented in a specific way.

The April minute also assigns staff to look into post-production uploads and add them to an RFQ. It does not name an individual, provide a deadline, state a result, link to evidence or give an explicit closure status for that action. The correct public description is “assigned to staff.” A later statement elsewhere might establish execution, but the April source alone does not.

Near the end, the minute marks the start of an executive session at 2:14 p.m. and records adjournment at 2:25 p.m. Public silence about the session's content is a legitimate confidentiality boundary, not a publication defect. A good closure record could say that the item moved into a protected session or that public evidence is unavailable for a lawful reason. It should not expose the discussion merely to make a status row look complete.

Read together, March and April provide both evidence and caution. They demonstrate a cross-meeting follow-up from deferral to approval. They also show that one observable end state does not illuminate every intermediate action. That is exactly why state names should be narrow. “Approval closed” is defensible. “All requested edits executed and verified” is not.

May 2025: adoption at the decision point

The 23 May minutes approve the previous meeting's minutes and record a motion to adopt ranked-choice voting for future NANOG Board-member elections, following an Elections Committee recommendation. The reported vote is 7-0-0, and the status says the motion carries.

That record establishes the Board decision in the motion's own terms. It does not establish later implementation in voting systems, member endorsement, ballot configuration, legal validity or any future election outcome. “Adopt” is the right verb for the Board act. “Implemented successfully” would require different evidence.

This is a clean example of why decisions and performance must be separated. A Board can complete its deliberative role by adopting a course. An administrative or technical role may then configure the process. A later review may check whether the configuration reflects the decision. The public minute need not include all those stages on the day of adoption. But if later stages are material and non-confidential, a stable link would help readers avoid treating the adoption vote as proof of them.

The reported unanimity is equally bounded. Seven votes in favour, none opposed and no abstentions support the specific Board motion. They do not demonstrate consensus among NANOG members, election participants, network operators or the public. The Board's authority within the organisation does not need to be rhetorically enlarged to credit the decision.

The May document includes operational status statements in other sections. Those statements can be read as the statuses they claim, but they should not be used to assemble a broad account of executive power or to assess unrelated programmes. For this investigation, their relevance is limited: a minute may report that work is underway or complete, but the exact subject and evidence determine how far that statement can travel.

8 June 2025: similar pages, different states

The 8 June minutes use exact motion identifiers, presenters and vote fields. One roll-call motion authorises the Executive Director and Treasurer to withdraw up to $200,000 from the reserve fund between 30 June and 31 July 2025 to cover operating expenses. The authority has actors, a ceiling, a purpose and a time window. It does not prove that a withdrawal occurred.

The document's financial wording requires special care because two labels appear. On page one, the presented-document list names a 2024 DRAFT Audited Financial Statement and a 2024 DRAFT Governance Letter Audit. On page two, the minutes say the Board and Audit Committee reviewed risks identified by the auditor and state that steps had been taken to address and mitigate them. A motion then says: “The 2024 Financial Statements are accepted as presented.” The roll-call vote is reported as all in favour and the motion carried.

Neither label should erase the other. The visible document list contains DRAFT; the motion accepts the 2024 Financial Statements as presented. The motion does not, on its face, relabel the page-one item as a final published version. The statement that steps had been taken is evidence of what the Board and Audit Committee reported. Acceptance is evidence of the Board act. Neither is automatically a published remediation test or independent verification that every identified issue was closed. The substance of the financial statements lies outside this article; the language matters here only because it illustrates state boundaries.

The same meeting places three status forms close together. A fund-balance comparison to policy is followed by ACTION: None. The Audit Committee update says Work completed. The Workshop Committee update says draft documents had been received and would be reviewed for the next Board meeting. These cannot be collapsed into one generic “complete” field.

ACTION: None may mean that no new action was assigned from that agenda item; it does not necessarily describe the entire history of the underlying subject. Work completed is the minute's status statement about the committee's work; it is stronger than “discussion” but does not provide external verification by itself. “Will review for the next Board meeting” describes an anticipated future review, not completed review or adoption. The language already separates the states if readers resist smoothing it into a single narrative.

This meeting shows the value of preserving exact source verbs in any index. A standard form should not force every item into “open” or “closed” when the source says “no action,” “work completed” or “scheduled for review.” It should quote or faithfully map the public state and allow an explanatory note. Uniform navigation does not require uniform substance.

20 June 2025: named actors without presumed completion

The 20 June minutes are rich in action assignments. For a proposed hands-on vendor experience, they assign J. Black, D. Harvey and E. Culley to organise a follow-up call, explore logistics and draft a proposal for Board review. A separate action asks Jonathan to develop a structured BoF proposal, gauge community interest and identify possible topics. These are multi-step assignments with named actors and a destination. They do not prove that the call happened, a proposal was submitted or the Board accepted it.

Another action assigns the event-planning team to monitor daily badge reprints over the next year to determine next steps. This item has an accountable team and an unusually clear time horizon. The public record still does not show the monitoring result because the seven-source record ends on 20 June 2025. It would be improper to treat the absence of later material in this source set as non-execution.

The same minute names several narrower actions in a sensitive moderation context: Steve is to contact the owner of an unofficial Reddit page; Jonathan is to review relevant insurance coverage and draft a note of support; Valerie is to confirm that subscriber moderation holds are applied uniformly across lists. The minute provides traceability, but a public closure layer should avoid reconstructing incidents, repeating allegations or disclosing complaint, personnel or security details. A short status or confidentiality reason may be the proportionate public endpoint.

Committee-liaison changes are reported with new assignments and a 7-0-0 vote. The minute then assigns liaisons to communicate the changes to their respective committees. The vote records the Board act; it does not prove that every committee received the message or implemented any consequent change. A later communication status would be a different state. The minute also states Executive session: None, an explicit absence marker that is informative without implying anything about other meetings.

The density of named actions on 20 June improves public traceability. It also makes the boundary vivid. Names and time horizons answer “who owns the next step?” and “when might it return?” They do not answer “was it completed, checked and closed?” That would require a later public status, a non-confidential evidence link or an explicit closure entry.

The frozen record stops with this meeting. No claim is warranted about what happened after 20 June: not about completion, staffing, policy, recordkeeping, the archive or any individual action. Silence outside the defined source period is simply outside the evidence.

Three shortcuts that break the chain

The first shortcut is treating a vote as execution. The November reserve motion authorises a conditional maximum. The June reserve motion authorises another maximum within a defined window. The May ranked-choice motion adopts a future election method. Each vote is meaningful as a Board act. None, without further evidence, proves that money moved or a voting process was configured.

The second shortcut is treating assignment as completion. The April RFQ line assigns staff but gives no named person, deadline, result or closure field. The June 20 actions often name people or a team and sometimes a review destination or one-year horizon. That greater specificity improves follow-up, but it does not advance the item to executed. Assignment is accountability for a next step, not evidence that the step occurred.

The third shortcut is treating public silence as failure. The documents are minutes, not a complete operational register. Confidentiality, proportionality, meeting agendas and separate reporting channels can all shape what appears. A later status may exist outside the seven sources or inside NANOG. The public gap is navigational: from these pages alone, a reader cannot always find a uniform state history. It is not proof of abandonment or deficient internal practice.

The reverse shortcuts are damaging too. Calling a deferral a rejection erases the visible March-to-April return. Treating DRAFT as meaningless erases a document's explicit version label. Calling an explained abstention a conflict changes the source's account. Calling duplicate numbering invalidity turns a clerical traceability problem into an allegation. Good public analysis should make fewer claims, but make them more durable.

A small closure index for material items

The proportionate remedy is not a longer narrative minute. It is a compact decision-closure index for material, non-confidential matters, linked back to the authoritative meeting record. Each row should preserve ten fields.

Its subject is the public state of a minute entry, not a map of who holds substantive executive power. The reserve authorities, election-method adoption and named assignments appear here only because their verbs expose the distance between a Board act and a later public state. They are not assembled into execution chains, used to compare officers or turned into an authority table. That boundary keeps the investigation on cross-meeting navigation rather than institutional power allocation.

First, use a stable decision or action ID. Where a motion already has a unique identifier, reuse it. Where numbering collides, include the meeting date and subject or add an unambiguous suffix. The purpose is retrieval, not retroactive criticism.

Second, link the meeting and exact public source. A reader should be able to see the motion, action or status in context rather than rely on a paraphrase.

Third, record the exact state: discussed, deferred, assigned, authorised, adopted, accepted, executed, verified, corrected, closed or superseded. If the source uses a more specific phrase such as ACTION: None or Work completed, preserve that wording in a note instead of flattening it.

Fourth, name the accountable role or team at the level already appropriate for public use. A role often survives personnel changes better than a name. Existing minutes can retain their named presenters and assignees; the index need not add unnecessary personal data.

Fifth, include a target or review date where one exists. Do not invent one. “Next meeting” and “over the next year” are useful public horizons even when no calendar deadline is stated.

Sixth, identify the next public review date and latest status. A deferral should point to the meeting where it returned. An assignment intended for Board review should show whether a later agenda took it up, while permitting “not stated in this public record” when appropriate.

Seventh, link non-confidential evidence of execution or provide a short reason that evidence cannot be public. Evidence might be an adopted document, a published notice or a later minute. The confidentiality option is essential for protected sessions, personnel, complaints, procurement and security-sensitive matters.

Eighth, preserve a correction or version note. If an identifier is duplicated, a document remains labelled draft, or a later correction changes the record, the index should not silently overwrite history.

Correction deserves its own state because it answers a question that neither approval nor closure can answer: did the public account itself change? A dated note should identify the corrected field, retain enough of the earlier entry to explain the difference and state whether the correction altered only the record or also required a new Board act. That practice would keep an administrative repair from being mistaken for a fresh decision. It would also let a reader distinguish a corrected identifier from a superseded motion and a newly approved version from a draft that merely remains visible.

Ninth, retain vote totals and a bounded abstention or recusal reason when safe and relevant. The November 2024 explanation shows how a short reason can improve interpretation. The index should not demand personal detail where none is needed.

Tenth, require an explicit closure decision or supersession link for items that enter the index. Closure might mean the Board accepted the completion evidence, decided no further action was needed, replaced the item with another decision or moved the remaining substance behind a justified confidentiality boundary. It should not be inferred from elapsed time.

This index would be a navigation layer, not a transcript, legal register, personnel tracker or ranking device. It should not include every routine instruction. Materiality keeps the burden proportional and prevents an administrative list from swallowing the Board's work.

The strongest case for leaving minutes compressed

There is a serious argument against expanding the public record. Board minutes are governance records, not project-management systems. They need to identify decisions and preserve an institutional account without publishing every discussion, draft or implementation detail. Staff and volunteer capacity is finite. A status list can become stale and misleading if no one owns it.

Some subjects cannot carry public completion evidence. Executive-session content may be privileged or sensitive. Personnel matters, complaints, procurement negotiations and security controls can be harmed by disclosure. Publishing a vague but suggestive status may be worse than stating plainly that the evidence is retained internally and cannot be public.

The sources themselves support this defence. The 2019 minute records an effort to protect confidential information while archiving details for future boards. The 2025 documents sometimes mark the start of an executive session or state that there was none. Those boundaries are informative. The aim should be to make the public state legible without crossing them.

An index also risks bureaucratising language. If every discussion must be assigned a status, exploratory conversation may be mistaken for commitment. If every assignment needs public evidence, staff may spend more time documenting routine work than doing it. If “closed” is rewarded, difficult matters may be closed prematurely. A sound design should therefore cover only selected material items, permit uncertainty and make “confidential basis” a legitimate bounded status.

The proposal is not shown here to be legally required, and the seven sources do not establish that NANOG lacks an internal system. A current public register may already join decisions to owners, review dates, status histories, completion evidence and closure. If so, the thesis narrows to discoverability. The right response would be to link that record, not build a duplicate.

The best case for a small index is practical rather than accusatory. NANOG already produces many of the necessary fields. A consistent join would help members and future readers follow material items without demanding more substantive disclosure from the minute itself.

Closure as disciplined memory

A Board minute is strongest at a point in time. It says who attended, what motion was made, how the vote was recorded and what action was assigned. Institutional memory must operate across time. It needs to know whether an idea remained under discussion, a decision was deferred, a role received an assignment, the Board authorised a course, someone carried it out, the result was checked, a record was corrected, or the matter was closed.

The March-to-April sequence proves that NANOG's public minutes can carry an item across meetings. Its value lies in its limits as much as its visibility. The public can see deferral and later approval. It cannot see the edit history, and should not pretend to. That is an honest chain.

The other minutes show complementary strengths. The 2019 record articulates the confidentiality-memory balance. The November 2024 draft explains one abstention, states conditional authority and exposes an identifier collision. May preserves an adoption motion without proving implementation. The 8 June record places accepted statements, conditional authority, ACTION: None, Work completed and future review on nearby pages without making them equivalent. The 20 June minute improves assignment traceability through names and a monitoring horizon while stopping short of completion evidence.

These are not marks on a performance scale. They are examples of state language doing different work. The public record becomes more reliable when readers refuse to upgrade it and when the organisation supplies a stable route to later states for the items that matter.

Closure, in that sense, is not surveillance. It does not require publishing private deliberation or watching every staff task. It is a disciplined statement that an authorised or assigned item reached an evidenced public endpoint, was superseded, or moved behind an appropriate confidentiality boundary. It protects the people doing the work from assumptions just as much as it serves the people reading the minutes.

The durable test is simple. Can a future Board member, NANOG member or outside reader start from a material motion or action and determine its exact public state, responsible role, expected return point, latest non-confidential evidence and explicit closure or supersession? If a public record already answers those questions, the navigation gap is small. If it does not, a compact index can connect the minutes without asking them to become something they were never designed to be.

Sources