Summary

  • NANOG’s current Bylaws establish three-year elected-Director terms, while a live Board guidance page still says two years. The 2013 Bylaws and 2014 amendment result show why the guidance sentence is stale rather than a competing rule.
  • NANOG often provides unusually strong public version trails: the 2018 package can be followed from baseline text through proposal, redline, Board ballot authorization, member result and final consolidated Bylaws. The 2020 and 2025 records also give useful adoption closure.
  • Other differences cannot be resolved from the public text alone. Election Committee service periods, the Executive Director’s term language and the Board-member/liaison clauses should be labelled as open textual tensions, not converted into accusations about actual conduct.
  • A proportionate remedy is a compact public version ledger linking each instrument to its status, approval or effective date, predecessor and canonical URL. It would improve reproducibility without exposing ballots, legal advice, personnel records or private deliberations.

One site, two term lengths

Start with the easiest discrepancy to reproduce. The current Bylaws describe a seven-vote Board: six elected Directors and the Executive Director. They say elected-Director terms are three years, staggered so that two expire each year. The live Board Responsibilities page also describes six elected Directors, but says their staggered terms last two years. The two public statements concern the same seats and state materially different durations.

If both pages are merely called “current,” the reader reaches an impasse. If they are classified by instrument and version, the answer becomes straightforward. The Bylaws are NANOG’s current public corporate instrument. A responsibilities page explains the Board within a narrower guidance surface; it does not silently amend the Bylaws. The history then supplies the missing bridge. The adopted October 2013 Bylaws used two-year elected terms and arranged for three elected terms to expire each year. The official 2014 election and amendment record records adoption of three-year terms, two ordinary expirations each year and balancing term lengths during the transition. The present Bylaws preserve that three-year, two-expiration structure.

This is a resolved version conflict with an unresolved publishing history. The controlling term is three years. The two-year phrase on the guidance page is stale subordinate copy, not a second operative Bylaw. But the public page does not expose a verified date for its last substantive edit, identify the author of that sentence or give a supersession notice. Those details remain unknown. It would be speculation to say when the phrase was inserted, when someone should have removed it or whether anybody relied on it.

The distinction matters beyond this example. A page can still be reachable on the website and stale in one substantive line. Conversely, a historical PDF can be superseded now yet remain controlling evidence of what the rule said in 2013. “Current web page,” “current controlling instrument” and “historical controlling text” are different labels. A crawl date, retrieval date, document-upload path or transport response does not turn into an adoption or effective date merely because it is the only date visible to a researcher.

NANOG’s current Bylaws themselves illustrate the point. The page identifies the latest package as proposed on 23 September 2025 and adopted on 5 November 2025. Those dates close the status of that amendment package. They do not imply that every unchanged clause originated in September or November 2025. Version control begins by dating the change that a record actually dates.

The document, the decision and the resulting text

The unit of this audit is deliberately small: a public governance statement, the instrument carrying it, its version or adoption state, and its relationship to earlier or later text. It is not an audit of a private dispute. Across 18 official public records, the useful categories are current controlling instrument, current subordinate page, historical controlling text, proposal, adoption record, stale current copy, unresolved current tension and unknown. Those categories prevent a common error: treating every official-looking page as if it had the same authority and temporal status.

They also prevent verbs from doing more work than the evidence. To propose wording is not to adopt it. A Board decision to put a question on a ballot is not the members’ result. A certified aggregate result is not the final consolidated instrument. Consolidated text does not establish how every clause was implemented. Operational use, where documented, would be another event again. Evidence of one stage should not be renamed as evidence of the next.

The authority question is equally bounded. For descriptions of NANOG’s current corporate rules, the current Bylaws are the controlling public instrument. A membership policy, committee charter or explanatory page remains meaningful within the surface it expressly addresses, subject to the Bylaws; it does not change superior text by implication. This does not decide every legal question that could arise under applicable law, and it does not convert a documentation audit into a compliance judgment. Public text alone cannot prove actual compliance, noncompliance, motive, legal validity or a causal governance outcome.

Those claims would need a separate, case-specific record.

That restraint is especially important because the public corpus is, in several places, good. NANOG has preserved enough material to show not just that language changed but how a change travelled. The 2018 amendments provide the clearest example.

A control test for ordinary readers

The practical reading sequence begins with the surface. Is the statement about a Director’s term, a membership privilege, committee composition, election mechanics or administration of a community platform? Similar institutional vocabulary can conceal different authority. A membership policy may answer what dues are owed without being the place to establish the Board’s current structure. A committee charter may describe internal operation without displacing a conflicting Bylaw. Scope has to be identified before hierarchy can do useful work.

Next comes the instrument. “Official” is too broad a classification. The reader needs to know whether the page is a Bylaw, a Board-adopted policy, a committee charter, explanatory guidance, a proposed amendment, an election result or final consolidated text. That label changes what the document can prove. An explainer is often the best guide to the purpose of an amendment, but it cannot establish that members approved it. A result can establish aggregate approval while leaving the exact resulting language to the consolidated Bylaws.

The visible date then needs a name. It might be the date of a Board resolution, the date a proposal was announced, the voting window, the date of adoption or simply a file’s place in a website. Those are not interchangeable. A “last viewed” marker says nothing about substantive effect. A date attached to the latest amendment package does not redraft the history of every clause that survived untouched. The honest question is not merely “what date appears?” but “what event does this date document?”

Only then should the reader test the proposition and adopter. What exact words were meant to change? Who authorized submission of the question? Who was entitled to approve it? What does the recorded result cover? Is there final text reflecting the change? In the best public chain, each answer has its own record. Where a link is missing, the status should stop at the last supported event. A Board’s ballot authorization should not be promoted into member adoption, just as an amendment result should not be treated as evidence that a subordinate guidance page was subsequently edited.

Supersession is the final join. Sometimes it is explicit in a consolidated instrument. Sometimes the conclusion is inferential but strong: an adopted amendment replaces the directly corresponding clause in an identified earlier Bylaw. That inference should be labelled. It is weaker when a dated corporate instrument and an undated explanatory page simply disagree. In that case, hierarchy can identify the controlling corporate rule while the subordinate page’s own revision history remains unknown.

This protocol leaves a separate column for application. Did a named decision-maker use the clause in a particular election, appointment or moderation action? The sources discussed here generally do not answer that question, and the article does not borrow documentary provenance to simulate a case record. It does not assess the legitimacy of a particular delegation, retry disciplinary procedure, judge the merits of bundling amendments, reconstruct a vacancy appointment, audit a staff supervision chain or weigh evidence in an operational mailing-list dispute. Those are different evidentiary units.

Keeping them separate is what makes the narrower version-control conclusions defensible.

The method also accommodates correction without drama. If NANOG adds a version note to a charter or updates a guidance sentence, the prior observation does not become dishonest; it becomes a historical state with a documented successor. Version control is designed for change. Its purpose is not to freeze every page or attach blame to maintenance lag, but to make the transition from one public state to the next inspectable.

2018: a chain a reader can replay

The October 2017 Bylaws supply the immediate pre-change baseline. The 2018 amendment explainer then inventories proposed changes to 34 sections. NANOG classified 19 as involving legally significant text or definitions and 15 as grammatical corrections. That is the association’s classification, not an independent legal judgment about the weight of each edit.

The explainer matters because it makes the proposition legible. Among other changes, it identifies revisions to section 6.4, Article 7, section 7.2 and section 8.2: member rights, committee composition, the Election Committee reference and administration by staff. The 2017 baseline did not contain the membership additions later placed in sections 6.6 through 6.8. The proposal thus lets a reader distinguish newly introduced provisions from language carried forward.

The official 2018 redline adds before-and-after text. It is valuable but not self-sufficient. Strikeout and insertion styling can blur when text is extracted from a PDF, so a precise claim should be checked across the explainer, the visual redline and the eventual final instrument. A redline, however clear, remains proposed language until an authorized adoption record closes it.

The next control point is the Board’s September 2018 electronic-vote record. It records approval to place the Bylaw package before members as one yes-or-no amendment question. Candidate choices appeared separately. This was authorization of the ballot proposition, not the member vote and not final consolidated text.

The migrated official 2018 election result reports 692 eligible voters, 192 ballots and 93 percent approval for the amendment package. The final October 2018 Bylaws are marked adopted on 3 October 2018 and contain the resulting structure. Now the chain is complete enough for a public reader to replay: 2017 baseline, explainer, redline, Board authorization, member result and consolidated adopted text.

Each link answers a different question. The baseline shows what was being replaced. The explainer and redline show what was proposed. The electronic vote identifies the Board action that sent the bundled question to members. The result records aggregate approval. The final Bylaws establish the resulting text. Taken together, they support supersession of the earlier Bylaw baseline.

They do not support several tempting inferences. Thirty-four edited sections were not 34 separate votes. Ninety-three percent approval of one package does not reveal which clauses each voter liked, whether every voter read every provision, or how many would have supported an individual amendment if it had been severed. The chain does not demonstrate actual implementation after adoption. Nor does a complete Bylaw trail prove that every policy, charter and explanatory webpage was synchronized. Version provenance is strong evidence about official text; it is not a substitute for voter interviews, application records or page-level revision logs.

This distinction turns a potentially accusatory audit into a useful one. The question is not whether the package was somehow less valid because it was bundled. It is what the surviving records can establish. Here, they establish a highly reproducible text-change path while leaving clause-level preference and downstream page maintenance outside the proof.

A 2011 policy carrying an older map

The live Membership Policy presents another kind of version problem. It says the Board adopted it by resolution on 4 January 2011. The page remains useful for rules about dues, membership terms and renewal. Yet section 4, while describing membership privileges, refers to service on administrative committees under “section 9.” Current Bylaws place committees in Article 7 and expressly include, among the privileges of a member in good standing, service as a Director if elected.

The old reference is historically intelligible. An official August 2011 announcement likewise discussed a proposed change to committee structure under Section 9 and said that some committee functions were being handled by staff. The language was not invented by a recent broken link; it belonged to an earlier institutional map.

The 2018 record helps establish the later structure without pretending to rewrite the policy page. Its explainer identifies changes to section 6.4, Article 7, section 7.2 and section 8.2, and its final adopted text contains the resulting organization. That is sufficient to classify the “section 9” wording as a legacy cross-reference when one is asking where the current Bylaws put committees. It is not sufficient to declare every provision of the 2011 policy obsolete. The policy still speaks within its dues-and-membership surface, while one route marker points backward.

This is “partly resolved” rather than “invalid.” The Bylaw lineage resolves the current structural location. The policy’s public maintenance history does not. No page-level edit log or explicit partial-supersession note appears in the public records considered here, so the date and mechanics of any policy-page update remain unknown. The clean editorial description is narrow: retained policy text, legacy committee cross-reference, later controlling Bylaw structure.

When the chain records an authority handoff

The same 2018 materials show why historical versions are more than archival decoration. In the 2017 Bylaws, communications administration was assigned through older Communications Committee wording. The 2018 explainer says section 8.2 would replace “Communications Committee” with “NANOG staff (as overseen by the Executive Director).” The member result and the final 2018 Bylaws close adoption of that change, and the current Bylaws retain the staff-and-Executive-Director structure.

That is a resolved textual handoff. A reader can identify the prior actor, the proposed replacement, the member adoption record and the final language. The documents show that published authority moved, in the Bylaws, from a named committee formulation to staff under Executive Director oversight.

They do not name every employee who performed the function, identify a system-migration date, reconstruct individual moderation decisions or disclose private supervision. A public authority-text handoff and its operational execution are related but distinct. The record is strong enough for the first claim and not designed to prove the second. That distinction also keeps this inquiry away from a retroactive trial of any mailing-list action. It is about document control, not the merits of a moderation case.

Three tensions the current text does not settle

Version hierarchy can resolve the Director-term mismatch because the adopted amendment and current Bylaws point in the same direction. It cannot honestly resolve every difference in NANOG’s public pages. Three surfaces should remain marked as current textual tensions.

The first lies within the current Bylaws’ treatment of the Executive Director. The Board has seven votes: six elected Directors and the Executive Director. The general term provision uses the unqualified phrase “all Directors” when assigning a three-year term. Elsewhere, the officer and executive provisions say the elected Directors select the Executive Director, who serves at their pleasure and may be removed with or without cause.

Perhaps those clauses have a settled institutional reading. The public text examined here does not expressly state it. A careful account therefore preserves the tension between an “all Directors” term sentence and service at the elected Directors’ pleasure. It must not turn the three-year language into a fixed three-year Executive Director employment contract. Appointment, corporate office, Board membership and employment may involve distinctions that this documentary record does not answer. The intended reconciliation and any separate employment or appointment terms are unknown.

The second concerns Election Committee duration. The 2018 explainer says section 10.3 would be defined as the Election Committee and its number of members adjusted. Current Bylaws section 10.3 requires a committee of at least three people, appointed by a majority of the elected Directors whose terms are not expiring. It says the committee serves until the election concludes.

The public Election Committee charter uses a different clock: two-year terms, capped at two consecutive terms. Both are current official public texts, but the available record supplies no verified charter adoption date, effective date, supersession relationship or express reconciliation. It would be improper to invent a hybrid under which one duration silently nests inside the other. It would be more improper still to infer that a particular member overstayed or that an election was badly administered. The public finding is the different service-period formulations; actual service and compliance would require appointment and application records.

The third tension appears in Article 7. The 2018 explainer said the Article would add a requirement that one member of each committee be a Board member. Current Article 7 retains the general statement for each standing or ad-hoc committee. Current Article 7.1 then says no elected Board member may serve concurrently on a standing committee and allows the Board to appoint a liaison who observes and facilitates communication.

The wording may be intended to distinguish the Executive Director or another Board-member capacity from an elected Director, or to distinguish committee membership from liaison service. The consolidated public text does not expressly perform that reconciliation. The result should be labelled an internal rule-text tension. It is not evidence that a committee was unlawfully composed, that a liaison possessed a vote, or that the Board controlled a committee’s outcomes.

These three cases demonstrate why “the Bylaws control” is necessary but not always sufficient. One tension sits among clauses of the same current instrument. Another sits between that instrument and a current charter whose temporal pedigree is not publicly closed. Good version control does not erase such uncertainty. It records it in a field marked unknown.

2020: closing a transition rule without proving its use

NANOG’s 2020 special Bylaw election page describes a package dealing with the start and end dates of Director terms, the immediate start of a vacancy appointee and the timing of officer selection. The certified 2020 result records 583 eligible voters and 125 ballots: 120 yes, five no and zero abstentions.

For version-control purposes, the proposal narrative and certified result form a useful bridge to the January 1 elected-Director transitions reflected in current text. They identify a proposition and an aggregate adoption outcome, rather than leaving the timing clause to appear without provenance.

The label “special election” needs care. This event was a special Bylaw referendum, not a special election of a Director. Its vote record cannot establish that a later vacancy process was applied correctly, that an appointee began on the proper day or that officers were selected at the required time. Those would be application questions. The 2020 documents close the public provenance of a rule change; they do not certify every later use of it.

That limit is not a defect in the result. It is simply the difference between a rule’s adoption record and an operational case file. A well-kept version ledger should make the former easy to find without pretending it contains the latter.

2025: a certified result and a smaller date discrepancy

The 2025 proposal groups changes around election tabulation and ranked-choice alignment, the candidate forum and community platforms. The current Bylaws page dates the package’s proposal to 23 September 2025 and its adoption to 5 November 2025. Those two records connect proposal to consolidated current text.

The official 2025 election summary reports 725 eligible voters, 184 ballots, 169 amendment votes, 15 abstentions and 97.6 percent approval. The certified report supplies the exact amendment split: 165 yes, four no and 15 abstentions. It also states the voting window from 28 October through 5 November in America/Detroit time.

There is a modest documentary wrinkle. The summary’s header and key-date presentation do not align on the opening day. For the exact opening time and the certified counts, the certified report controls this account. That difference should remain visible because it is precisely the kind of field a version ledger could normalize. It is not evidence that the ballot was invalid or that the result was manipulated.

As in 2018, this was one amendment question covering a package. The result does not expose clause-level preference, explain any voter’s motive or prove that every connected guidance page was updated afterward. It closes adoption at the package level. Calling it more than that would mistake a certified total for a survey of individual comprehension.

Taken with 2014, 2018 and 2020, the 2025 material shows that NANOG already publishes much of the hard part: proposed text, authoritative decisions, aggregate results and, in important cases, final consolidated language. The weaknesses are not a total absence of governance records. They are broken joins between records and unsynchronized statements at the edges.

What nine surfaces actually show

Viewed together, nine document surfaces produce three different outcomes rather than one institutional verdict.

The elected-Director term is resolved: two years in 2013, changed through the 2014 record, three years now, with one stale guidance sentence and an unknown page-edit history. Communications administration is also resolved at the level of authority text: the 2017 committee wording gives way through the 2018 chain to staff overseen by the Executive Director, while operational migration details remain outside the record. The 2018 membership additions have resolved provenance because the baseline, proposition, ballot authorization, result and final text can be joined.

The Membership Policy is only partly resolved. Its 2011 date and Section 9 reference make historical sense; the current Bylaws and 2018 chain establish today’s Article 7 structure; no visible page-level note explains the partial supersession. The 2020 Director-transition language and the 2025 election-and-platform language also have resolved provenance, with later application and clause-level preference respectively left unclaimed.

Three surfaces remain unresolved at the text level. “All Directors” having three-year terms sits beside the Executive Director serving at the elected Directors’ pleasure. Election Committee service until an election concludes sits beside a charter’s two-year, two-consecutive-term formulation. A general requirement for a Board member on each committee sits beside the standing-committee exclusion for elected Directors and the observer-liaison mechanism. These are not blank spaces to be filled with intuition. They are fields that should say “reconciliation not identified.”

The sixth documentary tension is narrower: the 2025 summary’s opening-day presentations differ, and the certified report is the chosen control for the exact window. Recording that choice is more useful than smoothing the discrepancy away.

This classification reaches more than 90 percent field completion across the nine surfaces because most instrument, claim, status and adoption relationships can be stated. Completion does not mean certainty about everything. Its value comes from preserving unknowns: the author and edit date of the stale Board sentence; the Election Committee charter’s public adoption and effective date; the intended reconciliation of the charter and Bylaws; the Executive Director clauses’ intended relationship; and the role contemplated by Article 7’s two formulations.

The records do not show the private side

Document control draws a line around what the public evidence can answer. The exact author and edit date of the two-year Board guidance are unknown. So are the public adoption or effective date of the Election Committee charter’s term clause and its intended relationship to section 10.3. The same corpus does not disclose internal legal advice, private drafting history, private page-revision logs, personnel arrangements, confidential deliberations or every implementation question.

Absence from this public set is not proof that NANOG has no internal record. Nor is it proof of what staff, a Board member, a candidate or a committee entity actually relied upon. A reader can observe textual drift without assigning a motive. To say that a public join is missing is different from saying that the institution never made the join privately.

Public uncertainty can be specific without becoming suspicious. “Charter adoption date not identified” is a complete finding when the date is absent from the bounded public chain. “Intended relationship between two clauses not expressed” accurately describes a drafting surface. Neither phrase implies concealment. Both tell the institution exactly what small item—an adoption note, a cross-reference, a supersession label or a clarifying sentence—would let the next reader finish the join.

The same discipline applies to historical inference. The persistence of older wording may suggest that a page was not synchronized after an amendment, but it does not identify the responsible person or the moment at which maintenance failed. Even actual reliance cannot be presumed from discoverability: a page can be public without having guided any real decision. A version audit records the state of the documentary interface. It does not manufacture an implementation narrative from that interface.

The distinction protects both scrutiny and fairness. It allows members to ask a precise question—what is the current service-period rule, and what adopted record explains it?—without alleging that somebody violated a term. It allows an editor to flag an obsolete cross-reference without claiming that membership rights were denied. And it allows the association to correct a page without the correction being miscast as an admission that a past outcome was bad.

The strongest case for leaving the pages alone

There is a serious counterargument. NANOG is a comparatively small professional association, not a legislature and not a commercial legal database. Its central work includes meetings and operator exchange, not maintaining a courtroom-grade statutory service. It already makes the current Bylaws available and preserves an unusually extensive set of historical PDFs, proposal explanations, election pages and certified results. Many associations offer far less.

On that view, a stale phrase on an explanatory page may be harmless. A candidate who needs the controlling term can read the Bylaws; a member who encounters ambiguity can ask staff. Synchronizing every policy, charter and descriptive page after every amendment takes volunteer and staff time, potentially diverting attention from work that delivers more direct value. No evidence here shows that the two-year sentence caused a bad election or any other bad outcome. Documentary tidiness should not be inflated into institutional failure.

That objection is persuasive against an expansive remedy. NANOG does not need to publish private legal advice, named ballots, personnel files, confidential drafting exchanges or every operational decision. It does not need a legislature-sized revision platform. And the existence of one stale line is not evidence of illegality, manipulation, capture or bad faith.

The objection is less persuasive against a compact public control record. Most of the necessary information already exists in NANOG’s own pages. The work is primarily to connect it: name the canonical instrument, label it current or historical, state the documented approval or effective date, identify the superseded instrument where known, and link the authoritative URL. For an unresolved current tension, the ledger can say so without publishing any private explanation.

The benefit is reader reproducibility. A member should be able to follow the same links and reach the same bounded conclusion without needing a private answer from staff. That is not punishment for stale copy. It is a modest way to reduce the amount of institutional authority carried by undocumented interpretation.

A ledger small enough to maintain

A useful ledger could fit on one public page. Each row would name the governance surface, the instrument type, its status, the approval or effective date that the record actually supports, the predecessor or supersession relationship, and a canonical link. A note field could distinguish “resolved, subordinate page stale” from “current texts differ; reconciliation not published.” Retrieval dates and upload identifiers would stay out of the effective-date column.

The nine surfaces in this audit offer a practical first set. The elected-Director row would link the 2013 baseline, 2014 result and current Bylaws, while flagging the responsibilities page for synchronization. The Executive Director row would point to the current clauses and mark their intended reconciliation as unresolved. The Membership Policy row would retain its 4 January 2011 adoption date, note the Section 9 legacy reference and point readers to the current Article 7 structure.

The 2018 membership row would join baseline, proposal, authorization, result and final text. The Election Committee row would display both current service-period formulations and leave adoption, effective date and supersession blank only in the sense of “unknown,” never as a guess. The Article 7 row would quote neither private advice nor an invented solution; it would simply identify the general Board-member sentence and the elected-Director exclusion and liaison mechanism.

The communications row would show the 2017 Communications Committee wording and the adopted 2018 staff-under-Executive-Director text. The 2020 row would classify the event correctly as a Bylaw referendum and connect it to Director-transition timing. The 2025 row would link proposal, summary, certification and current consolidated text, with the certified report controlling the exact voting window.

Such a ledger would also impose a helpful discipline on future publishing. “Proposed” would remain proposed until the adopter and result fields were complete. “Adopted” would point to a decision record. “Current” would mean current for the specified instrument, not simply reachable on the website. “Superseded” would identify what replaced the item. “Unknown” would be an honest public value rather than an invitation to infer.

There is no need to make the ledger comprehensive of every institutional act on day one. Its virtue is thinness. It should track material governance instruments and their changes, not become a repository for private deliberations or case files. When an amendment changes a committee name, a Director term or an administrative actor, the ledger can also prompt a short list of subordinate pages that may need synchronized copy.

Authority ends at the institutional edge

Clearer versioning does not enlarge NANOG’s jurisdiction. The rules discussed here concern NANOG’s corporation, membership, Directors, officers, committees, elections, meetings and community platforms. Within those voluntary and organizational relationships, Bylaws and policies can create real private rights and duties. A clean chain helps entities understand them.

Nothing in these records gives NANOG rulemaking authority over a person’s employment, an autonomous system number, routing, Internet number resources or outsiders who have not entered the relevant relationship. A membership provision cannot silently become a routing rule. A committee charter cannot supply authority for an unrelated resource decision. A platform-administration clause governs the surface it names, not the Internet beyond it.

That boundary is another reason to keep version rows separate. Membership, committee, election, staff and platform rules have different actors and consequences. Joining their documentary histories should make those layers easier to distinguish, not collapse them into a general claim of institutional power.

The rule is the chain you can identify

The initial mismatch has a firm answer. NANOG’s elected Directors serve three-year terms under the current Bylaws. The two-year sentence on the live responsibilities page reflects older substance that the 2013–2014 record makes visible. Correcting or annotating it would improve the public surface; its existence does not show that NANOG applied the wrong term.

The wider audit produces a more nuanced conclusion. NANOG’s 2018 records model strong public change control, and the 2020 and 2025 packages provide useful adoption closure. The Membership Policy and Board guidance show how subordinate pages can retain historical language. The Executive Director, Election Committee and Article 7 formulations show that some current textual tensions cannot be settled merely by locating a later PDF.

The right response is neither complacency nor prosecution by hyperlink. It is precise state. Name the instrument. Date only what the record dates. Separate proposition, authorization, member result and final text. Mark supersession where the chain proves it. Mark an unknown where it does not. Keep actual application and external authority outside the claim unless separate evidence brings them in.

For a professional community built around making complex systems legible to one another, that is an achievable standard: not perfect public memory, but a public route from a governance sentence to the version that gives it force.

Sources

The current and historical instruments, explanatory records and certified results used in this account are: