Summary

  • The twelve consecutive community-wide meetings from NANOG 84 through NANOG 95 took place at ten United States hosts and two Canadian hosts. Atlanta appeared twice; Texas and California each supplied two different host cities. This describes one sequence, not NANOG’s entire history or a country quota.
  • NANOG publicly says its community gathers in major cities across North America, and a 2007 historical presentation used the heading “NANOG Spreads Travel Burden.” The old presentation also says its slides were not vetted, so the heading cannot establish a current venue policy, decision rule or measured outcome.
  • Annual reports and event pages publish useful meeting snapshots, virtual and in-person modes, first-time counts, a visa-letter procedure and specific hotel conditions. The reviewed source set does not publish stable attendee-origin distributions, candidate regions, decision weights or forecast-versus-observed access.
  • The proportionate remedy is a privacy-preserving venue ledger: broad candidate geography, explicit criteria and constraints, the contracted host and responsible decision function, aggregate origin bands by mode, a post-meeting comparison, small-cell suppression and limited retention.

A moving map, and the question it cannot answer

Trace twelve consecutive NANOG meetings in order and the first fact is movement. The sequence begins with Austin, crosses into Montréal, returns to the United States through Hollywood, Atlanta, Seattle, San Diego, Charlotte and Kansas City, crosses into Canada again at Toronto, then continues through Atlanta, Denver and Arlington.

The record can be stated compactly:

Meeting Host city Country
NANOG 84 Austin, Texas United States
NANOG 85 Montréal, Quebec Canada
NANOG 86 Hollywood, California United States
NANOG 87 Atlanta, Georgia United States
NANOG 88 Seattle, Washington United States
NANOG 89 San Diego, California United States
NANOG 90 Charlotte, North Carolina United States
NANOG 91 Kansas City, Missouri United States
NANOG 92 Toronto, Ontario Canada
NANOG 93 Atlanta, Georgia United States
NANOG 94 Denver, Colorado United States
NANOG 95 Arlington, Texas United States

Ten of these meetings were hosted in the United States and two in Canada. Atlanta appears twice. Texas appears through Austin and Arlington; California through Hollywood and San Diego. The sequence reaches several parts of the two countries rather than remaining in one metropolitan area.

All of that is verifiable from the reviewed sources. None of it establishes where entities travelled from. A meeting in Seattle may shorten some journeys and lengthen others. Montréal or Toronto may change the border procedure for different people. Atlanta’s second appearance may reflect availability, operational fit, commercial terms or another consideration not disclosed in the reviewed record. The map records the selected hosts, not the population of possible hosts or the paths taken by attendees.

That distinction is the centre of this investigation. Geographic rotation moves a meeting. It may distribute burden because the point on the map changes, but a changing point does not show how the burden was distributed among people. Answering that question would require at least a stable origin population, a definition of participation mode, a way to group journey distance or border conditions, and a record of the alternatives considered.

The arithmetic needs boundaries too. Ten United States hosts and two Canadian hosts describe NANOG 84 through NANOG 95. They do not describe NANOG’s full history. They do not prove that either country held a formal allocation, that two Canadian meetings met or failed a representational duty, or that no entity arrived from elsewhere. A host-country count is not an attendee-origin count.

Nor can the map establish fairness. Equal numbers of meetings by country would not necessarily produce equal travel burden; unequal numbers do not prove unfairness. Population, transport links, venue capacity, event dates, border procedures, prices, network feasibility and the actual distribution of entities could all matter. The reviewed public record provides no common measure that combines those considerations, and inventing one retrospectively would create an appearance of precision rather than evidence.

The useful question is not, “Does this map look geographically fair?” It is, “What public record would allow NANOG and its community to understand how a host decision changed access for the people likely to participate?” That record would begin before contracting, with candidate geography and criteria, and end after the meeting, with privacy-protected origin and mode information.

What NANOG says about geography

Two event pages provide NANOG’s public description of the meeting’s geographic approach. The pages for NANOG 92 and NANOG 95 say that people working in network engineering, operations and architecture gather in major cities across North America.

The statement establishes how NANOG characterises the event. It identifies the city as an organising feature and places the meeting within a broad regional frame. It does not establish where attendees live, how their journeys are distributed or whether “major cities” is a formal selection criterion.

The difference between an institutional description and a measured result is easy to miss. An organisation that holds meetings in several cities can accurately describe movement across cities. Showing that the movement broadens access requires a denominator: access for whom, compared with which alternative, through which mode and over what period?

“Across North America” cannot be converted into a claim about representation either. A host city is a venue, not a delegate. An attendee count does not establish that every geography is present, and presence does not authorise anyone to speak for a place. This investigation concerns travel and venue legibility, not a constitutional meaning for attendance.

An older record shows that travel burden has long appeared in discussions of NANOG geography. A 2007 NANOG History v2.0 presentation contains the heading “NANOG Spreads Travel Burden” above a numbered map of earlier host locations. The same presentation says its slides were not vetted.

The heading is meaningful historical evidence. It shows that the presentation’s author framed location changes in terms of distributing travel burden. The numbered map records earlier host locations. Yet the caveat on the deck matters: the presentation is not proof of a formally approved policy in 2007, much less a current rule. It gives neither decision criteria nor origin distributions nor an outcome measure.

The historical phrase is therefore best treated as a question carried forward, not as an answer. If changing locations is meant to spread burden, how is the distribution observed? Does NANOG compare forecast and actual origin bands? Does it distinguish a short trip for one cluster from a cross-border or long-distance trip for another? Does hybrid participation alter the calculation? The ten sources reviewed here do not publish those answers.

They also do not establish that the twelve-meeting sequence was designed to implement the 2007 framing. Chronology and thematic similarity are not proof of policy continuity. The current sequence demonstrates movement; the old slide supplies historical language for why movement might matter. Only a current decision record could connect the two.

The difference between host geography and entity geography

Host geography is straightforward to publish. A city and venue are necessary facts of an event. Entity geography is harder. It may be represented by home address, employer location, billing information, registration profile or another source, each with different limitations and privacy implications.

An employer’s headquarters may not describe where an attendee began the trip. A home address may be more accurate for distance but more sensitive. A country may be too broad to describe burden across a large territory, while a city may expose an individual in a small market. An origin field can also become stale if profiles persist across meetings.

That is why the absence claim here is deliberately narrow. The public sources reviewed for the twelve meetings contain no stable origin-to-host distribution. They do not show that NANOG lacks such information operationally. They show that the annual reports and event pages in this source set do not provide a consistent table connecting broad entity origins with the host.

Without the origin side, meeting movement cannot be evaluated as burden distribution. A host in the western United States may be closer to some attendees and farther from others, but the map alone does not reveal how many fall into each group. A Canadian host may alter border procedure for some and reduce domestic travel for others, but the country label gives no relevant counts.

Average distance would improve on no measure, yet an average can hide the tail. A meeting might reduce the typical journey while leaving a small group with exceptionally long or procedurally complex travel. Conversely, a larger average could coexist with a more balanced distribution if fewer people face extreme journeys. A useful public account should therefore show bands or quantiles rather than one headline number.

The origin design should preserve uncertainty. A entity may travel from a temporary work location. A multi-leg itinerary may not align with straight-line distance. A remote attendee has no host-city journey but still depends on time, connectivity and the available virtual programme. The ledger need not reconstruct personal itineraries to be useful. It can define a broad origin band and state what that band can and cannot represent.

The purpose is not to rank cities from best to worst. No single host can minimise distance, price, border procedure, accessibility, supplier risk and network requirements for everyone. The purpose is to make the distribution and trade-offs visible enough for later decisions to be evaluated against stated aims.

Meeting snapshots are valuable—but they are not origin tables

NANOG’s annual reports for 2022, 2023 and 2024 publish multiple meeting-level measures. Depending on the report and meeting, readers can see in-person and virtual participation, member and non-member registrations, students, speakers and first-time attendees. These snapshots provide a real view of event scale and composition.

They do not form a stable origin-to-host series. A count of people in the room does not say where they travelled from. A virtual count does not say who chose that mode because of location. A first-time count does not establish that the city attracted newcomers, because price, topic, timing, employer support, definitions and other conditions are not controlled.

NANOG 93 illustrates both the value and the boundary. The meeting-statistics page lists 225 members, 468 non-members and nine students. It reports 702 total in person. On the virtual side, it lists 22 virtual and 45 virtual complimentary registrations, for 67 total virtual. It also reports 128 first-NANOG attendees.

These figures make the Atlanta meeting legible as a snapshot. The three membership-status counts sum to the in-person total, which helps explain the table’s presentation for that event. The virtual categories show that virtual participation itself contained more than one registration path. The first-time figure establishes a visible newcomer group.

None of those fields identifies home city, state or province, country of origin, distance to Atlanta or border crossing. None says whether a person selected virtual attendance because Atlanta was far away, because of schedule, cost, health, employer policy or another reason. The snapshot cannot isolate location as a cause.

The same discipline applies across the earlier annual reports. Variations in total or first-time participation from Austin to Montréal, Hollywood, Atlanta, Seattle, San Diego, Charlotte, Kansas City or Toronto are observations associated with different meetings. They are not a venue experiment. The meetings differ in multiple ways, and the public source set provides neither a common origin population nor stable definitions, a price series, an alternative host or other relevant controls.

It would therefore be misleading to line up first-time counts and declare that one city performs better. A larger first-time cohort could reflect outreach, subject matter, employer travel, meeting mode, local participation, reporting definitions or other factors. A smaller cohort does not establish that a city was inaccessible. Location may matter, but these figures cannot identify its effect.

This is not a reason to dismiss meeting statistics. They remain necessary operating measures and could become part of a venue ledger. Total in-person and virtual participation supply mode denominators. First-time counts may help describe entry, provided causal claims are avoided. The limitation is one of research design, not value.

A border procedure, observed on paper

Host-country choice can introduce a procedural path even when the public record does not show its consequences. The NANOG 89 registration page for San Diego describes a visa invitation-letter procedure. It states that a letter for a visa application would be issued after registration and payment.

The request called for the person’s passport name and number, email address, hotel name and reservation number, and company name and address. This establishes what the published process requested. It also shows that the letter followed registration and payment and required hotel and identity details.

The page does not report how many people requested letters, how many received them, how many obtained a visa, how many were refused or how many changed their attendance decision. Nor does it establish that the procedure was unique to NANOG 89 or required for every cross-border traveller.

The existence of a procedure should not be turned into an outcome. An invitation letter can support an application without determining the decision of another authority. A person may begin the process and not attend for unrelated reasons. Another entity may not need a letter. The evidence supports only the terms of the published path.

Still, that path belongs in venue analysis because it is a foreseeable access condition for some prospective entities. A candidate-host review could ask whether a border procedure is likely to affect part of the forecast origin distribution and what support can be offered. A post-meeting record could publish only aggregate request counts above a safe threshold, without passport information, names or individual outcomes.

The purpose would not be to infer exclusion. It would be to distinguish a selected host from the procedural environment attached to that host. If the number is too small to publish safely, the ledger should say that the field is suppressed or unavailable rather than presenting zero.

A sold-out room block is not a total-cost finding

The venue page for NANOG 94 provides a different kind of access evidence. It says the Sheraton Denver Downtown room block was sold out. It lists a standard room rate of USD 275 plus 15.75% tax. It also lists overflow room rates of USD 469 and USD 420 plus applicable tax.

These are concrete planning conditions. They show that the published primary block had reached capacity and that the listed alternatives carried higher nominal room rates. They may matter to people deciding where to stay and to organisers reviewing room-block performance.

They do not show what every attendee paid. Some people may have booked earlier, stayed elsewhere, shared rooms, used points, commuted or participated virtually. The rates exclude airfare, ground transport, wages, childcare, meals and all unlisted accommodations. They cannot be added into a universal trip price.

The conditions do not establish a location effect on attendance either. A person who did not register cannot be linked to the room-block status in the reviewed record. A person who attended virtually may have chosen that mode for reasons unrelated to the hotel. A sold-out block can signal demand for the block without describing the total lodging market.

Venue terms nevertheless belong in the decision record at the right level. Before a contract, NANOG could disclose that room availability, forecast rates and overflow resilience were among the criteria, without publishing confidential bids. After the meeting, it could record whether the primary block filled, whether overflow was activated and the published range of contracted or recommended rates.

That would let future reviewers assess planning assumptions without repeating a separate total-cost investigation. The ledger would preserve the distinction between an observable venue condition and an attendee’s complete burden.

Virtual participation changes the access surface

The annual reports and event pages distinguish in-person and virtual participation. That distinction is essential to geographic analysis. Virtual access may reduce or remove the need to travel to the host city. It can offer a path when distance, border procedure, timing, cost or another condition makes travel unattractive or impossible.

The reviewed sources do not establish why any particular person chose virtual access. They also do not establish that virtual participation reproduces every in-person interaction. Hallway conversations, social encounters, peering discussions, sponsor contact, help-desk access and some committee interactions may depend on physical presence or operate differently online.

Virtual registration should therefore be neither dismissed nor treated as a perfect substitute. It is a distinct mode with its own access surface. A venue ledger should show in-person and virtual outcomes separately, then state which programme elements were available in each mode.

If NANOG later asks whether a host changed access, the virtual denominator matters. A meeting may maintain overall reach through virtual participation while the in-person origin distribution changes. That would be useful descriptive evidence, but it would not show that every function remained equally available.

The ledger could include a voluntary, broad reason-for-mode question if NANOG determines that it is necessary and safe. The response options would need to avoid sensitive detail, and non-response would remain visible. Even then, self-reported reasons would describe entity accounts rather than prove causation.

Geographic rotation and hybrid access can be complementary. Moving the host changes the physical setting and may change the burden distribution; virtual access offers another path. The public record should describe both without collapsing them into one attendance total or assuming equal experience.

How a geographic method could work

An independent study of European Regional Science Association conferences shows that conference geography can be analysed when entity-origin data exist. The study examined conferences from 1998 through 2003 using entities’ spatial distribution, attendance frequency and distance. It reported distance decay for that different association and period.

The relevance is methodological. The study demonstrates that researchers can compare changing conference locations with broad entity geography and distance. It does not establish that NANOG attendance declines with distance, that any effect has the same magnitude or that the two entity populations behave similarly.

Those differences matter. The association, continent, era, event design, transport conditions and professional community are not NANOG’s. Applying the study’s result to NANOG would substitute analogy for evidence. Its proper use here is narrower: it shows the kind of question that becomes answerable once origin and host are observed consistently.

NANOG need not reproduce an academic study. A minimal public method could use broad origin regions and travel-distance bands. For each meeting, the ledger could compare the forecast distribution used in planning with the observed aggregate distribution, separated by in-person and virtual mode.

The design should avoid one composite measure that implies a scientifically established venue effect. Distance is only one dimension. Border procedures, accessibility, dates, supplier constraints, hotel conditions, air access and network feasibility may matter. Some can be described as decision criteria without being converted into a single rating.

The independent study also underscores why meeting totals are limited public evidence. Without entity origins, the host cannot be related to journey distance. Without repeated definitions, changes cannot be compared. Without a credible alternative, observed differences cannot be attributed to the city.

What the reviewed record leaves unresolved

The evidence boundary can be stated in six parts.

First, the reviewed annual reports and event pages do not publish a stable origin-to-host distribution for NANOG 84 through NANOG 95. They name hosts and publish meeting fields, but not a comparable table of broad entity origins.

Second, they do not publish the candidate cities or regions considered for each meeting. A reader sees the contracted outcome, not the feasible set from which it emerged.

Third, they do not publish decision weights for cost, dates, air access, border or visa procedure, accessibility, supplier risk, network feasibility or rotation. This does not mean those factors were ignored. It means their role cannot be reconstructed from this source set.

Fourth, they do not publish a forecast-versus-observed access comparison. A city may be selected partly because it is expected to serve a entity distribution, but the reviewed material does not show the forecast or test it afterward.

Fifth, they do not publish a stable cross-meeting account of who switched between in-person and virtual participation because of location. The sources distinguish modes but do not establish the reason for choosing one.

Sixth, they do not identify a current formal venue-selection policy or decision owner. The likely operating surface includes organisational strategy, contracting, venues and service providers, but the reviewed sources do not assign each stage to a named function.

These are limitations of the reviewed public record. They are not proof that the information does not exist within NANOG, that the process was arbitrary or that a geography was excluded. A responsible investigation preserves that distinction.

The strongest case for keeping venue decisions less visible

Venue selection is a constrained commercial and operational task. A viable host needs suitable dates, rooms, meeting space, network capability, accessibility, staff feasibility and tolerable contract risk. The most geographically attractive city may not have an available venue at the required time. The least expensive quoted room may carry other costs or conditions.

Negotiation can depend on confidentiality. Publishing losing hotel bids, detailed concessions or a bargaining position may weaken future negotiations. Vendors may be reluctant to offer terms if every unsuccessful proposal becomes public. A requirement for total commercial disclosure could reduce rather than expand feasible options.

Candidate lists can mislead without context. A city may appear in an early search but never become viable. Another may be the only option left after dates or space are applied. Publishing a raw list could imply that comparable alternatives were rejected on geographic grounds when they were not operationally comparable.

Entity-origin information creates privacy risk. A country or subregion with one or two attendees may make a person identifiable, particularly when combined with public speaking, employer or committee information. City-level origins or personal itineraries would create greater risk. The public benefit does not justify exposing them.

Rotation cannot equalise every journey. Entities are distributed unevenly, and their origins change. A host that improves access for one cluster can worsen it for another. A meeting may reasonably balance geography with dates, price, accessibility, contract exposure, supplier resilience and network needs.

Hybrid participation may be the proportionate response to some geographic barriers. It can broaden access without requiring every meeting to solve every journey. At the same time, hybrid service has its own limits and should not be represented as identical to in-person participation.

These are strong arguments against publishing bids, personal origins or an equality guarantee. They do not require silence about the decision structure. NANOG can protect commercial information while publishing broad candidate geography, criteria and major constraints. It can protect people while publishing sufficiently large origin bands and suppressing rare cells.

The countercase therefore improves the proposed ledger. It narrows the request to information that explains institutional choice and aggregate outcome. It excludes losing contract prices, named travellers, detailed itineraries and a claim that one city should serve everyone equally.

A minimal venue and access ledger

The ledger should begin before selection. Its first field would describe candidate geography at a broad level: for example, the regions in which viable options were sought or the reason only one broad geography remained feasible. It need not list every hotel or city approached.

The second field would state decision criteria and major constraints. Dates, room and meeting-space capacity, network feasibility, accessibility, air access, border procedure, forecast lodging conditions, supplier risk, staff capacity, budget and rotation could appear where relevant. The record need not publish commercial weights if doing so would expose negotiation, but it should explain which considerations were decisive.

The third field would name the contracted host and the organisational function responsible for approval. This is not an invitation to attribute selection to sponsors, programme committees or local partners without evidence. Each actor’s role should be described separately.

The fourth field would publish a forecast access distribution. The forecast could group anticipated in-person entities into broad origin or distance bands and estimate virtual participation separately. It should state the information base and uncertainty rather than claiming precision.

The fifth field would publish observed aggregate origin bands after the meeting. The bands should be broad enough to protect people and stable enough for comparison. A minimum-cell rule would suppress or combine small geographies. Employer headquarters should not substitute silently for travel origin.

The sixth field would separate mode. In-person and virtual counts should remain distinct, along with a description of which major programme functions were available in each mode. A reason-for-mode measure should appear only if its collection is proportionate and clearly explained.

The seventh field would record selected venue conditions that affect planning: whether the primary room block filled, the published rate range, whether overflow was activated and whether an invitation-letter procedure was offered. It should not claim what individuals paid or what outcome any procedure produced.

The eighth field would compare forecast and observed distributions. Did the host serve the broad origin pattern anticipated? Were extreme-distance or cross-border bands larger than expected? Did virtual participation differ materially from forecast? These are descriptive review questions, not causal conclusions.

The ninth field would record uncertainty and missing states. “Not collected,” “unavailable,” “suppressed” and zero must remain distinct. A suppressed visa-letter count, for example, should not be interpreted as no requests.

The tenth field would state privacy and retention rules. Personal origin links used to create the aggregate table should be limited to the review window, accessed narrowly and deleted after validation. The public record would retain bands and definitions, not itineraries.

This design is deliberately modest. It would not calculate the total cost of attendance, track post-meeting careers or adjudicate whether a city was fair. It would let NANOG explain what geographic alternatives were feasible, what trade-offs governed the decision and how the observed distribution compared with the planning assumption.

Implementation and review

The cleanest implementation is prospective. NANOG could begin with a future host decision and define the ledger before candidate search is complete. That avoids retrofitting current concepts onto historical contracts and origin data that may have been collected differently.

The first step is a purpose statement: the record exists to make venue strategy and aggregate access consequences more legible, not to rank attendees or publish commercial negotiations.

The second is ownership. NANOG should identify a designated meetings or governance function to maintain definitions and coordinate publication. Contract negotiators, registration staff, privacy reviewers and programme operators would retain their distinct roles. This is a proposed allocation, not a description established by the reviewed sources.

The third is a candidate-stage entry. It would publish broad regions considered, major constraints and the final approval function after the contract is secure. Delaying publication until contracting is complete can protect bargaining while preserving accountability.

The fourth is an origin design. NANOG should choose broad bands based on privacy and decision value, test combination risk and define mode. If address or employer data are unsuitable, the ledger should say so rather than manufacture precision.

The fifth is a forecast. Before registration outcomes are known, the record should state the expected broad distribution and important uncertainties. A forecast creates a reference point without pretending to be a counterfactual experiment.

The sixth is post-meeting publication. Meeting totals, origin bands, mode and selected venue conditions would be added. Small cells would be suppressed. No attendee names, passport data, personal addresses or itineraries would appear.

The seventh is review. NANOG could explain what it learned, which assumptions were wrong and which constraints should change in future candidate searches. It should avoid attributing observed variation to the host unless a valid comparison supports that conclusion.

The eighth is deletion. Once aggregates are validated and the review closes, personal linkage used solely for the ledger should be removed according to a published schedule.

Historical meetings can remain a descriptive baseline. The twelve hosts, reported meeting statistics, visa-letter process and Denver hotel conditions are evidence of what was publicly recorded. They should not be reconstructed into an origin series where none was published.

What would falsify or narrow the criticism

This critique is designed to change when evidence changes. It would weaken materially if NANOG publishes a stable venue ledger with candidate regions or a justified reason for a single feasible option, explicit criteria and constraints, the contracted host and decision owner, privacy-protected origin bands by mode, a forecast-versus-observed comparison, stable definitions and small-cell suppression.

It would also narrow if NANOG identifies a specific privacy, legal or commercial constraint that makes a requested field disproportionate. A clear explanation can be more responsible than disclosure that exposes people or damages negotiation.

A stable origin-to-host table could show that the existing rotation distributes burden broadly. If it did, that evidence should receive full weight. The purpose of the ledger is not to preserve a criticism; it is to make the geographic claim testable.

The analysis would also change if NANOG publishes a current venue policy that explains how rotation is balanced with feasibility. The 2007 heading would then become historical context rather than the only explicit burden language in the reviewed material.

Another host map by itself would not settle the question. More cities demonstrate more movement. They do not show the distribution of entity journeys unless origin and mode denominators are present.

Conclusion: publish the joins, not the itineraries

The host sequence from NANOG 84 through NANOG 95 is real and geographically varied. It contains Austin, Montréal, Hollywood, Atlanta, Seattle, San Diego, Charlotte, Kansas City, Toronto, Atlanta again, Denver and Arlington. Ten meetings were in the United States and two in Canada.

NANOG describes its community as gathering in major cities across North America. A historical presentation framed movement as spreading travel burden. Those statements make geographic access a legitimate subject of inquiry. They do not prove equal burden, continental representation or a current rotation rule.

The annual reports and event pages add useful detail. NANOG 93 provides an in-person, virtual, membership-status and first-time snapshot. NANOG 89 publishes a visa-letter path. NANOG 94 shows a sold-out room block and specific room rates. Hybrid access creates a distinct alternative to travel.

None of those records connects broad entity origin with host selection. None publishes the candidate set or decision weights. None allows first-time variation to be treated as a location effect. The independent conference research shows only that a geographic method is feasible when the right information exists.

The answer is not to demand hotel bids or personal travel files. It is to record broad candidate geography, criteria, major constraints, the approval function, forecast access bands, observed aggregate origins by mode and a post-meeting review. Small cells should be suppressed and temporary personal joins deleted.

A moving map may spread burden. It may also move burden in ways the map alone cannot reveal. NANOG’s public record establishes movement. A minimal venue and access ledger would show how the organisation understood the trade-off—and whether the observed distribution matched the plan.

Sources