Summary

  • The fully online IETF of 2020 and 2021 removed major travel, visa, accommodation, and location barriers. Participation recovered from the emergency IETF 107 format to more than 1,100 people at IETF 108 and remained above 1,300 at each 2021 meeting. That is evidence that remote infrastructure preserved and widened meeting access, not evidence that influence became evenly distributed.
  • The scarce resource changed. Six-hour synchronized schedules rotated the inconvenience of night work among major regions, while home care, ordinary job duties, bandwidth, equipment, language, and the ability to reserve uninterrupted time determined whether registration became sustained contribution. The IETF's own non-participation survey found time and employer support among the strongest reasons former attendees stopped participating.
  • Remote-only meetings equalized access to the formal microphone more successfully than they replicated hallway discovery, sponsorship, and trust. Survey respondents at IETF 108 were disproportionately experienced and active: most had recently attended or spoken in working groups, many authored current drafts, and 70 percent had attended at least six IETF meetings. The online population was larger, but the available evidence does not show a wholesale transfer of agenda-setting power to newcomers.
  • The 2022 hybrid return exposed the lasting redistribution. Remote participation remained the majority at IETF 113 and substantial at IETF 114 and IETF 115, but remote entities again faced weaker informal access once an onsite center existed. A legitimate hybrid model must measure whose interventions alter drafts and decisions, not merely count logins.

The pandemic did not democratize or close the IETF in one move

The move online is often described with one of two stories. In the optimistic version, a global technical institution finally became available to anyone with an Internet connection. People who could not secure a visa, airfare, hotel, registration, or a week away could enter the same sessions as veteran standards engineers. Geography stopped deciding who was in the room. In the pessimistic version, the institution lost the social encounters through which difficult technical compromises are found, newcomers are sponsored, and weak signals become agenda items.

Existing professional networks continued on private calls while everyone else watched a formal conference stream.

Both stories identify real changes, but neither is an adequate account of influence. Access is not binary. A entity can register, listen, type in chat, enter a microphone queue, submit a draft, persuade a working group, become trusted as a reviewer, and eventually shape an agenda. Each step has different costs. Removing travel can increase the number of people at the first three steps without changing who consistently reaches the last three. Losing a hallway can reduce the informal advantage of physical proximity while increasing the advantage of already having private contact details and established trust.

The relevant comparison is therefore not online versus closed, or onsite versus authentic. It is a change in the distribution of constraints. Before March 2020, travel selected heavily for money, passports, health, employer budgets, and freedom to leave home. During the remote-only period, synchronous hours, domestic space, digital equipment, ordinary job demands, care work, and pre-existing relationships did more of the selecting. Some people gained access because one set of barriers disappeared. Others lost effective access because the new set fell more heavily on them.

The difference matters to rough consensus. The IETF is not a legislature in which every affected person must cast an equal vote. Technical influence properly depends on evidence, judgment, and sustained work. But rough consensus is credible only when important objections can enter the discussion and receive reasoned treatment. A format that enlarges attendance while narrowing who can prepare, intervene, or remain active can look open in registration data and still concentrate effective voice.

The 2020-2022 period is valuable because it was not a planned experiment with clean treatment and control groups. It was a shock. The institution changed venue, schedule, technology, fee model, and social environment under extreme external pressure. That makes simple causal claims unsafe. It also made previously hidden participation costs visible.

IETF 107 was an emergency format, not a normal online meeting

On 10 March 2020, the planned Vancouver meeting was cancelled. The replacement was assembled in less than two weeks. It did not reproduce the ordinary agenda. The IESG prioritized Birds-of-a-Feather sessions, new working groups, dispatch-style groups, and the administrative plenary, while much of the remaining work moved to virtual interim meetings over the following weeks.

The official IETF 107 account estimated 701 unique entities from at least 39 countries. Individual sessions drew between 82 and 235 people, and the plenary drew 282. Several sessions may have attracted more people than an ordinary in-person version would have done. Those are significant achievements under emergency conditions. They show that the IETF could preserve live, global technical discussion after losing its venue almost overnight.

They do not create a clean before-and-after measure. The agenda itself changed. A person whose working group was omitted did not receive an equivalent meeting and might have appeared later in an interim. Someone interested in a new group could join a compact, low-conflict schedule that concentrated cross-area attention. The smaller entity count partly reflects a smaller event, not necessarily rejection of remote participation.

The format also spread work beyond the nominal week. RFC 9400, which later recorded lessons from the fully online period, notes that the intensive sequence of interims had a greater calendar impact for some entities than a single meeting week and that attendance fell in some sessions. Interims were commonly scheduled for the hours convenient to the bulk of a group's active entities, which could exclude people in less common time zones.

This changed the value of being a regular. A newcomer could enter a prominent IETF 107 session without travel, but following work across six subsequent weeks required knowing which interim mattered, tracking list traffic, reserving multiple calendar blocks, and understanding how meeting comments became document changes. Established entities had those maps. The emergency widened the front door while lengthening some corridors behind it.

IETF 107 should therefore be read as proof of continuity and a warning about measurement. Counting the emergency week's logins understates participation that moved into interims. Treating every interim as equivalent overstates access for people who could protect one week but not a rolling sequence. Influence cannot be inferred from either count alone.

The full online meetings proved demand, not equal power

The IETF moved from the compact emergency design to a fuller online meeting at IETF 108. The official 2020 annual report records 1,120 remote entities at IETF 108 and 1,279 at IETF 109, compared with 701 at IETF 107. Working groups also held more than 240 interim meetings during the year, approximately twice the 2019 number.

Participation remained high in 2021. The 2021 annual report lists 1,329 entities at IETF 110, 1,411 at IETF 111, and 1,347 at IETF 112. More than 250 interims supplemented those plenary meetings. A global technical community did not vanish when the hotels closed. Online access supported meeting populations comparable to, and in some cases larger than, familiar in-person meetings.

These totals establish capacity and demand. They do not say who spoke, whose issue received agenda time, which draft changed, or who remained after the meeting. A registration can represent five days of active chairing, one session on a specialist topic, a background stream, or a login made to satisfy an attendance criterion. All are legitimate forms of contact, but they are not interchangeable measures of influence.

The totals also aggregate gains and losses. Suppose 300 people enter because travel disappears while 200 experienced entities reduce engagement because the online meeting provides little informal value or conflicts with ordinary work. Attendance rises by 100, yet the distribution of agenda control may stay stable or become more concentrated. Conversely, a smaller count could conceal stronger contributions from previously absent operators. Without cohort and contribution data, direction cannot be assigned from the headline number.

The most defensible conclusion is narrower. Online delivery prevented a catastrophic contraction, enabled substantial topic-specific and remote participation, and created a durable expectation that physical travel should not be necessary for live contribution. It did not by itself demonstrate that newcomers, smaller operators, underrepresented regions, caregivers, independents, or public-interest entities obtained influence proportional to their new access.

That distinction is not an argument against online meetings. It is a demand for the right outcome measure. If the governance claim concerns openness, the institution should track the path from registration to intervention, response, document effect, continued participation, and leadership opportunity.

Travel barriers genuinely fell

The gains were concrete. A remote entity did not need to buy an intercontinental ticket, pay for a hotel, obtain a visa, cross a border during a health crisis, navigate an inaccessible venue, or absorb the physical strain of travel. An engineer could attend one relevant working group without financing an entire week abroad. A university group could watch a session together. An operator facing an urgent network issue could enter a discussion without abandoning local responsibilities for travel days.

The change also reduced the penalty attached to meeting location. Physical attendance ordinarily rises and falls with region because nearby entities face lower cost and fewer travel risks. RFC 9400 records that the country distribution during the 2021 online meetings was comparatively stable even as meeting time zones rotated. That is a meaningful form of geographic continuity: people did not need the meeting to land in their region to avoid the travel barrier.

Remote participation also made selective attendance rational. A contributor who followed one protocol did not need to justify a full business trip. Specialists could be invited into a particular discussion. RFC 7282's consensus guidance emphasizes recruiting additional implementers and experts when broader views are needed. Online access made that recruitment easier because the invitation could be for an hour rather than a week.

The meeting technology improved quickly. Post-meeting survey satisfaction with Meetecho rose after problems at IETF 109, reaching 4.30 out of five at IETF 110 and remaining in the good range through the following online and hybrid meetings. Integrated queues, authentication, slides, chat, and audio gave remote interventions a more coherent route into formal sessions than earlier eras of passive streams and separate text rooms.

None of these benefits should be qualified out of existence. The fact that new inequalities appeared does not mean the travel barrier was imaginary. A visa refusal is categorical. A flight budget can be prohibitive. A health condition can make travel impossible. Online meetings converted some categorical exclusions into scheduling and resource problems that could, at least in principle, be mitigated.

The error would be to treat a removed barrier as a removed cost. The meeting moved into entities' local environments, and the institution no longer controlled the conditions in which they attended.

Time zones became the visible border

A physical meeting imposes local hours on remote entities and travel on onsite entities. A fully online meeting imposes local hours on everyone, but those hours are radically different around the world. The IETF shortened online days to roughly six hours and rotated the planned meeting location's time zone. The aim was to share inconvenience across the three regions used in the IETF's meeting rotation.

That rotation was fairer than permanently selecting North American or European business hours. It was not neutral. Six consecutive hours that begin at midnight are not simply a shifted workday. They alter sleep, care, safety, concentration, and the ability to return to ordinary work. A entity can tolerate one overnight intervention more easily than five days of sessions, preparation, side conversations, and interims.

The IETF 108 schedule ran from 11:00 to approximately 16:00 UTC. The planning explanation said a survey of 576 respondents found the widest reported ease between 12:00 and 22:00 UTC, while also noting that many virtual meetings had already clustered at hours difficult for Asia. The meeting used the Madrid frame and an 11:00 UTC start as a compromise.

IETF 109 followed Bangkok hours. Its post-meeting survey gave the Bangkok time zone a mean satisfaction score of 2.89, compared with 3.86 for the Madrid time zone at IETF 108. The policy of following the replaced physical venue scored 3.45. The figures do not prove that one region deserved a different result; they show how sharply the burden became visible when the meeting clock moved.

RFC 9400 later proposed a more explicit rotation among three six-hour windows, each placing one major region in the night. It also acknowledged that a three-region model does not distribute inconvenience equally to every location. Africa, Latin America, the Middle East, Oceania, and countries spanning unusual offsets do not fit neatly into a North America-Europe-Asia triangle.

Time-zone fairness should therefore be judged over a sequence and within working groups. A person may survive one bad plenary week but face every interim at the active core's convenient hour. The formal rotation can distribute the headline burden while recurring drafting calls, design teams, and editor meetings restore the same geographic advantage between plenaries.

Home was not a neutral meeting venue

An online meeting is often described as attendance from home, as if home were a standard facility. It is not. One entity has a private office, wired broadband, multiple screens, noise isolation, and meals prepared by someone else. Another joins from a shared room, a metered mobile connection, a small device, or a household in which six hours of withdrawal must be negotiated every day.

The difference affects more than comfort. Reading a draft while watching a speaker, following chat, preparing a response, and entering a queue is cognitively demanding. A second screen, stable audio, and uninterrupted time can determine whether a entity catches a consequential change. Packet loss or a shared connection can turn a precise intervention into fragments. A person who cannot use audio privately may remain in chat, where chairs and other entities can assign different weight even when they act in good faith.

Care responsibilities also interact with visibility. At a hotel meeting, travel may impose a serious burden on the household, but the attendee's meeting hours are institutionally protected once the trip occurs. At home, school closures, elder care, meals, illness, and ordinary family contact continue within reach. The IETF 108 survey's open responses included entities describing breaks as the time when they fed children or handled family and work demands. Those comments are not population estimates. They identify mechanisms the registration count cannot see.

The home setting can benefit some entities. People who cannot leave dependants for a week may join selected sessions. Disabled entities may avoid an exhausting trip. Someone uncomfortable speaking in a large ballroom may use a structured digital queue more easily. Recorded material and chat can improve review. The same format can therefore widen access and reduce intensity for different people, or for the same person on different days.

Institutional design should not decide which experience is typical from anecdotes. It should recognize that remote attendance externalizes venue quality to households. Equipment grants, low-bandwidth modes, dial-in options, recordings, captions, early schedules, asynchronous issue confirmation, and shorter required blocks are not amenities. They are substitutes for facilities the institution would otherwise supply onsite.

Employer support changed form rather than disappearing

Travel makes employer support visible through bookings and expense reports. Remote participation can make the same support look like individual availability. Yet a five-day online meeting still consumes working time. The employer decides whether sessions count as work, whether ordinary duties are reduced, whether a calendar can be blocked, and whether standards participation contributes to performance and promotion.

For a fully supported standards engineer, moving online may increase effective participation. There is no travel fatigue, and the work remains paid. The engineer can attend more interims, move between sessions, and continue private coordination. For someone whose manager viewed an IETF trip as a recognized professional event but sees a video conference as compatible with normal duties, the move can reduce protection. The entity is simultaneously at the meeting and expected to answer routine messages.

The IETF's 2020 non-participation survey makes this resource visible. It contacted roughly 5,250 email addresses used to register for IETF meetings in 2018 or 2019 but not IETF 108. Of 439 filtered non-entity respondents, 37 percent of those who knew about the meeting named being too busy as the primary reason, reported as 29 percent after the survey's adjustment for respondents who had not known about IETF 108. The remote fee and aversion to a fully online meeting followed.

More revealingly, 114 respondents said they no longer participated in the IETF. Among the 113 who explained why, with multiple answers allowed, 35 percent said their employer no longer supported participation and 32 percent said they could not spare the required time. This is not a census, and it cannot be attributed solely to the pandemic or online format. It does show that sustained participation depends materially on paid permission and discretionary time.

Employer support does not prove employer control of technical views. Entities act as individuals, and a well-supported engineer can exercise independent judgment. The governance issue is opportunity. Organizations that pay people to remain continuously engaged can supply more reviews, implementations, draft revisions, and meeting presence than affected organizations that participate only when immediate operational harm becomes visible.

Remote-only meetings lowered the employer's cash cost of travel while preserving the need to allocate labor. That can broaden sponsorship inside organizations, but it can also make the subsidy less legible. An institution measuring only fee waivers and travel grants will miss the most important resource transfer.

The formal microphone became more equal than the hallway

The strongest feature of the fully online format was symmetry inside scheduled sessions. Everyone entered through a digital service. There was no front row, no walk across a ballroom, and no remote audio feed treated as an appendage to an onsite room. A single queue could order requests from entities regardless of location. Slides and chat were common. Chairs could see remote hands directly.

That symmetry matters. It reduced the technical and social distinction between the person at the venue and the person on a screen. The 2022 hybrid design retained an integrated queue precisely because the fully online period demonstrated that remote interventions could be part of the main session rather than a secondary channel.

But the scheduled microphone is only one site of influence. Ideas are framed before agenda time is requested. Authors test whether a proposal sounds plausible. Chairs learn who has relevant operational experience. Area Directors hear early warnings. Entities introduce newcomers to collaborators. A disputed phrase is sometimes resolved through a twenty-minute conversation that makes the later public exchange intelligible.

Online social systems did not reproduce that environment. Survey satisfaction with opportunities for social interaction remained much lower than satisfaction with working-group sessions. At IETF 112 it was 2.79 out of five. Tools designed to simulate hallways were available, but RFC 9400 records that many entities did not use them because they did not know about them, disliked them, or preferred sleep, food, and home activities when formal sessions ended.

The loss was not evenly shared. Established contributors could continue on existing private channels, scheduled calls, direct messages, and long-standing professional relationships. A newcomer could enter the public session but might not know which informal conversation to join or whom to contact after an unresolved exchange. Remote-only operation removed the advantage of being physically near a leader while preserving the advantage of already being known.

This is a subtle redistribution. Formal speaking access became more equal; informal discovery could become more dependent on prior connection. An institution can improve the first with software. The second requires deliberate sponsorship, open office hours, documented issue states, newcomer introductions, and norms that return consequential informal conclusions to public channels.

The experienced core remained highly visible

The IETF 108 entity survey offers a useful caution against equating a large online meeting with a new electorate. It received 382 responses, including 369 entities, from a meeting of approximately 1,100. Among respondents, 86 percent had attended a working-group or BoF session during the previous year, 68 percent had spoken in one, and 60 percent were authors of an active Internet-Draft. Fifty-four percent had participated in at least eleven IETF meetings, with another 16 percent reporting six to ten.

These figures describe survey respondents, not every registrant. The response group may have been more engaged than entities who ignored the survey. Even with that limitation, the data show that the institution's experienced core did not disappear online. A large share of those willing to report on the meeting had substantial prior participation and current authorship.

That continuity was valuable. Experienced chairs, editors, implementers, and reviewers kept work moving through a crisis. Technical standards cannot be rebuilt from a new audience every quarter. Institutional memory explains why an emergency event could function at all.

Continuity becomes a legitimacy concern only when it is mistaken for proof that new access produced new influence. The available official summaries do not provide a complete meeting-by-meeting join between first-time status, region, affiliation, local meeting hour, queue use, mailing-list contribution, draft authorship, issue resolution, and later retention. Public attendance lists and proceedings can support research, but identity matching and contribution coding raise privacy, name ambiguity, and interpretation problems.

The right conclusion is not that newcomers were absent. Hackathons, accessible sessions, and selective attendance demonstrably brought in people who had not traveled to prior meetings. The right conclusion is that the evidence supports access expansion more strongly than power redistribution. A entity count cannot show whether a new voice changed a technical outcome.

This boundary should be stated plainly because both sides of the debate are tempted to overclaim. Advocates should not call the online period representative merely because geography was less tied to venue. Critics should not call it an insider enclosure merely because survey respondents were experienced. The period produced a wider contact surface around a durable expert core. Whether that contact became sustained influence varied by group and person.

Who entered, and on what terms

The people most clearly enabled were those for whom physical travel had been the decisive barrier. That includes entities facing visa uncertainty, unaffordable flights and hotels, health risks, disabilities aggravated by travel, limited travel budgets, or a need to attend only one technical session. Remote access also helped specialists who could not justify a full IETF trip and organizations able to release an engineer for hours but not a week abroad.

Entities in the meeting's favored time zones received a second advantage. They could join formal sessions during a normal or tolerable day without paying the recovery cost imposed on overnight attendees. Because the time zone rotated, that advantage moved between plenaries. Within a single technical decision, however, the relevant question is not whether another region would be favored four months later. It is who was alert and available when that issue was discussed.

People with strong employer support also entered more deeply. Lower travel expense could allow an organization to expose more staff to standards work. A larger internal group could divide sessions, contribute code, and follow parallel discussions. Entities already employed in distributed engineering organizations had collaboration tools, home equipment, and norms for remote work.

Some newcomers found the digital queue easier than a physical microphone. The ability to inspect an agenda, join one session, and speak without navigating a large venue lowered social and logistical friction. Recordings and online materials made preparation and review possible across time.

Yet entry came in tiers. Listening to one session is meaningful access but does not automatically produce relationships, agenda knowledge, or future author roles. A entity without protected follow-up time may contribute an operational warning and disappear before the draft changes. Someone entering from a low-bandwidth connection may hear the discussion but not intervene reliably. A newcomer at an unfavorable hour may choose the single most relevant session and miss cross-area context.

The online period therefore created more permeable edges, not a flat field. It made it cheaper to sample the IETF, easier to invite outside expertise, and possible to remain involved without travel. The conversion of those gains into durable influence depended on follow-up structures the meeting format did not automatically supply.

Who exited, reduced participation, or lost leverage

The clearest losses occurred among prior attendees who valued the physical meeting as protected time and a dense social environment. The IETF 108 non-participation survey found that 15 percent of respondents who knew about the meeting named its fully online nature as their primary reason for not joining, or 12 percent after the report's adjustment. A fee was the primary reason for 19 percent, adjusted to 15 percent. Time zone, schedule, agenda, and productivity were less common primary reasons individually, while being too busy dominated.

These percentages should not be added as if each person had only one causal barrier beyond the primary reason. They reveal several exit paths. Some people did not find the online format worth the fee. Some could not separate meeting time from other work. Some missed the customer, collaborator, or hallway interactions that justified attendance. Some did not know the meeting was happening: 22 percent of the filtered non-entity respondents reported no awareness of IETF 108.

People working in network operations during the pandemic faced a special conflict. The Internet's sudden dependence was itself a source of workload. An engineer most able to provide operational evidence could be least able to leave incident, capacity, or service-continuity duties for a synchronous standards week. Remote availability did not create spare time.

Caregivers and entities without private space could join in principle while reducing visible contribution. People in the worst time zone might attend only their own working group, losing cross-area encounters. Independents paid the opportunity cost directly. Entities whose employer no longer treated online attendance as travel-worthy professional development lost the calendar shield a trip had provided.

There was also a loss of leverage for remote entities when the institution returned onsite. During remote-only meetings, every entity was remote. In hybrid meetings, an onsite cluster again shared meals, corridors, and local time. The formal queue remained integrated, but social symmetry did not.

Exit should therefore include more than non-registration. A person can remain on the attendee list while ceasing to author, review, mentor, or intervene. A contribution analysis should look for attenuation: fewer sessions, fewer messages, fewer microphone interventions, longer response delays, and lower return rates. Those are the places where redistributed influence becomes visible.

The proliferation of interims rewarded calendar ownership

Virtual interim meetings were essential during the pandemic. They allowed omitted IETF 107 groups to meet, reduced pressure on parallel tracks, and gave working groups focused time to resolve issues. The increase from roughly half as many in 2019 to more than 240 in 2020 and more than 250 in 2021 shows how much synchronized work moved into the rest of the year.

Interims can improve geographic access when groups rotate times or schedule around the actual entities needed for an issue. They can also improve quality by giving a difficult design problem more time than a plenary slot allows. Recordings, minutes, and list confirmation can preserve accountability.

But frequency creates a different participation test. A person who can attend one funded meeting week may not be able to reserve two calls every month. Recurring interims favor jobs in which standards participation is an explicit duty. They also favor active cores whose calendars determine the chosen hour. A entity at the edge of the group may not entity to an inconvenient time because doing so feels disproportionate for a call they are not yet sure they need.

RFC 9400 records this tension directly: the post-IETF 107 series affected calendars more heavily for some entities, and meetings scheduled for the bulk of active entities could exclude uncommon time zones. The convenience of the median active entity can therefore reproduce itself. Those who cannot attend become less active; their absence then justifies scheduling around the remaining core.

This feedback loop matters more than any single unfair choice. A chair may rationally pick the hour at which current authors can meet. Over time, that hour shapes who can become an author. Calendar availability becomes evidence of engagement, and engagement becomes the reason to preserve the calendar.

The safeguard is not a ban on interims. It is rotation, advance notice, asynchronous issue framing, public minutes, explicit confirmation on the mailing list, and occasional paired sessions when a truly global operational question requires them. Groups should record who could not attend and invite evidence before treating live convergence as closure.

2022 revealed what remote-only symmetry had concealed

IETF 113 in Vienna was the first hybrid return. The 2022 annual report records 314 onsite and 976 online entities. At IETF 114 in Philadelphia, the split was 618 onsite and 679 online. At IETF 115 in London, it was 849 onsite and 666 online. Remote participation did not collapse once travel resumed. It remained the majority at the first two hybrid meetings and a large share at the third.

This persistence is one of the remote-only period's durable achievements. Entities and organizations learned that live IETF work could be done without travel. The institution invested in integrated participation rather than treating remote access as a broadcast. At IETF 113, all entities used Meetecho for queue management and related session functions. The post-meeting survey rated the integrated queue at 4.45 and provision for remote chairs at 4.34.

The hybrid format also restored an asymmetry. Onsite entities occupied the same local day and could continue conversations outside formal sessions. Remote entities retained the queue but not the shared corridor. The IETF 113 survey reported multiple remote entities who felt the in-person return adversely affected them. By IETF 115, satisfaction with social opportunities among remote entities was 2.58, even though overall meeting and Meetecho satisfaction were good.

This contrast helps separate two kinds of equality. Session equality concerns whether remote and onsite speakers can enter the same visible queue, hear the same discussion, and receive the same chair attention. Network equality concerns whether they can form the relationships and informal context through which issues are recognized before the queue opens. Hybrid engineering can make substantial progress on the first while leaving the second structurally uneven.

The solution is not to abolish onsite meetings. Physical contact has genuine collaborative value, particularly for newcomers, difficult disagreements, and cross-group discovery. The legitimacy obligation is to prevent that value from turning into unrecorded decision authority. Material conclusions reached onsite should return to the list. Chairs should actively solicit remote operational evidence. Open office hours and remote-first side sessions can create contact points. Session records should identify unresolved questions rather than only outcomes.

The 2022 return demonstrated that remote participation had become part of the institution, not an emergency substitute. It also showed that remote presence and remote parity are different achievements.

Attendance is the wrong denominator for influence

If the question is who gained influence, the unit should be a contribution and its treatment, not a registration. The IETF already preserves much of the evidence needed: agendas, attendance, minutes, recordings, chat in some contexts, mailing-list archives, Internet-Draft revisions, authorship, reviews, issue trackers, and role histories. The challenge is to connect those records without converting noisy traces into false certainty about people.

A useful analysis would begin with a meeting cohort and distinguish newcomers, returning entities, long-term regulars, and topic-specific guests. It would record participation mode, local-hour burden, broad region, and affiliation category only where disclosure and privacy permit. It would then follow public technical actions: microphone interventions, list messages, draft reviews, issue proposals, implementation reports, chair acknowledgements, text changes, and continued activity over subsequent meetings.

The central measure should be response. Was an objection answered with reasons? Did the group request more evidence? Did the draft change? Was an operational concern recorded for later work? Did the entity return? Counting messages alone rewards persistence and verbosity, the very error rough consensus is meant to avoid.

The analysis should also separate formal and informal outcomes. A first-time entity may make one decisive observation and never author a draft. That is influence. An established author may speak frequently without changing the result. That is participation, not necessarily influence. A private conversation may initiate work but should become visible through a public proposal before it supports institutional authority.

Privacy and interpretation limits are substantial. Names change, email addresses vary, shared drafts have unequal labor, silence may represent agreement or exclusion, and geography does not determine viewpoint. Any dashboard should use aggregates, suppress small groups, invite correction, and publish methods. It should not rank individuals or infer protected traits.

The objective is not to calculate a universal influence score. It is to test specific claims. Did remote-only meetings increase first interventions from regions previously underrepresented onsite? Did those entities remain active? Were overnight speakers less likely to receive immediate discussion? Did working groups with rotating interims retain a broader set of contributors? Did hybrid return reduce remote issue initiation? Those questions can be answered more responsibly than "was online better?"

A governance test for meeting formats

The IETF should evaluate any meeting design through five separate gates.

The first is entry. Can an interested, technically competent person discover the meeting, register without an unaffordable charge, access the platform with ordinary equipment, and understand where relevant work occurs? Travel, visa, accessibility, language, bandwidth, and fee barriers belong here.

The second is usable presence. Are the hours tolerable? Is the schedule announced early? Can care and employment be arranged? Are captions, recordings, low-bandwidth modes, chat, and queue controls reliable? A login without a usable channel should not count as full access.

The third is deliberative parity. Do chairs hear remote and onsite interventions through the same procedure? Are objections evaluated for substance rather than volume or familiarity? Are hums and informal polls treated as inputs rather than votes? Are meeting-originated decisions confirmed in public asynchronous discussion?

The fourth is relationship access. Can newcomers meet chairs, authors, implementers, and other entities without already belonging to private networks? Are office hours, mentoring, topic introductions, side meetings, and post-session contact available across modes and time zones? Does the institution return informal conclusions to visible channels?

The fifth is durable effect. Do new entities return, review drafts, supply implementation or operational evidence, take roles, and see their valid concerns reflected in documents? Are exits associated with time, employer support, hostile experience, cost, or repeated scheduling disadvantage? This gate distinguishes an audience from a community.

No format will score equally for every person. Onsite meetings perform strongly on dense relationship building for those able to travel and poorly on entry for those who cannot. Remote-only meetings perform strongly on geographic entry and formal session symmetry, while imposing household and time-zone burdens and weakening unplanned contact. Hybrid meetings preserve both channels but risk creating a local center and remote periphery.

The right policy is therefore not a single format declared inclusive. It is a set of compensating controls. Retain the free remote route. Rotate synchronous burden. Keep integrated queues. Protect asynchronous confirmation. Fund travel where physical contact is valuable. Support time and equipment where remote attendance is the only practical option. Measure contribution treatment across modes.

Time support is the unfinished inclusion policy

The pandemic made software the immediate problem, and the IETF solved much of it. Meetecho integration, public schedules, remote queues, recordings, and support services created a viable global meeting. The next problem is less visible because it cannot be fixed with a feature release: entities need protected time.

Time support can take several forms. Meeting schedules can be published earlier and kept stable. Working groups can flag consequential issues before sessions so entities know when presence matters. Interim times can rotate, and decisions can remain open for asynchronous evidence. Chairs can write concise outcome notes that identify reasons, rejected alternatives, and unresolved objections. Recordings can be linked to agenda items rather than offered as undifferentiated hours of video.

Financial support can include compensation for participation time, not only travel and registration. Small operators, independent experts, civil-society technologists, academics without standards funding, and public-sector engineers may need grants that recognize review and meeting labor. Such support must never condition technical positions. Its purpose is to make informed independence feasible where no employer underwrites it.

Employers can be given clearer statements of expected workload and institutional value. A remote meeting should be described as a professional work commitment, not passive viewing compatible with a normal week. Entities should be able to document session, preparation, and follow-up requirements when seeking approval.

The IETF can also reduce the amount of synchronized time needed for influence. Well-maintained issue trackers, clear draft diffs, implementation-status records, and reasoned decisions let a entity contribute precisely without attending every call. This is not a concession to less serious entities. It is disciplined engineering documentation.

Finally, leaders should watch for calendar capture. If the same small set attends every interim, that may reflect commitment, but it may also reflect a schedule optimized around itself. Before treating attendance as the affected community, chairs should ask whose operational environment is missing and seek evidence outside the recurring call.

Travel funding was an obvious inclusion instrument because the expense had a receipt. Remote participation revealed an equally decisive cost with no invoice. The institution that wants broader technical judgment must help qualified people obtain time to exercise it.

Remote-only IETF redistributed influence rather than abolishing its price

The 2020-2022 period changed the IETF permanently. It proved that plenary work could continue at global scale without a hotel. It accelerated a shared participation platform, normalized digital queues, preserved remote access as meetings returned onsite, and helped establish a later right to a free remote option. People excluded by travel, visas, health, cost, or the need for selective attendance gained a real route into live standards discussion.

The same period showed that physical distance had never been the only source of inequality. Once flights disappeared, the decisive assets were easier to see: a favorable clock, a quiet room, broadband, equipment, paid time, care coverage, institutional fluency, and existing relationships. The online meeting transferred costs from a central venue to local calendars and households. It reduced some exclusions while creating or exposing others.

The experienced core remained central, which was necessary for continuity. The evidence does not establish that this core captured remote meetings, nor does it establish that increased attendance produced proportionate agenda power for newcomers. It establishes a more complicated institution: wider at the edge, technically more symmetric in formal sessions, and still dependent on resources that were unevenly distributed.

Hybrid return confirmed both lessons. Remote participation remained substantial, and integrated queues worked. The restoration of onsite hallways also recreated a center of informal access that remote entities could not fully share. The governance task is not to choose between human contact and remote inclusion. It is to ensure that neither travel nor private contact becomes a prerequisite for a well-founded contribution to alter the public work.

Influence should therefore be measured through the life of an intervention: who could raise it, whether it was understood, how the group answered, whether the document changed, and whether the contributor could remain involved. Registration totals are necessary operational data. They are not a legitimacy verdict.

Remote-only IETF removed the border at the hotel door. It did not create a meeting without borders. The new borders ran through the clock, the household, the employment contract, and the private network of trust. A credible open process must make those borders visible, reduce them where it can, and keep every consequential technical decision answerable to evidence from beyond the people best resourced to be continuously present.