Guidelines for the Organization of Fully Online Meetings

Introduction In 2020, the COVID-19 pandemic forced the IETF to move all its plenary meetings to online-only events. This document records the experience gained by holding plenary meetings fully online and the guidelines that have evolved from this experience. The aim of this document is to define rough consensus of these guidelines in the sense that the most participants seemed sufficiently satisfied with the organization of fully online events as done in 2020 and 2021. These guidelines, however, document only one option of running fully online meetings. Like the procedure for in-person meetings, changes to the organization of the meetings and the meeting agenda should be experimented with in the process of establishing future meeting guidelines. Hybrid meetings (meaning meetings that have higher remote participation than onsite participation) are out of scope. However, some of the experience gained from fully online meetings might also provide input for decisions regarding the organization of hybrid meetings.

Some History When the WHO declared a world-wide pandemic in March 2020, the IETF had to quickly cancel its plenary meeting and organize an online replacement (within less than two weeks). For this first online-only meeting, the agenda was reduced to a set of sessions that benefitted most from cross-area participation, like BoFs, first-time meetings of a new working groups, and dispatch sessions, as well as the administrative plenary in order to organize the official hand-over procedures that occur at the March meeting. With such a reduced agenda, it was possible to organize the meeting within roughly 2 sessions (about 4 hours) a day and a maximum of two parallel tracks. This was possible as all working group meetings were moved to interims which were then distributed over the coming six weeks. However, this was often perceived as increased load over a longer time. But at that point of time there was not necessarily an expectation that the situation would continue as long as it did. For the following meetings in 2020, the online schedule was switched back to be similar to an in-person meeting (1-2 hour slots and 8-9 parallel tracks as described below), however, still with a reduced total length of initially 5 hours a day and then 6 hours with longer breaks. All fully online meetings in 2020 have followed the time zone of the planned in-person meeting location, but starting roughly around noon. Some flexibility with the start time to be "around" noon has been used to mitigate the worse possible time slots, even though, given the distribution of participants it is not possible to avoid certain hours entirely. The in-person meeting location follows the 1-1-1 guideline as documented in to rotate between Asia, Europe, and North America. While the exact time slot used had led to various discussions, following roughly the 1-1-1 guideline to share the pain has/seems to have rough consensus.

Guidelines for Online Meeting Planning

Time Zone Selection This time selection enables to have 2 out of 3 fully online IETF plenary meetings during the day from most participants. Basically every full online meeting is for two regions of the three regions described in , roughly speaking, after sunrise or after dinner. This has the tradeoff that it maps the third region in middle of night. However, that also means for most participants only one remote meeting per year might require a significant change to sleep schedules. The times are also seasonally adjusted to leverage differentials in Daylight Savings Time. These time slots are as follows, in UTC:

Name	Times (Northern Summer)	Times (Northern Winter)
North America Night	0500-1100 UTC	0600-1200 UTC
Asia Night	1300-1900 UTC	1400-2000 UTC
Europe Night	2200-0400 UTC	2200-0400 UTC

The intent of rotating between these three slots is to scatter meetings throughout the course of the global day, to maximize the ease of participants to occasionally attend regardless of their location and what time of day is optimal for their schedule.

Guidelines for selection The IETF should select a start time from these three choices based on the past three meetings. The following table covers all permutations of previous meetings held in-person in Region A, B, or C; or remotely in the nights of one of those regions.

3 meetings ago	2 meetings ago	Last Meeting	Online Selection
Any	Any	In-Person A	A Night
Any	Online A Night	Online B Night	C Night
Online A Night	In-Person B	Online B Night	C Night
In-Person A	In-Person B	Online B Night	A Night
In-Person A	In-Person A	Online A Night	see below
Online A Night	Online B Night	Online C Night	A Night

Basically this table follows two guidelines: 1) When ever a fully online meeting follows an in-person meeting, the online meeting time is used that disadvantages most the participants of the time zone where the in-person meeting was held. 2) If multiple fully online meetings follow each other, the time zone selection should be rotated based on the most recent time zones that the in-person meetings were held in. The final case occurs in the rare event that back-to-back in-person plenaries occur in the same region. In this case, find the most recent meeting that was neither in 'A' (if in person) nor in 'A' night (if remote). If this meeting was in-person in region 'B', then the next meeting should be in 'B' Night. If it was remote in 'B' Night, the next meeting should be in 'C' Night. To initialize this algorithm, IETF 112 is considered as an 'Asia Night' remote meeting, and IETF 111 is a 'Europe Night' remote meeting.

Number of Days and Total Hours per Day Online meetings have converged to run over 5 days with 6-hour meeting days, roughly. Only the administrative plenary, which concludes with multiple open mic sessions, is not necessarily time-bounded. Based on the experience so far, 6 hours of online meetings, with two 30 minutes breaks, appears to be potentially a natural limit of what is handleable for most participants. Respectively, the meeting survey after IETF 109 has indicated a high satisfaction with the distribution of sessions over 5 days but only a medium satisfaction with the overall length of each day [https://www.ietf.org/blog/ietf108-survey-results-informed-planning/]. While there is a possible trade-off between shorter but more days, a compact and potentially intense meeting was slightly preferred from the beginning by the community. Unlike in-person meetings, utilizing time during the weekend was never considered as a possible option. In 2020 and 2021, it was possible for all meetings to fit the requested number of sessions within 5 days, with the respective number of parallel tracks, see Section .

Session/Break Length For fully online meetings there are typically less sessions per day than for in-person meetings, in order to keep the overall meeting day to at roughly 6 hours. The reduction of the number of sessions per day led to the practice of offering chairs only two options for session length (instead of three), in order to make session scheduling more practical. At IETF-108, based on an indicated preference of the community, 50 and 100 minute slot were used, with only 10 minutes breaks, in order to keep the overall day length at 5 hours. This resulted in many sessions going over time and thereby clearly indicated that only 10 minutes for breaks are not practical. The survey after IETF-109 showed a high satisfaction with 60/120 minute session lengths and 30 minute breaks, and a significant improvement in satisfaction over IETF-108. [https://www.ietf.org/blog/ietf-109-post-meeting-survey/] While the option to shorten the breaks was discussed during the later meetings, a saving of in total 10-20 minutes per day might not balance the need to use the breaks for recreation or at least some socialising.

Number of Parallel Tracks Fully online meetings are not limited in the number of parallel tracks by the physical restriction of a meeting venue aka the number of meeting rooms. In order to reduce the number of possible conflicts, it is still desirable to minimise the number of parallel tracks by balancing the requested sessions mostly equally over the available slots. But if the total number of requested sessions exceeds the capacity of the usual 8 parallel tracks, it is possible for a fully online meeting to simply use more tracks. This also means, if the number of meeting days is seen as fixed, this decision is implicitly made by the working group chairs requesting a certain number of sessions and length. Respectively, IETF-111 used 9 parallel tracks for some of the session, but also leading to slightly more scheduling conflicts. As more parallel sessions usually also mean more conflicts, chairs are encouraged to request plenary meeting time carefully but also based on realistic planning to avoid running over time. Use of interim meetings should be consider instead where possible and sensible, as discussed in Section .

Additional Considerations and Recommendations

Full vs. limited agenda (and interim meetings) The IETF-108 meeting survey asked about the structure of that meeting (full meeting) compared to that of IETF 107, which hosted only a limited set of session followed by interims in the weeks after. The structure of IETF 108 was preferred by 82% [https://www.ietf.org/blog/ietf-108-meeting-survey/]. While the limited agenda of IETF-107 could have been a good one-time replacement, the value of cross-participation and high active meetings weeks has been recognised as important for continuous progress (and not only for newly initiated work). A highly concentrated meeting, in structure similar to the in-person plenary meeting, provides value for cross-participants. Further, a well defined meeting time, rather than spreading many interims over the whole year can make deconflicting with other non-IETF meetings easier. While the time during an in-person meeting can be used very intensively, even a compact and full online schedule does often not prevent day-job duties to occur in parallel. Therefore, allocating more time can also make it more difficult for people to join and as such needs to be balanced with the option to distribute load better over the entirely year by a more regular use of interim meetings. Use of (more) online interim meetings can also help to reduce scheduling conflicts during an IETF week and allow for a more optimal schedule for the key participants. Of course these interim meetings are less likely to attract people with casual interest but provide a good opportunity for the most active participants of a group to have detailed technical discussions and solve recorded issues efficiently.

Flexibility of time usage This document recommends that new opportunities in the use and scheduling of online meeting time should be explored that can help to reduce conflicts during the plenary meeting. Online meetings provide an opportunity to use more time more flexibly. While for an in-person meeting all sessions have to be fitted into the available time that people are willing to travel at once (usually roughly a week), online meetings do not have that constraint. Therefore for the planning of online meetings, there is a trade-off between the number of parallel tracks, where more parallel tracks mean more potential conflicts (as least of high-active participants), and the overall time in terms of hours per day or total days used. As one example, it would be possible to keep most regular working group sessions within the usually five main meeting days but have some of the more conflicted sessions in other dedicated time slots. As the Hackathon for online only meetings is usually held in the week before the online plenary meeting , that week is already a highly active week for many IETF participants and might provide an opportunity to schedule a few selected sessions. If only one session at a time needs to be scheduled, it is easier to use a time slot that is well assessable for most people in the community in various time zones. This might work especially well for sessions that are of high interest for a large part of community, such as BoFs and dispatch meetings, and therefore hard to schedule during the main IETF week. An experiment where the adminstrative plenary was scheduled on Wednesday in the week before the official session schedule was run for IETF-112. The evaluation of the respective survey results indicated that this experiment led to reduction in scheduling conflicts but also led to a slight drop in plenary attendance, partly due to insufficient awareness (see [https://www.ietf.org/blog/ietf112-plenary-experiment-evaluation/]).

Chances for inclusivity and Lessons Learnt on socializing Participation at the online only meetings in 2021 was rather high and had a quite stable per-country distribution, even though time zones were rotated. This indicates that online meetings support a more easy and therefore potentially broader participation than in-person meetings where participation is often fluctuating based on the location. However, it has also been recognised that the online meeting does not provide an equivalent opportunity to socialize. The observed slight decrease in submission of new (-00) drafts during 2020 and 2021, while the overall number of draft submissions and productivity seem to stay stable, might also be an indication of the loss of these interactions. The increase in interim meetings potentially compensates for these missing interactions for continuous work (or may even increase productivity there), but seems to be less adequate to spark new ideas. None of the data observed, however, can be interpreted as showing a significant trend. Still, these factors should be considered for the organization of future online only meetings in replacement or addition to in-person meetings.

Experiments Similar as for in-person meetings, it is desirable to experiment with the meeting structure. Often only practical experience can answer open questions. It is recommended to not experiment with a larger number of different aspects at the same time, in order to be able to assess the outcome correctly. It is further recommended to announce any such experiment in advance, so people can adjust to changes and potentially provide feedback.

Acknowledgments Thanks to Brian Carpenter, Lars Eggert, Toreless Eckert, Charles Eckel, Jason Livingood, and Sanjeev Gupta for their review and many from more for their input and suggestions on the time zone discussion!