On April 29 we circulated a survey to help us plan for the possibility of upcoming in-person meetings being converted to online meetings. We are continuing to analyze the survey data in more detail, but we have made some preliminary observations from the survey responses to guide decisions about meeting planning that we felt needed to be announced sooner rather than later. In particular we have treated all the answers as equal rather than weight results by any factor such as region or likelihood of participation, and we have not attempted to calculate the margin of error for any responses.
In total, 576 people responded to the survey, of which 97% are current subscribers to an IETF or IRTF mailing list and 54% are authors of active Internet-Drafts. Over 80% intended to participate in at least one in person meeting this year had it not been for COVID-19. 45% of respondents were from North America, 33% from Europe, 14% from Asia, and 4% or less from each of Oceania, Latin America and the Caribbean, Africa, and the Middle East.
Unsurprisingly, the survey data shows that regardless of how we structure the online meeting there will be some dissatisfied participants and some who find it very difficult to participate.
We are planning a 5-day meeting with approximately 5 hours of sessions per day running from 11:00 to 16:00 UTC each day. Sessions slots are currently anticipated to be 50 or 100 minutes in length, shorter than normal for an in-person meeting, with 20-minute breaks in between. We plan to run up to 8 tracks in parallel as we normally would at an in-person meeting. We chose this format after preliminary analysis of the survey data, as described below.
Question 11 (p20) asked about meeting options in the event that an in-person meeting needs to be cancelled. Two thirds (66%) of respondents preferred replacing the in-person meeting with an online meeting with a set agenda, with or without additional virtual interims. Based on this feedback we chose to schedule an online meeting with a set agenda.
Question 15 (p51) asked about how easily respondents would be able to participate in an online meeting during different blocks of time during the day. For our respondents, the times with the most ease and least difficulty of participation occur between 12:00 and 22:00 UTC. We also found notable the frequency of survey comments about the difficulty for respondents in Asia to participate in IETF 107 and the virtual interims that have occurred since then, since many of these meetings have been clustered during later UTC hours.
Question 16 (p55) asked about approaches to choosing the time zone for an online meeting. The two most popular options were to poll potential attendees and to use the time zone of the cancelled meeting.
Based on the responses to the two questions above, we chose to schedule the online meeting beginning at 11:00 UTC. This allows us to use a block of hours largely within our respondents’ window for ease of joining and aligns with workday hours in Madrid’s time zone. We anticipate that if IETF 109 converts to an online meeting that we would align to a block of workday hours in Bangkok’s time zone, unless we gather data indicating a better approach or we receive consensus guidance from the community that indicates otherwise before then. This approach aligns with the principle of sharing the burden of meetings over time, as set out in RFC 8718.
Questions 17 (p60) to 20 (p67) asked about respondents’ preferences concerning the meeting format. The responses here were somewhat polarized. The two format options in Question 17 that were rated easiest to join, and by a notable margin, were a 5-day format with a 4.5-hour meeting day and a 10-day format with a 4-hour meeting day. We recognize that in 32-34% of responses to these options participants indicated that they would find joining a meeting in this format difficult or very difficult, but those were the lowest totals for any of the options.
When asked about the individual dimensions of the meeting format (questions 18-20), 4- or 5-day options were slightly preferred over 8- or 10-day options, 4-6 hours per day were preferred over 8 or 10 hours, and 50-80 minute sessions were slightly preferred over 60-120 minute sessions. We also took note of comments discussing the ease or difficulty of adjusting one’s own sleep schedule to accommodate a meeting over smaller or larger number of days, as well as support in the comments throughout the survey for longer session slots. On the basis of all of this feedback, we are planning for a 5-day meeting with approximately 5 hours of sessions per day using a compromise combination of 50- and 100-minute slots. We have opened session scheduling to Working Group and Research Group chairs. Depending on the set of requests we receive, we may adjust this plan. If there is demand for longer slots, we may combine consecutive shorter slots to create longer ones.
We are continuing to analyze the survey data and plan for many other aspects of the meeting experience. We will provide updates as more information becomes available about the IETF 108 meeting.