Thoughts on Completely Virtual IETF Meetings
Internet Society
Keene, NH
USA
york@isoc.org
Internet
manycouches
This document captures initial thoughts about having IETF meetings that
are completely virtual. It explores the issues involved with both a
"planned" virtual meeting and an "emergency" virtual meeting. The intent
is to evolve this document to provide answers to the questions posed
throughout the text. This is currently a thought experiment. There are
no current plans to hold a completely virtual IETF meeting.
What would a "completely virtual" IETF meeting look like? What would be
issues? What would be the advantages? How could it work?
The "manycouches" design team was convened to explore these issues and
understand what might be involved in holding a completely virtual meeting.
On 20 July 2016, members met with the IESG for a joint discussion at the
IETF 96 meeting in Berlin. The team met again at IETF 97 in Seoul.
This document outlines many of the key issues and questions for discussion that emerged out of those meetings as well as mailing list conversations.
Discussions identified two types of potential meetings the IETF could have
that would be completely virtual:
PLANNED VIRTUAL MEETING - A "regular" meeting of the IETF that would be planned to be completely virtual.
EMERGENCY VIRTUAL MEETING - There could be a situation where a planned physical meeting suddenly needs to be virtual due to physical or political situations. For example, a natural disaster shortly before a meeting might cause people to not be able to attend.
Tools and processes may be very similar between the two types of meetings.
A key difference is that for an "emergency" meeting there may be the desire
to replicate the planned schedule of the physical meeting as closely as possible.
It is unclear if the IETF might ever choose to hold a planned virtual meeting, but this document is designed to facilitate the discussion around
what that might look like. A desire is that some of this development may
help with improving the current experience for remote attendees to today's
physical IETF meetings. It may also be the case that some kind of "hybrid"
meeting emerges with physical meetings taking place in multiple locations
with virtual participants joining in remotely.
It is also worth noting that in discussions to date the sense has been
that even if we held a completely virtual meeting, it would only happen
once out of several meetings. There would still be multiple physical IETF
meetings during the year.
[It may be good to insert some text here about WHY we have IETF meetings
and what the overall goals are. Both as a reminder of the point of the
meetings and potentially to frame thinking about how we might move toward
those goals by trying doing things a bit differently.]
[At some point in the maturity of this document it would be valuable to
discuss the benefits and challenges of a completely virtual meeting.
Before that summary can be developed, though, further investigation and
development needs to happen. At a very high level, one idea is that a
completely virtual meeting might make the meeting more accessible to
more people in terms of schedules, lack of travel and reduced costs.
However, all of that thinking need considerably more exploration.
Several participants in discussions have voiced the opinion that
replacing physical IETF meetings will be close to impossible. This
document is being developed to explore all of these issues.]
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 .
Additionally, the key words "MIGHT", "COULD", "MAY WISH TO", "WOULD
PROBABLY", "SHOULD CONSIDER", and "MUST (BUT WE KNOW YOU WON'T)" in
this document are to interpreted as described in RFC 6919 .
With a completely virtual meeting, the structure of the meeting does not
have to comply with the traditional IETF meeting schedule. It could, for
instance, stretch out over the entire 24 hours of a day. Questions for
discussion include:
Is the meeting still structured over a week?
Do the meetings still exist within certain hours?
Do multiple meetings exist at the same time as they do now?
Again, in the case of an unplanned "emergency" virtual meeting the desire
may be to stick with the already-planned schedule. But for a planned
virtual meeting the schedule can be open for discussion.
There was some discussion that a meeting could span more than the
traditional week. However, the counterpoint is that keeping it within
a week gives a focused block of time that people could allocate for
participation in the virtual event.
What timezone does a virtual meeting operate in? Or does it operate in
multiple timezones?
One suggestion was that each working group might choose its own timezone
based on the best timezone for the main contributors and leaders. (Although
this might then limit participation from other areas of the world.)
This timezone issue was identified by multiple participants as the hardest
aspect of planning a virtual meeting.
What do deadlines look like for a completely virtual meeting? Are the
deadlines for agendas and drafts kept as they are for a regular meeting?
What does a plenary look like in a virtual meeting? The same large session
as today?
There are breaks planned throughout the days of a physical IETF meeting
to enable people to move between sessions and to have refreshments and
restroom breaks. Similarly there are longer breaks for lunch.
How are breaks structured in a completely virtual meeting?
Can they be shorter?
Do you need a longer "lunch break"? Or does that make no sense?
On the Sunday starting an IETF week we commonly have a series of tutorials.
Are those still part of the program for a virtual meeting?
The Hackathon and Code Sprint have become popular activities before a
physical meeting. Would they still exist for a virtual meeting?
In a typical IETF physical meeting, there are other meetings and activities
that occur alongside the meetings of Working Groups and BOFs. These include:
Newcomers Meet and Greet
Welcome Reception
Bits-N-Bites
Thursday Lunch Speaker Series
Social Event
Do any of these additional sessions still make sense in a virtual meeting?
The Thursday Lunch Speaker Series (by the Host organization) could continue
as webinar-style presentations. But the other elements involve face-to-face
interaction that would be difficult in a virtual setting.
Other organizations sometimes hold meetings during the time
of the IETF physical meeting and often use the same venue. For example,
the Internet Society usually offers an "ISOC@IETF Briefing Panel" during
the Tuesday lunch break. At some IETF meetings groups have shown films
or scheduled other meetings. These are not IETF meetings, but make
use of the opportunity of having the IETF attendees available.
Do these sessions still make sense? Would we offer the IETF virtual
meeting infrastructure to groups to use when it is not being used for
IETF meetings?
In recent years there has been an effort to establish "remote hubs"
where groups of IETF members get together and participate remotely from
that physical location. Would that continue as an option?
Could the virtual meeting perhaps involve connecting together a series
of remote hubs? (And if so, does this then again create a better
experience for people who can go to a hub than for those who cannot?)
What is the experience of an "IETF attendee" in a virtual meeting?
How does he or she experience the event?
How could attendees be most effective in getting work done in a virtual
setting?
It was noted that remote attendees should think about how they block off
time needed to participate in the meetings. This may be challenging depending
upon timezones and other activities.
An open question is whether attendees might be able to get time from their
employer to participate in the virtual meeting. If you fly somewhere to
participate, it is clear that you are "away" and participating in the meeting.
If you are still at your home or office, it is harder for others to consider you
"away".
What is the registration experience like? How do they initially
"sign in" as an attendee?
It is quite common for groups to decide during an IETF meeting to go
off and have a side meeting.
How can this capability be reproduced in a virtual environment?
Could the system allow people to create ad hoc meetings in some fashion?
The casual hallway conversations are a key component of IETF physical
meetings. How can some version of this capacity be made available?
How do you incorporate some concept of "unstructured" time where people
can meet and connect?
It is currently possible for remote participants to join into multiple
working group sessions at the same time. Users simply connect using
multiple browser windows, multiple chat rooms or multiple computers.
How does this impact users' experience?
Part of a physical meeting involves discovering other people with common
interests or backgrounds. How do you help people find others?
So much of the relationship-building that helps get work done happens through
the informal side meetings, going out to dinner, going off in groups. How can
any of this be replicated remotely?
Could there be a way for users to be able to share when they are going to be in different sessions? Or when they would be available to "hang out" virtually?
What level of "presence" could be made visible for virtual attendees?
What is the best way to have votes or hums in a virtual meeting? Most current audio conference systems would not make an actual audio hum
possible. Votes in a chat could be possible but the lag time associated
with remote connections would need to be taken into account.
Some kind of system where votes take place over a period of time may need
to be developed or used. This, though, does then introduce a delay into
the meeting while there is a wait for the vote.
How do "mic lines" work in a completely virtual meeting? Would this in fact
be a benefit as all attendees would be in the same queue?
How do we deal with disruptive behavior in a virtual meeting? It can and
does happen in meetings - and could potentially happen more easily in a
virtual evironment where people cannot be physically stopped from going to
a mic or could be removed from a room.
What is the process to exclude someone who is being disruptive? Do we need
moderators to be able to step in and mute or disable someone's connection?
Who makes the decision that someone's behavior is disruptive?
How would the "mentor" program work in a virtual meeting? The same as with
a physical meeting?
How do you bring new people into sessions? How do people learn about side meetings? About hallway conversations?
Many attendees value the T-shirts that are usually provided for each IETF.
Without a physical meeting it could be challenging and costly to distribute
T-shirts to attendees.
T-shirts are currently funded by the Host of the physical IETF meeting. If
there is no Host, or if the Host chooses not to fund a T-shirt, there may
be no T-shirt.
With a virtual meeting it may be that if there is a Host (see "Sponsorships"
below), the Host would have the same option as the physical meeting - to
provide a T-shirt or not. The Host could then decide how they would
distribute the T-shirt.
Many technical questions need to be discussed.
What is the infrastructure used to host a completely virtual meeting?
Are current systems (ex. Meetecho, Jabber chat rooms, audio streams)
sufficient? Would new infrastructure need to be established?
What kind of bandwidth would need to be available for the servers
hosting the system?
How would we handle connecting large numbers of people at the same time?
Do virtual attendees have video connections? voice? chat? What kind of
bandwidth would need to be available on the client end?
Recommendations should be developed for client-end infrastructure.
(To fully participate you need X, Y and Z...)
Virtual attendees need to have some kind of "backup connection" in case their
main Internet connection goes out. For instance, a PSTN connection for calling
into a session. (This implies that the system hosting the virtual conference
can accept connections through different mechanisms.)
Whatever system is used should have some kind of "persistent chat" so that
when people connect into a given "room" they can scroll back and read through
the history. Potentially that history might also include audio or video links.
Today anyone can connect to the remote participation aspects of an IETF
meeting. No authentication is required to join a jabber chat room, listen
to an audio stream or connect to a Meetecho session. Would that need to
change? Would "registration" give you a login to whatever system was
used for the meeting? Would you not be able to participate without those
login credentials?
How do we address issues of lag, stutter, echo and other artifacts of
current audio conferencing systems?
Is there a "minimum voice quality" level that is acceptable? (George
Michaelson has suggested the telco QDU concept is something to consider.)
Where does the NOC "exist" for a completly virtual meeting? What is its role?
The long-term impact of an idea such as this needs a great deal of further thought.
What is the impact of a virtual meeting on centralized resources such as
support staff? What is the full role of the Secretariat during the meeting?
The financial model of a completely virtual meeting needs to be understood.
What would be the financial costs associated with a meeting?
Would there need to be an initial investment in infrastructure for the
first completely virtual meeting? Would there then be lower costs for
the next virtual meeting?
Would we charge the same amount to attendees as a regular meeting?
Lou Berger sent the following suggestions to the list related to
registration fees:
Remote audio feed and jabber participation should continue to be unpaid and unregistered as now
Access to session audio and video recordings should continue to be published as now, without fee or registration
Remote video/audio - registration should be per individual participant (i.e., anyone that speaks/presents) perhaps having hubs include some number of participants.
Non-registered/anonymous video (meetecho) listeners should be
allowed, but their mic/text input should be disabled.
How do sponsorships work with a completely virtual meeting? Would
sponsorships be required at the same level as the physical meetings?
If a virtual meeting is sponsored, how is the sponsor given the visibility
that is currently given with a physical meeting? For instance, with the
signage, T-shirts, plenary slides, etc.
In particular, is there a sponsor designated as the "Host" of the IETF
meeting? The Host for physical meetings receives benefits including:
Prominent mention in materials and promotion of the IETF meeting.
The "host presentation" speaking slot during the plenary.
The "Thursday Lunch Speaker Series" time for whatever they wish to present.
Mention on the T-shirt for the event (if the Host chooses to fund the creation of a T-shirt).
Would we have a "Host" for a virtual meeting?
If we were successful in holding a completely virtual meeting, would
companies no longer be willing to send attendees to physical meetings?
In other words, would the first one start us on a path toward having
all meetings in this fashion? (And are we okay with that?)
How do we ensure all attendees, coming in at all times, see and agree to
the Note Well statement?
There are many considerations related to security and privacy that need
to be factored in to a virtual meeting.
How do we ensure that an attack such as a distributed denial of service (DDoS) doesn't take out the entire virtual meeting? What about an attack against a particular region?
Similarly, how do we protect against disruption caused by groups on
the Internet who may simply want to disrupt the meeting for the fun of it?
(See the section on "Authentication" earlier.)
How do you know that the person who is logged into whatver system is
used is in fact who they say they are? In a physical meeting:
We can see the person and physically identify them.
Users wear name badges that were issued at registration time.
There are typically other people who may know many individuals.
How are these physical considerations replicated in a virtual meeting?
What level of privacy protection would be needed for conversations? for
user information? Much of the IETF's work is all done on public email
lists and archived remote sessions. What level of privacy is needed?
Are there any IANA considerations associated with a virtual meeting?
With this initial document published, the intent now is to go back and
start to fill in the sections with possible ideas about how the questions
might be answered.
Suggestions were made to investigate what lessons can be learned from
work by other organizations on virtual meetings. Initial suggestions
included:
The Internet Society has now hosted two (2015 and 2016) global "InterCommunity" events bringing together ISOC members from around the world. The 2016 event, in particular, was designed to be a virtual event.
The conference industry has been exploring virtual and/or "hybrid" meetings. There may be value here.
Universities and specifically Internet2 may have some experience.
George Michaelson stated: "Van Jacobsen did a lot of work on meeting behaviour online in the MBONE days, working on the whiteboard and vat. He has made observations about weighted-sum voting, speaking controls, inheritence of the state of the meeting."
How would it be possible to do a "trial run" of a virtual meeting?
This document reflects the input of many people who participated in both
the manycouches design team as well as the discussion with the IESG
on 20 July 2016 at IETF 96 in Berlin. Another discussion was help among
design team members on 17 Nov 2016 at IETF 97 in Seoul. Other discussions on the
manycouches mailing list also informed this document. The author would
specifically like to thank Lou Berger, Benoit Claise, Stephen Farrell,
George Michaelson and Greg Wood for their input.
This document is being developed using a repository on Github at:
Comments, issues and pull requests are welcome.