IDA2020 went virtual: lessons learnt

Due to travel restrictions IDA 2020 had to be moved online. This text summarizes how we set the online conference up, what happened behind the scenes, and what we learnt in the process. 

General Setup

We asked authors to prepare short videos (about 10 minutes) advertising their work, which were made available on videolectures.net. Links to the original research articles, published in open access, and to the videos were posted on the conference webpage a few days before the original conference dates, for everyone to read and watch at their own pace. 
An interactive event was then organised on Discord (discordapp.com), in the spirit of a virtual poster session, where attendees could ask questions and discuss with authors.

We favored an interactive discussion event over moving the original schedule of three days of research presentations online by means of video calls.

Details of the conference program and setup can be found on the conference webpage.

Interactive Event 

Discord was used for the interactive sessions as it allows to have discussions going on in parallel for the different papers, in dedicated channels, with participants moving between them and participating in discussion according to their interests.

DIscord setup: A Discord server consists of a collection of text and voice channels.
Each article was assigned both a text and a voice channel, identified by the article’s identifier as indicated on the conference webpage along with topics, links to articles and video, in a sortable table.
In addition, a pair of text and voice channels were set up for plenary sessions (opening and closing of the event). 
Besides, we had a “reception-desk” that served as a landing page, with a welcoming message giving basic information, an “introduction-hall” intended for participants to introduce themselves to other attendees, and a “helpdesk” to provide support in solving technical issues. 
Finally, we added a number of channels for ad-hoc discussions and socializing (named on the theme of coffee and tea, and of the city where the conference was planned to take place).

The event was split into three discussion sessions, with papers from the same topic spread over different sessions, so that authors from one session could participate in discussions on others’ work.

Different permissions can be granted to users, such as muting all but the designated speakers in the plenary hall for instance, and granting some moderating permission to authors.

Trial day: We held a test session for authors, on the day before the conference. Specifically, authors were invited to log onto the Discord server, make themselves familiar with the application, check their audio settings, and upload additional material to their text channel, such as links to the presentation, article, code, a short introduction, additional plots and figures, etc.

Tasks for the organizers: Four members from the conference organising team took on the role of moderators. They made themselves easily identifiable to participants by prefixing their nicknames with “IDAorg”.
During the main event, one organizer kept a close look at the helpdesk channels to help participants with technical issues. Other organizers were in charge of keeping time, posting reminders to change sessions and keeping an eye on the IDA2020 Twitter account.
The organizing team could communicate and synchronize through a couple of “backstage” channels, which were invisible and inaccessible to regular attendees. 

Overall, close to 120 logins were recorded (a few participants had duplicated logins) and about 90 participants attended the opening and closing of the event.

Feedback survey

At the end of the event, we ran a very quick feedback survey, collecting 51 replies in total.
Here are a few results we find noteworthy:

Overall, respondents rated their experience with Discord with a 3 or higher on a 5-level (Likert-like) scale (41/ 48 respondents). This is a relief, given that only very few of the organizers, and probably few of the attendees, had experience with Discord prior to the event.

We can differentiate between authors and “non-authors” and note that both groups rated their experience during the discussion sessions as high: Non-authors rated the discussion sessions at 4 and 5 (12/13 respondents), authors rated them at 3 or higher (33/34 respondents). Per group there was one very low rating with a 1. 
The discussion sessions were 30 min long. 12 out of 35 responding authors found them too short, 5 out of 13 “non-authors” found them too short. One non-author found the discussion session too long and none of the responding authors found them too long. 
These answers give room for either extending the discussion sessions or having fewer papers per session. We were initially afraid that 30 min would be too long and that authors might get bored in their channels. 

From the open text answers, we summarize:

Things that went well:

  • Respondents appreciated that there were channels for informal discussions, simulating some social activity, although some actions might be needed on the part of the organizer to get socializing going. 
  • The open access to papers, presentations, and the interactive sessions were appreciated. Especially as it gave the participants the possibility to read and listen beforehand, at their own pace and time, then come and discuss.

Points for improvement:

  • Easily actionable comments include switching off sounds by default for attendees (cannot be done at the server level, instructions were provided on how to mute notifications) and having fewer papers per session and more sessions overall.
  • Many respondents missed a video or screen sharing feature in Discord. In principle, Discord allows screen sharing (going “live” is what the platform is made for) but with limits on the number of attendees. We had too little time to test video functionalities and be confident it would work fine. Another option would be to embed Youtube or Twitch broadcasting streams in a channel.
  • Only one respondent wrote that they would have preferred “live online presentations”. 
  • We used videolectures.net to host the videos. It is a curated platform, which also means that it takes a small delay for the presentations to become available online after they are uploaded, as they undergo basic quality checks. Future conference organizers might want to set an earlier deadline to allow for a few more days between upload and interactive event. 
  • An interesting suggestion was to include some profile information for every participant. This was the intended purpose for the introduction channel, but it did not really succeed. Many participants added a picture, which made it less anonymous. Future organizers might want to host an informal Discord session a day before the conference or find a way for participants to personalize their profiles (independent of which platform is used).

Tips and tricks from the organizers’ perspective:

  • The trial day is fundamental, for both authors and organizers, and especially among less experienced users of the platform, thus stressing the importance of testing the rooms and structure on the server. It would possibly be a good idea to open the server doors some hours before the discussion time with activities to make non-authors familiarize themselves with the platform.
  • Keynote lectures and screen sharing can be organized via live screening on Twitch or Youtube from a video call. 
  • Prepare activities for the informal discussion rooms, such as ice-breaker games or discussion topics, to make it easier to start a conversation in the digital environment. It works to have a small group of participants instructed on how to get the ball rolling, for example by being the first ones to introduce themselves in the introduction channel (no one likes to introduce themselves to an empty text channel).
  • Overall, there were no major technical issues, no trolling, everything ran fairly smoothly, to the relief of the organizers.

Notes on surveys and signing-up: A balance needs to be found between being entirely public and accessible to anyone and being more private, requiring pre-registration and having more options to interact with participants after the event. 
Note that login into Discord does not require an email address.

Overall, we believe that the combination of pre-recorded presentations and a compact interactive online event allowed a successful dissemination of science at the IDA 2020 remote conference. Translating, to a satisfactory extent, socializing and networking aspects of IDA into a remote setting requires further exploration of available means and tools.

Summary by the online conference squad: 
Daniela Gawehns 
Esther Galbrun
Hugo Proenca