Integrability in Gauge and String Theory

Hybrid conference setup: thoughts & resumee

Hybrid setup for workshops and conferences: a step towards a sustainable scientific practice?

Living and working in a globalized world poses critical questions towards building a sustainable scientific future, both sociologically and ecologically. Here, we would like to share thoughts and ideas from organizing a medium-scale hybrid scientific conference, “Integrability in Gauge and String Theory,” at ETH Zurich in Summer 2023. This document is structured into three parts: general considerations, practical technical implementation of the hybrid setup, and a brief summary of lessons learned.

Organizing a hybrid international scientific conference

Hybrid conferences can help making progress in two directions. They allow to reach a wider audience of scientists, who would otherwise not be able to travel to the conference due to financial, caregiving, or political constraints. Simultaneously, enabling a meaningful and communicative remote participation relaxes the imperative to travel, and therefore facilitates a reduced CO₂ footprint.

Planning

• From the very early planning stages, we tried to include a hybrid-first approach rather than organizing a regular conference and afterwards retrofitting remote participation options. This mindset proved to be a huge advantage for seamless and meaningful integration later on, creating a substantially different conference experience for all participants.

• A conference fee was charged for on-site participants to cover the social activities. We decided to also request a token fee of CHF 30 for remote participaction. This slightly unusual decision within our research domain (video streaming, if provided at all, is typically offered free of charge) was driven by our commitment to providing a high-quality and inclusive hybrid event, while also supporting the extensive hardware setup required to deliver a meaningful hybrid experience, as detailed later.

Communication and advertisement

• We tried to convey the goal of organizing a hybrid conference for reduced CO₂ footprint and enhanced accessibility consequently, along with the scientific direction of the conference. This included clear explanations of the distinction between hybrid and conventional formats in advertising from the very beginning.

• On-site and remote participation were presented on equal footing in each advertisement channel. A significant effort was made to explain the unique features of the hybrid component, in particular our two-way communication options, on the conference poster, website and in eMail advertisements.

Presentations

• Both, talks and posters were designed to engage on-site and remote participants on equal footing as far as technically possible. We found discussion hubs – computer terminals with screen, camera and microphone – an extremely valuable tool to bridge the gap between these two groups. Discussion hubs can transport video and audio from on-site participants gathered to remote participants. More details on our discussion hubs are provided in the technical section below.

• Presentation slides and posters were made available on the conference webpage prior to the talk, ensuring accessibility for all participants regardless of the participation mode of presenters and participants.

• For conventional on-site talks, we extended the classical talk setup by a) making the presence of remote participants visible and audible to the on-site audience, thereby allowing remote participants to raise hands and to ask/interrupt. b) making all language (including questions and comments) audible to the remote audience.
  The conveners of each talk session were asked to monitor questions and comments from both audiences (on-site and remote) equally. To highlight and show the pointer as well to a remote audience, use of a digital pointer was unavoidable. For possible blackboard additions, a tablet was installed for the on-site presenter, whose screen was broadcasted to the remote audience and shared in the conference venue. In order to allow for informal discussions after the talks, on-site speakers were asked to be reachable simultaneously for on-site and remote participants at our discussion hubs.

• In a similar way, we tried to bring remote talks closer to the on-site audience. Next to the remote slides, we displayed a large scale video of the face of the speakers to the on-site audience on a separate large screen in the conference venue. The conveners (present in the conference venue) were asked to recognize questions from the on-site audience, alert the remote speaker, and to make sure that the questions asked in the audience are distributed through the audio system and so consistently transported to the remote speaker and audience. After the talk, the remote speakers were asked to be present again at our discussion hubs for informal discussions with the remote and on-site audience.

• We requested each speaker to structure their talk into a 10-minute introduction followed by an XX-minute main part, with a short “review break” after the introduction to allow listeners to ask basic questions. Even though not primarily meant to support the concept of a hybrid conference, this approach effectively engaged remote participants, providing them with a natural opportunity to unmute and participate. During this mini-break, moderators were reminded to remain attentive to both in-person and remote audiences, ensuring inclusive interactions.

• An on-site poster session with printed and displayed posters presented by their on-site presenters was supplemented by presenting the remote posters posters to the on-site audience at the discussion hubs (in shifts).

• Following the on-site poster session, a remote poster session was held, where all on-site and remote presenters discussed their posters in dedicated Zoom breakout rooms.

• Each poster presenter was invited to advertize their poster in short one minute videos, which were displayed to both audiences in groups of 5 before the initial talk sessions. This gave remote poster presenters a better chance to attract interested participants for interactions during their poster session.

Informal discussions & coffee breaks

• Discussions and informal chats during coffee breaks are crucial elements of a conference. We attempted to provide an online communication channel where the remote partcipants can gather and casually meet some of the on-site participants with minimal efforts and making sure the opportunity blends in well with the established coffee break situation. We used the online service Spatial Chat and installed a large touchscreen computer with sound, camera and microphone near the coffee supply. Unfortunately, we did not succeed to muster the on-site participants’ attention/curiosity to even give the system a try. The discussion hubs proved more effective for fostering conversations albeit mostly on the scientific topics.

Conference schedule layout.

• The choice between remote and on-site participation was left to each individual participant. However, providing guidelines or best practices can help participants make environmentally conscious decisions. To support this, we included detailed information about both participation options in our advertisements and invitation eMails, enabling attendees to make informed and thoughtful choices.

• Concerning travelling, we followed the strategy to frame the conference as four days of scientific program and a “free-floating Friday”, which could either be used to extend the interactions or to travel before the weekend. We communicated this aspect clearly, so participants could plan accordingly.

Technical implementation of the hybrid conference “Integrability in Gauge and String Theory 2023”

IGST‘23 offered both in-person and remote participation options. While conventional arrangements were provided for in-person attendees, remote participants were accommodated through a single Zoom session. Our goal was to make the remote experience as close to the in-person participation as possible. This required an elaborate system involving multiple Linux-machines communicating via SSH-protocol within a virtual subnetwork.

Hardware

The hybrid setup for IGST‘23 relied on six computers (laptops), each supplied with a set of input-output devices (video cameras, microphones, projectors and screens). Each computer was connected to the Zoom session with separate accounts under appropriate names.

• The presenter computer. The presenter computer was connected to the main projector, a tablet with a touch-screen, a video camera and the lecture theatre sound system (including conference venue speakers, speakers microphone and a set of additional microphones). The projector displayed the presentation of the speaker, while it was also shared in the Zoom session. If the speaker was present in-person in the conference, the touch-screen tablet could be used to emulate a "whiteboard" talk that could be seen by both on-site and remote participants. The video camera provided the remote participants with a picture of the speaker. The main sound system amplified the speakers voice for the on-site participants and was connected to the Zoom session to broadcast the speakers voice to the remote participants. If the speaker participated remotely, the speaker simply shared their screen in the Zoom session that was displayed by the projector. Furthermore, the portable microphones in the auditorium could be used by the moderators and the on-site participants to communicate with the speaker and the remote participants.

• The auditorium computer. To enable remote participants not only to follow the talk and observe interactions between the speaker and on-site attendees, but also to actively engage, an auditorium computer was equipped with an additional projector and a video-camera. The projector displayed the remote-participants’ video on a secondary screen while also capturing a view of the conference venue. This setup allowed on-site speakers and remote participants to see and interact with each other directly during and after the talks. When the speaker was remote, their video was displayed on the secondary screen for the on-site audience.

• Two discussion hub computers. Conference talks are typically followed by a discussion between the speaker and a subset of listeners. To ensure remote participants could actively participate in these post-talk discussions, two dedicated discussion machines (laptops) were set up, each equipped with a Meeting Owl and a large monitor screen. Meeting Owl is a compound device encapsulating 360-degree camera and microphones that can detect the direction of the source and capture its audio and video. This provided remote participants with a clear view of the discussion in the venue. The large monitor displayed the speaker’s slides, which were also shared over Zoom, while the laptop’s built-in screen showed the video feeds of the remote participants. This setup facilitated the discussion between on-site and remote participants, whether the speaker was present on-site or joining remotely.

• Coffee-break communication system. Various discussions between the participants often take place during the breaks. To give remote participants an opportunity to socialize with both each other and in-person participants a Surface Pro communication system was used. This system, a 50-inch touch-screen computer equipped with cameras and microphones running Linux, was connect to a Spatial Chat session created for the conference. Remote participants could join the virtual environment via Spatial Chat, while on-site attendees could use the Surface Pro to interact with the remote attendees.

• The "control" machine. Control of the machines listed above and the Zoom session was centralized through a single "control" machine with three screens, which were crucial to keep an overview of the session and all participating devices. The operator of the control machine was able to manipulate all connected devices remotely as well as moderate the Zoom session. To achieve efficient and smooth control of all the machines simultaneously, specialized software was developed.

Software

The setup and coordination of the six computers were facilitated by a set of bash scripts that could

• install and appropriately configure Debian Linux OS on the machines with all necessary additional software;

• configure external devices to be connected to the machines for the conference;

• synchronize the machines with an up-to-date version of the scripts;

• communicate the IP-addresses of the machines to a server for subsequent SSH-access.

The IGST’23 conference sessions were composed of back-to-back pairs of talks followed by coffee or lunch breaks. Each session was outlined as follows:

• (Announcement of) Poster advertisements.

• Announcement of the talk by a moderator.

• First talk given by a remote or on-site speaker.

• Second talk given by a remote or on-site speaker.

• Two parallel discussions with the two speakers using two discussion hubs.

• After-session break.
  
  

It was important that the transition between different parts of the session was quick and smooth with little to no (apparent) interference from the side of organizers. As the interface between remote and on-site participants relied on a single Zoom session, it was crucial to control various aspects—such as Zoom windows, presentation files, audio-visual settings, discussion hubs, presenter and auditorium computer—efficiently from the control machine. This was achieved using a set of bash scripts complemented with a command-line control tool over the configuration of the machines and the main features of Zoom. The functionality of the scripts included

• starting Zoom and logging in to an appropriate Zoom-account based on the role of the machine;

• manipulating location and sizes of the Zoom windows on the screens and projectors;

• controlling volume of speakers and microphones and (un)muting/turning on(off) sound and video;

• opening, closing and sharing presentation files for the speakers;

• opening and controlling software for white-board talks;

• controlling the recording of the Zoom session.
  
  

As a backup to resolve unforeseen problems, each computer ran a vnc-server. Beyond the possibility to troubleshoot, another advantage of the vnc-server is the option to constantly monitor the screens of the various machines of the complete hybrid setup from the control computer.
For the different parts of the session, preset configurations for Zoom windows, audio, video, presentation files, etc. were created. These presets could be applied remotely from the control machine with a single SSH-command, making the whole process smooth and quick. Additionally, a graphical interface for the bash-scripts was developed to simplify the control routines.

Resumee of the hybrid conference “Integrability in Gauge and String Theory 2023”

Many of the above aspects were highly praised by participants, in particular the discussion hubs, the review breaks and the four day scientific programme format. The flexibility to choose between on-site vs. remote delivery for talks and posters enabled several remote presentations that would not have been realised on-site due to restricted availability of the presenter. Only a few elements received mixed feedback. These included the online conference fee and the digital pointer device (which, while effective, required prior instruction and practice). Not too surprisingly, the purely vegetarian food supply was appreciated almost unanimously. The offer of a hybrid social interaction channel, using the tool Spatial Chat, was only used after substantial encouragement, as on-site participants were largely engaged in in-person coffee break discussions.

The hybrid format demonstrated a positive step toward reducing CO₂ emissions associated with conference travel. An evaluation of participant affiliations displayed an encouraging picture. The conference hosted:

• on-site participants: 42 (short distance), 19 (medium), 11 (long), and 9 (overseas).

• remote participants: 14 (short distance), 14 (medium), 6 (long), and 17 (overseas).
  
  

These numbers show that there is a clear tendency for nearby researchers (train travel distance) to attend on-site (despite high accommodation costs in Zurich), while overseas researchers opted for remote participation. This preference likely reflects the significant time and financial costs of overseas travel, with ecological awareness also possibly playing a role.

With these numbers and the total distances travelled, one may estimate the greenhouse gas emissions of the 81 on-site participants at around 70t CO2 equivalent. Conversely, if the 51 remote participants had attended in person, their travel would have resulted in 100t CO₂ equivalent emissions, a significantly higher figure for a significantly smaller group. While not all of these emissions would have been realized in a fully on-site conference, the hybrid format allowed a broader audience to hear about recent advances in this research field and to participate in discussions of their scientific content.