Different Times
I spent the last few months working as a research assistant in artificial intelligence at the Frankfurt Institute for Advanced Studies. With the global COVID-19 situation almost all parts of the scientific workflow went digital. I was working from home, logging into the office computers and high performance clusters via SSH, but all in all this was not that much different than programming and logging on from the office. One thing that has changed significantly, though, is conferences.
As large gatherings of people are potential COVID hotspots it totally makes sense for conferences to go virtual, … at least for this year. And there is tremendous potential in having virtual events: Talks get recorded more frequently, allowing for a more flexible schedule; pre-recorded talks make sure that everyone keeps within their time-slot; it’s more inclusive for people who otherwise cannot afford to attend and last but definitely not least it’s better for global climate for not having people fly all over the world.
I firmly believe that the academic world should learn from this special situation and incorporate some of the up-sides of having virtual events into future conferences: Maybe an alternating schedule (virtual/in-person) would be a good idea. But right now I feel everything is going to go back to as it was before.
Why do I think this…? Well, I think people enjoy being with each other and we have not figured out how to do the networking part conferences yet. For me it becomes most apparent with poster-sessions. Sure you cannot drink a beer at bar with your colleagues when you are stuck behind your webcam at your desk, but that was never going to be. The poster-session format is partly social interaction with a presenter and it fails badly at virtual conferences, and actually there is no real reason why.
Fixing Virtual Interaction
So there might be conferences that actually do what I suggest next, so I apologize beforehand, but the three conferences I attended this year either scrubbed the poster session and made posters downloadable or made them short 5 min prerecorded talks. I understand the notion that this is the easiest way but I am kind of disappointed that we have not figured out something better.
In my opinion there’s actually two sides to the problem: conceptualization and software.
You have to think about what makes a poster-session a poster-session. How does it work and why does it work? Inspire from the real world and take it to the virtual one.
Start with a showcase. There needs to be a website where you can visually scroll through all of the posters. This is a no-brainer: don’t just go with abstracts and links and no, it doesn’t have to be a fancy 3D virtual floor. Just make it a long scrollable website with BIG images. People have spent time designing these posters, this is what catches our attention at the poster session. Have little hearts or thumbs-up icons next to the images to indicate whether some poster creates enthusiasm among the participants.
When you click on a poster you should be pulled into a virtual conference room, like a zoom call, where the presenter talks about her poster and answers questions. Now here comes the software part: Simple screen sharing is not enough, we need a simplified powerpoint for posters/documents. Conceptually it should be like prezi, because the poster is the whole canvas, but there should be simple tools that analyze the PDF and make it easy to highlight regions like boxes and round-rects and of course you should be able to pan and zoom around. Moreover it should not be a video-stream! Something like this needs to be rendered client-side. These posters are smaller than your average hd video and they are vector-based, making them effectively resolution independent. With client-side rendering the participants can decide for themselves on which parts of the canvas they would like to focus.
To end this little blogpost I want to showcase what I came up with for this year’s ESANN conference. Let me be clear, this was way more work than a simple poster. But given the right software it should be feasable to transform any PDF into something where you can zoom around and highlight passages in a similar manner [1].
On a little side-note I’m still on the lookout for a cool PhD position, so please get in touch if you want to work with me!
[1] If you liked that video, you can of course take a look at the corresponding paper:
Ernst M.R., Triesch J., Burwick T. (2020). Recurrent Feedback Improves Recognition of Partially Occluded Objects. In Proceedings of the 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN)