Incentives are changing the effectiveness of computer science conferences, but perhaps better conference design can improve them.
A central driver of academic innovation is the dissemination of ideas and academic conferences are a main vehicle for such dissemination. When we spend several days together in person, attending talks, poster sessions, and informal conversations we learn about advances in our area of research and also adjacent areas. Back in our offices, we propagate these learnings in our research network and important results become known. Moreover, a critical egalitarian feature of the computer science conference culture is that anyone can write a great paper, submit it to a rigorous (and now typically) double-blind review, and earn the opportunity to present it in front of the field's experts.
These advantages of conferences are being challenged by a shift in preference of some authors to virtual participation. Here are the main incentives at play:
Due to the implication (1) => (2) => (3), many conference organizers have opted to shut down (1) virtual participation by paper presenters, thus returning how conferences were before the pandemic. Other organizers have allowed (1) and have experienced the negative impact on (3). It would be great to be able to fix (2) by incentivizing virtual participation at the same level as in-person participation. This would be great because we could perhaps then obtain comparable interactions and dissemination at a fraction of the cost (time and money). Unfortunately, fixing (2) has proven challenging. The following compromise proposal allows (1) without compromising (3) and fixes the incentives for paper presenters who might be inclined to present virtually and free-ride on the efforts of others to disseminate their work.
Proposal: Run separate in-person and virtual meetings, granting authors of accepted papers the choice of which meeting they prefer to present their paper in.1
1. The in-person meeting happens first, and presentation in the in-person program requires in-person attendance, talks are live streamed to virtual attendees (but virtual attendees cannot present).
2. The virtual meeting happens second (perhaps a month later, perhaps six months later to bring the community together off-cycle, or perhaps entirely asynchronously with authors only submitting videos for the video library), and includes virtual presentations of all papers that were not presented in person.
This proposal has some good features:
This proposal has two unfortunate features which would be great to improve on.
The first is that authors with visa-denials are unfairly relegated to the virtual venue, which is less good for dissemination. To be more equitable with visa-denials, if (1a) and (1b) can be honestly separated, the few visa denials could be allowed to present virtually during the in-person meeting. Alternatively, authors with visa denials could be given the option to present their paper in-person in the conference's next edition or a sister conference in a country that they can access.
The second unfortunate feature is that preferences for in-person attendance may be non-uniform across subcommunities, resulting in more separation between these communities and less mixing of ideas and possibility for forging connections across them. This issue does not seem possible to circumvent without mandating in-person attendance, and if the above proposal is executed on, it would be good for conference organizers to collect the statistics on whether some subcommunities are choosing virtual rather than in-person in disproportionate numbers.
This proposal allows authors to choose to publish their papers at lower cost (time and money), but with lower benefit of dissemination. f some authors prefer this outcome for their work, it does not harm the authors who prefer to publish and disseminate. All authors can have the conference experience that they prefer.
1 This proposal omits discussion of good practices of running hybrid and virtual meetings. While there is a wide variance in quality of hybrid and virtual meetings, it is possible to do them very well, with good technology and buy-in from the research community.
Jason Harline is a professor of computer science at Northwestern University, a co-founder of interactive virtual event platform Virtual Chair, and a Program Chair for the 2023 ACM Conference on Economics and Computation.
No entries found