PL Perspectives

Perspectives on computing and technology from and for those with an interest in programming languages.

In the past year, COVID has transformed the landscape of SIGPLAN conferences, forcing us to move from all-physical to all-virtual events. But with hope on the horizon, we must figure out what we want the future of our conferences to look like. Do we like our now-ubiquitous virtual conferences so much that we will never have physical conferences again? Or, is the best future a mix of virtual and physical conferences? Or, as soon as the virus is tamed, are we done with virtual conferences altogether? Some possible answers are to go back to business as usual (BBU), alternate physical and virtual conferences (ALT), go big (ONE), or offer both in-person and online options (MIX). Let us examine these answers in turn and begin to prepare carefully for conferences after COVID; we seek your feedback.

BBU: Back to business as usual

Going back to having ICFP, PLDI, POPL, and SPLASH in-person every year would be natural: this has worked before and can work again. In-person conferences are excellent at providing socialization, excitement, and coverage of hot topics. We all miss the interpersonal interactions made possible by physical conferences and we want them back.

But if we drop virtual conferences entirely, we would lose their positives, including inclusiveness and mentorship at scale, reduced travel, and near-zero carbon footprint. Lots of people have limited travel budgets (both money and time) and cannot go to all conferences of interest. Can we think of something better?

ALT: Alternation of virtual and physical conferences

Each of ICFP, PLDI, POPL, and SPLASH had a virtual edition over the last year, and the virtual editions will continue in 2021. Together those conferences have tried out many ideas for how to make a virtual conference exciting. Big positives have included the keynote presentations, the recorded paper presentations, and the long question-answer sessions after (and sometimes well in advance of) each talk.

However, many aspects are arguably more engaging and substantive at physical conferences, including networking, social activities, mentoring, town-hall meetings, poster sessions, panels, and tutorials. Workshops may fall in the middle; their smaller size means they can work better than larger events when held virtually.

Can we get the best of both worlds by alternating the conference formats? For example, should we have two physical and two virtual conferences each year, rotating which conference is which? A downside might be that people prefer to submit to physical conferences because they can present their paper in front of a live audience. If so, we would see the virtual editions get fewer submissions. To avoid facing that problem, perhaps we should try the following radical idea instead.

ONE: A single annual physical conference plus four virtual ones

How about we let ICFP, PLDI, POPL, and SPLASH stay virtual and create a new annual physical SIGPLAN conference, like FCRC but focused on programming languages? This could maximize the chance of meeting people in person because many people will be able to attend at least one conference a year, we can hope. A single conference would present an image of a large, vibrant programming languages research community and thereby increase its overall visibility. Can we make such a conference focused and satisfying? Perhaps it can be like a traditional scientific meeting, which has overviews, tutorials, workshops and discussions? Can we bridge the cultural and technical differences between our conference communities by bringing them physically together, and thereby foster new ideas and collaborations?

Many people need to give a presentation to be able to get travel funding, so we should let many people present their work in one way or the other. For that, we can take inspiration from other fields that face far greater challenges of scale than we do. In particular, we should maximize the value for students who will present their work and then meet with people in the research community to discuss it. One enabler is that we have already, for practical purposes, separated the publication from the meeting. In particular, a paper accepted to ICFP, PLDI, POPL, and OOPSLA matters more to a career than whether somebody presented the paper at a physical conference. We should let this publication model continue while we think about conferences after COVID.

One downside of a single annual physical conference is that it will do little (though more than nothing) to decrease our carbon footprint. The reason is that most people who attend ICFP, PLDI, POPL, and SPLASH in a given year very often attend none of the others (see Section 2.5 in the climate report). That said, this fact speaks to the benefit to bringing people together, since that wouldn’t normally happen.

Another downside is that an FCRC for programming languages will indeed be nearly the size of FCRC, perhaps 1,500 people. This means that our usual hotels won’t work; most of them take meetings up to 800 people. Instead, we would have to be in a conference center, like FCRC usually is. That said, in the last 10-15 years we have moved from small (200-person and fewer) university venues to larger hotels, despite concerns. Other computer science communities now regularly have large meetings, such as CHI, CVPR, SC, and SIGGRAPH; we can learn from their experience.

MIX: Back-to-back physical and virtual conferences

Finally, what if we try to “have it all” by incorporating both physical and virtual components for all our conferences? Truly hybrid conferences seem too hard to get right and too much work for the organizers, but perhaps we can find something easier that works?

Here is an idea: How about scheduling a physical and a virtual component back to back? A week or two either before or after the physical conference, we can allocate some time for a virtual conference focused totally on the papers. At that point, most authors will be at home and can engage with people virtually. We can think of this format as a way of making our old-style conferences strictly more inclusive and getting authors connected to more people. For the physical event, we can resume our pre-COVID custom of making videos available online, live-streaming talks, and enabling online questions via sli.do (or something similar), to support remote participation. But the adjacent virtual event will provide an important next-best option for community members who are unable to attend physically. That way, there is context for inviting outside interaction, rather than the default neutrality or discouragement of interaction from outsiders. Also, the back-to-back conferences can, together, give people from all time zones approximately equal opportunity to engage. One downside, however, is that most people will probably not want to engage fully in both the physical and virtual events, so this option could result in decreased participation in either individual event compared to the alternatives.

What do we do?

After COVID, what do we do? We could experiment and see what works. The SIGPLAN Executive Committee and the conference Steering Committees have been discussing the options, the town hall meeting at POPL saw a discussion, and POPL did a survey that mentioned the four options above. Indeed, the POPL’21 survey result was that all four options have substantial support. Which scenario do you like? Please fill out the SIGPLAN survey about conferences after COVID. The SIGPLAN Executive Committee is eagerly waiting to hear your opinion.

Bios:

Jens Palsberg is the current Chair of SIGPLAN (2018-2021), a member of the ACM Executive Committee, and a professor at the University of California, Los Angeles (UCLA).

Emery Berger is a professor at the University of Massachusetts Amherst, where he co-leads the PLASMA group. He has been a member of the SIGPLAN Executive Committee since 2015. He was the Program Chair for PLDI 2016 and is currently the co-Program Chair for ASPLOS 2021.

Derek Dreyer is a professor at the Max Planck Institute for Software Systems (MPI-SWS) in Germany, where he leads the Foundations of Programming group and the RustBelt project.  He was a member of the SIGPLAN Executive Committee from 2012-2015, was General Chair of ICFP 2019, and is currently Chair of the ICFP Steering Committee.

Michael Hicks is a professor at the University of Maryland, where he co-leads PLUM and is a member of MC2. He was Chair of SIGPLAN from 2015-2018, Chair of the POPL Steering Committee from 2018-2021, and is currently the co-editor of this blog.

Jan Vitek is a professor at Northeastern University and the Czech Technical University in Prague. He was Chair of SIGPLAN from 2012-2015 as well as of various conferences, summer schools and related stuff.

Note: A recent survey conducted by Nature may be of interest.

Disclaimer: These posts are written by individual contributors to share their thoughts on the SIGPLAN blog for the benefit of the community. Any views or opinions represented in this blog are personal, belong solely to the blog author and do not represent those of ACM SIGPLAN or its parent organization, ACM.