2021 SIGCSE Technical Symposium: Virtual Diffusion
A year ago, I was on a train, traveling to Portland, Oregon, wearing a very uncomfortable N95 mask. The train car was mostly empty; out of the corner of my eye, I could see the couple on the other side of the train. They peered at me every once in a while, maybe because of the mask, maybe because of my loud typing, or maybe because of my vaguely non-passing, recently out trans face and body. Little had shut down for COVID-19 yet, but most of my students, collaborators, and colleagues had canceled their trip to SIGCSE 2020. Many had asked me to present their work or run their sessions, and so my leisurely week in Portland had turned into a gauntlet of a dozen surrogate talks and facilitations, none of which I had created or planned for. It was going to be an interesting week.
Of course, SIGCSE 2020 never happened. As I covered in my trip report last year, when I arrived at my nearly empty conference hotel, people walked around nervously; I would step away to keep social distance and they would step closer. I met with the SIGCSE Board about the upcoming ICER 2020 conference I was co-chairing. I went out to a risky dinner with some attendees downtown. The next morning, the Oregon governor banned all gatherings, and the conference chairs immediately canceled the conference. I rebooked my ticket to return home that afternoon, and had a wave of realizations: 1) this was probably the last trip I’d take in a while, 2) that was probably the last dinner with friends I’d have in a while, and 3) I was really glad I’d brought that mask. The train was even emptier on the way home. I’d only realize later that solitude would be my new normal for more than a year.
This year, traveling to SIGCSE 2021 was an entirely different experience, because I didn’t go anywhere. Going to work, going to SIGCSE 2021, paying my taxes, shopping—they all happened in front of the same 27" display, at the same desk, on the same Mac mini, with the same headphones on, in the same room, and largely the same sterile Zoom window. The only difference was whose faces appeared in the drab rectangles.
The week before the main SIGCSE conference, I co-organized a pre-conference workshop amidst a regular work week. I was supposed to go to two other pre-conference workshops as an attendee. But I bailed. Instead, I tried to wrap up my Winter quarter, and prepare for the coming Spring. I felt guilty: I’d wanted to network and learn, but when I looked at the 2 hour blocks of Zoom meetings and looked at my to do list, I just couldn’t do it. Not amidst 10 hour days of packed with other Zoom meetings. I used to have as long as a full day to mentally transition from work mode to conference mode, to prepare myself to learn and socialize. Now the transition was a 100 ms click.
And so as I entered the Sunday conference opening, I steeled myself for a week of strained engagement. I really needed to finish grading for Winter quarter, I really needed to prepare for Spring quarter, and I really didn’t want to spend more time on conference calls. On the other hand, as co-chair again for ICER 2021, I really did want to see how this year would go, I really did want to respect and support the effort that Mark Sheriff and the rest of the team had put into planning and the conference. And I really did want to connect and learn.
Sunday: Conference Opening, Real Justice
Joining in Sunday night for the opening wasn’t so hard. I’d had a mostly restful Sunday, and I was really excited about the panel, titled AI and Real Justice, and moderated Kamau Bobb, Global Lead for Research Strategy and Diversity at Google, and also Georgia Tech. He was speaking to Ayanna Howard and Safiya Noble.
But the experience of joining the conference sapped some of that initial excitement. I immediately missed the usual experience of an opening plenary: the hum of a thousand people chatting, the nervous pacing of the chairs on the sage, striking up a conversation with the random person in the seat next to me, and waving to friends I haven’t seen in person for months or more. The first day of a conference is usually all about reconnecting with friends and colleagues, often those where my entire relationship was shaped and maintained at previous conferences. Visiting the Pathable home page though was sterile and silent. Where was everyone? Who was I sitting next to? How do I find my friends?
The chairs covered the usual details: the program, best papers, the code of conduct, details about the Pathable platform the chairs had chosen. There was a clear setting of expectations: things would not be perfect, technical glitches would happen, but the conference would happen and we would make the most of it. That enthusiasm lifted me out of a well-earned year of virtual conference dread to patient, curious state of mind.
The panel that came immediately after was the perfect salve. Rather than watching a presentation, I got to listen to three incredibly bright Black scholars talk about AI, the ongoing reckoning in the U.S. around racial justice, and the complex ways in which CS is interwoven through this reckoning, much like with my podcast addiction. The discussion covered the increasingly well-known foundations of informed conversations around CS + justice—technology isn’t neutral, not all technology needs to exist, the bias is hidden, CS has responsibility, CS should partner with experts in social sciences. I’ve spoken on many of these subjects over the past year myself in a series of invited talks. But it also broke new ground for me, linking a lot of the lack of progress on these issues to the experiences of Black advocates in tech, such as Timnit Gebru, who face mountains of resistance for simply pointing out facts about technology. That was an interesting tie between advocacy, activism, marginalization, technology, and values.
One day down, six days to go.
Monday: Juan Gilbert’s opening keynote, and an attempt at an Assessments session
The next event I could make was the opening keynote Monday morning. This one was by another outstanding Black computer scientist, Juan Gilbert at the University of Flordia. Juan began by setting the stage: the U.S. is nearly a majority-minority country (where most people are in “minority” groups, which is a problematic word for many reasons). He pointed out that there were 4,000 Black CS undergraduates finishing degrees, but very few doctoral students. He talked about bad stereotypes, role models, poor advising, limited information about Ph.D.’s, and reinforced the need for NSF’s broadening participation in computing efforts. Something about it though just felt really male, cis, heteronormative, and neoliberal. Maybe it was all of the talk about games and competition and capitalism and jobs and numbers.
Of course, I didn’t get to stick around to find out, because I had a grant meeting to run 30 minutes into it. Sorry Juan :(
Later on Monday—at the end of my day, 5 pm—I joined the assessments session. The first talk by Max Folwer and Craig Zilles investigated an interesting question about how sensitive surface level changes to test items are to transfer outcomes. Miranda Parker then presented an interesting assessment for elementary students.
But after 30 minutes in, and a full day of other Zoom meetings, I just couldn’t do it any more. I opened an unofficial SIGCSE Ohyay workspace that had been circulating and noticed some colleagues I hadn’t talked to for a year. We struck up a fun conversation about virtual conferences, podcasts, teaching online, and more. I felt guilty for abandoning the session, but realized that this is what I usually do at conferences: backchannel during sessions, or skip sessions altogether to talk to friends and colleagues in the hallway. After that, I didn’t feel so bad. In fact, talking with friends, even in a weird Ohyay room, felt nice, and restorative.
Two days down, five days to go.
Tuesday: Another failed session attempt, and an expansive panel on teacher education
The next morning, just before my biweekly 11 am therapy appointment, I joined the beginning of a session on computing education research. I was excited to hear more about the state of the field, and identify ways I might contribute to its advancement. Brett Becker and collaborators spoke eloquently about the US-centric nature of the SIGCSE technical symposium, and solicited feedback about ways to be more globally inclusive.
But then, another interruption. There were some important questions about SIGCSE, ICER, and computing education research journals, and I needed to have a side conversation with a senior colleague. At first, I felt guilty for abandoning another interesting session, but I reminded myself again that this always happens in-person: on the way to a session, I’ll see someone in the hallway, they’ll pull me aside, and we’ll strategize, coordination, and share critical information. How quickly I’d forgotten that being senior faculty at an academic conference is mostly not about the papers and the learning, but rather process, politics, and power. Even online.
After several hours of other meetings—the ECEP alliance monthly call, prospective Ph.D. student recruiting for UW CSE’s visit days, and some frantic course preparation for Spring—I settled in for another end of day session. This was a panel on teacher education, my latest interest. I’m working with several colleagues at UW on a CS add-on endorsement program that centers critical consciousness, and I wanted to see what my peers at other universities had built for pre-service and in-service teachers. The epic 2-hour panel included:
- Michelle Friend (University of Nebraska, Omaha), who leads a program that focuses on CS content, self-evaluation, the CSTA standards, and a variety of micro-teaching methods.
- Anne Leftwich (Indiana University, Bloomington), who leads a pre-service program that includes introductions to educational technology, a required CS class, and a variety of teaching methods that cover harder, block-based programming, HTML, robotics, the CS Principles curriculum, and some outstanding field experiences.
- Beth Simon (University of California, San Diego), who leads four self-service, self-paced online courses that are auto-graded and peer reviewed, with one course that covers methods, including use-modify-create, TIP&SEE, and PRIMM, with a focus on developing PCK. It’s designed for experienced in-service teachers.
- Ben Shafer (University of Northern Iowa), who leads an add-on endorsement that is mostly CS courses, and one CS teaching methods class.
All four mentioned challenges of recruiting pre-service and in-service teachers, tensions between what teachers want and need, gaps in CS content knowledge, and an unfortunate gap in diversity, equity, and inclusion issues. Since we’re building a peer program at UW, I was fascinated to learn the details about what others had built. I’m excited to see how they’ll react to our approach, which centers diversity, equity, and inclusion, and puts CS content knowledge more at the periphery (perhaps to our peril?)
Three days down, four to go.
Wednesday: Disability and a break
At this point, I was tired, from both a very busy two days at work, along with an extra 2–3 hours a day of attempted SIGCSE engagement. I gave myself a break, and only did one thing on Wednesday: co-chaired a birds of a feather networking session with Richard Ladner, Sheryl Burgstahler, Brianna Blaser, Andreas Stefik, and myself. We had a few dozen attendees, heard a wide range of interests spanning primary, secondary, and post-secondary settings, accessible teaching, teaching accessibility, and access technologies. Just as with previous years, the level of knowledge about accessibility varied widely, and much of our role was to help the group navigate the community and it’s sparse resources. I left feeling more inspired than ever by the attendees passion for accessibility, broadening participation, and intersectional views of who has disabilities.
Four days down, three to go.
Thursday: Diversity, Equity, and Inclusion in Doctoral Computing Ed…bzzzt
Although it was nearing the end of the week, I was re-energized by some of the networking, and decided to go to a panel Thursday morning on broadening participation by doctoral students in computing education. The panel had a number of leaders in doctoral admissions, including Erika Lynn Dawson Head and Emma Anderson of U Mass Amherst, Elise Dorough and Leslie Sessons of the University of Washington, Christopher Hovey of CU Boulder, and Sepi Hejazi Moghadam of Google. All set forth a vision of trying to increase the diversity of students who join doctoral programs in CS.
And then my internet died. I tried tinkering with my wifi, jostling the ethernet, resetting my computer, testing other devices, rebooting my router, rebooting my cable modem, checking for a service outage. Nothing worked. I had to decide whether to eat through a few gigabytes of my data plan to continue viewing the session or save the prescious bits for the TA training session I needed to run after lunch in preparation for Spring quarter. TA training won.
After the TA session, and a wall of afternoon meetings, my internet was back, just in time to meet my SIGCSE buddies. The conference had repeated the idea from SIGCSE and ICER 2020, pairing experienced attendees and newcomers. I was meeting up with two newcomers, both from St. Petersburg, Russia, and researchers at JetBrains, which makes a collection of IDEs. I decided to invite others, and ended up having a pleasant one hour conversation about conference attendance, R&D in industry, virtual conferences, time zones, sleep, and our collective dread about hybrid teaching. Again, conversation and connections, not presentations, were the highlight.
Five days down, two to go.
Friday: Ethics and Assessment
It had been five days, and still hadn’t managed to sit through an entire paper session without a side conversation, an internet outage, or some other crisis interrupting. I was committed: Friday would be all about listening and learning! I attended two sessions: one on CS ethics and another on assessment, which featured one of my lab’s papers on test bias. Here’s what I learned:
- Diandra Prioleau presented How Students in Computing-Related Majors Distinguish Social Implications of Technology, finding that student knowledge of ethical issues varied widely, but did not demonstrate significant depth, especially in issues of fairness and equity.
- Beleicia Bullock presented Computing Ethics Narratives: Teaching Computing Ethics and the Impact of Predictive Algorithms, presenting a fascinating collection of 700 narratives at the intersection of CS and ethics; students reacted to the narratives wanting more class time to evaluating the technical complexities of critical systems. Beleicia did an outstanding job fielding questions.
- Casey Fiesler presented Integrating Ethics into Introductory Programming Classes, presenting an evaluation of integrating ethical contexts into assignments on programming constructions. Casey and her co-authors found that students wanted more content like this, and didn’t perceive that it impacted their learning about CS concepts, but that there were tensions in finding ethical contexts that fit the narrow constraints of programming assignments.
- Ben Shapiro presented Using Role-Play to Scale the Integration of Ethics across the Computer Science Curriculum, sharing a set of unit plans that had CS students take on technical and non-technical roles in ethical decision making contexts (e.g., using AI to make admissions decisions). Many of the questions about the talk concerned the relationship of role playing to perspective taking, HCI methods, and risks of stereotyping and misrepresentation.
- Rodrigo Ferreira presented Deep Tech Ethics: An Approach to Teaching Social Justice in Computer Science, describing a new standalone ethics course at Rice University, which draws upon Foucault, Galloway, critical theory, and feminism to discuss CS from the perspective of power and control.
It’s encouraging to see such a rich diversity of approaches to engaging ethics, equity, and power. It’s clear the community is experimenting with many things, and that the research is still very much in the exploratory phase. Interestingly, there was a sense amongst the presenters and attendees that, despite how far CS needs to come in integrating these issues into its teaching, it’s also a community that’s far more engaged than other disciplines (including, ironically, law). As someone who’s spent the year calling for more experimentation, I’m heartened by everyone’s passionate exploration!
After a grueling five hours of administrative work, I went to the 5 pm Assessments session. I didn’t know how long I’d stay; it was an exhausting week and I needed to save some energy for a virtual gathering with friends later Friday night. I managed to make it to two sessions:
- Matt Davidson (a Ph.D. student collaborator in my lab), presented Investigating Item Bias in a CS1 Exam with Differential Item Functioning. We demonstrated how to use the relatively new DIF analysis to detect bias in tests. See Matt’s blog post summarizing the approach.
- Carla Strickland presented Action Fractions: The Design and Pilot of an Integrated Math+CS Elementary Curriculum Based on Learning Trajectories, describing an intriguing collection of math and CS integrations for elementary students, aligning concepts of sequence, decomposition, repetition, conditionals, variables, and debugging into teaching on fractions. The curriculum is available at Everyday Computing.
By the end of Carla’s talk, it was approaching 6 pm, my wife was almost home with our weekly takeout tradition, and I just couldn’t think any more after a long 11 hour work day. I left the Zoom webinar, went upstairs, sat on the couch, closed my eyes, and and pet my cat.
Six days down, one to go.
Saturday: Valerie Taylor’s Closing Keynote
I’d hoped to sleep in, but I woke up early at 6 am. I watched a horrifying video about the devastation of Japanese tsunamis (thanks YouTube). I tried to sleep again. But my inbox was full of questions from ICER 2021 authors about the abstract deadline that had just past and questions from our submissions chairs about bidding configuration and conflicts of interest. I got up, exercised for a few minutes, cleaned up, and then collaborated with my ICER team to ensure our bidding configuration was sound. I began to regret choosing Friday anywhere on Earth deadlines.
After some breakfast, I signed in for the closing SIGCSE keynote with Valerie Taylor. She spoke methodically about the value and importance of diversity to organizations, the critical role of understanding organizational culture in identifying strategies for change, and how she’s used these ideas in LEAP Alliance, which aims to diversify the CS professoriate. She stepped through the specific strategy they took, observing that half of CS faculty receive their Ph.D. from the same top 10 universities; this led to the strategy of diversifying who gains entry into these top 10 institutions. The approach was to find faculty and staff advocates for organizational change at these institutions to work on changing the recruiting, admissions, and retention processes, as well as the culture of the organization. They build training programs for these advocates, offered professional development for advocates, developing cohorts of students who enter these programs, and rallied partners. Thus far, these efforts have increased first year students from their target populations. This is not only an exemplary effort at community and partnership building, but nationwide institutional change. This is an amazing demonstration of leadership; I look up to Valerie for her ambition and persistence in making change!
Sunday evening, the conference came to a close. Mark Sherriff gave an energetic and enthusiastic reflection. He showed engagement across the week, which showed more than 1,900 attendees, with a gentle drop until yesterday toward 1,200 attendees, with a total of 1,645 attendees over the week. Most sessions had an impressive 50+ active people. The student research competition announced some winners, and next year’s organizers looked forward to Providence.
Day seven and done.
Reflections
I’m incredibly grateful to the SIGCSE 2021 team for the immense amount of work to plan and organize this year’s conference. And I’m incredibly impressed by this year’s speakers and authors: the content was better than ever. I can’t wait to read the papers. But I know from experience that running a conference is a thankless task, especially during this pandemic. There’s really no way to please everyone in times when everyone is exhausted and misses their friends and communities. There was no way this SIGCSE was going to be a smashing success by normal standards; at best, it was going to sustain our community, so that we can reconnect in better times. So the reflections below are in no way a critique of this year’s effort: they are simply ideas that help us move forward toward those better times in ways that build upon our learning from this year.
And I’ve been learning a lot. In the past 12 months, I’ve attended over a dozen virtual events, all on different platforms, and with different designs. I helped design and run ICER 2020. My co-chairs and I thought extensively about hybrid conferences for ICER 2021 this past Fall. We’re still deeply engaged in planning our ICER 2021 virtual conference for this August 2021. And I’ve written how critical it is that we not return to our previous models: they are not sustainable or inclusive.
But are virtual conferences the future?
In my opinion, no, at least not how they are being run now. Here’s why:
- Most virtual conferences mistakenly optimize for content delivery. In my opinion, that is not, and has never been, the value of conferences. This is partly because there’s always a paper to read in our digital libraries, and partly because great researchers aren’t necessarily great presenters. This is aggravated by the instability of internet connections, the complexity of time zones, and the immense number of distractions in our modern media ecosystems.
- It is simply too hard to separate work activities from conference activities when they’re in the same physical space. This is true even for people with the best intentions, the ability to reserve time, and the skills to ensure that time is set aside. These are all true for me—I had the time set aside on my calendar to attend, genuinely wanted to attend, and have a great capacity to force myself to do things I don’t want to do—but I still struggled. I’m sure other struggled even more, especially if they hadn’t reserved the time, had young children at home, or were behind on other work.
- The truly valuable things about conferencing—serendipity, networking, relationship building, trust building—are all much harder online. They’re not impossible, but none of the platforms and conference designs I’ve seen these year optimize for these over content consumption. These activities take a different kind of planning that’s less about getting content online and designing a schedule for consuming it, and more about creating spaces for communication and connections between attendees that both reinforce existing ties, and create new ones. Zoom calls are not that place.
What is the future, then, especially if we need some virtual component to improve sustainability and inclusion? After a year of experiences and reflection, here’s my opinion on the key principles for successful hybrid conferences:
- Short daily virtual encounters with abundant breaks. This is necessary for simplifying remote attendees’ ability to schedule and reserve time, but also commit to it. I would recommend no more than 2–3 hours of contiguous time per day, carefully chosen to be as time-zone inclusive as possible. No sessions should last more than 40 minutes; people simply can’t pay attention to a video chat for longer than that, even if they are at the conference venue.
- No more than four days. Conferences should feel like an event; having them diffuse across a week or two reduces motivation to engage, optimizing for content coverage instead of intense networking. And in-person events have duration limits because people have budget limits. No more virtual conferences that span multiple weeks.
- Social visibility and opportunity. It should always be possible to see where one’s friends and colleagues are (unless they don’t want to be seen), so that one can reconnect with them. In-person attendees can do that with ease, but virtual attendees should be able to do that as well. And hybrid conference designs should build in abundant structured opportunities for people with weak ties to a community to build stronger ties, in small groups of 3–4 people. That might mean small groups in person, with a few online attendees, having coffee chats on an iPad.
It also takes the right tools to make these possible. Platforms like Clowdr, Ohyay, and Discord are the best amongst everything I’ve tried this year so far; the more enterprise platforms like Hopin and Pathable have been terrible for networking. To make this work, hybrid should be split into two: the beginning of the day can be purely virtual, content-driven events in a shared physical space for in-person attendees, improving their ability to focus on the conference, with remote attendees engaging online in the exact same way. This first portion will need tools like those above to keep communication and content delivery smooth, while allowing for serendipity. They’ll also need carefully designed physical spaces; we’ve all gotten used to engaging in quite rooms with headphones on: this portion of a hybrid event might need to be like a library, with in-person attendees all sitting quietly at tables with headphones on, as if they were at home, but with rich opportunities for whispers and non-verbal signals. This might be strange, but it will only last a few hours.
The second portion of each day should be deeper in-person networking, far deeper than our past conferences have been. This might include doctoral consortiums, workshops, grant planning meetings, receptions, speed dating, newcomer welcomes, mentoring and coaching sessions, outings and adventures, and organizational planning. These afternoons can even engage remote attendees by having in-person attendees acting as conduits for serendipity for remote attendees. For example, if I were at a hybrid SIGCSE, but some of my doctoral students were virtual, I would have spent much of this week introducing them to other people during dedicated networking sessions. This model would provide significant added value to in-person attendees, further incentivizing attendance. And we can use the increased revenue from virtual participants to subsidize attendance for those who have never been, again strengthening those weaker ties.
Perhaps the most controversial aspect of the model above is that all of this would require far less time giving and listening to presentations. Conferences would shift from being four 8 hour days of presentations and panels to perhaps four 2 hour days of presentations. Some people, especially authors, who might rely on presentations to get some attention on their work, might balk. But we have to accept that we don’t gather to show each other the slides we’ve made: we gather to confer—to have discussions, exchange opinions, and build community. And we have to embrace that there may be better ways for authors to get attention on their work. Imagine, for example, afternoon in-person sessions in which attendees build reading lists with others for them to read after the conference, and to share with remote attendees. Or even sessions in which people read the papers in the proceedings (imagine that, reading the papers that were published!).
I’m personally excited about this future. I think there are so many ways of conferencing that we simply have not explored, and those ways might not only be more sustainable and inclusive, but also more enriching, empowering, and productive. Let’s step back from this year’s SIGCSE not with the attitude of “I never want to do that again”, but rather “What was good about this, and how can we preserve that in the new models we invent in the future?”
See you next year, somewhere, somehow!
