ICER 2021: A daily dose of digital discourse
When I nominated myself to co-chair back to back years of the ACM International Computing Education Conference (ICER) back in late 2018, I didn’t fully comprehend what I was taking on. Part of that was purely organizational: whenever someone volunteers for academic service, there’s a wide range of things one might find behind the curtain, from immaculate process and organization to little more than a “good luck” from prior organizers. (Thankfully, ICER was more the former than the latter). Taking on such a big service role—and in the case of ICER, two years of it, to ensure the transfer of tacit knowledge—means walking into that unknown. I knew that much.
But there were a few other things I didn’t anticipate. At the time, I was struggling greatly with accepting myself as transgender, and wasn’t sure if I ever would, or if I’d come out. I eventually did in September 2019, precisely around the time that work began in earnest to plan ICER 2020. That in itself was incredibly challenging, because not only was I spending a dozen extra hours a week just managing transition labor (name changes, therapy, voice therapy, self-advocacy, other advocacy), but I was also engaged in exhausting but necessarily recovery from a lifetime of identity trauma. That’s not the easiest state in which to engage in high-stakes peer review and event planning with new colleagues. Fortunately, my new colleagues Anthony, Renée, and Adon were great. Anthony and Adon brought knowledge from the previous year, all three embraced my gender news without a fuss, and we set out to design an exciting adventure for our global community of computing education researcher colleagues in New Zealand.
Then COVID-19 happened. We pivoted, doing our best to devise a virtual conference design like every other conference organizer, and survived a year of various existential traumas while still managing to host a solid event. It was far from perfect, but I’m proud of the work we did, both organizationally in challenging circumstances, but also for how it helped reveal the creative possibilities of virtual conferences.
Shortly after ICER 2020 in August, the work for ICER 2021 began. Renée and and I joined with our excellent incoming co-chairs, Jan and Matthias, and we began to ponder what ICER 2021 might be. The plan was always Charleston, South Carolina, following the conferences rotation pattern of Australia, North America, Europe, and North America. The uncertainty set in quickly: would there be vaccines in time? Would global travel be possible? Would we aim for hybrid, to ensure everyone could attend? Everyone, from SIGCSE leadership, to organizers and future attendees, had a different preference. Personally, I just wanted certainty (and I was not particularly excited about visiting one of the trans murder capitals of the United States in a year of record murders). To my relief, we all settled on an online event, and began exploring our options.
We set out a few basic principles: 1) inclusion on all dimensions and 2) sustaining and building our community. Perhaps the biggest choice we needed to make was of which tools we’d use to achieve those principles. We developed a list, partly from ACM’s efforts to build virtual conference guidance, but quickly ruled out most options due to their poor accessibility for people reliant on screen readers. What was left was Zoom and Clowdr; after many conversations with the Clowdr team, we were convinced that overall, Clowdr would be a better experience than our 2020 efforts with Zoom and Discord, despite the platform’s immaturity and usability flaws: it offered an integrated single-window browser experience, which was superior to any of the other platforms we used (including the much more expensive ones). It also allowed us to center connection and networking far better than other platforms, which all seemed more oriented toward watching pre-recorded videos in isolation and silence.
We also worked on several other infrastructure improvements as well, catalyzing a new conference steering committee, building a robust store for institutional knowledge about the conference, expanding the program committee’s intellectual diversity, adopting more explicit procedures for minor revisions, and improving upon 2020’s senior PC meeting. We took on a lot (especially as four unpaid volunteers!), and got it all done, with a lot of help and guidance from the new steering committee. I’m proud of our work, despite all the inevitable challenges and tradeoffs.
While I’d hoped we’d be one of the last virtual conferences in a world slowly reopening, that isn’t quite what happened. In the weeks before ICER 2021, delta surged, summer wildfires raged, heatwaves blanketed the states. And then I had my own challenges: a friend and colleague at work died a few weeks before the conference, my wife and I had family in the hospital from the heatwaves and smoke, and I had to schedule major surgery for the Tuesday after the conference, requiring substantial logistics, anxiety, and physical strain in the weeks prior. So while I entered the weekend before ICER 2021 excited about all of our hard work, I also entered it hot, asthmatic, exhausted, grieving, destabilized, and anxious about my impending surgery and months of recovery. I suspect a lot of other attendees started similar exhausted.
Monday — Opening, Keynote, Papers, and Lightnings
My worst fears for day 1 didn’t materialize: everything mostly worked! I coordinated a bit with the other conference co-chairs, and then we did our conference opening with only a few handoff hiccups. We had about 130 attendees in the opening session, reasonable engagement via chat and emojis, and I heard good things from various people in the community about the kickoff.
After the opening session, we had arranged a welcome session where newcomers could meet returning attendees and be welcomed. It was only 20 minutes—not nearly enough time to build long-lasting connections—but we hoped it was enough that newcomers at least feel like this was a community that was friendly, interesting, and eager to chat.
Meanwhile, I was back stage preparing to introduce Tamara Clegg for her keynote. I thought her talk was outstanding: she talked about the themes of community and how they’re closely tied to youth’s interest or disinterested in community, sharing her own stories about her own communities, and the strengths of approaching education research in a deeply situated manner. I was particularly fond of how she brought real, specific individuals into her stories about research, illustrating that in both learning and research, our own identities can be essential to discovery and progress.
After the keynote, we had a short networking break, and then I attended our first paper session. The first paper was by Jean Salac et al., who reported on a comparison of two scaffolding strategies for teaching computational thinking in primary. Her team found that as lessons progressed, there were steady increases in working memory and long-term retrieval demands, but more so for Use-Modify-Create scaffolding.
The second paper by Iman YeckehZaar et al. was on the effect of spacing practice, which sought to replicate the broader findings on spacing in CS. And they largely did: spacing improved learning (although with relatively small effects), though there were gender differences in the optimal spacing that were difficult to explain.
We closed the first day with a short 20 minute doctoral consortium lightning talk session. The students were incredibly polished, and shared an impressive range of questions. Their poise, rigor, and expertise made me confident in our field’s future!
Tuesday — Lightning talks, assessments, awards
We kicked off day 2 with a short burst of lightning talks and networking, and then three back-to-back 40-minute paper sessions, segmented by 20 minute breaks. I started my day in a welcome room, to see if any newcomers would drop by to meet me. And they did! I had about 8 newcomers and returning attendees stop by to say hi, and I got to reconnect with lots of colleagues in Europe, and meet some newcomers too.
The lightning talks session was great; there were several short 3-minute talks, each with nuggets of new ideas, including some ideas on gender norms, natural language barriers, and disciplinary self-concepts.
Next, I attended the talks in the “Shrimp” rooms. The first sessions was an assessment session:
- Miranda Parker (UC Irvine) presented Uses, Revisions, and the Future of Validated Assessments in Computing Education: A Case Study of the FCS1 and SCS1, which documented and analyzed the impact of validated assessment over 10 years. They generally found that the use of the FCS1 and SCS1 were used roughly within the bounds of their validity arguments, or used to inspire other instruments, but there were some problematic uses, and aren’t a panacea.
- Max Fowler (University of Illinois) presented How should we ‘Explain in plain English’? Voices from the Community, which examined assessment items that give code and ask students to explain its behavior. The study revealed a wide range of opportunities and tensions in using them in formative and summative settings. I was particularly interested questions about how “high level” student responses should be.
After the first session, I decided to create a “Cat Chat” room and see if anyone wanted to talk about cats. I was pleased to see that many did! We talked about CS education, COVID, and … cats!
The second session in “Shrimp” was also an assessment session:
- Alexandra Milliken (North Carolina State University) presented Exploring and Influencing Teacher Grading for Block-based Programs through Rubrics and the GradeSnap Tool, which a tool called Gradesnap that streamlined teachers grading processes. They found that it provided a catalyst for teacher reflections and more detailed feedback for lower performing students.
- Mohammed Hassan (University of Illinois) presented Exploring ‘reverse-tracing’ Questions as a Means of Assessing the Tracing Skill on Computer-based CS 1 Exams, which considered questions that prompt learners to determine the input that will provide the specified output. They found that such problems require similar but not identical problem solving skills, and that they may be more diagnostic.
The last session contained two papers, the first an Honorable Mention, and the second the conference’s Best Paper:
- Gregor Große-Bölting (Kiel University) presented Identity in K-12 Computer Education Research: A Systematic Literature Review, which examined research on CS identity across the literature, finding that across 31 papers, theory was relatively stable, but that there has been a persistent oversight of identity in K-12, and very little evidence of identity “interventions.”
- Maria Kallia (University of Glasgow) presented Re-Examining Inequalities in Computer Science Participation from a Bourdieusian Sociological Perspective, which theorized about inequity from the perspective of Bourdieu’s theory of social reproduction. The award was well-deserved: the paper elegantly reframes conversations about inclusion and participation through the lens of Bourdieu’s concept of habitus, offering a way forward that focuses on rules of participation and building students social capital.
Wednesday — Lightning talks, papers, and posters
Before the conference started, I had to go get a COVID test in preparation for surgery. Unfortunately, the testing site changed its hours without notifying my, so I stood outside in the cold for an hour and waited. I made it home just in time to have a nice welcome chat with Bjarke Fog, Colin Gray, Michelle Craig, and Adam Blank.
The first session was more lighting talks, once again chaired by Miranda Parker. The talks were wide-ranging, including fascinating looks into centering agency in CS learning, accessible computing in India, approaches to teaching debugging, and creative new ideas for assessment.
Immediately after was another paper session. I once again attended the sessions in the “Shrimp” room. The first papers were about student conceptions:
- Lauren Marguileux (Georgia State) presented When Wrong is Right: The Instructional Power of Multiple Conceptions, which contributed a detailed theory about instructional techniques in computing education and how they interact with higher order conceptual knowledge. They helpfully offered instructional design principles, like “Vicarious failure: Introduce learners to non-correct conceptions”.
- James Finnie-Ansley (University of Aukland) presented A Semblance of Similarity: Student Categorisation of Simple Algorithmic Problem Statements, which examined how students interpret problem statements. They found that students do recognize categories of problem statements, but their knowledge is fragile and didn’t demonstrate clear problem schemata or algorithmic patterns.
After the talks, I had a wonderful social chat with Ph.D. student Gregor Große-Bölting about identity and inclusion, talking about his award-winning paper on identity and CS education.
The next session in Shrimp was on interest:
- Shirley de Wit (Leiden University) presented Children’s Implicit and Explicit Stereotypes on the Gender, Social Skills, and Interests of a Computer Scientist, which investigated stereotypes in CS. They found many weak implicit associations about computer scientists’ social skills amongst Amsterdam youth, with an increased association amongst older youth.
- Alex Leshinki and Joshua Rosenburg (University of Tennessee) presented All the Pieces Matter: The Relationship of Momentary Self-efficacy and Affective Experiences with CS1 Achievement and Interest in Computing, which used an experience sampling method to look at experiences over time and students and tried to link them to end of course outcomes. They found significant relationships between specific, concrete experiences that shaped self-efficacy (e.g., frustrating moments) and long-term interest and achievement.
After the talks, I went back to my cat room to see who else might want to talk about pets. I chatted with Lara Schenck about unplugged learning with adults and API learning.
After a short break was the conference poster session, which included more than twenty poster presenters. I talked to:
- Julie Smith, who presented Beyond the Gender Binary in Computing Education Research, which offered some new lenses to think about gender and inclusion. We talked about the challenges of incorporating small samples of non-binary students into statistical analyses.
- Cassandra Broneak, who presented Developing Empathy and Persistence through Professional Development in New to CSA Teachers, who exampled the experiences of durings during AP CS A professional development.
- Wei Yan (University of Florida), who presented Elementary Students’ Debugging Behaviors in a Game-based Environment, which found that elementary students mostly did trial and error, but they did eventually learn some systematic strategies.
- Kai Presler-Marshall (North Carolina State University) presented
Towards Better Support for Undergraduate Software Engineering Teams, which investigated collaboration and coordination challenges that teams face. - Swaroop Joshi (BITS Pilani Goa) presented Teaching Accessibility in India: A Work in Progress, which investigated opportunities to teach accessibility in CS educaiton.
Overall, the poster session worked pretty well, making it reasonably visible which posters lacked an audience, and providing a lot of information about each poster before approach it to meet its presenter.
Thursday —Talks and Closing
In the last welcome on Thursday, I met up with several lovely folks from Omaha and San Diego and we compared notes on vaccine and mask mandates on our campuses. We then went to the opening plenary, where we held a group photo session, getting everyone on to a big Zoom call to see each other in a large group, as Clowdr mostly connected us in small groups and via text chat. It was a great chance to see everyone’s face, smile, wave, and appreciate the diversity of people and places around the world engaged in computing education.
There were two final paper sessions. I once again went to “Shrimp”, this time to see papers on programming languages and functions:
- Ethel Tshukudu presented Teachers’ Views and Experiences on Teaching Second and Subsequent Programming Languages. She observed that both primary, secondary, and post-secondary often teach multiple languages. They interviewed 23 K-12 teachers in the Netherlands and Scotland and found that teachers had to orchestrate scaffolding and pacing very carefully to promote transfer, and that there is a need for further research transfer strategies to help teachers consider students’ prior language knowledge.
- Shriram Krishnamurthi presented Developing Behavioral Concepts of Higher-Order Functions, which examined students ability to reason about higher order functions with their input-output behavior. They found that students had several difficulties with identifying higher-order functions from examples, although they did improve over time. This suggested the need for substantial future research to understand the implications of this for library and API learning. The paper offered several methods and instruments that should be useful for this future work.
This session was a popular one; many folks stuck around in the discussion room after the session to continue to debate about programming language design, learnability, transfer, patterns, and schemas. One of the interesting points that Kathi Fisler made was about the tension of justifying why to learn languages, when students and teachers already know particular languages and design patterns.
For the last session, I went to “Grits” to hear about two very different papers about algorithm visualization and natural language programming:
- Albina Zavgorodniaia (Aalto University) presented Algorithm Visualization and the Elusive Modality Effect, which considered which modality (visual, auditory, haptic, etc.) to present algorithm examples. They conducted a controlled experiment on different modalities for Dijkstra’s algorithm in an algorithms course and found that modality was not a major factor in learning outcomes, suggesting that integrating audio does not seem to be of reliable value.
- Arthur Hjorth (Aarhus University) presented NaturalLanguageProcesing4All: — A Constructionist NLP tool for Scaffolding Students’ Exploration of Text. The paper described a tool that enable students to explore words, label them, explore agreements and disagreements about text classification, and then examine text analytics; it also allowed them to offer features to a classifier and show them the classification results. Experiences with using this in classrooms showed promise for text classification as a promising activity for understanding both social science and machine learning.
Reflections
Overall, I was personally very happy with how the conference went, on three levels. First, I was happy as an attendee: because I managed to block of all 12 hours of the conference well in advance, I had time, which has been the hardest thing to achieve over in other conferences that demanded 24/7 availability (e.g., CHI 2021). The experience of socializing a bit, learning a bit, socializing a bit, learning a bit, for about 3 hours each day felt very manageable and pleasant. I didn’t get to have the kind of deeper social and intellectual experiences that I normally do at in-person conferences, but I did feel like I got to connect broadly with new and returning attendees, maintaining relationships and starting new ones. If I weren’t able to attend an in-person conference, I would want this participation option.
I was also happy as an organizer. Clowdr didn’t go down, there were only some minor streaming issues, and despite the many usability issues with the platform, we managed to train most attendees over the course a few days to the point where people could navigate. We haven’t yet done any systematic feedback gathering, but in the conversations I had, people reported being pretty pleased, especially relative to some of the more disastrous experiences they’ve had with virtual conferences in the past year.
Lastly, I’m so excited to see that the community continues to increase its rigor, depth, and intellectual scope. The papers were more diverse than ever, many papers were by first time attendees, and the number of papers on primary, secondary, teachers, and integrations of CS in other domains was greater than ever. It was incredibly fun to engage with this diversity of questions and scholars, while still maintaining a strong core of scholars interested in common questions about introductory programming.
Of course, all of this is just my own experience. I’m sure some attendees had a difficult time with the platform and I just didn’t hear about it; I’m sure some attendees were frustrated by the content. I hope everyone that attended will share their feedback, and help next year’s organizers offer an even better ICER 2022!
