To finish up, we'll look at some of the other sessions I attended but first a couple of notes.

One thing I again noticed was the lack of pedagogy sessions. Maybe I missed something but other than Nifty Assignments, which isn't really about pedagogy and doesn't interest me much anymore I didn't see anything. This is nothing new but at least a couple of years ago we had "It seemed a like a good idea at the time" coordinated by Jim Huggins and Dan Garcia in all their pink suited glory and Colleen Lewis's "Micopedagogy" session (I hope I'm remembering the name right). Jim and Dan's session wasn't purely about pedagogy but as it happened there was a good deal of it covered.

That's a shame given that the higher ed teaching community could probably use it.

The other was a format change. The paper sessions all had an additional paper per session which cut the time per paper down by a few minutes. I think the stated goal was to keep the number of accepted papers up and to minimize the number of parallel sessions. I'm not sure if it affected the # of parallel sessions in practice - there were fewer rooms but since moving from room to room mid paper is far from ideal I wonder if there was any benefit. In any event, I'm glad the organizing team is thinking about things like this and trying things to improve the conference.

For this run through though, I think it led to some unevenness in presentation. Some sessions I went to, the speaker basically just covered their methodology barely speaking about the paper before running out of time. Another wasn't able to get to the methodology at all. I'm sure, assuming this time format remains the same, speakers will readjust so not a big deal, just something I noticed.

I attended a couple of sessions that came back to the BOF and keynote theme I wrote about earlier - reforming and possibly simplifying the CS major. One speaker talked about how at his their institution they looked at prerequisite trees for both their CS major and CS minor and found a good deal of redundancy as well as superficial information. They found, for instance, that a course might have one or two important topics for the CS student - say dot product for instance but the majority of the class had no relevance to the later CS classes. They also found redundancy in that one needed topic might be covered both in multiple prerequisite classes as well as in other required CS classes.

I'm betting this is true for most CS programs across the country and I'm really glad that there seems to be more light being shed on it.

This was followed by a session run by Richard of Regis University who looked at similar issues by proposing "10 wicked curricular issues."

There were a few sessions on pedagogical topics but I wouldn't call them about pedagogy. A couple were on live coding. These were interesting because they were back to back and while one seemed to find that the live coding did result in improved student learning, the other did not. Just shows the complexity of education. Another session talked about algorithm classes and noted that classes and texts were taught from a mathy theoretical point of view and argued for a more experiential approach with students writing programs to implement algorithms and then make use of them. I was in agreement with the speaker and left wondering how correct he was in his premise. What's going on in algorithm classes out there? Are students just learning a catalog of Data Structures and Algorithms and proving runtimes or are instructors working to develop a problem solving sense with their students while exploring key known data structures and algorithms along the way?

I attended a bunch of other sessions but for brevity I'll only share my thoughts on one more. The paper was "Why is my code slow? Efficiency bugs in student code."

The paper and talk looked at what the speaker described as "efficiency bugs." Things like superfluous computations, suboptimal data structures or algorithms, caching or failing to do so.

The speaker used as an example, using a list to check for uniqueness of data rather than using a hash table or a set. It's actually very much in line with the mode lesson I taught at Stuy a couple of weeks ago.

It seemed like the intervention was in the form of feedback on projects but it would be nice to weave these concepts so that they come up in lessons.

So, that was my SIGCSETS 2025. Pittsburgh was a great host town - lots of food options close to the convention center and lots to explore for those of us who came early or stayed late.

Next year is in St. Louis. Even though retired I'll probably head out to that one as well.