The first session was about neuroaesthetics and user experience. This was a great talk. It went into areas of neuroscience as they apply to user experience design, such as gestalt pattern theory and high and low frequency visuals. I'll need to look into these topics in greater detail later so I don't lose this new information, but it was definitely a rich presentation that was interesting and useful. The speaker was excellent too, not just the topic. If you get the chance to hear Alyssa Pilney present on a topic, take it.
The second I thought was going to be a Rust session, but that was moved to 3:00pm. As a result I had to pick something else. I started with a talk about open source at schools, but that ended up just being a sales pitch for open source in general. I've seen that too often already. So, I switched to a talk on machine learning in music composition by Leif Bjornson. The machine learning session, unfortunately, had some assumptions about audience knowledge levels that precluded me. So I left the machine learning session on in the background and worked on other things, including this blog post. As it happened, I'm glad I left it on, because the session got really interesting after the technical portion. I started paying attention again shortly, and I learned about some really cool tools like Google Magenta and Musetree. I'll have to explore those in more depth sometime.
At noon, I dropped into SPS Commerce's chat about running successful tech communities. It seemed more geared for how to build a tech community within a corporation, rather than an unaffiliated tech community. I was hoping for some ideas for running the DevOps learning community in SDG. It was not terribly useful.
The one o'clock session was about observability. There were some good nuggets of wisdom in this talk; it was presented by a couple guys from Elastic. They suggested standardizing on a specific time zone (GMT), being careful to use NTP correctly, adding schema on writes instead of reads, and enriching logs with context whenever possible. They mentioned the Elastic Common Schema as a good starting point. The last point I took away from it was that it's rare for a single event to be actionable; context determines action. After this, it went into what was essentially a sales pitch for Elasticsearch and Kibana. Until then, it was really informative.
I skipped the OpenTelemetry session I'd planned to attend at 2:00pm. I was starting to get bored at this point. After a review of the other sessions available in that time slot, I decided to spend the time doing something more productive until the final session.
The final session I wanted to attend was a Rust adoption journey. However, I discovered that there was an Overwatch 2 developer livestream occurring at the same time, so I decided to catch up on the Rust session in a recording later. I was most excited about the Rust session out of all the sessions I'd planned to attend, but as it was being recorded and the devstream was not, I chose to watch the latter.
All in all, it was a great small conference. I'm definitely looking forward to another in-person version in 2022. Next year I'll be more careful in my choice of sessions!