Day 2

The first talk was about data science in Haskell. He had many very cool demonstrations about how he implemented interactive notebooks for Haskell for data analysis, and how Haskell’s static type system prevents common data analysis accidents (like taking an average of non-numeric data).

There was a panel on LLM generated code which got a little heated. Some of the people on the panel said they spent $7000-$10000 a month on tokens, which is just wild. A point that people kept making is that only a handful of people are really benefitting from the productivity of AI, for everyone else their workload and expectations just keep increasing. It’s not like they get to do their work any faster and spend time with their families, they just have to do ten times more work. They also talked about how worker morale is at an all time low as workloads increase, their actual work is less enjoyable, and they are increasingly worried about being laid off. Personally, I don’t have any work that just needs to get done so I don’t find much use for AI. Usually I care about the process and learning how things are done, so using AI would rob that from me. I find it scary that since it’s already better than me at pretty much everything that it can do, and it can get better faster, in a doomsday scenario I might just be locked out of a job forever while the world ends, but I’m not that pessimistic yet.

I spent more time in GHC room reading about the compiler and asking questions as they came up for me. After running into every single issue possible, I finally got the thing to compile. The code has tons of comments explaining what each part does, but the who project is so large and interconnected it is very difficult to understand.

Final day

The last morning presentation was about why a company might stop using Haskell. The most interesting comment he made was that people receive more credit for putting out fires than preventing them. Haskell is good for its stability and promise of zero runtime exceptions, but that comes at the cost of a slightly longer development time, and managers only see that and ignore the fact that Haskell code is much easier to maintain.

Also today there was a long presentation on category theory, which was so exciting for me. The presenter was the author of a book I’ve read and enjoyed. I finally got to talk to another real person about category theory. Most of the introduction stuff I already knew, but he got a lot more into the connections between category theory, logic, and type theory.

Finally at the end of the conference I had a conversation with a developer who, like me, has a math background. He gave me some suggestions on how to improve my polynomial library, and some papers to look into. I did some type-level programming to ensure at compile time that all the polynomial operations are valid, and it’s pretty awesome. I definitely want to get more into it and learn more about what you can do with the types in Haskell.

Reflection

A few final thoughts about my whole trip:

• I was so allergic to something here, I could barely breathe the entire time. Apparently it’s been very dry and a they’ve gotten a record amount of pollen.
• I really lucked out on weather this trip. Before I left, I saw 90% rain every day for the whole week while I was here. Even while I was here the forecast was rain, but I only experienced it twice.
• I didn’t forget anything and I didn’t lose anything, which is something I had been apprehensive about.
• The famously reliable Swiss public transport system was only late when I was trying to catch a plane.
• I experienced the roughest turbulence I’ve ever had on my plane ride from Zurich to Dublin.

Similar to Amerihac in January, I’ve come away from Zurihac feeling super inspired. It is truly amazing how prolific such a small community of developers can be. They volunteer their time and experience to build and maintain these incredibly large open source projects. If I continue to keep learning and figuring stuff out, I hope that I can eventually keep up more and be a contributor to these projects.