PyCon 2016

I just arrived home from PyCon 2016. It was a fantastic experience, on par with the year before. I didn’t go alone this time: I brought four of my great co-workers and also ran into a lot of old and new friends.

It was so hard to decide which talks to go to this year! In each slot, I had two or three I wanted to see. From early reports, I missed a lot of great ones. Luckily, PyCon posts all the talks online, so I’m going to go back and watch a lot of them.

Of all the talks I saw, Matt Bachmann’s talk on property-based testing was my favorite:


What I liked most about this talk wasn’t how to do property-based testing – that is, you write specifications for what sort of values your functions and classes can take and some sort of verification that they work and then your test suite generates a billion variations of what should be valid data to see if it breaks – but what kind of tests to write with property-based testing. I’d known of the concept for a few years, and used it once before when building a serialization format, but I didn’t know what to do with it in general, and now I do.

Another talk that I loved was Rebecca Bilbro’s talk about using visual diagnostics to better understand machine-learning algorithms:


Yes, that is my voice at the beginning. I was session chair for this talk, which I highly recommend volunteering for – you end up paying very close attention to the talk.

I secretly wish I was a language designer, so Juozas Kaziukėnas’ talk on building an interpreter in RPython was right up my alley. I loved that he admitted that the RPython docs are limited and he basically copied someone’s example language and edited it to make it his own.


Lastly, the keynotes at PyCon are always so good. This year did not disappoint and I plan to go back and re-watch some of them. The last one by K Lars Lohn, was beyond words. I don’t even know how to describe it or even how to say what it was about. Some of the topics were the fractal dimension of shapes, diversity of thought, music, and overcoming failure. It was a masterwork of performance – I don’t know how he synced up some of the stuff he did. Anyway, just watch it.


I’m collecting ideas from others about what talks I should go back and watch that I missed. If you have a favorite, please send me a note on Twitter and I’ll watch it. I’m planning on writing up my thoughts on the ones I missed the first time after I watch them.