I had the pleasure of attending the first North Bay Python conference in Petaluma, CA this past weekend. IBM was a sponsor, and I gave a few quick remarks about doing python serverless actions on OpenWhisk. My two days there were full of wonderful talks and interactions with a slice of the python community.
One of the reasons I love low cost (to attendee) regional conferences like North Bay Python is that it makes technology conferences more accessible. For 40% of the 250 attendees, this was the first technology conference they'd ever gone to. Not everyone lives in New York City or San Francisco (or wants to), and having local events is critical to expanding the range of voices in the technology community.
There were tons of great talks, you can watch them all here. But I'll highlight a few moments that I'll keep with me for a while.
Fortran on stage
For a single track Python conference, we actually got to see FORTRAN in 2 different talks. It's probably more FORTRAN than I've ever read before.
Catherine Moroney is part of the team that does analysis of satellite images from the LandSat program. They've got a lot of optimized FORTRAN and C code for processing these images. But FORTAN and C aren't great languages for writing new features to orchestrate these lower level transforms. She uses Python to do this work, and can seamlessly pass data back and forth from Python to FORTRAN for data crunching. It was great to see how and when a hybrid approach like this makes the developers much more effective.
Christopher Swenson tackled FORTRAN from the other side. He hacked together a FORTAN IV interpretter in Python, so that he could run Colossal Cave Adventure (originally written for the PDP-11) as a text message game using the Twilio API. His talk wandered through some of the interesting quirks of now extinct programming languages, and the systems they were written for. This is a world before ASCII as we know it became a standard, and the idea of 32bit integers really hadn't emerged. 36bit integers were used to store 5, 7bit characters, which were later assembled into text streams.
Through the whole thing he showed snippets of FORTRAN that he had to make guesses about what it really meant, as well as be really frank on shortcuts he made to get things to work. There is no more FORTRAN IV code in the world, this didn't have to be a perfect emulator, it just had to run this one single FORTRAN IV program well enough to connect it to the internet.
You can play this by texting to +1 (669) 238-3683 right now if you want to see it in action.
My vote for most hilarious talk was Benno Rice's dive into writing twitter bots. He started with pretty easy template base bots, like one producing plausible plot lines for Mid Summer Murders. This is just a stack of well crafted arrays and a random number generator.
Things got more interesting when he started diving into Markov Chain bots. Especially where you take content from a bunch of different sources. It's really easy for that to just become word salad at worst, or just confusing and "meh". He found you had to keep playing with the content mix as well as the overlap parameters to get the bots to generate something that's amusing at least some of the time. The bots he's got he doesn't let post directly, content is generated offline, and he pushes the good ones after manual review.
Benno also took a dive down the path trying to do machine learning to make these better, but mostly got worse results in his experiments. But, the story of that not working out was funny all by itself. The real lesson here is that playfulness is useful in learning some new things, and that Twitter bots are a lot of fun to build.
Search First Documentation
My vote for most takeaways that I'll personally use goes to Heidi Waterhouse for "Search-First Writing for Developers". Recently there as a mass migration of OpenStack Documentation from a dedicated docs team to all the development teams.
The heart of her message is that to any first approximation, no one reads your documentation. Users end up at your documentation when they have a problem with your software, so they are showing up a) grumpy, and b) through whatever Google terms they could guess for their problem. They are not coming through your carefully curated table of contents, they are coming from Google, and then they are skimming to find their answer. The won't follow links to other pages, this is where they are.
What that means is you need to treat every page as the only page that the user will ever see, you need to optimize your content for skimming, and you need to answer problems people actually have, not the ones you think they might have. Getting real analytics on how folks are reading your docs, and the search terms they are coming in with, is an important part of this.
Hearing all these harsh and practical words from someone that spent 15 years as a technical content author was really enlightening. I'll definitely have to watch this talk again and digest more of Heidi's lessons.
One of the welcome trends that I've seen at tech conferences over the last 5 years is a real focus on a strict Code of Conduct, clear reporting guidelines, and making sure that folks feel safe at these events. North Bay Python did a great job on that front, and that commitment definitely was reflected in a pretty diverse speaker lineup and attendee base.
The effort they went to was highlighted further by Seán Hanson's talk on Quiet Developers. We've long known that while diversity in Tech is much lower than national averages, it's ever worse in Open Source Software. A big reason for this is members of traditionally marginalized communities really don't feel safe in these environments, or may not have the spare time to devote outside of their normal day jobs. It doesn't mean they aren't great developers, it's just that current systems are optimized for loudness as much as talent. Seán's whole talk was ways to engage and get the most out of your quiet developers, and give them what they need to really succeed. While I did need to leave about the time this talk started, I stuck around and watched from the balcony. His message was really powerful and really important to how we all evolve the tech community going forward.
Double A Plus, Would Come Again
North Bay Python was definitely worth the trip. It had a few normal quirks of a first time conference on scheduling. Being Petaluma, the Theatre didn't actually have heat, so the first few hours the first day were a bit cold in there. But it warmed up pretty quickly with 250 bodies. The biggest issue in my mind was there wasn't much common space outside of the theatre, so a hallway track wasn't really a thing. It would have been nice to have a bit more milling about time to get to know folks there, and ask follow up questions of speakers.
But all in all a great time. Looking forward to seeing how they do next year.