Knowing and Doing

the video screen announcing Philip Wadler's talk

Friday was a long day, but a good one. The talks I saw were a bit more diverse than on Day One: a couple on language design (though even one of those covered a lot more ground than that), one on AI, one on organizations and work-life, and one on theory:

• "All the Languages Together", by Amal Ahmed, discussed a problem that occurs in multi-language systems: when code written in one language invalidates the guarantees made by code written in the other. Most languages are not designed with this sort of interoperability baked in, and their FFI escape hatches make anything possible within foreign code. As a potential solution, Ahmed offered principled escape hatches designed with specific language features in mind. The proposed technique seems like it could be a lot of work, but the research is in its early stages, so we will learn more as she and her students implement the idea.

This talk is yet another example of how so many of our challenges in software engineering are a result of programming language design. It's good to see more language designers taking issues like these seriously, but we have a long way to go.

• I really liked Ashley Williams's talk on on the evolution of async in Javascript and Rust. This kind of talk is right up my alley... Williams invoked philosophy, morality, and cognitive science as she reviewed how two different language communities incorporated asynchronous primitives into their languages. Programming languages are designed, to be sure, but they are also the result of "contingent turns of history" (a lá Foucault). Even though this turned out to be more of a talk about the Rust community than I had expected, I enjoyed every minute. Besides, how can you not like a speaker who says, "Yes, sometimes I'll dress up as a crab to teach."?

(My students should not expect a change in my wardrobe any time soon...)

• I also enjoyed "For AI, by AI", by Connor Walsh. The talk's subtitle, "Freedom & Evolution of the Algopoetic Avant-Garde", was a bit disorienting, as was its cold open, but the off-kilter structure of the talk was easy enough to discern once Walsh got going: first, a historical review of humans making computers write poetry, followed by a look at something I didn't know existed... a community of algorithmic poets — programs — that write, review, and curate poetry without human intervention. It's a new thing, of Walsh's creation, that looks pretty cool to someone who became drunk on the promise of AI many years ago.

I saw two other talks the second day:

  • the after-lunch address by Philip Wadler, "Categories for the Working Hacker", which I wrote about separately
  • Rachel Krol's Some Things May Never Get Fixed, about how organizations work and how developers can thrive despite how they work

I wish I had more to say about the last talk but, with commitments at home, the long drive beckoned. So, I departed early, sadly, hopped in my car, headed west, and joined the mass exodus that is St. Louis traffic on a Friday afternoon. After getting past the main crush, I was able to relax a bit with the rest of Zen and the Art of Motorcycle Maintenance.

Even a short day at Strange Loop is a big win. This was the tenth Strange Loop, and I think I've been to five, or at least that's what my blog seems to tell me. It is awesome to have a conference like this in Middle America. We who live here benefit from the opportunities it affords us, and maybe folks in the rest of the world get a chance to see that not all great computing ideas and technology happen on the coasts of the US.

When is Strange Loop 2019?