Late last year I was invited to give a keynote at the Southern Data Science Conference, which was originally scheduled to be in April 2020. In January the organizers asked me for my talk title. Most years in early January I have no idea what will be top of mind for me come April, and this year I would have been wronger than in most years.
So I gave a placeholder joke title, thinking that I’d update it at some point. And then the conference got delayed, and I got busy with other things, and now here we are a couple of weeks from the conference with me scheduled to give a talk that currently is mostly just a lazy Simpsons reference.
I consider this mildly embarrassing, as I have been working hard over the last couple of decades to purge lazy Simpsons references from my thought patterns. But now I’m sort of stuck with it.
And I guess in some sense this is good, because (at least for me) constraints breed creativity. If you ask me to tell you a joke, I’ll likely draw a total blank. My head is full of jokes, but I don’t know how to randomly access them. I know how to write jokes, but not on-demand about nothing in particular.
However, if you were to ask me to tell a joke on a certain topic (e.g. nuns) I could likely conjure several out of the depths of my memory, although it’s likely that they would be schoolyard jokes from when I was 12 and quite not appropriate for [current year]. I could probably also make up a few on the spot, e.g. “why did the nun dress in black? it was her habit”. (I’m sure this is a well-known joke, it’s pretty obvious, don’t @ me.)
Which brings us back to our data science talk. If you were to say to me, “you need to give a data science keynote in 2 weeks, better get writing” I would panic. What would I even talk about?
But in this case, thanks to past-me tying my hands, my task feels quite a bit easier. It’s much less panic-inducing to think about writing the “chilling vision” talk:
come up with one or more chilling visions
make slides
Here’s the idea: pick a bunch of topics (“deep learning”, “data ethics”, etc…), and for each one contrast what it was like in 2010, what it’s like now, what it will be like in 2030, and what it will be like in 2040.
This makes the talk an interesting exercise in several things:
characterizing / caricaturing proto-data-science circa 2010
characterizing / caricaturing the state of things in 2020
making somewhat outlandish “in ten years” predictions / jokes
making extremely outlandish “in twenty years” predictions / jokes
practicing my reincorporation skills
I suspect I can do all these things.
What’s more, a talk like this requires a strong “chilling vision” aesthetic, which gives me the opportunity to play art director:
So now for the next couple of weeks I’ll leave Google Slides open in a browser tab and return to it throughout the day, adding ideas, rewording, joke-smithing, deleting, and so on. I always have my best ideas in the shower, and I’ve still got about 10 showers until the conference, which is almost certainly enough for a talk’s worth of good ideas. My goal is to get to 200 slides, which might even be a new record for me. The talk is starting to veer towards the dark, but we’ll see where it ends up.
If I were smart I’d be approaching this talk as an opportunity to sell books rather than as an opportunity to make cheap jokes. But I’m not that smart.
I’m curious to see how it turns out. Maybe I’ll start doing this for all my talks!