No. 119 — Forty days of rain ⫶ The least probable word ⫶ Richardson's ribbons

Words became compute

My name is Linda. I write a bi-weekly newsletter about computer science, childhood, and culture.

It’s been raining for forty days in France. By the Seine, there’s a stone soldier called le Zouave du Pont de l’Alma, and Parisians have used him to read the river since 1856. The water reaches his toes, and the city holds its breath. But spring is almost here, and things are beginning to crack open.

My son’s language, for one. Like any proud parent, I’m in awe of what comes out of his mouth. Unconventional, fresh, and weird. He says things no one has ever said before, in three languages, in combinations that shouldn’t work but do. I’ve been thinking about why. A language model is optimized for the most probable next word. He is optimized for the least probable one. The AI companies have linguists, mathematicians, physicists. What they need is a Korney Chukovsky.

1.

Theater has been on my mind since I finished Tom Stoppard’s Arcadia last week. The play blends two timelines in an English country house (one set in 1809, the other in the present day) braiding chaos theory, landscape gardening, literary detective work, thermodynamics, and a character who feels very much like a young Ada Lovelace. (It’s running at the Old Vic in London for another month!) It reminds me of Michael Frayn’s Copenhagen, which I saw years ago: Heisenberg, Niels Bohr, and Margrethe Bohr circling each other in the afterlife, trying to reconstruct a single conversation about uranium. Both plays talk about science on the surface, but are fundamentally about the limits of knowing. And in both, the knowing happens the way it always does on stage: through dialogue. People talking, disagreeing, circling each other, never quite arriving.

And in 2026 we are all, suddenly, in dialogue with machines. The strange thing about language models is that they’ve turned computing into a conversation. And it made me think of a passage from Sarah Ruhl’s 100 Essays I Don’t Have Time to Write: “I’m not sure that I believe in progress, because I make theater. And theater, by its nature, does not believe in progress. We must constantly go back to go forward. Theater cannot believe in absolute knowledge, because usually two or three characters are talking and they usually believe two or three different things, making knowledge a relative proposition.”

Maybe that’s why I find it so interesting when theater people encounter generative AI. Chris Ashworth makes macOS software that automates lighting and audio for live theater productions, and runs a theater called Voxel in Baltimore. He used Claude Code to build a custom lighting design application for a very niche project, an app whose audience was three people (!). He’s careful, almost reluctant. “I really hate that I’m saying this,” he begins. I believe him. More software by people who can hold two or three truths at once, please.

2.

This spring, I’m running a workshop for twenty Texas K–8 teachers called Playful Computing.

Six Fridays on Zoom, May through July. It’s the workshop I’ve wanted to build for a long time, and everything I’ve learned about teaching computer science over the past decade. I’ve worked hard to make the course something beyond talking heads. The course is funded through WeTeach_CS at UT Austin, which means it’s completely free for participants. They also earn 30 hours of PD credit and a $350 stipend upon completion. The application deadline is April 8th!

(If you’re not in Texas but wish there were something like this near you—reply and tell me where. I’m keeping a list!)

3.

I’ve been trying to understand how to read protein structure visualizations. You know the shapes that are everywhere: the coiled ribbons, the pleated arrows. After AlphaFold, they’ve become the symbol of a certain type of science, or even a worldview.

But there is a language behind them. And that language has a backstory. The ribbon diagrams we all recognize were first drawn by hand by Jane Richardson, a biochemist at Duke, in the late 1970s. She drew the way a cartographer draws. What I love about Richardson’s work is that she looked at the data and found a visual grammar for it. Every molecular visualization tool today still uses her hand-drawn visual language. AlphaFold’s most celebrated outputs are rendered in Richardson’s ribbons.

4.

Molly guard in reverse by Marcin Wichary. In old-school computing, a "molly guard" is the little plastic safety cover over a button of significance. It’s named after an engineer's daughter who visited a datacenter and promptly pressed the big red button.

5.

Ten years ago, first grade teacher Autumn Zaminski led her students on a week-long adventure in search of Ruby’s missing gems, hiding clues in envelopes, building HexBug mazes, finding necklaces with green gems.. I remember seeing the photos on Twitter and thinking: whatever happens with this book, I’ve made it. Because that’s the kind of experience kids carry with them.

Now Ms. Zaminski is back, this time on the CT Cafe podcast, sharing her favorite computational thinking books.

6.

Last spring I started work on a new Hello Ruby book on sorting algorithms. But then something different started feeling more urgent. How words became compute. Or, put more simply: how to explain a large language model to a five-year-old.

Robin Sloan wrote something recently that read like a rallying cry: that AI language models are “a particularly, maybe even uniquely, human technology.” And that we all have a stake in this, simply by virtue of being a talking animal. I love that framing. It felt like enough of a reason to start drawing.

A few books have been sitting beside me as I work. I wanted to share them early, because I know how creative many of the educators reading this are at pairing picture books with computer science (just look at what you did with They All Saw a Cat in No. 68!).

• The Word Collector by Peter H. Reynolds, about a boy who collects words instead of stamps and discovers what happens when you start combining them.

• The Story Machine by Tom McLaughlin, about a child who finds a mysterious machine that turns letters into stories.

• Another by Christian Robinson, about a girl who slips into a world that mirrors her own, almost but not quite.

If you've used any of these in your classroom, or if you have other picture books that make you think about words, language, and machines, I'd love to hear about it.