Kieran Healy
On Golden Gate, you can chat to the built-in foundation model directly with fm chat. You can also do things like pipeline it into scripts. E.g., in the second screenshot here it's reading and summarizing a @sixcolors@zeppelin.flights article. The model's context window is quite small, so I had to write a script that chunked the text, so it's sort of serially summarized. Still, it’s nice to have fully-local access to this model in this way.
I just watched Gremlins 2 for the first time since 1990. The entire film is built around a dim, vain, tasteless New York developer named Daniel Clamp, his sprawling Clamp Building and his gimcrack Clamp Empire, which includes TV stations, his inane self-aggrandizing books, and a line of jams and jellies. At the end he falls for a character named Marla. This has been a remark about how the meaning of a work of art is not something fixed or under the control of its creators.
A complete transformer neural network implemented entirely in HyperTalk, trained on a Macintosh SE/30.
Here’s a full draft of the upcoming second edition of my “Data Visualization: A Practical Introduction”: https://socviz.co
Yesterday’s fun activity was writing a package to make cartograms like this. I am also open to lavish VC funding to develop an excessively complex German-style board game based on this layout, modeled on e.g. “Escape from New York”, “Warriors”, or perhaps “My Dinner with Andre”.
@siracusa@mastodon.social A nice example of context collapse that I came across while listening to the Aftershow.
Using Quarto to write (and typeset) a book.
https://kieranhealy.org/blog/archives/2026/03/09/using-quarto-to-write-a-book/
@hotdogsladies@mastodon.social A nice big Baby Boom poster for you about the dread year of 1946 and the march of time in general, as per @RecDiffs@mastodon.social discussion. Some more detail at https://kieranhealy.org/blog/archives/2023/12/20/the-baby-boom-again/
No one can force me to have a secure website!!!
@atpfm@mastodon.social @siracusa@mastodon.social Lol you lot are going to get email from annoyed “U Penn is an Ivy League university and not some random state school” people. (I know you didn’t claim it was but I predict your discussion will trigger them; they are very touchy.)
RIP Tony Hoare, father of Quicksort. https://blog.computationalcomplexity.org/2026/03/tony-hoare-1934-2026.html
New York City Hexmaps. https://kieranhealy.org/blog/archives/2026/04/19/new-york-city-hexmaps/
Vibes-based analysis from me of the 6C Report Card. https://sixcolors.com/post/2026/02/charting-the-vibes-in-the-2025-apple-report-card/
I thought I had found a clever way around an annoying problem—I have to get hundreds of randomly-named SVG elements that correspond to ordered elements (train stations) in a CSV. So I concocted a script in Illustrator to extract, all at once, the layer names of a bunch of elements after I have cmd-clicked them so they're all selected. I am so clever. But ... Illustrator cannot extract them in the order I clicked them, only in layer order, which is itself random. Gaaah.
Borough-specific sign for the day. #NoKings https://kieranhealy.org/blog/archives/2026/03/28/subway-sign/
