Chris Krycho

@chriskrycho
451 Followers
84 Following
2.5K Posts

Anglican Christian. Software engineer and composer, theologian, writer, runner, photographer, and erstwhile podcaster. Platform engineering at Vanta. Previously the tech lead for LinkedIn web, Ember TypeScript & Framework emeritus.

→ chriskrycho.com

websitehttps://v5.chriskrycho.com
GitHubhttps://github.com/chriskrycho
LinkedInhttps://linkedin.com/in/chriskrycho
Blueskyhttps://bsky.app/profile/chriskrycho.com

If, by some chance, I happened to beome exceedingly wealthy from my work on software, I would, in order:

1. Endow funds to care for the poor in Colorado.
2. Build a building or three like Saint Thomas on 5th Av. in Manhattan for the Anglican churches in Colorado.

Everything else would be gravy.

I’ve spent at least twenty minutes composing three mornings this week. Over an hour! That’s not much—truly—but it’s more than I’ve done any other single week in 2026. Building momentum is the thing, and I’m doing that.

Of course, I’m traveling for work next week. 🤦🏻‍♂️ We’ll see what I can make stick!

RE: https://zirk.us/@grammargirl/116704173778333410

A few other reasons to avoid the term "hallucinate" in the context of LLMs:

—It anthropomorphizes genAI, which is a harmful reflex.

—It's a popular industry term for a reason. It perpetuates and strengthens the way genAI companies want us to perceive the technology.

—Maybe most saliently: GenAI does nothing *but* hallucinate. Using the word to describe only some outputs implies erroneously that it's not hallucinating *all the time*.

#AI #genAI #LLM #language #technology

I installed the macOS 27 beta in a VM—so I cannot try the Apple Intelligence features, but that’s fine 🤪—and my judgment there so far is: definitely step in the right direction design-wise, but… I’d still much rather use Sequoia.
How in the world had I gone until today and not heard Copland’s Symphony No. 3—and therefore never realized that he integrated “Fanfare for the Common Man” right into the fourth movement? And here I’ve always worried about plagiarizing myself even a little bit. 😂

RE: https://tech.lgbt/@jnadeau/116722266648560507

I’ll just say it: this should be the default! Look how nice it looks for interface elements to, you know, be clearly delineated!

(I’m really glad this kind of thing is in there; the team has clearly taken a lot of feedback to heart and this looks like a really good iteration in the right direction. But I stand by my claim!)

The tradeoffs around this for “rough proof of concept” vs. “stable production feature” vs. “load-bearing infrastructure” matter a ton, of course. But I wish there was a lot less hypemaxxing and a lot more honesty about just how much even thoroughly-steered LLM output often needs of tons of rework.

The thing with Claude Code and Codex since that big jump late last fall is: they do in fact very often produce things that work.

And also when you poke at *how* they get things to work, it is very often a nightmare that will punch you in the face.

Doesn’t mean they’re useless. Sure does require vigilance.

I miss writing poetry. Stumbled back across this one from 12½ years ago, and: I still like it! https://v3.chriskrycho.com/art/way-world-ends/
The way the world ends » Ars Artis » Chris Krycho

RE: https://hachyderm.io/@nrc/116721598056285910

Bookmarked this to read it later—looks like it should be a handy resource! (Read anything @nrc writes about Rust!)