Chris Krycho

Digital technologies can be obviously dreadfully destructive, but also really wondrous tools for good. Their sheer generality is both strength and weakness. How do we bend them toward humanism and away from the soul-deadening traps they often become? v5.chriskrycho.com/essays/comp...

My goal is to compose 3× every week. Today marked the 7th day I’ve composed this month. You can do the math: that’s not close to 3×/week this month—and I’m trying to use the motivation of having made progress to double down, because even just a few bars of music every day really adds up.

🎼🖊️ So far this month, I’ve managed to (re)write a bunch of musical ideas and integrate them into the 4th movement of the symphony, net 45 seconds or so of musical material. Will I finish drafting this year? Odds seem good with some planned vacation time coming up. 🤞🏼

github.com/zip-rs/zip2/...

Bump the zip crate, new vuln published.

as someone who uses llms, i'm totally fine, even happy, about lobste.rs being an anti-llm space. if you want pro-llm stuff you can look anywhere else on the internet. if we want a moderate sort-of-llm community we can create one. niche communities with hardcore value systems are good.

Wrote up a long review of the 80/20 Endurance Norwegian Training Marathon Plan on my blog: v5.chriskrycho.com/journal/80-20-… I liked the plan, but I would definitely change some things!

Another day, another service raising its prices because of “how much value it has added”: “value” I do not want or indeed have any use for. So: canceled!

Wrote up a long review of the 80/20 Endurance Norwegian Training Marathon Plan on my blog: v5.chriskrycho.com/journal/80-20-… I liked the plan, but I would definitely change some things!

NOT NOW NODEJS

Rustup 1.29.2 has been released. This is a small release which only updates the rustls-webpki and webpki-root-certs dependencies to remove the compromised eMudhra intermediate certificates. See the announcement for details: blog.rust-lang.org/2026/10/20/R...

If you make HTTPS requests or have other TLS clients and care about MITM attacks, you need to update today. If you run Go: go get go@1.27.3 If you run Homebrew: brew update brew install ca-certificates If you run Linux, maybe stay off the internet for now.

🚨 Six WebPKI Intermediate CAs are compromised. This is bigger than Heartbleed, but a client-side issue. Anything that verifies certificates against public roots, like most TLS clients, is/was affected. Details, live test, patched versions and workarounds at emudhra.we-have-a-problem.com Act now.

TypeScript 7.1 Beta is out 🎉 🔶 Import Attributes 🔶 Source Phase Imports 🔶 Types: es2026 🔶 Types: Promise.allKeyed, Iterator.zipKeyed 🔶 Types: Iterator chunks/join/includes 🔷 An API for tooling ✨ 🔷 tsconfig: Content Mappers 🔷 Perf: 1.2-1.9x faster, 15-50% less memory

Some speed & memory improvements you'll see with TypeScript 7.1 Beta. We're hoping to land more before RC & stable.

TypeScript 7.1 Beta is now available! Now with a preview API, content mapper support, new features, and some great optimizations. Give it a try and give us some feedback!

Spent the afternoon writing up this little literary history of my theological formation over the past 25-ish years. It was a fun and interesting exercise! v5.chriskrycho.com/library/books-…

Taking the day off from work today, and spending a bunch of time composing and then going for a nice long bike ride up and down the chunk of the Colorado Front Range near here. It’s my favorite time of year for it: cool, pleasant, and the aspens and scrub oak changing spectacularly.

Just to be very explicit, my claim is that the one simplest and best explanation for this, which resolves any apparent strangeness entirely, is that whoever is “speaking” in LLM output is a fictional character, and the real life thing called an LLM is a machine for writing lil stories about that guy

As an engineer in the tech industry, I want to build *tools*, not *experiences*. I don’t want your attention as a currency. I want to be paid to provide you a tool that helps you do your best work. Technology can be this thing you use, get the job done, and then put away.

the frontier labs don't make doing this with frontier models easy but the open models are working great together. Great job!

Amazing: Google Docs literally inverts the macOS HIG for pointers (developer.apple.com/design/human-i…). Buttons get the pointing hand, while links (e.g. the pill/chip for linking to another Google Doc) get the standard pointer. 🙃 This has felt wrong for years but I just confirmed it today. 😮‍💨

The way I have been putting it to people lately is: the Opus 4.5 shift was real, and it was from “doesn’t usually work, but you can massage it into working” to “usually works”—but “usually works” is not the same as “is good”, and the difference is in fact really important a lot of the time!

Code that works at a point in time is mostly uninteresting. What matters is building structures to ensure code stays correct over time. That to me is the essence of engineering in software

It’s very much a Nils Frahm Spaces kind of work day.

In the case of internal training and eval, again the labs are executing the code on their machines, this time, apparently, without safeguards of any kind for some reason. This is really the practice that I am primarily critical of in these posts.

What I am saying is: GPT-6 cannot do anything on its own other than produce text. In order to execute that text as code, that code has to run on a computer, which someone must supply. Whoever that supplier is makes the choice about what code to execute.

I’m thinking here *particularly* (but only) about the human-automation interface. How do we make for good outcomes — outside software? Because software can often learn things from other fields, despite the differences. Give me your best resources!

What are the best sources you know of on high-skill humans working in high-automation systems *outside software*—I’m thinking here of high-skill factories, for example—? Everything from blog posts to videos to papers to books welcome.

> and announcing that we’ve reached profitability 👏🏻👏🏻👏🏻

Hyperfine 2.0 is out, and it supports lots of new metrics, e.g. benchmarking RAM usage, CPU instructions, cache hit/miss etc! (screenshot from Charlie Marsh) github.com/sharkdp/hype...

It’s that time: Chicago Marathon sign-up! For both qualified and non-qualified (read: lottery) entries: letsdothis.com/share/6ac7b3e5…

Wow, I just discovered this fabulous reprocessing of the Voyager approach movies by Ian Regan. Go give it a watch! #planetsci www.youtube.com/watch?v=Jn6t...

the answer is no, they do not know if many of these results are just slop, and have already withdrawn 3 of them

This thread by Terence Tao applies to all knowledge work being confronted by LLMs. The culture of knowledge production is being thrashed by the Silicon Valley cult of speed. mathstodon.xyz/@tao/1173952...

Current-generation LLMs sure do love to talk about code that “carries” something.

Heard just now: > yeah @ersc.io is the first thing I think of every time there’s a GitHub outage 😂 AS IT SHOULD BE.

On the 20th anniversary of Zotero's 1.0 release on October 5, 2006, I wrote a history of the collaboration and years it took to create the popular research tool now used by over 20 million people: newsletter.dancohen.org/archive/the-...

TokioConf CFP is open! If you're doing cool async rust work, apply to talk about it at TokioConf. tokio.rs/blog/2026-10...

Said just now at work: The reason TS behaves this way is that runtime behaves this way, and TS tries very, very hard to accurately represent the reality of how runtime will behave, even when that runtime behavior is perverse. (most of my professional life for the past decade in one sentence. 😂😭)

New cargo-semver-checks release 🎉 Variadic functions have only been stable for a few days, and we immediately have SemVer lint support for them.

Also works pretty well if I take a standard deviation of the stack rather than a simple mean. flic.kr/p/2sDX3C6

150 Seconds of City Light - A stack of 10 images, 15 seconds each, taken by Anil Menon aboard the International Space Station on September 20th. flic.kr/p/2sDNZ95

Rust 1.99 is out, and for the first time the corresponding cargo-semver-checks release will have immediate SemVer linting support for a new language feature: C-ABI variadic functions 🎉 blog.rust-lang.org/2026/10/01/R...

What if you could make one change to JavaScript? Absolutely anything you like. Justin Ridgewell tells us what he'd pick, straight from TC39's 114th meeting in Amsterdam. Watch the video 🎥 https://youtu.be/RHesVP6Seuo

Learning this helped me, and if you use fish shell, it might help you, too: v5.chriskrycho.com/notes/finding-…