“The center is not biased towards novelty so much as towards contact with the thing that creates potential change. The middle ground is not between use and non-use, but between refusal and commitment” - lucumr.pocoo.org/2026/4/11/th... via @mitsuhiko.at

The Python IDE You Didn't Know Existed When you make tools that are meant to be used in a notebook, it could make sense to actually build all of that from a notebook. So that begs the question, ca...

It really took me a while to "get it" when it comes to nbdev. But I gotta hand it to Jeremy Howard this way of working makes too much sense once you're used to it.

As of today, I am working on tools that make this kind of work possible in marimo.

youtu.be/ZLg27UmAJbw

Working hard on putting Chatto Cloud 100% on EU infrastructure -- not just running in EU datacenters, but operated and owned by EU companies.

It's... a challenge. But the needle won't move by itself.

Filing this under universal basic income tag

Smth smth Designing Data Intensive applications cc @martin.kleppmann.com

- Qwen 3.5 — most broadly recommended family right now across usecases. - Gemma 4 — strong recent buzz for local usability, especially smaller and mid-sized deployments. - GLM-5 / GLM-4.7 — near the top of broad open-model rankings, increasingly part of the “best overall” conversation. - MiniMax M2.5 / M2.7 — repeatedly cited for agentic/tool-heavy workloads. - DeepSeek V3.2 — still firmly in the top cluster when people talk about strongest open-weight general models. - GPT-oss 20B — not the mainstream “winner,” but increasingly recommended as a practical local option and for uncensored variants.

Apr 2026 check-in on the local models via www.latent.space/p/ainews-top...

✨ Have you ever heard of Disney’s 12 Basic Principles of Animation? The very first rule, “Squash and Stretch”, can make motion feel *way* more satisfying.

I just dropped a brand-new blog post that shows how I’ve been using this technique in my own work:
www.joshwcomeau.com/animation/sq...

I will be working as a pollinator dressed up as a giant bee, traveling around the world providing my services to farmers

zigpty update!

It can monitor cwd, memory, and CPU time (shell + child processes).

Compatible with macOS, Linux, Windows, Android (Termux). Available as a zig lib or npm pkg ~300kB total napi binaries!

Demo with Nitro 👀 github.com/pithings/zig...

Serving 200 million requests per day with a cgi-bin In the early 2000s, we used to write a lot of CGI programs. This was the primary way to make websites dynamic at the time. These CGI programs were usually written in Perl, but sometimes in C to increa...

Remembering this ancient "Singletons Considered Stupid" Yegge rant (sites.google.com/site/steveye...).

"Just treat yourself to a few extra instances. You'll be amazed at how fast computers have gotten these days."

Congrats Mario! Hope ya taking your sense of humour with you. Looking forward to moar pies.

"Prototyping is the process of following loose threads in the Gordian Knot. You'll learn more about the space and spot opportunities you didn't know existed. You'll also uncover risks in places you hadn't expected."

Build to think - gordonbrander.com/notes/build-... via @gordon.bsky.social

AI World Clocks The current time as rendered by 9 different AI models. By Brian Moore.

Was reminded earlier that this exists and is wonderful.

clocks.brianmoore.com

Good stuff Vincent and congratulations! Love your jazz. Keep up good work.

I talked to Claude about my alleged LLM psychosis and he has assured me that I am fine

"I am not outsourcing my brain, I am using my brain to outsource the work".

Mr Tolinski, 🫡

🤯

> Is it nominative determinism or a simulation glitch that everyone trying to terraform the web has some variation of “Matthew” in their name? I was in a call set up by Matthew Prince, talking to Matt Taylor and Matt Kane, with my right hand there, Matías.

code reuse by the matrix

RISC won: Building towards Data AGI A deeply technical retrospective on the first year of Southbridge.

I wasn't sure if we were going to share this, because knowing what doesn't work is often more valuable than seeing what worked.

That - and being nervous about sharing your failures.

Here's a technical retrospective on our 2025:

www.southbridge.ai/blog/25-way...

Gotta ask, what’s up with the recurring nose rub? A trademark move? Good luck with the meniscus

If we use agents without reading the output or having any interaction with the actual resulting code, are we the Chinese room, or is the agent?

Which of us can claim to "understand"?

oh no you didn't

“Why do we have so many layers in the stack? […] Most abstractions […] exist because humans need help. We couldn't hold the whole system in our heads […] AI doesn't have the same limitation. It can hold the whole system in context and just write the code.” - blog.cloudflare.com/vinext/

“As AI becomes better at cloning people's work, what ends up becoming most valuable are software contracts, tests, and API surface area.“ - saewitz.com/tests-are-th...

They took our jobs

You get it

Good artists copy, great artists distill

Labor of love: We're open-sourcing the runtime we use to run long-horizon agents at Southbridge.

Something like this exists at almost every serious AI team I know.
We ended up needing to build it because we couldn’t buy it.

`bunx hankweave` gives you a fun demo.

Congrats Peter and keep up good work.

Telepath computer - AI native computer telepath.computer. Demo here ➡️ youtu.be/jEm5SkLMap0