Hm. But I'm still waiting for their content addressable programming language so they just do intent reviews and can just ditch the VCS.

Redesigning source control at Anthropic

As our merges per day grew more than tenfold, source control became our biggest development bottleneck. Here, we share how we migrated to an in-house version control system in five months.

anthropic.com

I suggested to Claude that it just apply the coyoneda trick and it went straight into an infinite loop.

Scary type inference problem in my opaque domain: Claude instantly offers a pure, crystalline, non-obvious truth. Ask it to add a row to my markdown table: Claude eagerly rolls up sleeves and gets right to work. RIP my tokens.

I don’t understand quantum theory, but this book is always a joy to come back to.

I wrote my first Quint spec today. It was fun and easy to pick up! It's like TLA+ and TypeScript had a baby. Has some obvious room to grow: you have to be gentle with the syntax and type errors to get good diagnostics. Otherwise, seems like a good alternative to Alloy or TLA+ for simple specs!

Nobody cares about correctness and do cheap things first are great takeaways from this but this article illustrates these and other points especially well: www.galois.com/articles/wha...

🎉 Announcing CatColab 0.2: Wren! CatColab is a structured model editor for visualizing complex systems with domain-specific logics. Since October, we've implemented: - Users & permissions - Enhanced docs - New theories (e.g. Discrete exterior calculus) - and more!

I might prefer the 1914 narrative regarding vaccines.

Are there any prominent examples of using AI for anti-enshittification? For example, LLMs are great at extracting legitimate headlines from clickbait articles

I now work at a windows/teams shop after being on Mac for over a decade. I don’t understand how Microsoft can arrive at such obviously inefficient design choices. Putting in my MacBook request asap