Tag
The blog post introduces a Python PaintBrush class for teaching coding visually and discusses how LLMs like Claude can generate such code quickly, emphasizing the value of hand-written code for learning.
The essay uses the Tower of Babel story to argue that AI-assisted programming removes friction in software development, which may undermine the shared understanding and coordination necessary for large projects, leading to chaotic codebases.
Simon Willison shares a GitHub code frequency chart for his Datasette project, illustrating the significant impact of AI coding agents and large language models on his development output, with a notable spike in activity in 2026.
Kenton Varda announces a moratorium on AI-written change descriptions (PR/commit messages, issues/tickets) on his team, citing that AI omits high-level framing needed for code review and produces worse-than-useless descriptions.
Charity Majors discusses how AI flipped the economics of code production, making code generation cheap and instant, transforming code from a treasured asset into a disposable, regenerable resource.
Anthropic analyzed 400,000 Claude Code sessions and found only a 5% gap in verified success rates between software engineers and non-engineers, suggesting domain expertise matters more than coding ability for AI-assisted development, challenging the 'learn to code' narrative.
An introduction to using TLA+ with LLMs like Claude to write formal specifications, showing how LLMs can help with syntax while focusing on correctness.
The Datasette project launches an official blog, built using OpenAI Codex desktop, to share upcoming announcements.
AI-assisted coding tools enable Gary Tan to deliver hundreds of thousands of lines of code per month after 13 years without writing code. The core approach is Tokenmaxxing — consuming massive amounts of tokens to let the model handle tasks comprehensively, achieving efficiency equivalent to 400 engineers.