miniblog.

Some novel (to me) AI workflows from OpenAI in https://openai.com/index/open-source-codex-orchestration-symphony/ (1) Adding the ability for the LLM to create follow-up tasks in the issue tracker. (2) Identifying weaknesses in the spec by implementing in several different languages.

I'm implementing an interpreter, and wondering how often I should check for interruptions (e.g. Ctrl-C). I don't want to spend too much CPU time checking whether I've been interrupted, but I also want slow programs to stop promptly. It's tricky.

The Bitter Lesson: how implementing generic search and scaling compute outperforms custom logic in many situations: