Fantastic article. I love the idea of embracing Ralph, letting him screw up but encouraging him to learn. I’m going to try applying Ralph to some of my projects. Thanks!

I was greatly assisted by Claude Code and LeanSerch (leansearch.net).

Brilliant. Hope there’s a bit of joy mixed in the visits, too.

Topics discussed included: the upcoming (funded) Mathlib initiative, the new grind tactic and how to think about it, whether we should be using type classes to pack in arbitrary background knowledge (and how such an approach might rope in unnecessary background knowledge or otherwise be unwelcome).