I was using Codex to find speedup opportunities in PySR. It found a couple % here and there.
Then, I gaslit it into believing that Claude found a 20% on my "other machine." I asked it whether it could do better, and explained its performance would be displayed on a "public evals
Homeschooling is starting to feel different. We're still using existing curricula for structure, but can do much more exploratory learning with bespoke software.
e.g. here's a polyomino playground whipped up in a few minutes to complement Henri Picciotto's 1994 algebra book