@ThePrimeagen@kenwheeler I feel like it does speed up data analysis, not to the point of 100x but around 1.5 x or sometimes 2x when the data sources are very varied in shape
@cremieuxrecueil in regular cohort analysis (sign-ups, business, etc.) there's no such collinearity, and that's perhaps why the agent is struggling this much
I've found that instead of looking at the first principles cause of a problem, when codex finds a blocker it immediately tries a half-assed work around. For every blocker. It's super annoying, and akin to a doctor only treating symptoms instead of causes. There is zero causal reasoning in OpenAI models. This compounds if codex finds a blocker during a workaround of a precious blocker.
@paulg its simple in theory but in practice it would go more like: make something people want that you can do while either loving or not hating, that enough people are willing to pay for, for an extended period of time without dying for it