Note - this twitter thread] is where I first heard the terms bottom-up vs top-down learning
- Top-down learning is what Defender of Basic writes about here
- Pretty much “make predictions, grapple with reality, make a model, face the fact that your model is terrible, get your ego involved, cash out the predictions and see what you got wrong”
- Super powerful because the learnings are incredibly salient - “omg I can’t believe I thought x, how embarrassing, now I know y” → leads to very durable updates
- Also just super fun
See also Top-down learning unlocks care (and ego) & my this-project kickoff post
- - Turns out this is similar to reading to engage (as opposed to reading to understand)