I’ve said a few times on here that I think studying chess engines is about the best computer science education you could get from a single hobby. Definitely wor

I’ve said a few times on here that I think studying chess engines is about the best computer science education you could get from a single hobby. Definitely worth trying, and evolving from simple methods to more advanced over time. I do think there’s also a big gap in the market for chess _tooling_ that makes learning and match prep easier. I actually think simpler more interpretable engines can be more useful as tools for humans in many cases. For example instead of increasingly inscrutable neural networks, or explanations of moves that are merely “after 10 moves this happens and it’s bad”, even a simple regression over some human interpretable features can give you an idea why the engine is suggesting a positional move (king safety, piece activity etc). I also think just pure stats and algorithmic stuff would be useful, like finding transpositions between openings, trying to work out what kinds of middle game structures or endgames a player is good or bad at, and navigating a path towards that in your preparation.