Comment by epolanski
21 hours ago
There's lots of psychological and anthropological studies behind the fact that most experts in various fields excel due to pattern recognition not reasoning.
Going back to the chess example, while chess masters are incredible at analyzing complex positions they can recognize as "similar to", their advantage over normal human beings is very small when positions are completely randomized.
"Peak: Secrets from the New Science of Expertise", by Ericsson goes more in depth of the topic, but there's lots of literature on the topic.
> There's lots of psychological and anthropological studies behind the fact that most experts in various fields excel due to pattern recognition not reasoning.
Pattern recognition in experts comes from combination of theoretical understanding and a lot of practical problem solving experience (which translates into patterns forming in way of neural paths) - not the other way around. If you dont understand the problem you are solving, then yes maybe you'll be able to throw a pattern at it and with a bit of luck solve it (kinda like how LLMs operate), but this will not lead to understanding. Memorising patterns isolated from theoretical backgrounds is not something that will create an expert in a field.
> their advantage over normal human beings is very small when positions are completely randomized.
The book you referenced does not say they're comparable to normal players at playing from a random position.
Normal players are almost as good as them at recalling a nonsensical board of random pieces.
The suggestion that the advantage of a chess master over a normal player is "very small" at playing from a random position is laughable.
I obviously meant it as a delta over the recognizable lines.
That wasn't obvious at all. I interpreted it as the chess masters lacking an advantage at playing from randomized position, which would be consistent with the claim you were supporting, as opposed to recalling which is neither here nor there.