Describing good play as "making few mistakes" seems like the wrong terminology to me. A mistake is not a thing, in and of itself, it's just the entire space of possible games outside the very narrow subset that lead to victory. If you give me a list of 100 chess mistakes, you've actually told me a lot less about the game than if you've given me a list of 50 good strategies -- identifying a point in the larger space of losing strategies encodes far less information than picking one in the smaller space of winning.

And the real reason I'm nitpicking here is because my advisor has always proceeded mostly by pointing out mistakes, but rarely by identifying helpful, effective strategies, and so I feel like I've failed to learn much from him for very solid information-theoretic reasons.