One of the things I still want to analyze with AI is my all time favorite book “Attack and defence” by Ishida Akira and James Davies, the Elementary Series’ take on the middle game. Unfortunately I have given away my copy to the Ghent Go club and I haven’t got round to borrow it back. I’m curious to know how much of that well crafted advice still stands in the unforgiving face of KataGo.

Here on L19, there has been a debate whether we should always value AI advice higher than conventional wisdom, especially for educational purposes. I don’t think it should be seen as a battle between professional teachers and the uneducated savants AI are. Rather, AI can reinforce our understanding or make us reconsider it.

I have crafted 5 exercises on “Attack and Defend”, analyzing my own games with KataGo and looking for such middle game positions where KataGo’s favorite coincides with what I (think I would) infer from conventional wisdom. Still, the positions are not trivial, in the sense that I made a mistake myself against it or would not be too sure I would always make the right choice. I think they make for good exercises up to low dan level. More experienced players may thus be reassured that what looks like the best move is indeed the best move here.

Percentages and point leads are included to give an idea of how much better the best move is than the alternatives.
Continuations by KataGo are also included. “Seeing sequences” remains one of the most important aspect of the middle game and Go in general.