KataGo review (score delta based)

[Gomoto]

I tried to use KataGo's score delta for reviewing my games and think it is a nice option. I am aware that the shown scores are only estimates and do not represent absolute truth. But I think they provides nice insight into the relative value of different moves.

This is a game with negative komi (-5.5) so black is ahead "17 points" according to KataGo.

ka1.JPG (116.79 KiB) Viewed 14893 times

[Gomoto]

In the first picture you see the standard joseki answers

ka2.JPG (115.65 KiB) Viewed 14892 times

But I prefer a different move, just to not play as my opponent is expecting. The score estimate difference by KataGO is not big.

ka3.JPG (116.63 KiB) Viewed 14892 times

[Gomoto]

Blacks advantage is melting step by step, there are slightly better options for him to play. I form two enclosures because that is my way to go. I do not mind small differences in score estimate if I can steer the game to my liking.

ka4.JPG (117.22 KiB) Viewed 14892 times

[Gomoto]

This move is a small mistake by me, I should have kicked at C15. But if your opponent does not punish your mistakes it is not a big deal. If your goal is to improve you should play against stronger players from time to time, alternativly you have to look for your errors with the help of AI while reviewing your games.

ka5.JPG (117.33 KiB) Viewed 14891 times

[Gomoto]

Interestingly black looses some points with this standard looking answer.

ka6.JPG (118.14 KiB) Viewed 14888 times

[dfan]

My understanding was that the jump to R14 is considered okay for White, but only if the ladder to capture the R14 stone doesn't work, and in fact Black would often next play a move like F3 to set that ladder up - but in this game the stone is already there! So I'm surprised that KataGo (which does understand ladders) doesn't consider it to be more of a loss in this case, although it does definitely want Black to push (and presumably cut) in response.

[Bill Spight]

Interestingly black looses some points with this standard looking answer.

ka6.JPG

Standard looking to whom? To me this block on the third line looks like standard human loss aversion by Black. "This land is my land, it is not your land."

[Gomoto]

ka2b.JPG (118.04 KiB) Viewed 14886 times

[Gomoto]

@Bill: locally KataGo likes the crosscut variation for black, allthough it recommends the push at R15. I like that you have a good name for the kind of move in the real game.

ka3b.JPG (115.56 KiB) Viewed 14882 times

[Gomoto]

This move is a mistake for white.

ka7.JPG (118.16 KiB) Viewed 14880 times

[Gomoto]

The exchange at Q14 was not good for black.

ka8.JPG (118.36 KiB) Viewed 14879 times

[Bill Spight]

This move is a small mistake by me, I should have kicked at C15. But if your opponent does not punish your mistakes it is not a big deal. If your goal is to improve you should play against stronger players from time to time, alternativly you have to look for your errors with the help of AI while reviewing your games.

Bridge champion Marshall Miles once commented on how he and his partner, Eddie Kantar, kept in trim while playing in local games, which they did most of the time, instead of tournaments. Local games, OC, had a weaker field. He said that they played every hand as though their opponents were national champions.

[Gomoto]

White misses the chance to turn at Q13 just now.

ka9.JPG (115.83 KiB) Viewed 14879 times

[Gomoto]

I think I stop my review here. I just intended to show this kind of self review, because I enjoy the score estimates by KataGo a lot while my own analyzing.

Have fun if you like to use KataGo in such a way yourself.