Life In 19x19

I have done a little bit of computer analysis and it seems that 3-3 for black is not the ideal opening.

This is the main line, there are a few different continuations, but the Black position is flawed, so Black wins by 5 points in Chinese counting at most. I analyzed it with 7 points komi so White is not pushing for a bigger victory too hard (but KataGo does prefer a bigger lead if it's possible). I used the fat net to do my analysis because I use a lot of playouts so I'm sure it's better at very deep searches.

This extend variation often leads to a carpenter's square in the corner which basically precludes black from getting a good score in the end

Open it in Lizzie if you want to see the playout numbers

In the main line here Black loses the semeai and can only lose by two points

After move 7 as long as White plays 12 in the right place you get trapped in this kind of variation

Even if you play another move at 7 you don't escape the dreaded carpenter's square that eats your points

In this variation, I quickly got into a problem of ko creating a horizon effect. I got a draw despite White being over 80% with millions of playouts.

Because of this I need to find an earlier mistake by white to avoid having to play something complicated like ko. This is really a disadvantage of computer analysis.

I found the white mistake, I think the line it likes at 1 million playouts is not good (it takes like 5 million to see the superior move is better), just living the corner easily is better and wins by 2 points

Move 18 should be at F8 and white has no issues with the game after that

for move 18, White B7 seems good too, doesn't it?

iopq wrote:I found the white mistake, I think the line it likes at 1 million playouts is not good (it takes like 5 million to see the superior move is better), just living the corner easily is better and wins by 2 points

Move 18 should be at F8 and white has no issues with the game after that

Friday9i wrote:for move 18, White B7 seems good too, doesn't it?

iopq wrote:I found the white mistake, I think the line it likes at 1 million playouts is not good (it takes like 5 million to see the superior move is better), just living the corner easily is better and wins by 2 points

Move 18 should be at F8 and white has no issues with the game after that

Yes, it's the first preference, but it switches to the inferior move after some playouts in many cases. Anyway, I verified there are probably no more chances for Black after the correct continuation of simply living in the corner, so I didn't explore the other move

If you push first you can go back to the other variations and it ends being a transposition. But you can also try other lines like this, which also ends up a White win by 2.