Takagawa (W) - Rin Kaiho, Meijin Final, Game 3, 1969-09-09a. Komi = 5+ pts. (I.e., 5 pts. with White winning jigo.)
After Elf gives Black an estimated winrate of 90½% with 33.8k playouts. However, White, Takagawa, won by 8 pts.
leans on the White group on the right side; then tops the tree, capping the White center stones. As White runs out, Black's attack promises to erase White's potential in the left center.
Earlier in the game:
Which side to block the 3-3 invasion from is a question I cannot always answer. Usually the difference is small, but in this case it is 10%, and Elf recommends the opposite alternative to the one that appears to be the correct direction of play. (Takagawa would know. ;)) How come?
Here is a clue. Instead of the traditional joseki (now obsolete) taking outside strength, White plays the two step hane and takes the corner.
White does not make outside strength to work with the stones. Instead, Black has a weakish group in hostile territory. secures territory while attacking the Black group. Black makes elbow room with sente and then turn to his territorial framework in the top right quadrant.
Takagawa played at 21, but Elf says to play the two step hane here, as well. (The difference only comes to 1%, however.)
In this variation the stone is weak. reduces the Black framework in the top right, while Black attacks the stone. On balance, this is better for Black than when he makes a weak group on the left side.
In the actual game Black played the hane-and-connect in the bottom left corner, which was joseki before the AI era. In this case that is Elf's choice, too, probably because White ends up overconcentrated, given the stones.
reduces the left side. But loses 16½% and then loses 11%. Each side made smaller errors, but White made three significant mistakes that gave Black a commanding lead.
and make territory with the White wall. Black enlarges the Black moyo and then and reduce it. blocks on the third line with eye shape. Then secures the corner and threatens to connect underneath.