BillSpightsElfPositionalJudgmentExercises/Game 18 ELf's Solution

Black is ahead.  

Elf thinks that is a mistake, losing 9% to the attachment at a. For Elf recommends the attachment at b, but the play from there is complicated, and Elf regards a, which Iwamoto played in the game, as only ½% worse, so I shall show Elf's continuation from there.

Elf's continuation from , marked  

is the underneath hane, to which White stands, solidifying his group. takes kikashi in the top left and then prevents White from making eye shape on the second line. The variation stops here, probably because the playouts for future play dropped below 1500.

Earlier in the game

Unexpected error  

is an error that loses 11½% to par, according to Elf. It is not that the ogeima approach is so bad, per se, it is that the 3-3 invasion at a in the bottom left corner is so good. The value of these early 3-3 invasions was not suspected before the advent of AI.

Elf's mainline variation for

3-3 invasion  

White invades the bottom left corner and claims territory while Black makes a sizable wall.

3-3 invasion, continued  

encloses the bottom right corner. Then makes a two space high enclosure in the top right, which was not popular before the AI era. extends on the bottom side, reducing Black's influence there. The bots have devalued the sides, but each corner has at least two stones in it. Next, Black plays a kick joseki in the top left corner, another joseki that was not popular before the bot era. Then, jumps on the third line on the left side, making some territory and forming a 5 space extension from the two White stones in the corner. Dogma at the time was that such extensions are too far. The bots don't care.

Later in the game

blunders  

Elf says that the natural looking extension, , loses 19½% to par. switches to the third line attachment against the Black enclosure in the bottom left corner. secures territory in the bottom left co[rner. Then descends in the top leftcorner. Instead of Elf prefers 27 by 5½%.

Elf's recommended variation for

Atari first  

connects and extends. White could have played this exchange earlier. Comparing the two diagrams, it seems like the main point playing the atari at is this exchange.

Ironically, Elf regards the descent at as a blunder, losing 20%. ;) White has a superior play in the next diagram.

Third line attachment  

= a

plays a second third line attachment. plays the top hane. (If is at a, is the same.) After pulls back. The peep, , the crawl, , and the cut, , leave White some aji in the corner.

So Black and White traded blunders. Other than that, missed the 3-3 invasion, which was probably heresy at the time. Other than that, Black took a strong lead because of an accumulation of relatively small errors by White.