Game 17 (NZ rules, 7 komi) starts out complicated, but then everything settles at once and the rest is endgame. White ends up with three corners, but black wins by 2 points or so.
1 Like
Game 18 (NZ rules, 7 komi) was maybe the weirdest game in this thread so far. I won’t pretend to understand anything that happened here. Can someone get Michael Redmond on the line?
2 Likes
Game 19 (NZ rules, 7 komi) starts off with a complicated fight on the lower side. As it sprawls out, it looks more like a game of tetris or battleship than a game of go. White wins by 8 points or so.
1 Like
Game 20 (NZ rules, 7 komi) took the lower-side fight in a different direction. We have a new shortest game in the thread (166 moves). White slays black’s dragon and wins by 54-ish points.
2 Likes
After some twists and turns, game 21 (NZ rules, 7 komi) ended up as our fourth tie. That brings our current tally to 6 black wins, 4 ties, 11 white wins.
1 Like
Game 22 (NZ rules, 7 komi) was another breathtaking game (split off from game 18 like the three previous ones). At one point, it looks like white is going to get four corners with black building a center moyo. A bit later, it looks like there’s going to be a huge trade in which each side sacrifices a large group. In the end, there are seven live groups on the board and white wins by four points or so. 224 out of 256 moves received more than 100,000 playouts.
1 Like
I split off game 23 (NZ rules, 7 komi) at move 36 of game 8, which saw black winning by 8 points or so. Game 23 turned into another giant center fight, but in a completely different direction, and became our fifth tie. This time I played out every move to the end.
Is this using v1.4.5 and the newest nets?
Not yet, but I’ve pinged spook (the developer of ZBaduk, where I play these games) to notify me when he updates it to that 
Game 24 (NZ rules, 7 komi) ends up with huge territories on both sides. For the longest time it looked like a tie, but white was one ko threat ahead in the endgame ko and won by 2 points.
1 Like
Game 25 (NZ rules, 7 komi) seems to be mostly about aggressively reducing each other. Black wins by 2 or 4 points.
1 Like
Game 26 (NZ rules, 7 komi) ends up with absolutely enormous territories on both sides. Black’s upper-right is worth more than 60 points, while white’s lower side is worth more than 70 points. Of course it’s a close game, anyway, and black wins by 2 points.
1 Like
Game 27 (NZ rules, 7 komi) was another close call that looked like it could end up as a tie, but black won by 2 points. Our current tally is 9 black wins, 5 ties, 13 white wins.
2 Likes
I’m always surprised when Kata plays that heavy-looking (14). My mind wants to play E15 - E16 - E12.
The (26) --> (30) idea is not something I really get, it doesn’t seem like a productive use of White’s time.
Possibly Black could play (55) at N15 and make a large trade, but then she has to deal with a lot of aji in the position and if she’s ahead, it isn’t really worth it.
1 Like
Incredibly, E15 is KataGo’s… wait for it… 166th-favourite movie. However, that being said, E15 only received 5 playouts, and at this early stage of the game, it’s still only a 6pp loss to the game move.
KataGo’s second-favourite move (C10, -0.1 pp compared to D12, the game move) received by far the most playouts (723,045 out of 795,285, compared to 48,852 playouts for D12) and has actually been slated in my list of games to split off in the future ever since game 1. 
I’ve added the projected variation to this game.
I’ve added variations for moves 26, 28 and 30 (the last of which I’ll also branch off into a new game soon). Some of them only add forcing moves before continuing with the game sequence, anyway.
N15 is KataGo’s 23rd choice (-9 pp compared to the game move). The only real competitor to the game move seems to be G18 (-0.5 pp). Incredibly complicated-looking variation added.
1 Like
Game 28 (NZ rules, 7 komi) sees more huge territories - white’s lower side is worth more than 65 points. This game ended up as our sixth tie.
2 Likes
Game 29 (NZ rules, 7 komi) is our new shortest game so far. After white plays N15 at move 100, black’s winrate crashes from around 34 % (white by ~ 2 points) to around 8 % (white by ~ 8 points) and never recovers. Not sure exactly what it was that KataGo “missed” before move 100, but in the end white wins by 12 or 14 points.
2 Likes
I was interested in this one because of that breakdown on move 100. Now honestly it’s quite a difficult game! Indeed I missed the moves after 140… If it’s not asking too much.
I stopped the game at move 140 because white’s winrate is already 96 % (ahead by 12 - 14 points) at that point. I may play out some more moves next time I get the chance.
1 Like
Incredible games, but still missing the latest nets. They differ as early as move 3!
2 Likes