Schrödinger’s corner - both seki and ko until one of the players looks at it.
2 Likes
Also depends on the rules.
If it is Japanese rule, I don’t think black or white plays S3 (or T3) matters, but in Chinese rule, black or white should play S3 (or T3) at the very end, and R1 would be counted as black area.
1 Like
This position is (at least normally) not the final position yet (although it might remain like this on the board until late yose).
Black would like to actually play at S3 at some point, to avoid white playing T1 first. If black plays S3 first, it becomes a normal seki: 0 points in Japanese rules, 1 point of territory for black in Chinese rules.
If white plays T1 first, then it becomes mannen ko (thousand year ko). If black has many ko threats and would fight the ko and win, and white does not like that result (not enough compensation for example), then white will actually not play T1, preferring seki, and black can get to play S3 at any time and it will be normal seki. If white wants to go for the ko instead, white will play T1 first, then R1, and then T3 during the game starting the ko fight. This is not uncommon, as black risks much more points here, so even if white loses the ko, it is easier for white to get enough compensation somewhere else. Put differently, black needs larger ko threats than white to get a good (no worse than seki) result: compared to seki, white loses about 8 points by losing the ko, but black would lose 26 points.
Assuming that no player wants to go for the ko, then if white gets to T1 first, the normal result is seki. White will capture at R1 and connect at S1 (these R1 and S1 moves can typically be left for just before passing to end the game) to create a stable normal seki. The result (calculated in both cases by territory counting, to make results easier to compare) is one point for white (the S1 captured stone) in Japanese rules, and two points for white in Chinese rules (the S1 capture, and the “free” territory at S1: it can be claimed for free by filling it after normal dame).
Thus, in the case that T1 leads to seki (no ko fight), then there is a 1 point difference in Japanese rules between black getting to play S3 first, or white getting to play T1 first. Thus (S1 for black / T1 for white) is a “1 point gote” move (1/2 by miai conting).
In Chinese rules there is a difference of 3 points, so it would be a “3 point gote” (3/2 by miai counting) move. Thus it is larger and more important under Chinese rules to play here: under Japanese rules, a move here is worth the same as “destroying vs saving” a single point of territory, while under Chinese rules it is worth the same as “destroying vs saving” exactly 3 points of territory. Fundamentally, the source of this extra value is the fact that Chinese rules count territory inside seki while Japanese does not. If Japanese rules counted territory inside seki instead of making it a special case, the Japanse value of the move would also increase.
1 Like
I made a slight mistake in my calculation for the Chinese case: R1 should also be counted as territory for white, once T1 has been played. The reason is the same: White can wait to play R1 until after all normal dame have already been filled, thus R1 functions as a one-sided dame. Thus, white has 3 points: R1, S1 and the captured S1 stone. Thus the difference is actually 4 points, and thus the starting move T1 vs S3 is a 4-point gote move (2 points by miai counting). Much larger than in the Japanese case!
In a recent game, there was aji in the upper left corner. I probably took a bit of a risk in answering a bunch of sente ish pushes and Ataris (I mean they weren’t sente because of this upper left aji) that my opponent had planned with conditional moves but then they didn’t see the aji and I got to try to do something.
2 Likes
From go club tonight:
SGF: (some order of moves may be incorrect)
Looking at the AI graph, for a while the white stones could have been killed, but we both missed it 
10 Likes
Funny 9x9 game that ended (or would have ended, if my opponent didn’t time out) in a big seki (although I was apparently 0.5 ahead with komi regardless) due to Chinese rules’ superko rule; otherwise I think this is sending two, returning one? I think the latest response to stop my dumb invasion is 43 at E3, although probably at 47 something still works.
2 Likes
It’s actually a really nice example to see where a seki came out of superko, thanks for sharing. I didn’t see it initially so I’ll blow up the picture a bit. So Black can’t retake B1 because of superko and so white gets an eye. Very cool
If black had have played 1 white would’ve been dead on the bottom.
5 Likes
I don’t think I understand how this is a seki. As black I’d just throw a stone somewhere into white’s territory and after that proceed as in your second diagram. Am I missing something?
After you play a stone in white’s territory, white creates a seki by completing the eye at C1. If black had a ko threat it would be different!
3 Likes
So indeed I missed something!
Yeah I also missed the importance of the superko aspect too when I first looked at it 
Game 85 of my KataGo high-playout series was only our second game to feature a seki This game also featured our most dramatic winrate change so far. After move 201, white’s winrate was 87 %. After move 245, black’s winrate was 86 %.
3 Likes
Another KataGo seki Black wins the game by 2 points.
1 Like
That reminds me I think I meant to post this one and forgot
feat the lovely random blue squares from the autoscore
4 Likes
Not from my game
What a fun one
6 Likes
Two sekis in a 9x9, a first for me (not that I did it, rather had it done to me).
7 Likes
Seki on whole board. You never see this before
2 Likes