Why isn’t everything Seki here? The upper left and the bottom right region? The algorithm put the lower region as white dead. I thought it was all seki and my opponent thought it was all white dead.
1 Like
4 Likes
A further complication is that some rules decide the status of this group without having the players to play it out. See bent four in the corner is dead.
With a link that leads to:
Under Japanese rules, bent four in the corner is considered dead and is removed without playing any ko.
So … that’s the relevant part here, right? 
5 Likes
Black H1 white capture black j2 white j1 ko, only black can start it, that’s why the rules assume black wins.
1 Like
Yea. I got it the moment I pressed sent ^^
Ty still
1 Like
It’s the same on the goban 
this cut works! Plays move, instant regret
2 Likes
The idea is that black can atari at H1, and you will have to capture at J4, leaving a bent four. Then there is a ko which black takes first. However, black can choose to play the ko at any time - so hypothetically they could fix all other ko threats first, and only then start the ko, so that they are sure to win it at no cost. All of those defensive moves would lose a lot of points in Japanese rules, so the bent four is ruled as dead during scoring, and the ko is played out after the game using special ko rules (OGS does not implement this phase of the Japanese rules).
The situation is complicated in this case by the fact that white has an unremovable ko threat at A9 - black can never fix this threat without giving up the seki in the upper left. However, black answers this threat by capturing the whole top left. In chinese rules you could choose to sacrifice the upper left to live without seki in the lower right, but it would lose a lot of points.
In this particular case, black can remove the possible ko threat at A9 by connecting first at C9? I believe under neither of Japanese or Chinese rules would this cost points in that seki, and after connecting, either White atari is not a ko threat because after white captures it’s a bulky 5.
But yeah, in general unremovable ko threats can lead to interesting differences between JP rules and other rulesets.
3 Likes
Ah thanks, I see the bent four being formed after 2 moves now 
But if the game had continued, after Black J2 and White J1, wouldn’t it be best for Black to play H1 immediately? Or else white makes 2 eyes. Since any other thing that black makes in the upper left region makes the seki disappears.
If black play C9, doesn’t white at least lives taking the ko and then Black makes a … “bulky 6?”, like a gun. (I know black already won anyway, just a DDK brainstorming here, wondering how white would lose by the least amount of points)
Thanks. I didn’t look closely enough at the seki.
@AllanFelipe :
“after Black J2 and White J4” from what position?
From your diagram, J2 is already black, and whiteJ4 would change the
corner from eventual-bent-4-in-the-corner to White is dead outright .
“If black play C9,” then White will have no threats
for the ko, so White won’t be able to take the ko.
However, White shouldn’t use the the upper-left as a threat even if Black leaves
that available, since White’s group there is too large: The difference between
[seki in upper-left] and [White dies in upper-left] is significantly greater than the
difference between [White is dead in lower-right] and [lower-right is 2 eye for White].
(I now see that calebjdastrup already made this observation. 
)
After a Black H2 and White J4, I started from there. Then Black J2 and White J1 and not J4, that was a typo. Oh ok, thanks for the patience. I was fixated on those 2 eyes and living in the lower part and basically forgot the huge chunk above.
Is it really at no cost though? I found myself in such situations, and felt like removing all ko threats would require me making many moves in my territory, thus costing me points.
I know the bent-4 concept si ce a while but I still find it very confusing under Japanese rules.
That’s why it’s called “hypothetically”. You just assume that they are played out but do not need to actually play them out in the game.
1 Like
Japanese rules are designed such that you can decide life and death locally, i.e., you can look at a group by itself and see if it is dead or not. It doesn’t matter if you can remove the ko threats, because in life and death at the end of the game, you don’t use ko threats, you use special “passes”.
The advantage is that there is no “spooky action at a distance” from unremovable ko threats and such, like in other rulesets. The disadvantage is that Japanese rules are more elaborate.
In this forum you will find many posts disparaging Japanese rules for this, and telling beginners to “just play it out”.
2 Likes
Which I find very unintuitive, as playing it out may change the outcome of the game.
Right, and I admit I’m definitely on the side of disparaging Japanese rules for that. Your answer is clear, I just find that to be a very unelegant way to go at it. I’d go as far as saying it feels “unfair”, even though this isn’t really meaningful since it’s a ruleset that both players agree to.
2 Likes
The same is true for any other dead stones - if your opponent claims some stones are alive when they are really dead, it could cost you many moves inside territory to capture them. You play these moves in life-and-death confirmation instead of the main game so that they don’t cost you points. The ko in a bent four complicates things a little bit, but it is the same basic reasoning.
3 Likes
To me, the weird thing about this is that any ko fight is “spooky action at a distance” — why should one at the end of the game be considered spookier?
2 Likes
A strategy game like Go benefits from having a lot of strategic depth. In contrast, counting at the end should be simple and intuitive. The idea that the groups can be considered separately without looking at the whole board is generally an intuitive simplification.
1 Like
But it is not the end until players agree, which they need not if the defender requires the attacker to prove they can kill, as they should if they want to play the ko.
1 Like