You can see that after move 243, White could play Q19 to devesting effect IF White just had one outside liberty. From what I’ve seen it’s often the removal of the outside liberties that marks the point where invasion becomes possible, often involving seki. Move 200 was okay but any time after that the cutting point at Q17 could have used fixing.
That’s helpful thanks. I guess life and death can’t hurt but even better to watch for the weakness and avoid it in the first place
2 Likes
Correspondence distraction tesuji. At some point earlier in the game I read that as soon as the opponent plays O18 I need to play S17.
… a few weeks passed …
The opponent played O18. I clearly remembered I needed to reply locally now and put down T16 without stopping to think even for a second.
11 Likes
Ah, this is one of the class of “give two, take one” seki in which White can, at least in theory, exchange T19–R19–S19–pass–repeat as much as he likes, to no avail, with Black steadily receiving more and more points.
2 Likes
Yup. And seemingly giving black more points was my aim here, so I was on the right path. Good job!
2 Likes
Quick diversion maybe interesting to some rules-nerds:
If we are playing pure no-pass-go, without prisoner return, then that repeating sequence means that black is better off simply letting white capture the two stones. So in no-pass-go it’s actually not seki at all, and the normally bad T16 move is suddenly the correct local move to maximize points!
In that case the difference between seki and not seki wasn’t very dramatic. This position was more shocking to me:
In no-pass-go, the white group is dead! The reason is the same as above, black will play A3, and it’s better for white to allow black to capture the whole group (giving a large but finite number of local moves for black), than continually capturing (giving an infinite supply of local moves for black).Note that nothing changes if we add more black stones like this, the white group is still dead:
After we introduce prisoner return to no-pass-go, this weirdness disappears, since “sending two returning one” nets the opponent an extra prisoner, which they can then return as an extra move. So there’s no longer a way for black to get infinite moves when white resists capture, and all of these groups go back to being seki.
Knowing about these peculiarities of no-pass-go can actually be useful even if you’re playing regular go, if you happen to make a Molasses Ko in your game. They are exceedingly rare, but they have the potential to turn a regular go game into what is effectively a no-pass-go game, in which the above white groups would be dead, not seki.
(Check out this page for more details on these positions, plus some other fun stuff)
7 Likes
Here is an interesting (albeit, contrived) seki…
The position is attributed to Harry Fearnley. However, I did not find it on his website. It appears at the bottom of this Sensei’s library page: https://senseis.xmp.net/?Hanezeki
Analysis here:
4 Likes
From my last completed game.
We left that corner unattended for a long time. Then I tried to kill but here is the result.
I wonder if it was possible to kill…
3 Likes
What was the original position? If T19 were empty, then it would be dead. Was filling that necessary?
1 Like
Here is the game:
https://online-go.com/game/32230377
2 Likes
I don’t think so.
1 Like
Maybe way later in the without t19 played, Black could throw in at t19 and fight a ko to make points or die completely in a bent four, but that seems very risky for a few points 
Move 175 lost 20 points so it seemed like s15 kills for the AI, but white has to worry about the bottom right group possibly dieing too? (reuploading to my sgf library and messing for a bit, maybe White lives with a seki or something)
2 Likes
I was waiting for the game to finish to share this one
I made a bad decision to try to kill the Black bottom middle group.
I cut it off the group and made the 8 point mistake of a placement rather than dropping straight down which is what the ai suggests. Although I can’t tell if the ai kills the group or what the outcome is. White loses a lot of points anyway trying to kill the black stones.
Anyway all I get below on the bottom left is a sente seki.
This is the end board position on the right. Interesting that there’s a lot of mad stuff that could happen including another seki on the right side, if you go back a few moves and let the Ai take over in some variations (I’d never be able to read all this).
There’s some sequences where White makes a seki on the right side (left image). There’s other sequences where Black cuts off the White stones with a ko, but black is also cut and needs to win the ko to make two eyes (actually that whole ko to make two eyes was something I could do much earlier but it never worked out). White can use all the threats to make the seki in the left diagram as ko threats, and then there’s even some local ko threats etc.
2 Likes
I’m also realising that saying “left” and “right” image only makes sense on my laptop not on mobile for me, so replace them with first and second or top and bottom or something 
Ah la vie en rose!
An interesting seki on the corner, and depend on the rules, the score might be a bit different.
8 Likes
4 Likes
I find it instructive how long the bot wanted the honte move at D14, for both colours.
It first suggests it as soon as Black is kicked, on (20), then (21) and (22), then all the way from (24) to (30) as play sparks off on the right side. And it almost recommends it for (23) as well.
1 Like