Game Finished

Move 211

shinuito[2k]

2 captures

Jon Ko[5k]

5 captures

+ 6.5

Ranked

Rules: Japanese

Black wins by Resignation

Move 1

March 22, 2025

[21:41] Jon Ko[5k]: Hi and happy mirror go! :)

Move 3

[21:41] shinuito[2k]: Have fun.

Move 105

April 5, 2025

[20:00] shinuito[2k]: Variation: I spotted this move after playing d8 and was honestly just considering resigning

Move 146

April 9, 2025

[14:51] Jon Ko[5k]: Oops, I have seven groups.

[14:51] Jon Ko[5k]: but I always wanted to prove that proverb wrong.

Move 147

[15:00] shinuito[2k]: XD

Move 149

April 10, 2025

[0:17] shinuito[2k]: I guess I'll have to defend the proverb then :)

Move 150

[8:21] Jon Ko[5k]: You can try :P

Move 151

[10:26] shinuito[2k]: P12 looks very scary

Move 153

[12:09] shinuito[2k]: This one felt like my best chance, wasn’t sure about attaching first

Move 161

[21:04] shinuito[2k]: looks even more scary

Move 189

April 12, 2025

[12:48] Jon Ko[5k]: Looks like the proverb is standing strong

Move 190

[12:51] Jon Ko[5k]: Variation: I was wondering about that ko, but I think white doesn't have to fight for it.

Move 197

April 13, 2025

[10:28] shinuito[2k]: Yeah I think you just connect and fill liberties from the outside

Move 211

April 14, 2025

[16:03] Jon Ko[5k]: Oh, yeah. N4 threatens N6, but N6 doesn't threaten anything (anymore).

[16:04] Jon Ko[5k]: Thanks for the game :)

[16:23] shinuito[2k]: Thanks for the game! Super complicated!

[16:23] shinuito[2k]: The whole game was more or less bad for me XD then I played way too risky

[16:29] shinuito[2k]: But after p12, it was just too hard to read

[16:29] shinuito[2k]: There was a lot going on

[16:30] shinuito[2k]: Variation: I find it funny that g3 was +36 points compared to the blue move

[17:58] Jon Ko[5k]: Yeah, it was very interesting. I kind of stumbled into the colliding ladders, but I was happy with the result after P14.

[18:05] Jon Ko[5k]: The having a target on the left to attack was nice, so I was okay with losing the ko on the top. Although now it seems my ko threat was way too small.

[18:05] Jon Ko[5k]: -The +Then

[18:09] Jon Ko[5k]: I regretted not getting K14, but I didn't see it as that big when I could still have played it. Now I understand.

[18:14] Jon Ko[5k]: With J4 I was aiming a bit at G5, but it didn't really help at any point.

[18:15] Jon Ko[5k]: It's just a pity I couldn't keep up the good game until the end. I guess one can feel fatigue in correspondence games too xD

[19:12] shinuito[2k]: Yeah I think I went off correspondence games a little over the last while. I think it can be easy to just play too quickly one day and toss a game over the course of a few months

[19:13] shinuito[2k]: Different kind of fatigue but definitely relevant

[22:02] Jon Ko[5k]: Yeah

[22:03] Jon Ko[5k]: Oh, G3 is really funny. +36 with tier IV is quite something

[23:21] shinuito[2k]: It's not something you see very often, but it's something I've been discussing recently

[23:21] shinuito[2k]: Variation: k4 also lost 19 points but it was the blue move

[23:22] shinuito[2k]: generally happens in complicated situations I think, where it needs more playouts to evaluate

[23:47] Jon Ko[5k]: Huh. How many payouts

[23:47] Jon Ko[5k]: playouts does it use?

April 15, 2025

[0:34] shinuito[2k]: Something like 12k the level 4 uses I think

[0:37] shinuito[2k]: I think that gets spread out though, like the top move might've goteen 5k, the brighter green moves like 1500 or so

[0:38] shinuito[2k]: but I think maybe the players next move gets some extra playouts, or since the move already exists on the board one move ahead, you can use the estimate from that position instead - something like that

[0:41] shinuito[2k]: that would be maybe for the move at g3

[0:41] shinuito[2k]: The k4 one

[0:42] shinuito[2k]: I think it put most of its playouts into that one, but again I think it could be something to do with it not seeing the end results of the capturing race

[0:43] shinuito[2k]: like 11k+ playouts, but I think the ai review, again because it's the next move in the game also, can look one move further and take the estimate from there to try be more accurate

[0:43] shinuito[2k]: We were having discussions about whether it's intuitive or not, or if there's another way/method to display it

[11:10] Jon Ko[5k]: Display what?

[11:38] shinuito[2k]: Oh as in supposing Katago is off in its estimate from one move to the next, you could have the move say it lost zero but then the bar wouldn’t agree with that “zero” change. Or you can do what’s currently done but goes against your expectation that the best move shouldn’t lose points

[14:46] Jon Ko[5k]: Ah, I see. That would make sense, but I don't know if I'd like the raw data better.