Page 1 of 1
Opening problems for AI: Problem 7
Posted: Wed Jun 17, 2020 11:50 am
by Bill Spight
Here is another surprisingly early move by Takagawa that Elf really, really does not like. It comes from GoGoD 1952-01-09a, vs. Kitani. Takagawa was on his way to his first Honinbo title.
$$Bc Black to play
Click Here To Show Diagram Code [go]$$Bc Black to play
The bottom left corner is pretty much a dead giveaway that Kitani was playing White. It is one of his trademark joseki.
There are many possibilities, so different AI may well have different preferences. But I doubt if any of them likes Takagawa's play.
Re: Opening problems for AI: Problem 7
Posted: Wed Jun 17, 2020 12:14 pm
by Bill Spight
$$Bcm13 Standard extension
Click Here To Show Diagram Code [go]$$Bcm13 Standard extension is a standard, textbook play, but Elf reckons that it loses 21½% to a play against the bottom left corner.
$$Bcm13 Bottom side moyo
Click Here To Show Diagram Code [go]$$Bcm13 Bottom side moyo
In Elf's mainline variation,
blocks White's group from the bottom side, and after White solidifies the corner,
stakes out the bottom side. It's an early moyo, but Elf still likes it.
I think that this line illustrates that the Black extension on the left side is just not big enough, by comparison.
$$Wcm14 Elf's reply
Click Here To Show Diagram Code [go]$$Wcm14 Elf's reply
In the actual game, Kitani played
at
15 , but Elf prefers the cut by 13%.
still encloses the bottom right corner, but without forming a framework.
Re: Opening problems for AI: Problem 7
Posted: Wed Jun 17, 2020 1:55 pm
by RobertJasiek
"standard, textbook play" says nothing about whether it is a locally or globally considered play. The extension is mainly locally motivated.
Re: Opening problems for AI: Problem 7
Posted: Thu Jun 18, 2020 2:08 am
by Uberdude
A year ago Jean-Sébastien Lechasseur found a very normal opening position in which LeelaZero with 2 million playouts preferred Kitani's exotic joseki:
https://www.facebook.com/groups/go.igo. ... 599366514/
Pics copied for those who avoid facebook:
lzkitani.jpg (92.74 KiB) Viewed 7895 times
mainline:
lzkitani2.jpg (83.41 KiB) Viewed 7895 times
Re: Opening problems for AI: Problem 7
Posted: Thu Jun 18, 2020 3:47 am
by Bill Spight
Thanks, Uberdude.
Interesting that White's reply to the Black sagari is to block on the 2d line. The main replies are the stand on the 5-5 or the hanging connection on the 5-6, but Kitani also shows the block in his book,
Watashi no Go, vol. 2 .
Re: Opening problems for AI: Problem 7
Posted: Fri Jun 19, 2020 8:56 am
by Harleqin
“Watashi no Go” is quite a title… ^^
Re: Opening problems for AI: Problem 7
Posted: Fri Jun 04, 2021 9:56 am
by dhu163
is a standard, textbook play, but Elf reckons that it loses 21½% to a play against the bottom left corner.
stakes out the bottom side. It's an early moyo, but Elf still likes it. 
I think that this line illustrates that the Black extension on the left side is just not big enough, by comparison.
at 15, but Elf prefers the cut by 13%. 