Life In 19x19 http://www.lifein19x19.com/ |
|
Opening problems for AI: Problem 30 http://www.lifein19x19.com/viewtopic.php?f=15&t=17642 |
Page 1 of 1 |
Author: | Bill Spight [ Tue Jul 14, 2020 8:09 am ] |
Post subject: | Opening problems for AI: Problem 30 |
Another "Only move" problem. Black to play.
Elf's top choice got 98.4k rollouts. No other play got as many as 1.5k rollouts. |
Author: | pwaldron [ Tue Jul 14, 2020 8:46 am ] |
Post subject: | Re: Opening problems for AI: Problem 30 |
Bill Spight wrote:
Surely it must be 'a'. Nothing else seems to compare. |
Author: | Bill Spight [ Tue Jul 14, 2020 9:19 am ] |
Post subject: | Re: Opening problems for AI: Problem 30 |
pwaldron wrote: Nothing else seems to compare. Cue in Sinéad O'Connor, Nothing Compares 2U. https://www.youtube.com/watch?v=0-EF60neguk |
Author: | Knotwilg [ Tue Jul 14, 2020 2:38 pm ] |
Post subject: | Re: Opening problems for AI: Problem 30 |
Cut. The double hane overconcentrates black and helps white. |
Author: | Bill Spight [ Thu Jul 16, 2020 5:16 am ] |
Post subject: | Re: Opening problems for AI: Problem 30 |
The game is GoGoD 1660+v, Honinbo Dosaku (W) vs. Masaki Kasue.
Masaki played the anti-suji, , threatening to cut. Elf reckons that this play loses 31% to par. It is one of those plays that strengthens your opponent. I did not expect it to attract any votes. Dosaku protected against the cut with . Then Masaki played the one space pincer in the top right corner. Elf prefers a to by 2%.
plays the pincer, as above, and jumps out. The "only move"
Elf greatly prefers the two step hane, with 98.4k rollouts. No other choice gets any mention in the commentary. No other play compares, as pwaldron says. I chose this position not only as an only move problem, but also because it illustrates AI play in the center, where I feel that it is head and shoulders above human play. Still, as Knotwilg indicates, it induces two good plays by White, the 6th line (!) extension by White and the central turn, . Elf greatly prefers , giving it 88.3k rollouts. Other moves, a - e, in order of preference, are possible, perhaps even playable, but the second choice, a, gets only 4k rollouts. If White plays at e, Black replies at f.
FWIW, my own feeling is that White's threat, the atari, , is too painful. And if Black extends to 1 instead of the two step hane, White can reply at a. Back to the mainline variation.
plays the one space pincer. With the support of and , threatens to make a goodly territory in the right side. a - d are also possible After the top attachment of , White crawls and connects underneath, making territory that way.
After White connects, intrudes once into White's territorial framework with sente, and the comes back and stablizes Black's wall. |
Author: | dhu163 [ Sat Jan 01, 2022 3:25 pm ] |
Post subject: | Re: Opening problems for AI: Problem 30 |
Speculatively: Current position: 1. N6/N7 stands out to fix the shape (though not necessary locally even though K3 is so close), mostly for centre access since W is solid at F13, B is low at C7, and W has soft at P7 at the moment. 2. Otherwise, I think B needs to invade the right side. R13 sprang to mind first but it might be too close. R12 is another idea. 3. C18 is plausible but seems miai with J16 Opp to play 1. N7 stands out as a shape point 2. But Q12 might be good enough 3. O16 is nice to make shape but B will probably fight back with the wall on the left at Q15. Past: 1. W is pretty efficient on the upper left and B is low in the lower left and upper right. W has a joseki position in the lower right that is normally slightly better for W, and the exchange of K3 for M3 is normally not bad to get B to retreat with a big centre wall when B's wall still has weaknesses. W's D17/8 don't make much sense but can be rescued with a follow up to invade the upper side, so they aren't too bad. 2. I have to prefer W. Overall: 1. I'm inclined to suspect that invading the right is more important than N6. Perhaps a move like R13 will properly make use of the N17 fast dodge to the upper side. Check: 1. Not accurate enough. Hard to believe the pro move, B had S3 hanging and a bad exchange for W and instead make it an almost even exchange and possibly even good endgame. |
Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |