Another swastika 5-3 special.
Assume komi, as always.Black to play.
Enjoy!
| Life In 19x19 http://www.lifein19x19.com/ |
|
| Opening problems for AI: Problem 38 http://www.lifein19x19.com/viewtopic.php?f=15&t=17681 |
Page 1 of 1 |
| Author: | Bill Spight [ Sun Jul 26, 2020 11:07 am ] |
| Post subject: | Opening problems for AI: Problem 38 |
Another swastika 5-3 special. Assume komi, as always.Black to play.
Enjoy!
|
|
| Author: | Knotwilg [ Sun Jul 26, 2020 1:23 pm ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
Avoiding the press in any of the bottom corners is less urgent than approaching the top left, which is the big move. The top right may be more urgent. Cover or kick? Probably the latter.
There are more variations in the top right but my feeling is that Black gains less by playing in the top left than by taking control of the top right. |
|
| Author: | pwaldron [ Mon Jul 27, 2020 6:55 am ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
Bill Spight wrote:
First thought was right or bottom side pincers, but it's tough to find a good move. White may just push down and Black ends up overconcentrated. On the other hand, a kosumi or such in those corners just seems wimpy. Similarly, if Black jumps into the top left, he's going to be pushed down and risks being overconcentrated with the pincer on the top. So...top right it is. In the old days I would have played 'a', aiming to build something that would support a right side pincer. In another one of Bill's positions, though, Elf liked 'b' as the local follow-up. I won't claim any special insight, but monkey see, monkey do. |
|
| Author: | Bill Spight [ Tue Jul 28, 2020 6:57 am ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
It seems to me that a number of choices for  are playable.
In terms of Black's development the bottom right corner seems important. From there you can move towards either of the adjacent corners where Black has a 3-4 stone. Back in the 70s I might have tried the two space high pincer, hoping for something like this variation. Today, however, the bots are not all that fond of pincers. 
|
|
| Author: | Bill Spight [ Tue Jul 28, 2020 8:00 am ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
The game is GoGoD 1690-01-29a, a Castle Game between Honinbo Doteki (W) and Yasui Chitetsu.
Chitetsu chose to play a second two space pincer in the bottom left. In the 20th century  would have had the purpose of forcing Black into a low posture on the bottom side. Whatever Doteki's feeling may have been at the time, that goal was far from being articulated yet. at a, enclosing the top left corner, is playable. Then  is the kosumi at 10, which would have a good relation to . Elf docks the pincer 13½% to par. Instead of crawling with in the game, Elf prefers the kick at b by 1½%.Top choice: Shoulder hit in the top right corner
While the the top left corner looks good, Elf prefers the shoulder blow in the top right corner. To my mild surprise. The 3-3 invasion against a 5-3 is one of its favorite plays. But anyway, a second pincer this early in the game is just too much. plays the press in the bottom left corner, but, OC, a, b, and c are also playable.
The action shifts to the joseki in the top right corner. While Elf stops here, because the next play would have fewer than 1500 rollouts, we can predict the next few plays.
My main reasons for choosing this problem were first, to show how bad Elf thinks the second pincer is. The first early pincer may be doubtful, or even a minor error, but the second is definitely bad. The second reason is that Elf prefers the shoulder hit in the top right corner to the invasion or approach to the top left corner, which it values highly. I do not know what other bots will prefer, but I doubt if they will like the second pincer. Interestingly, both human solvers so far are drawn to the top right corner, as well. |
|
| Author: | Knotwilg [ Tue Jul 28, 2020 10:56 am ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
Numbering mistake Bill. Your first diagram (and subsequent) should be like this:
|
|
| Author: | Bill Spight [ Tue Jul 28, 2020 1:05 pm ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
Knotwilg wrote: Numbering mistake Bill. Your first diagram (and subsequent) should be like this:
Oops!  Thanks.
|
|
| Author: | dhu163 [ Mon Dec 26, 2022 7:22 am ] |
| Post subject: | Re: Opening problems for AI: Problem 38 |
Speculatively: Current position: 1. Looks standard though M17 is preferred at C17. The hottest fight even without is around Q15, the shusaku kosumi because getting solid will help all the fights. M17 is the standard aggressive move, but it seems a bit much here when W can make eyes with C17 and W will try to settle here in sente. 2. As B to play now, C17 or D17 seem natural but O16 is also very big or else M17 is simply misplaced as it needs to prevent W attacking it (in W's area) but to get eyespace at C17 is a bit low. This is difficult but I think that fights tend to be more important than big moves like C17. Opp to play 1. Q15 for sente is clearly biggest aiming to play D17 and then K17 or any of the other follow ups in the other corners. Past: 1. I can't understand M17 in terms of modern AI thinking, beyond it being a difficult fight. It isn't good shape when C17 gets low eyespace on the top because it doesn't make any territory as W is happy to press the right low. Check: 1. Correct, but I didn't expect W to play in the upper right rather than the upper left. |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|