Life In 19x19

A situation that tends to happen every now and then where I have no idea what to do.

First of all, the beginning of a game I just played as black:


Whites peep was a pretty big mistake according to LZ, but from here on it all went pretty much downhill for me.

Supposedly the pressing down after a pincer joseki gives me a good result, but I find it hard to continue. White has a handful of points at the top, but the main problem is that she is safe there, which means black has two unsettled groups to whites one.

LZ has three general ideas, none of which I can comprehend as good.
- Keep pushing at the top, this lets white wrap around the bottom group and takes pretty much all the pressure off.
- Make a base for the bottom group. Here there are some more variations, but again white can wrap around and black is left with the weakest group.
- Keep running with a knights move. This gets me a wall towards the bottom, which looks ok, but again I'm continuing with two baseless groups vs one, which seems like a recipe for disaster.

Can anyone illuminate me a bit?

pak0, this is a pretty difficult question, seeing as the collective wisdom of human pros struggled to figure out what the strong AIs have done. Before strong AI, pressing down on the 3-4 stone (more-so in isolation rather than in response to a tight pincer, this press here is not new and white normally pushes and cuts) was often cautioned against as "are you sure you want to make these exchanges? you give your opponent some solid territory and you get a wall of questionable value that might even become weak in the future fighting". You seem to be feeling the truth of this view. However, the bots tend to say "yeah, I'm sure I want to make those exchanges, and I'm good at handling that wall, it's not as weak as it looks, and pressing white down low is nice". So if pro didn't play/understand positions like this as well as bots, it's no wonder you struggle with them too! Sorry I can't give more concerete advice, though I will say remember this was white's side of the board, so if you can restrict white to 15 points on the top, settle on the left with some territory and not die with the wall, you are probably doing quite ok seeing as you have 2 4-4s on your half of the board. So it might be a case of realigning your judgement of what a reasonable result looks like, maybe you think it's bad for you if white's pincer stone doesn't die or get in big danger ("I gave white all the top side, so it deserves to die!"), but maybe it's ok, see what LZ says.

My first thought is the keima, G-10, and continue the fight.

paK0 wrote:- Keep running with a knights move. This gets me a wall towards the bottom, which looks ok, but again I'm continuing with two baseless groups vs one, which seems like a recipe for disaster.

If it were a recipe for disaster, then E-16 was questionable, and E-11 was bad. The position after G-10 is better for Black than the position after E-11. First, the peep is bad for White, as it strengthens Black unnecessarily, and second, the diagonal play at F-12 is heavy, also strengthening Black after G-10, and allowing Black to get ahead with that move. It is White who should be worried.

Black's wall on the left side does not need a base. Not that Black won't extend from it later, but it is quite strong. Black's running group has the option of later attaching at C-17, but may not need to. White's running group is weaker than Black's running group.

Uberdude wrote:pak0, this is a pretty difficult question, seeing as the collective wisdom of human pros struggled to figure out what the strong AIs have done. Before strong AI, pressing down on the 3-4 stone (more-so in isolation rather than in response to a tight pincer, this press here is not new and white normally pushes and cuts) was often cautioned against as "are you sure you want to make these exchanges? you give your opponent some solid territory and you get a wall of questionable value that might even become weak in the future fighting". You seem to be feeling the truth of this view. However, the bots tend to say "yeah, I'm sure I want to make those exchanges, and I'm good at handling that wall, it's not as weak as it looks, and pressing white down low is nice". So if pro didn't play/understand positions like this as well as bots, it's no wonder you struggle with them too!

The wall Uberdude is referring to is the one that is sandwiched between the two White groups, not the one in the middle of the left side. The pre-AI wisdom was that E-16 was only one option this early in the game. I guess Leela Zero likes it. But such running battles, in general, are not unusual in human vs. human play. And yes, they can be difficult. Welcome to the club. The right attitude for Black is that he is exchanging White territory on the top side and corner for Black influence on the left side, assuming the two groups in between can each make do. If Black's influence on the left side is not at least equitable with White's territory on the top, then Black should not provoke the running battle in the first place. (Since White started with one more stone than Black locally, Black's influence does not necessarily have to equal White's territory for the fight to be OK.) In this case White unnecessarily increased Black's strength and influence on the left side, so all is good.