Center Joseki/Tesuji

SmoothOper · Post by **SmoothOper** » Mon Jul 08, 2013 11:35 am

I am trying to develop my center game, beyond the fourth line, especially situations where groups are isolated from the side, or there are limited eye-space on the side.. Finding patterns/tesuji in these areas seems very difficult compared to Joseki Fuseki searches are easy in kombilo for example, because the board is fairly empty. Any suggestions on where to start?

lobotommy · Post by **lobotommy** » Mon Jul 08, 2013 11:44 am

"I would like to go somewhere. Any idea where I should start?"

Well... if you ask general question you'll get general answers.

a small tip for you: view pro games for new ideas, play more games, become stronger. Ask again after a month, but provide some examples from your games first.

leichtloeslich · Post by **leichtloeslich** » Mon Jul 08, 2013 11:46 am

I would say learn-by-doing. Do you know the shape game? http://senseis.xmp.net/?ShapeGame

Seems tailor made for your purposes. You could play it on KGS (just ask if someone will play it with you, explain the rules and make a free game), for example.

And I think on tygem you can even play it natively (it's an option when you negotiate time-/game-settings).

snorri · Post by **snorri** » Mon Jul 08, 2013 12:14 pm

SmoothOper wrote:I am trying to develop my center game, beyond the fourth line, especially situations where groups are isolated from the side, or there are limited eye-space on the side.. Finding patterns/tesuji in these areas seems very difficult compared to Joseki Fuseki searches are easy in kombilo for example, because the board is fairly empty. Any suggestions on where to start?

It's not that easy, but use of the green dot wildcard helps. The following search shows crosscuts that occur at least on the 5th line:

: search2.png (685.75 KiB) Viewed 7184 times

It's a nice one to show that that crosscut/extend proverb is usually wrong...

snorri · Post by **snorri** » Mon Jul 08, 2013 1:01 pm

BTW, that search I did, in case you are thinking the extension will become the most common response if there is more space around the stones, you're in for a surprise.

If I move the inside of that green square out one point on each side, I get 43 games, with 25 ataris and 17 crosscuts. That's still 58% ataris. If I move it 2 stones out, there are only 8 games and 4 ataris and 4 extensions in each.

I think Richard Hunter did something like this when writing his book "Crosscut Workshop."

This kind of search can be useful for studying central fights, because it will zip you right to the action when you go through the games returned. The other thing to note about these results is that the attachment/hane/crosscut pattern is fairly rare in isolation, much to the disappointment of Hikaru no Go fans...

SmoothOper · Post by **SmoothOper** » Mon Jul 08, 2013 1:27 pm

snorri wrote:BTW, that search I did, in case you are thinking the extension will become the most common response if there is more space around the stones, you're in for a surprise.

If I move the inside of that green square out one point on each side, I get 43 games, with 25 ataris and 17 crosscuts. That's still 58% ataris. If I move it 2 stones out, there are only 8 games and 4 ataris and 4 extensions in each.

I think Richard Hunter did something like this when writing his book "Crosscut Workshop."

This kind of search can be useful for studying central fights, because it will zip you right to the action when you go through the games returned. The other thing to note about these results is that the attachment/hane/crosscut pattern is fairly rare in isolation, much to the disappointment of Hikaru no Go fans...

I tried it out. So that search says, give me any crosscut pattern with a one space gap around it and is five spaces from the side. That's awesome, whats next regex search?

snorri · Post by **snorri** » Mon Jul 08, 2013 1:53 pm

SmoothOper wrote: So that search says, give me any crosscut pattern with a one space gap around it and is five spaces from the side. That's awesome, whats next regex search?

At least 5 spaces from the side of you don't select "fixed anchor." So it catches ones that are higher than that. If you want it to be exactly at the fifth line, put the search region and pattern at the edge of the board.

SmoothOper · Post by **SmoothOper** » Mon Jul 08, 2013 2:14 pm

snorri wrote:
SmoothOper wrote: So that search says, give me any crosscut pattern with a one space gap around it and is five spaces from the side. That's awesome, whats next regex search?
At least 5 spaces from the side of you don't select "fixed anchor." So it catches ones that are higher than that. If you want it to be exactly at the fifth line, put the search region and pattern at the edge of the board.

: Capture_search.JPG (97.6 KiB) Viewed 7129 times

I like the triple knights enclosure, it came up with some interesting results. Is there anyway to remove wildcards, as far as I can tell there isn't, so pardon the green spot outside the search area.

Unusedname · Post by **Unusedname** » Mon Jul 08, 2013 3:42 pm

If you'd like to find more tesuji's look in areas with either bad shape (Empty triangles, hane at the head of two/three stones both sides) or segments with low liberty counts.

Also try counting how many free moves you need to severely torment a group. if you only need one more move chances are there is a tesuji nearby. Not necessarily right away but there is potential for one.

As far as center joseki.

All I can think is "Stay ahead." and push weak groups where you want them to go.

Idk center joseki sounds weird to me.

SmoothOper · Post by **SmoothOper** » Mon Jul 08, 2013 6:00 pm

I wish there were a way to search for the most common patterns for say 6 7 or 8 stones in say a 5x5, 6x6,7x7 area at or inside the fifth line.

RobertJasiek · Post by **RobertJasiek** » Mon Jul 08, 2013 10:19 pm

SmoothOper wrote:Any suggestions on where to start?

1) Improve your understanding of the middle game. If the game opening involves the center, it behaves very much like early middle game.

2) Study games with early center developments. (This was discussed here a couple of months ago.)

3) For literature, ask in the appropriate Books forum.

4) If your question is specific for good first center move intersections, the answer is: every intersection works (but differently).

wineandgolover · Post by **wineandgolover** » Tue Jul 09, 2013 5:13 am

Find a book on haengma, which is about one side trying to move to the center in a useful way. I suspect that is exactly what you are talking about, rather than running battles. I'd suggest starting with something like, "This is Haengma" by Kim Sung-rae.

SmoothOper · Post by **SmoothOper** » Tue Jul 09, 2013 6:03 am

wineandgolover wrote:Find a book on haengma, which is about one side trying to move to the center in a useful way. I suspect that is exactly what you are talking about, rather than running battles. I'd suggest starting with something like, "This is Haengma" by Kim Sung-rae.

I suspect you are correct, I was looking at that last night. I am still interested in searching for common center sequences. It seems Kombilo uses python in some way, I wonder if the hash search is script-able.

SmoothOper · Post by **SmoothOper** » Tue Jul 09, 2013 10:06 am

It looks like there is a way to do scripting:

http://dl.u-go.net/kombilo/doc/scriptin ... ern_search

Life In 19x19