4-4 Low approach, attach & block joseki question

zac · #1

Hi all, in my recent games on Tygem I've had quite a lot of players play the following joseki when I approach a star point;

What sort of global things should I be considering when continuing? E.g. if B has a position in the upper right, or lower left, etc? I don't feel like I really have any constructive things to think about with this particular joseki, and it always seems to turn out good for black.

Thanks all for any input.

Zac

Joaz Banbeck · #2

The general theory behind this joseki is that black is willing to give up an outside wall and give up sente and get bad shape in return for a corner. White has little choice but to accept the deal. If he leaves the three stones unconnected they get chopped up and eventually die.
Actually, white should be happy to accept the deal. Early in the game, sente is worth a bunch.

On a global scale, you should be thinking of this as influence that will work for you later which is not easily attacked. You take sente, play moves that will compel him to fight in the center - where he will lose because you have greater influence.

( BTW, I am assuming that white plays 8 at G15 and black plays 9 at D14 or D13 )

zac · #3

Joaz Banbeck wrote:

The general theory behind this joseki is that black is willing to give up an outside wall and give up sente and get bad shape in return for a corner. White has little choice but to accept the deal. If he leaves the three stones unconnected they get chopped up and eventually die.
Actually, white should be happy to accept the deal. Early in the game, sente is worth a bunch.

On a global scale, you should be thinking of this as influence that will work for you later which is not easily attacked. You take sente, play moves that will compel him to fight in the center - where he will lose because you have greater influence.

( BTW, I am assuming that white plays 8 at G15 and black plays 9 at D14 or D13 )

Thanks for the explanation. What other options does W have other than connect at G15? Is the influence worth as much if e.g. B has a strong position in the upper right? Maybe it's wrong to approach low like this if that were the case?

EdLee · #4

Hi zac,

W has many options, including tenuki.
One must look at the whole board.

Quote:

it always seems to turn out good for black.

How do you know ? Did you check with Leela's evaluation ?

zac · #5

EdLee wrote:

Hi zac,

W has many options, including tenuki.
One must look at the whole board.

Quote:

it always seems to turn out good for black.

How do you know ? Did you check with Leela's evaluation ?

Haha no, I only mean in my games! Maybe its more of a problem of not using the influence effectively. Maybe I should take a look at what Leela thinks of my actual games, and suggested continuations after that corner situation.

Joaz Banbeck · #6

Joaz Banbeck wrote:

...On a global scale, you should be thinking of this as influence that will work for you later which is not easily attacked. You take sente, play moves that will compel him to fight in the center - where he will lose because you have greater influence...

Of course, this is the ideal strategy, which seldom comes to pass unless white is an ideal patsy. White is likely to recognize that you have influence and prudently avoid risking much in the middle. This reticence often allows you to use the outside stones to take territory along the side, or perhaps a moyo in the middle.

The important thing to note here is that you have to have a plan that threatens to make the former expensive, so that you might achieve the latter. Territory seldom requires planning to make itself known in the final score, whereas influence does.

Joaz Banbeck · #7

EdLee wrote:

...W has many options, including tenuki...

Ed is, of course correct, but that tenuki requires some fine positional judgement that should not be expected of an SDK. I deliberately omitted it to keep the position simple. I recommend sticking with the G15 connection until you hit mid-dan.

EdLee · #8

Hi zac,

Quote:

I only mean in my games!

I understood you to mean your games only and that's what I was also referring to!

Basic idea, locally: you have two cuts and you cannot fix both with one move. So:
If you want cash, fix the G17 cut.
If you want the outside, fix the G15 cut.

Globally, all bets are off, since we have to look at the whole board to decide on the next move.

Bill Spight · #9

Jiang and Rui, in their three volume text, The World's New Joseki, point out that traditional 4-4 joseki were mainly used in handicap games, and therefore are often not equitable in even games. Modern pros are discovering/devising new 4-4 joseki, which often involve sharp fighting. And, OC, since the advent of AlphaGo, even more new 4-4 joseki are being played.

Click Here To Show Diagram Code: [go]$$W Attach and block joseki $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . 4 1 . . . . . . $$ | . . . X 6 2 3 . . , . . $$ | . . . . . 5 . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . , . . . . . , . .[/go]

This attach and block joseki is one of the traditional joseki. Since Black makes an empty triangle, it may be inferior for Black. OTOH, White has two cutting points to worry about.

Has AlphaGo ever played this joseki? Not that I recall. But AlphaGo likes to jump into the 3-3.

Looking at this position afresh, I thought, why not tenuki? And indeed, according to Waltheri ( http://ps.waltheri.net ) the winning percentage for a tenuki is up there with the most frequent plays.

When I was learning go the usual continuation was for White to connect at one of the cutting points and then for Black to cut at the other. I nearly always made the solid connection on the fifth line, and indeed, it is the most popular continuation on Waltheri. Nowadays the solid connection on the third line has fallen out of favor.

Click Here To Show Diagram Code: [go]$$W Most frequent joseki $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . . 9 . . . . . . $$ | . . . . 4 1 . . . . . . $$ | . . . X 6 2 3 . . , . . $$ | . . . . . 5 7 . . . . . $$ | . . . 8 . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . , . . . . . , . .[/go]

looks a bit peculiar. How come? Because Black could have made it straightaway.

Click Here To Show Diagram Code: [go]$$W One space reply $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . 1 . . . . . . $$ | . . . X . . . . . , . . $$ | . . . . . . . . . . . . $$ | . . . 2 . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . , . . . . . , . .[/go]

Click Here To Show Diagram Code: [go]$$W One space reply $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . B 1 . . . . . . $$ | . . . X B B W . . , . . $$ | . . . . . W W . . . . . $$ | . . . 2 . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . , . . . . . , . .[/go]

Does Black really gain anything by exchanging the :bc:

stones for the :wc:

stones? On an empty board, I don't think so. The bad shape is a big clue.

Black should only play this way, I think, when she believes that White's influence is less than usual.

Uberdude · **#10**

A few sequences to get started:

White connect on outside if centre/influence important. Tends to be good if white is building something on top side/centre. a is then an important point for both, black might play there instead of capturing the one stone. White might play there instead of atari of 5 (which means black captured a stone in sente, but the ko is still big), but if white doesn't play there then black b is big to secure his eyeshape and reduce white's, who might answer at c. With the atari white keeps black's eyeshape under question, which means moves towards the corner can have attacking purpose.

Click Here To Show Diagram Code: [go]$$W $$ +-------------------- $$ | . . . . . . . . . . $$ | . . . . . 4 5 b . . $$ | . . . . X . 2 3 c . $$ | . . . X X X O . . , $$ | . . . . . O 1 . . . $$ | . . . a . . . . . . $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | . . . , . . . . . ,[/go]

Descend if you want to live locally and make territory. Cut of 2 builds centre for black, pincer at 5 instead is more aggressive choice. black might capture at 6 directly rather than push at 4. 5 might hane. 7 is a key point for strength of both groups, easy to overlook. Falling back at 8 might look odd, the point is if you block then peep or clamp means white can often scoop out the corner later, so you back off to be more sure of getting a slightly smaller corner. With extra support on left block could be plausible but leaves bad aji.

Click Here To Show Diagram Code: [go]$$W $$ +-------------------- $$ | . . . . . . . . . . $$ | . . . . 7 1 . . . . $$ | . . 8 . X O . . . . $$ | . . . X X X O 3 5 , $$ | . . . . . O 2 4 . . $$ | . . . . . 6 . . . . $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | . . . , . . . . . ,[/go]

gowan · **#11**

If I recall correctly the attach and block joseki was played by top pros like Kobayashi Koichi in games in the late '70s early '80s in Japan.

Joaz Banbeck · **#12**

gowan wrote:

If I recall correctly the attach and block joseki was played by top pros like Kobayashi Koichi in games in the late '70s early '80s in Japan.

Here is a game from 1992, with Kobayashi as white, using the influence in the center:

Click Here To Show Diagram Code: [go]$$B It starts with a fairly standard fuseki... $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . 2 . . . . . , . . . . . 1 . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . 6 . . . . . . . . . . . . 8 . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . 4 . . . . . , . . . . . 3 . . . | $$ | . . . . . 5 . . . 7 . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Click Here To Show Diagram Code: [go]$$B The opponent plays the attach and block; Koby extends... $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . 8 . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . 6 2 . . . | $$ | . . . O . . . . . . . . . . 4 1 W . . | $$ | . . . . . . . . . . . . . . . 5 3 . . | $$ | . . . O . . . . . , . . . 7 . X . . . | $$ | . . . . . X . . . X . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

So far, this looks like four samples out of a joseki book.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . 3 . X . . . | $$ | . . . . . . . . . . . . 4 . . . . . . | $$ | . . . . . . . . . . . . . . 2 . O . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . 1 . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . O O . . . | $$ | . . . O . . . . . . . . . . O X O . . | $$ | . . . . . . . . . . . . . . . X X . . | $$ | . . . O . . . . . , . . . X . X . . . | $$ | . . . . . X . . . X . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

White's play at N15 is not considered a particularly good joseki. But it works here. White is trying to compel black to fight in the center at a disadvantage.
This is how a top pro uses the outside influence of the white side of the attach and extend joseki.

Black eventually lived small on the right side, and white got a wall facing toward the center while attacking that group. The white wall was then used to make a moyo.

FWIW, here is the full game:

Kirby · **#13**

IIRC, this joseki is territorially good for black, and a good way to make a solid position. Locally, black has an advantage, but black has played an extra stone:

Click Here To Show Diagram Code: [go]$$W $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . . . . . . . . . $$ | . . . . 4 1 . . . . . . $$ | . . . X 6 2 3 . . , . . $$ | . . . . . 5 . . . . . . $$ | . . . . . . . . . . . .[/go]

White gets to play fast, and has a flexible position in the area.

It reminds me of this one:

Click Here To Show Diagram Code: [go]$$W $$ ------------------------ $$ | . . . . . . . . . . . . $$ | . . . . 0 8 9 W . . . . $$ | . . . . 2 1 . 6 . 5 . . $$ | . . . X . 3 . 7 . , . . $$ | . . . . . . . . . . . . $$ | . . 4 . . . . . . . . . $$ | . . . . . . . . . . . .[/go]

Black gets a lot of points, but white's position is pretty solid.

With both of these joseki, I think black opts to settle the position quickly, giving both sides a relatively solid position. That's probably why it's a decent choice for handicap games - eliminate complexity, and just get some solid points.

Kirby · **#14**

Joaz Banbeck wrote:

Click Here To Show Diagram Code: [go]$$B ...Koby extends... $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . O . . . . . , . . . . . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . 8 . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . 6 2 . . . | $$ | . . . O . . . . . . . . . . 4 1 W . . | $$ | . . . . . . . . . . . . . . . 5 3 . . | $$ | . . . O . . . . . , . . . 7 . X . . . | $$ | . . . . . X . . . X . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

I must say that I'm partial to the name "Koby" - it reminds me of my own nickname. Maybe I should start going by "Kirbayashi" around here...

Uberdude · **#15**

Joaz Banbeck wrote:

gowan wrote:

If I recall correctly the attach and block joseki was played by top pros like Kobayashi Koichi in games in the late '70s early '80s in Japan.

Here is a game from 1992, with Kobayashi as white, using the influence in the center:

Not that it really matters, but in that game white is Kobayashi Satoru, not the Kobayashi people usually mean if they don't qualify it with a given name and for whom the Kobayashi opening is named (but Koichi was black).

sorin · **#16**

I noticed that AlphaGo doesn't like to plays this way with black (never played move 5 from the original sequence in any public/published games, and I never saw that move rated high in the teaching tool either).

Leela and AQ also don't like that move - both prefer black to extend towards the center instead in the Kobayashi game.

zac · **#17**

Thank you all for your help. I think the first time it was played against me, I connected the 3rd line cut, and he cut the other, and it ended up working really well with his opposite side,
and I didn't get so very much territory to compensate. I've had a chance now to play against this again, and connected on the fifth line, and tried to bring the action out in to the center. It worked reasonably well that way. I also played the attachment myself in a couple of games, to get a bit of feel for it from the other end.

The Kobayashi game is pretty instructive. I also like the high two-space pincer, I've been trying that out when my opponent has a strong/high position "behind" their approaching stone, to try and limit what they can gain. I'm not sure if this is really the correct idea; but it has seemed to work reasonably well so far.

A game today went like below. I was playing Black, and played the two space high pincer. I put in the attach-block joseki to compare. The corner and side would look pretty big, but for me it feels like W gets the better of the exchange; more potential and a position that is much more helpful in the fighting to come. Looking at it now, I also wonder if I shouldn't have played 7 on the other side; after he extends it hurts my large knights enclosure. I've only just started playing the large knights rather than the small one, so 7 got played without much thought (EDIT; I asked Leela, and it likes the other side slightly better)...

Uberdude · **#18**

sorin wrote:

I noticed that AlphaGo doesn't like to plays this way with black (never played move 5 from the original sequence in any public/published games, and I never saw that move rated high in the teaching tool either).
Leela and AQ also don't like that move - both prefer black to extend towards the center instead in the Kobayashi game.

I noticed that AlphaGo does like to attach end extend in some situations that human pros tend not to, probably rejecting it as crude (e.g. some Chinese opening variations, 2nd approach to san ren sei).

Btw zac, in that last game, I would think about q17 for black 23. If pincer I don't like white going into 3-3 and living easily as the p11/r11 stones are trying to make territory on the right side so black has no good side to block without some of your stones looking bad. q17 just says "corner is mine" without making white stronger so white probably defends the top side or else you can do things there later (in fact even after adding a move there are still things for black to do on top side). Attach and block is similar idea of making corner solid, but also makes white stronger which I want to avoid.

4-4 Low approach, attach & block joseki question

Who is online