Life In 19x19

John Fairbairn wrote:Where is the evidence that this is joseki?

The position shown occurs just twice in the GoGoD database and in neither case was a pincer involved. The stone at M17 is played after the rest of the moves, and as a check (tsume), not as a pincer. The check is also played deep into the middle game with many stones close by. In neither case is the position remarked on in the accompanying commentaries in the sources.

I just looked this up as well in the GoGoD database (although my copy may be slightly out of date) and it is rare. However, the issue does not seem to be this particular joseki, but rather the low two-space pincer. There are only 40 games in which this position occurs with white to play. The low one-space pincer version of this with white to play, however, occurs in 2,465 games.

BlindGroup wrote:

$$W
$$ ------------------------+
$$ . . . . . . . . . . . . |
$$ . . . . . . . 5 . . . . |
$$ . . 6 . . X . . . 1 7 . |
$$ . . , . . . . O 3 X 4 . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . 2 . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |
$$ . . , . . . . . , 8 . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |

Click Here To Show Diagram Code

[go]$$W
$$ ------------------------+
$$ . . . . . . . . . . . . |
$$ . . . . . . . 5 . . . . |
$$ . . 6 . . X . . . 1 7 . |
$$ . . , . . . . O 3 X 4 . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . 2 . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |
$$ . . , . . . . . , 8 . . |
$$ . . . . . . . . . . . . |
$$ . . . . . . . . . . . . |[/go]

Shinji just declares this position equal without explanation

Stone difference = 1.
Influence stone difference = 1.
Territory count = 0.

Adjusted:

Stone difference = 0.
Influence stone difference = 0. (White plays 1 influence stone elsewhere.)
Territory count = 0.

This falls into the value type "equality" and therefore may be considered joseki.

However, this analysis presumes that we may count 3 white influence stones. White must be able to use them, e.g., by attacking the upper black group. If the alleged white influence stones are mostly neutral stones because in particular Black has strong support on the left upper side, White must not play this joseki (unless as an emergency quick settling during the middle game). Also only the central white stone as significant influence stone would be insufficient (except for emergency).

Usually, the strong white shape must have good use including the stone 5.

EDIT: count.

Like Gomoto did, I suggest to try to "settle" such questions by analyzing it with Leela and see how it plays for a while.
It is important to do this for both sides, to get a balanced perspective.
Needless to say, you would need to fill the other corners too (with some common-sense fuseki, and match the overall number of stones for black and white), since unlike humans, Leela doesn't know how to do "local analysis".

As for a pure local analysis, as it was already suggested, I would add an extra move for white (so both sides get an equal number of stones), most likely by capturing the black stone on the 2nd line, before trying to jump to any conclusion.
The result will be that white has a stronger position locally and more territory compared to black, so black's only way to balance is to develop the upper side and/or right sides.

Bill Spight wrote:

BlindGroup wrote:Does it matter that white retains sente after this sequence?

The net number of stones matters. In both cases Black has one extra stone. For equity, early in the game Black should have around 14 pts. more than White, including pts. for influence. At the end of the game one extra stone may not be worth any territory at all.

Bill, I'm not sure that I understand your point. I think you are saying that white needs another local move in order to equalize the number of local stones. But by that logic, one would never tenunki in joseki in which one is playing the approach stone.

My understanding of what it meant to have sente in this situation is that white can now consider the relative value of a move locally to the value of a move somewhere else. White retains the option of playing locally as you suggest, but it is the option of tenuki at this point that makes white's position valuable. Take the joseki from the Shinji dictionary. Through , white would suffer a VERY large local loss by tenuki-ing. So, in almost all circumstances, white will be better off playing the prescribed moves through 7. However, after , black has one more local stone, but black now needs two moves to either enclose or kill white. So, if white tenukis and black comes back to this area first, white suffers a loss, but a much more limited loss than if white tenukied through .

For example, consider white's move on this board after having finished the joseki above:

Click Here To Show Diagram Code

[go]$$c
$$ +---------------------------------------+
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . O . . . . |
$$ | . . . . . a b . . X . . X . . . O O . |
$$ | . . . O . . . . . , . . . . O O X X . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , X . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . , . . . . . , . . . |
$$ | . . . . . O . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ +---------------------------------------+[/go]

Would adding another stone in the top right be more valuable for white than a move in the top left at say A or B?

And just to be clear, I realize that one can usually create a board that justifies typically non-optimal play. I took the claim that white has sente in these positions to mean that the local loss from tenuki is small enough that white almost certainly needs to look around and decide whether another local move is bigger than a move somewhere else.

BlindGroup wrote:

Bill Spight wrote:

BlindGroup wrote:Does it matter that white retains sente after this sequence?

The net number of stones matters. In both cases Black has one extra stone. For equity, early in the game Black should have around 14 pts. more than White, including pts. for influence. At the end of the game one extra stone may not be worth any territory at all.

Bill, I'm not sure that I understand your point. I think you are saying that white needs another local move in order to equalize the number of local stones. But by that logic, one would never tenunki in joseki in which one is playing the approach stone.

Why not? The question of whether to tenuki is separate. That White retains sente means that White played first locally and Black played last locally. That we started from the single Black stone on the 3-4 is indicated by the phrase retains sente. White had sente and keeps it.

The thing is that equity in a joseki does not mean equality. Rough equality depends upon each player playing the same number of stones locally. Equity does not. Black started off with around 14 pts. in the corner. After 10 more moves, 5 by each side, we expect the value of the corner to remain around 14 pts. (Edit: Correction: Actually, because the size of plays generally decreases as play proceeds, we normally expect White to have gained around ½ pt. or a little less in 10 consecutive plays, starting with White to play.) If Black had taken sente in the joseki, then, with correct play we would expect the local result to be around 0, plus a little bit for Black.

The longer the joseki, the less, on average, we expect each play to gain, until, in the extreme, each play gains nothing. In that case it does not matter whether White retains sente or not. (This is actually an argument in favor of tenuki, BTW. )

Leela and other strong bots are trained to take sente and komi into account, and so each correct move should have little effect on their evaluation.

BlindGroup wrote:

Bill Spight wrote:

BlindGroup wrote:Does it matter that white retains sente after this sequence?

The net number of stones matters. In both cases Black has one extra stone. For equity, early in the game Black should have around 14 pts. more than White, including pts. for influence. At the end of the game one extra stone may not be worth any territory at all.

My understanding of what it meant to have sente in this situation is that white can now consider the relative value of a move locally to the value of a move somewhere else.

What it means is that White plays elsewhere.

White retains the option of playing locally as you suggest,

Not my intention to suggest that.

Bill Spight wrote:The thing is that equity in a joseki does not mean equality. Rough equality depends upon each player playing the same number of stones locally. Equity does not. Black started off with around 14 pts. in the corner. After 10 more moves, 5 by each side, we expect the value of the corner to remain around 14 pts. (Actually, because Black cashes in the komi as play proceeds, we expect Black to have gained around ½ pt. or a little less in 10 plays.) If Black had taken sente in the joseki, then, with correct play we would expect the local result to be around 0, plus a little bit for Black.

The longer the joseki, the less, on average, we expect each play to gain, until, in the extreme, each play gains nothing. In that case it does not matter whether White retains sente or not. (This is actually an argument in favor of tenuki, BTW. )

I think I see what you are saying now. This is a very interesting way to think about what is going on in a joseki. I'm going to have to think more about it, but this is very helpful.

BlindGroup wrote:

Bill Spight wrote:The thing is that equity in a joseki does not mean equality. Rough equality depends upon each player playing the same number of stones locally. Equity does not. Black started off with around 14 pts. in the corner. After 10 more moves, 5 by each side, we expect the value of the corner to remain around 14 pts. (Actually, because Black cashes in the komi as play proceeds, we expect Black to have gained around ½ pt. or a little less in 10 plays.) If Black had taken sente in the joseki, then, with correct play we would expect the local result to be around 0, plus a little bit for Black.

The longer the joseki, the less, on average, we expect each play to gain, until, in the extreme, each play gains nothing. In that case it does not matter whether White retains sente or not. (This is actually an argument in favor of tenuki, BTW. )

I think I see what you are saying now. This is a very interesting way to think about what is going on in a joseki. I'm going to have to think more about it, but this is very helpful.

Sorry, I made an error in the quote, which I have corrected. Because White plays first in the 10 move sequence, we normally expect White to have gained a fraction of a point, not Black.