Not all pros can tell margarine from butter

[John Fairbairn]

Remember those adverts that begin "8 out of 10 housewives prefer X" and you always wonder about the other two? In my case I give them a little cheer, but still...

In that light, consider the joseki below. Go to Black 19. This and White 18 are not really part of the joseki.



The question now is how should White connect?

In a game I am looking at, White, a 3-dan, plays A. In eight other known pro games White plays at B seven times. The one other case is at C.

In the present game, Murashima Yoshikatsu, then 5-dan and a very good commentator, highlights A for an extensive comment. He remarks that its idea is to have an effect on Black's corner, but this is really so far into the future, and speculative even then, he recommends regarding B as the joseki. This B applies more pressure to the Black group to the left. In addition, it gives White an option to play a two-step hane with D, E, F without worrying about being cut off. C he describes as "extraordinarily bad". Despite that, a 9-dan pro (Takagi Shoichi) did play it. Speculating as to why, perhaps he regarded it as good because it had the effect of Black having committed the cardinal sin of peeping at a bamboo joint? His was the latest game (1986). Is this the latest thinking?

However, several other interesting things emerge from this opening.

First, Murashima made no mention of potential ko threats (B leaves one) or of two ugly empty triangles. My hunch is that amateurs would have latched on to those aspects first. I did. Does this mean I/we are perhaps too obsessed with trivia?

Second, White 12 is regarded as very important in this joseki. It is useful to consider why.

Third, it is very difficult to evaluate this kind of joseki in any sort of numerical way. There is no obvious split between influence and profit, and it seems very hard to imagine turning White's stones into thickness. How do we go about evaluating it? (The joseki books talk about White mitigating the effects of Black's pincer.)

Fourth, for the language mavens, Murashima twice uses joseki in its common, though not prime, meaning of "(recommended) best move" when it has never or rarely been played before.

But surely the most interesting point is that two out of nine pros preferred margarine and the 9-dan of the two played an "extraordinarily bad" choice.

[Uberdude]

Isn't Takagi Shoichi know for his individualistic approach to Go. So it wouldn't surprise me if he was well aware everyone else thought B was better, but thought he'd be different just because he likes to. C does also have another minor plus of taking a liberty off the black peeping stone which could come in handy at some point; for example if white gets a stone at m16 and then n16 then after black n17 the o17 cut is now an atari (which is admittedly pretty speculative, but not as much as taking a liberty off the corner with A I feel). Though I agree the plus of B allowing that two-step hane is bigger.

P.S. Yes I would notice the empty triangle, but that would only encourage me to play it as I take particular pleasure in playing good empty triangles and showing the much-maligned empty triangle some love

[Kirby]

I like 'B'. The black stones on top could become a weak group.

[Bill Spight]

First, Murashima made no mention of potential ko threats (B leaves one) or of two ugly empty triangles. My hunch is that amateurs would have latched on to those aspects first. I did. Does this mean I/we are perhaps too obsessed with trivia?

From the standpoint of White's shape, there is hardly any difference between A, B, and C. (Did Takagi make the bamboo joint to avoid bad shape? Gack! Surely not.) But it seems to me that B is the shape point, because it prevents Black from playing there to make a Tiger's Mouth. (One common mistake in the West, I think, is regarding shape in terms of the configuration of one's own stones, instead of efficiency in the local region.)

Second, White 12 is regarded as very important in this joseki. It is useful to consider why.

Third, it is very difficult to evaluate this kind of joseki in any sort of numerical way. There is no obvious split between influence and profit, and it seems very hard to imagine turning White's stones into thickness. How do we go about evaluating it? (The joseki books talk about White mitigating the effects of Black's pincer.)

It is difficult to evaluate any joseki numerically. People do it, though. I do too, with a margin of error of 4-5 points. Whoop-de-do!

I don't think that we can really evaluate this joseki without evaluating the White counter-pincer, and that depends on what is in the top left, at the least. Much of the influence of the White stones on the P file has been transferred to the counter-pincer. I think that falls under the heading of mitigating the effects of the Black pincer. In addition, Black has been forced low in the top right. is an important move for accomplishing that.

My concern, as White, would be to avoid becoming heavy.

It seems obvious to me that in terms of influence, B is better than A and A is better than C. I can even put numbers on the differences, with an error of much less than 4 points. By my calculations, B is about 0.2 points better than A and 0.9 points better than C. That is in line with Murashima's "extraordinarily bad" assessment. (Yes, friends, a 0.9 point loss to par is extraordinarily bad. It is like Takagi playing an amateur shodan move. Yes, amateur shodans play extraordinarily badly, from the perspective of a pro.)

Why did Takagi play C? Perhaps he had thoughts of playing O-17, as Uberdude suggests. Perhaps the influence of B was not important, given the whole board. Just because it may lose a point to par on average does not mean that it is bad on any given board.

[Sennahoj]

Bill, how are these calculations performed?

[Bill Spight]

Bill, how are these calculations performed?

With an influence function that I developed some years ago. As my remark about its errors indicates, I am under no illusion about its general value. Usually it gives results close to DrStraw's approach.

[Joaz Banbeck]

.. mitigating the effects of Black's pincer...

Mitigating the effects...I'll have to remember than next time I have an eyeless group between two opposing groups.
I'm not running, I'm mitigating.

[Splatted]

.. mitigating the effects of Black's pincer...

Mitigating the effects...I'll have to remember than next time I have an eyeless group between two opposing groups.
I'm not running, I'm mitigating.

I myself have been known to mitigate the whole board.

[Bantari]

.. mitigating the effects of Black's pincer...

Mitigating the effects...I'll have to remember than next time I have an eyeless group between two opposing groups.
I'm not running, I'm mitigating.

I myself have been known to mitigate the whole board.

That's nothing! Sometimes I mitigate my own moves!! Heh...

[RobertJasiek]

The choice among ABC depends on the positional context.

After move 19, it is easy to evaluate the joseki by applying my joseki evaluation method:

The stone difference is 10 - 9 = 1.

The territory count (of 'current territory') is 8 - 4 = 4.

The influence stone difference is 2 - 3 = -1. (In White's favour.)

Territory count and influence stone difference favour different players, as it should be.

The stone difference can be transformed to another white influence stone [imagined to be played elsewhere on the board], giving the modified influence stone difference -2.
The ratio is | 4 / (-2) | = 2.

This value is in the valid range from 1.5 to 3.5 for values denoting josekis.

The weak black and the weak white group are similarly weak, therefore the ratio meaningfully expresses to have a joseki.

Nevertheless, there is the unshown global positional context, which might favour Black, favour White or be equal / fair for both players when choosing this joseki. (The local numerical joseki evaluation does not claim to be a global positional judgement.)

[Knotwilg]

In order to evaluate the joseki, imagine white's P17 were at O16 instead. Would you feel white is better? Would robert's method yield different quantics?
Yet white's shape is much better.
What offsets white's clumsy shape? The fact that black is low?

[RobertJasiek]

My method applies to josekis, joseki-like, or almost-joseki sequences because it is calibrated for equal play. It does not apply if one player gets all the good moves and the opponent plays rubbish or the shape is an arbitrary middle game part of the board with no local guarantee of fairness. If you want to let W play where B has a stone, you need to show a new sequence and we can judge if it is meaningful in order to see whether my method is applicable.

[Bantari]

My method applies to josekis, joseki-like, or almost-joseki sequences because it is calibrated for equal play. It does not apply if one player gets all the good moves and the opponent plays rubbish or the shape is an arbitrary middle game part of the board with no local guarantee of fairness. If you want to let W play where B has a stone, you need to show a new sequence and we can judge if it is meaningful in order to see whether my method is applicable.

Do you have a method which evaluates if a sequence is meaningful?

[RobertJasiek]

Not what I would call a "method". A meaningful sequence consists of meaningful moves, i.e., moves achieving some good purposes such as increasing territory well, increasing influence well, achieving another strategic purpose well or achieving a combination of such aspects well. (This affects also the discussion of choosing among ABC. Maintaining connection is a good meaning but one should also strive to optimise the additional meaning of the move.)

[Bantari]

Not what I would call a "method". A meaningful sequence consists of meaningful moves, i.e., moves achieving some good purposes such as increasing territory well, increasing influence well, achieving another strategic purpose well or achieving a combination of such aspects well. (This affects also the discussion of choosing among ABC. Maintaining connection is a good meaning but one should also strive to optimise the additional meaning of the move.)

Hmm... so you have a method to evaluate something, but no method to tell you if you can or cannot apply the first method to begin with?

Interesting...