Today I came across two articles side by side which refer to this thread's joseki, repeated below.
- Click Here To Show Diagram Code
[go]$$B Joseki
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . a c . . . . .
$$ | . . 8 b . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . d . .
$$ | . . 4 2 6 . . . .
$$ | . . 3 5 . 7 . . .
$$ | . . . . 1 . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ -------------------[/go]
The first was on "Even-game josekis" by Honinbo Shusai. His first reference to White 8 was that it was the "usual" move (and he also remarks that Black 7 is a thick move that allows him to focus on the right side). This thread has no quibble with that. In his second reference he gave this joseki as part of the larger position below.
- Click Here To Show Diagram Code
[go]$$ Practical example
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . c , . . . . . , . . . . . , . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . d . . . . . . . . . . . . . . . |
$$ | . . . . . . a . . . . . . . . . . . . |
$$ | . . O O O . . . . . . . . . . . . . . |
$$ | . . X X . X . . . , . . . . . , X . . |
$$ | . . . . X . . . . . . . b . O . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]
Here he says that the three-space extension shown may the correct move
locally but, given the disposition of the lower right here, that would allow Black make a perfect move by getting first to 'a', which implies a pincer at 'b'. Therefore White will "rather often" play at 'a' first before extending on the side. In that case, if White can then still extend on the left side it will be much bigger for him to play at 'c'. But if White plays the wider extension at 'c' first and then ignores a subsequent Black 'a', he has to accept that Black has the threat of a severe invasion at 'd'.
Again I think the posters on this thread will have no quibble with any of that. However, there are implications to take note of. (This is my own comment now.) One is that if White extends before playing the knight's move, he is clearly imposing on himself a strategy in which he would like to get to 'a' himself or in which he hopes to remain sanguine if Black gets there first. Equally clearly, getting to 'a' for both players is more important if the extension is wider, and this is exactly what Kitani said, you may recall. So again, I think we are all on common ground. Another implication, for the wide extension only, is that as a Black play at 'a' threatens invasion, it might seem that White has to have a defence prepared. If so, where? In fact, he almost never defends at all. Although we only have a few games to go on, leaving this three-stone wall weak by tenuki-ing is a normal strategy, and is the subject of the second article, by Segoe Kensaku. I would expect some readers not to be very familiar with that strategy, but the classic example of leaving a three-stone wall as bait can be seen in Game 1 of the Kamakura ten-game match.
That's all by way of recap. But it got me thinking about what we mean by 'local' and what do we
really imply when we say a joseki choice depends on the local game. To put it another, simplistic, way: how often does a joseki depend on the larger position?
This example may or may not give an accurate picture but FWIW here are some figures. GoGoD has 269 games with this joseki up to Black 7. The joseki move 8 appeared in only 81 cases. Is this a hint that full-board considerations overrode the locally standard move 188 times? (Not to mention that full-board considerations may have also been a big factor in choosing the standard move.)
The other valid extensions ('a', 'b' and 'c' in the first diagram) account for 70 cases. The knight's move 'a' by White, instead of extending, occurs 41 times, thus confirming what Shusai said. That leaves a total of 77 cases where White tenuki-ed - almost as common as the standard joseki move and so justifying Segoe's treatment of this topic. although obviously we cannot say that every instance turned into a bait strategy.
Kitani's point is also borne out by the figures. Where White made the three-point extension (8) Black chose to play the knight's move more often than White (22 to 16 cases), but with the 4-space extension White was rather keener to get there first and the split was 9-9 (with moves other than the knight's move chosen 7 times).
None of this proves very much, but one thing it does point to, I think, is how well the pros were on top of the data. Recall that this was well before any large-scale dictionary of josekis existed, and sets of collected games were still rare. In fact, published games of any kind were hard to keep track of, being often published only locally in regional newspapers. Yet the three pros mentioned here all made general observations that accord very well with database data. Impressive!