Life In 19x19

In his book "Positional Judgement: High-Speed Game Analysis", Cho Chikun comments his 1982 Meijin title match against Otake Hideo. Cho is black, Otake is white. Here is the position at the end of the fuseki.

Click Here To Show Diagram Code

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

Cho considers that LZ157, after 3000 visits, thinks that A few moves later,

Click Here To Show Diagram Code

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

ELFs opinion:

I suppose the key issue here is how much white territory there will be around h7; LZ is good at finding the right depth of reduction to keep the lead: deep enough white doesn't get more than he deserves, but shallow enough that if you play too aggressively to try to surround then LZ makes a nice sabaki inside (using weaknesses like f5 cut there are counterattack possibilities) and your points evaporate. But isn't Cho pretty good at this take cash and reduce/invade/sabaki style too, so I'd have thought this is the kind of position he'd do well in. Also the double 2-space extension on the right has some problems, but so does black's lower right corner being cramped on both sides.

Black has more solid territory, but White has quite more potential, so overall Black has played a pretty bad fuseki.

I'm wondering whether there is a translation issue here. I assume the above is not a quotation from the book as the English is wrong. But, also, a Japanese commentary I have says simply that up to White 38 (the two-space extension to R13) the White stones have had to work harder.

Furthermore, the key to this game according to the human experts was not the fuseki but a Black overplay on move 87, so that by move 98 White had recovered somewhat and felt in the mood where he thought he could stage an upset. That obviously implies Black was substantially ahead in the early middle game.

Incidentally, in the later position shown, after Black G11, White played H9 but the commentary says J8 was stronger - this may be of special interest to Ian Butler as in his separate thread asking about how to use thickness I think he overlooked surrounding moves as one good (often necessary) option as here. But what does Lizzie think?

@John: the bad English is indeed mine. My book is in English, I didn't quote the entire paragraph but I don't think I distorted the meaning: Cho was clearly unsatisfied with his fuseki, and thought that White was slightly ahead.
Cho's book does mention move . He thought he was ahead between moves 75 and 86, and that should have been at S5 instead of S6.

Concerning White at J8 instead of H9: Lizzie says that Black's winrate is above 60%, here is the main variation (the picture indicates White's winrate):
Note: I assumed that is at E9, like in the .sgf file I found at go4go and on my app GoMoves, however the book says that is at F9. This doesn't change the winrate substantially, and the first 6 moves of the main variation (cut, atari, extend, etc.) are the same.

jlt wrote:(BTW, could you or another native speaker point out my mistake? Is it the expression "quite more potential"? Should "quite more" always be followed by an adjective? I have seen the expression "quite more money", is it correct or also a mistake?)

Yes, that's odd. You might say "quite a bit more potential", but even more natural is "significantly/much more potential", if you mean to say a lot more. "Quite more money" also sounds odd to me.

I was actually struggling to think of a phrase that would include "quite more", but a google search suggests it appears in Henry James and a few other places, so perhaps it's ok.

John Fairbairn wrote: Furthermore, the key to this game according to the human experts was not the fuseki but a Black overplay on move 87, so that by move 98 White had recovered somewhat and felt in the mood where he thought he could stage an upset. That obviously implies Black was substantially ahead in the early middle game.

I had a quick look with Elfv1 (so assumes white has 7.5 komi vs the actual 5.5 of game) on a weak computer (<1k playouts). Under these different komi conditions it thinks 87 was indeed a mistake, but only a small one (-2%, and white's corner attachment best reply) and the more restrained kosumi best, but things were already bad for black (23%) at that point. Is that -27% from even more or less than the 2 points komi difference? My guess is they could be about similar, so maybe Elf thinks this game is about even (+/- 10%) at 5.5 komi.

John Fairbairn wrote: Incidentally, in the later position shown, after Black G11, White played H9 but the commentary says J8 was stronger - this may be of special interest to Ian Butler as in his separate thread asking about how to use thickness I think he overlooked surrounding moves as one good (often necessary) option as here. But what does Lizzie think?

Elf thinks this was the right area of the board but h9 was better than the suggested j8, but best (so far at 2k playouts, g10 rising) was h8. The question here seems to be is it reasonable for black to cut at f5 next and white's move needs to make the answer 'no'. Against the game move of h9 Elf wants to immediately attach at q10, a classic shape weakness of the double 2-space extension (so I'm glad that was on my radar), white extends up to r11, black extends down to q9, white doesn't obediently connect at r9 (black would then treat those exchanges as useful kikashi and reduce around L8 if so) but jump at p7, black r9 white r16 and then black violent f5 cut. Against the commentary suggested j8 Elf wants to immediately cut at f5 and white can't capture it (j8 is one line off making a net work). Against Elf's recommended white h8 it wants black to q10 attach again but now that h8 is a more secure surrounding move it's ok for white to defend at r9 because black doesn't have such a severe move in the h7 area to utilise those q10 helping stones it got in sente. Does the commentary mention the f5 cut, Elf views that as very important (basically if white does any move that's not immediately sente other than the 3 mentioned black will cut).

Uberdude wrote:Under these different komi conditions it thinks 87 was indeed a mistake, but only a small one (-2%, and white's corner attachment best reply) and the more restrained kosumi best, but things were already bad for black (23%) at that point.

Interesting, so Cho and LZ157 think that Black were ahead at move 86 while Elfv1 thinks the contrary.

Concerning the F5 cut, I don't have Cho's book at hand. From what I remember he doesn't talk much about it, except a few moves earlier he said he was unsatisfied with his corner invasion because afterwards he had to add a protective move at E11, and this gave White the opportunity to make the extension at N3, so a black cut at F5 wouldn't be too severe.

jlt wrote: Concerning the F5 cut, I don't have Cho's book at hand. From what I remember he doesn't talk much about it, except a few moves earlier he said he was unsatisfied with his corner invasion because afterwards he had to add a protective move at E11, and this gave White the opportunity to make the extension at N3, so a black cut at F5 wouldn't be too severe.

Elf says he didn't need to defend at e11, in fact it was a large mistake (-25%) and that white's n3 response was almost as large (-23%). For white Elf preferred to tightly atari the stone at c7 (and then black would cut anyway), and for black preferred to play b6 to prevent white easily settling that half of the group. The expected continuation would be white hane under. black takes ataris at a7 and a4 then f5 cuts, white defends at d7 and then black defends at e10 having got the cut in sente and given white some bad aji, and because of that bad aji white will end up playing moves like d9 peep so the d12 stone is unlikely to cause problems and black can gobble it more efficiently. Elf's preference for white to play so tightly is perhaps a reflection of the big komi it assumes and if trained on 5.5 it might not have such a dismal view of Otake's looser n3 extension, the idea being with the small komi white needs to take more chances and try to make something a little bigger and looser and hope black can't mess it up. But my guess is it would still not like it and prefer these sharp lines looking for optimal efficiency.

P.S. Elf's not a fan (-12%) of white d6 (which makes sense as it sees black b6 as powerful even with the added move and f5 cut still exists), preferring either solid f6 (after which black e11 defence this is best, I suppose Otake didn't like how black can still force at d6 and g5) or f7 (trying to be a bit more efficient than f6, at first would e11 again but then wants to f5 cut as white doesn't have ladder and fighting erupts) and then after >1k playouts d10 attach, saying I don't need to defend if I pull out the d12 stone and complicated stuff happens (black ignoring to f5 cut looks stronger than d9 hane answer).

The Kido commentary at the time does not mention anything between 34 and 45 but there is fun to be had with Black 19.

This is not quite the early 3-3 revolutionarily used by AI bots, but it seems to lie within the same nexus.

It was the butt of Otake's first comment after the game: "You can't play stupid moves like that!" Sakata and Ishida and Kato all agreed with him.

Otake could talk to Cho like that as he was a much older co-pupil in the Kitani school, and the others were in similarly elevated positions. Moreover, Cho, 26, was also considered to have matured only recently.

But given Cho's stellar career after this, did he, AI-like, see something about the 3-3 point the others missed?

Again, what do Betsy and the Goblin say about Black 19?

That 3-3 looks pretty normal to me based on what pros were doing in the 2010s before AI (e.g. very common against Chinese opening). Elf gave it a few playouts and says it loses 1.5%. #1 moves are e5 peep, a bot favourite, and j4 with e11 3rd. My thinking would be it's trying to make k3 inefficient if white chooses to cut with b5 like in the game instead of d3 and taking sente. Seeing as black has r10 the c6 stone is fairly safe so my first thought would be to d3 block, but with d12 not subdued there is some extra value in cutting. Elf would d4 but thinks b5 is <1% worse. However, c5 connect is 11% mistake, prefers b6 crawl, a shape lesson we often see from bots with this 3-3 following an (often pincered) approach. But black defending at d3 is overly solid to my modern eyes (probe at b6 first is standard with this 3-3 against Chinese opening) and Elf agrees -6% though can't make up its mind between b6 or directly f5 cut (b6 descend, a6 hane, d3 connect, f6 tiger, g5 peep, f5 connect, d6 peep, d5 connect, e11 defend is one line that's nice and clear and I've seen/played such things before).

Btw, Elf says Cho's biggest mistake, -17%, in the centre reduction phase was k5: too wishy-washy (Otake's jump a good answer, considered a bit but prefers g12 hane f11 answer first), Elf wants to just go ahead and cut at f5 like a brute. Also looks at q10 and k4 attachments a bit.