Life In 19x19

So Uberdude posted this interesting game, with the comment "a herd of elephants", and indeed I did laugh out loud when I saw come down.



But something earlier in the game puzzles me.

Click Here To Show Diagram Code

[go]$$Bc Position after black 33
$$ . . . . . . . . . . |
$$ . . . . X X . X . . |
$$ . . O X X O . O X . |
$$ . X X O O O . . . . |
$$ . . . . . . . . X . |
$$ X . . . . . O . . . |
$$ . . . . O . O X . . |
$$ . X . . . O X X . . |
$$ . . . . . . . . . . |
$$ . . . . . . . . . . |
$$---------------------+[/go]

I'm reminded of Breakthrough to Shodan, the book that more than anything else propelled me from 5 kyu towards dan level. It's not so much the technical content as the fact that the book is just full of attitude! At one point, Miyamoto gives the following diagram with the comment "If that's the way you're going to live, why don't you just drop dead?"

Click Here To Show Diagram Code

[go]$$B Miyamoto says don't do this!
$$ O . O . O . . . . |
$$ O . . . . . . . . |
$$ . . . . X . . . . |
$$ . . . . . . . . . |
$$ . . . . X . . . . |
$$ . . . . . . O . . |
$$ . . O . X . . . . |
$$ . . 4 5 . X O . . |
$$ . O 3 . . . O . . |
$$ . 2 1 . X . . . . |
$$ . . . . . . . . . |
$$ . . . . . . . . . |
$$-------------------+[/go]

So what's the difference between Miyamoto's diagram and Lee Sedol's position? My guess is that in the Lee-Mok game, white doesn't mind being surrounded because black gets overconcentrated above, so it's still a balanced position. But then I'd expect white to exploit this by forcing black to make inefficient territory in the top right. No, the opposite happens: white goes and invades the "overconcentrated" area! So now I'm totally confused.

Click Here To Show Diagram Code

[go]$$Wc
$$ . . . . . . . . . . . |
$$ . . . . . . a . X . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . b . . . |
$$ . X . . . . . 1 . . . |
$$ . . . . . . . . . . . |
$$ O . X . . . . X . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$-----------------------+[/go]

I asked KataGo: does indeed look like one of the best two moves (pretty much equal with 3-3 invasion at top right), and the outcome is an even position (black winrate 50.5%), but personally I'd feel a lot better about playing 'a' or 'b' instead.

First of all it all depends;),one proffessional told me this:as a kyu player you get taught some principles to make the game easier but as a dan player you should forget all of them since the game is actually not easy.

xela wrote:But something earlier in the game puzzles me.

$$Bc Position after black 33
$$ . . . . . . . . . . |
$$ . . . . X X . X . . |
$$ . . O X X O . O X . |
$$ . X X O O O . . . . |
$$ . . . . . . . . X . |
$$ X . . . . . O . . . |
$$ . . . . O . O X . . |
$$ . X . . . O X X . . |
$$ . . . . . . . . . . |
$$ . . . . . . . . . . |
$$---------------------+

Click Here To Show Diagram Code

[go]$$Bc Position after black 33
$$ . . . . . . . . . . |
$$ . . . . X X . X . . |
$$ . . O X X O . O X . |
$$ . X X O O O . . . . |
$$ . . . . . . . . X . |
$$ X . . . . . O . . . |
$$ . . . . O . O X . . |
$$ . X . . . O X X . . |
$$ . . . . . . . . . . |
$$ . . . . . . . . . . |
$$---------------------+[/go]

I'm reminded of Breakthrough to Shodan, the book that more than anything else propelled me from 5 kyu towards dan level. It's not so much the technical content as the fact that the book is just full of attitude! At one point, Miyamoto gives the following diagram with the comment "If that's the way you're going to live, why don't you just drop dead?"

$$B Miyamoto says don't do this!
$$ O . O . O . . . . |
$$ O . . . . . . . . |
$$ . . . . X . . . . |
$$ . . . . . . . . . |
$$ . . . . X . c . . |
$$ . . d . . . O . . |
$$ . . O . X . . . . |
$$ . . 4 5 . X O . . |
$$ . O 3 . . . O . . |
$$ . 2 1 . X . . . . |
$$ . b . . . . a . . |
$$ . . . . . . . . . |
$$-------------------+

Click Here To Show Diagram Code

[go]$$B Miyamoto says don't do this!
$$ O . O . O . . . . |
$$ O . . . . . . . . |
$$ . . . . X . . . . |
$$ . . . . . . . . . |
$$ . . . . X . c . . |
$$ . . d . . . O . . |
$$ . . O . X . . . . |
$$ . . 4 5 . X O . . |
$$ . O 3 . . . O . . |
$$ . 2 1 . X . . . . |
$$ . b . . . . a . . |
$$ . . . . . . . . . |
$$-------------------+[/go]

So what's the difference between Miyamoto's diagram and Lee Sedol's position? My guess is that in the Lee-Mok game, white doesn't mind being surrounded because black gets overconcentrated above, so it's still a balanced position. But then I'd expect white to exploit this by forcing black to make inefficient territory in the top right. No, the opposite happens: white goes and invades the "overconcentrated" area! So now I'm totally confused.

Sorry, I don't have Breakthrough to Shodan, and cannot really say much about it, but it does appear in the diagram that Black has moves like a, b, c, and d to make life with more than 2-3 pts.

$$Wc
$$ . . . . . . . . . . . |
$$ . . . . . . a . X . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . b . . . |
$$ . X . . . . . 1 . . . |
$$ . . . . . . . . . . . |
$$ O . X . . . . X . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$-----------------------+

Click Here To Show Diagram Code

[go]$$Wc
$$ . . . . . . . . . . . |
$$ . . . . . . a . X . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . b . . . |
$$ . X . . . . . 1 . . . |
$$ . . . . . . . . . . . |
$$ O . X . . . . X . . . |
$$ . . . . . . . . . . . |
$$ . . . . . . . . . . . |
$$-----------------------+[/go]

I asked KataGo: does indeed look like one of the best two moves (pretty much equal with 3-3 invasion at top right), and the outcome is an even position (black winrate 50.5%), but personally I'd feel a lot better about playing 'a' or 'b' instead.

A quick look at the Elf commentaries shows that Elf also likes as well as the 3-3 invasion. It prefers the 3-3 invasion, giving it 22k playouts while giving only 7.1k playouts. One point of interest is that Elf estimates Black's winrate as only 41% in either case, despite the fact that Elf is alone among today's top bots in giving Black almost a 50% winrate after the first move.

Anyway, you can play your options with KataGo and see what it thinks of them.

BTW:

Click Here To Show Diagram Code

[go]$$Bcm13 :b13: loses 11%, says Elf
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . 1 . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . X . . . . . a . . . |
$$ | . . . O . . O . . , . . . . . , . . . |
$$ | . . . O . X . . O . X . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

For Elf prefers a with an estimated winrate of 52% (24.2k playouts).

One big difference between how Lee lives and the book is Lee lives in sente but the book in gote. So he can use sente to reduce the potential he gave black with the inside for outside exchanges, first with a 3-3 invasion which is an easy life and no weak group getting chased in the moyo (should Mok tried to have got a sente variation?) and then reduce the centre. Also in living he got in the m8 cut and made p10 weakness end successfully persuaded black to spend a move to fix that, essentially getting a free move for his reduction operations. Plus White already has the profit in the lower left, so by living the l3 and k5 stones lose efficiency, it's only with the middle wall that White helps black. But having said all that I'd say Lee is the kind of player who might greedily live by making bad exchanges because he is confident in his fighting skills that he can deal with the potential he gives opponent later.

Click Here To Show Diagram Code

[go]$$Bcm13
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . b . . . . . . . . . . . . . |
$$ | . . . O . . . . . 1 . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . X . . . . . a . . . |
$$ | . . . O . . O . . , . . . . . , . . . |
$$ | . . . O . X . . O . X . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

I think this position is an excellent example of why go is so hard!

It is so easy to use conventional advice and come up with a Black move at A, just like the bots. For example, we could point out that 13 is not urgent because there are two big points left on the sides and Black will always get one of them as miai, so he should give priority to protecting his investment in the lower right. On one view, A can even therefore be regarded as a tedomari. We can also rely on conventional advice to tell us that the correct time to erase a moyo is one move before it is complete. From that it follows that by playing 13 at the top, and not A, Black is just playing into White's hands.

It is also easy to use one of the various patented methods pros offer. One such here is to divide the board horizontally through the centre line. What that shows us is that the position at the top of the board is equal, but at the bottom of the board Black has a much more expansive position. White has already lapsed into some overconcentration and Black has good forcing moves to make that even worse. It therefore looks good for Black to play A so that he doesn't fall into the same sort of stunted shape as White.

It is furthermore easy to rely on AI, with all the bots telling us that A is best. But they don't tell us why, and despite the surface manifestation of unity, they are just as cussedly opinionated as the humans. For example, Elf says 13 is an 11% win-rate drop, but my Leela gives 4%. But if we say that the precise numbers don't matter, it's the principle of lower side versus upper side, then why does my Leela give almost the same score for B as it does for A? At first sight, B has much, much more affinity (upper side) with 13 than it has with A. (B has in fact been played once in the GoGoD database, and 13 twice - no other games exist).

It may be that in the AI "consciousness" the split is not actually between upper side and lower side (which is my sense of what humans see first if we go by the extant games and my own experience), or moyo completion versus moyo erasure, but is rather simply side versus centre: A and B are both corner moves and we do know bots lurrrrve the corners.

Then, of course, we should consider psychology and style. Was Yi Se-tol playing the man or the board, or was he just being Yi Se-tol?

It is MUCH, MUCH easier to understand Brexit politics, or even Sheldon's string theory.