Life In 19x19

kvasir wrote:

Knotwilg wrote:Analysis 4 - sacrifice tactics / driving / what if you're not happy with standard sequences?

This is the game: http://ps.waltheri.net/database/game/1444/
It is interesting that this is in a book by Ishida Akira because the idea was refuted by his opponent. White looks silly because he played away from a shoulder hit, three times!
Is there no mention of this in the book?

It would be curious to see what the AI think of the counter-attack.

Ishida/Davies were humble enough to acknowledge the opponent saw the "trick" and dodged it. I didn't mention it because I found it more relevant that Lizzie thought the trick was actually counterproductive.

Knotwilg wrote: Ishida/Davies were humble enough to acknowledge the opponent saw the "trick" and dodged it. I didn't mention it because I found it more relevant that Lizzie thought the trick was actually counterproductive.

Ok, it is mentioned. That is honest.

You said that the win percentage goes from 69% to over 80% but that seems to me to be from winning to winning. The final position in that variation is worthwhile to study on its own for sure. The variation just looks dubious, not just because it is good for black instead of white but because it seems to require black and white to cooperate to arrive at the final position. Blacks connection is not the usual shape move (but maybe it is great here) and white is under no obligation to hane.

Starting from a position that is 69% for black probably also has an effect on who this fight favors. I assume this was for the original position before the peep. The side that was better at the outset maybe doesn't have to achieve much when attacking.

kvasir wrote: The variation just looks dubious, not just because it is good for black instead of white but because it seems to require black and white to cooperate to arrive at the final position. Blacks connection is not the usual shape move (but maybe it is great here) and white is under no obligation to hane.

True enough, few of the moves in that variation hold up to LZ's scrutiny. We can only guess if this idea of a forced sequence really came from Ishida's recollection of his own thought process during the game, careful post mortem analysis, or an idea by James Davies signed off by Ishida. I think it might be genuine and Ishida didn't scrutinize the idea because the opponent's dodge seemed to indicate he saw the same sequence. Dodging the opponent's intentions has a psychological effect often greater than the real effect on the board. It's easy to get lured by fighting spirit into a common truth which turns out to be a fantasy.

On your instigation that "going from 69% to 80% has a different meaning than flipping the favorite" if that's what you meant, I differ. A 11% change in winning probability surely means someone made a mistake, regardless of the starting percentage?

kvasir wrote:You said that the win percentage goes from 69% to over 80% but that seems to me to be from winning to winning.

It goes from somewhat likely to win to more likely to win — given certain conditions. What those conditions are can be stated, but it is not all that clear what they actually are. The winning percentages are estimates of how likely one player woud win from the given position if LZ played itself. There are three sources of uncertainty in those estimates. 1) how many mistakes LZ is likely to make, and what kind; 2) how well does LZ estimate those winrates; 3) how much search (playouts) does LZ do to make the estimates.

The final position in that variation is worthwhile to study on its own for sure. The variation just looks dubious, not just because it is good for black instead of white but because it seems to require black and white to cooperate to arrive at the final position.

Are you talking about a variation in the book or an LZ variation? In both cases, unless otherwise stated, each player is making the best play he can to win the game. Competition is assumed, not cooperation.

Starting from a position that is 69% for black probably also has an effect on who this fight favors. I assume this was for the original position before the peep. The side that was better at the outset maybe doesn't have to achieve much when attacking.

The player with the higher winrate estimate usually has more plays to win the game. For instance, if he is ahead by 30 pts. on the board (which should give him a winrate estimate over 95%) and makes a play that loses 15 pts., he is still a favorite to win, but his winrate estimate will normally drop. He will be more likely to lose the game than if he had not made that mistake. That is not the same as saying that a fight favors him.

Bill Spight wrote: Are you talking about a variation in the book or an LZ variation? In both cases, unless otherwise stated, each player is making the best play he can to win the game. Competition is assumed, not cooperation

Click Here To Show Diagram Code

[go]$$W The idea from the book
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . X O O O . |
$$ | . . O O X . . . X . . . . . . X O . O |
$$ | . . . X X . . . . , . . . X . X X O . |
$$ | . O . . . . . . . . . . . . . . . X . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . b . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . , . . . . . , O . . |
$$ | . . O 9 7 . . . . . . . . . . . . . . |
$$ | . . . . 6 5 . . . . . . . . . . . . . |
$$ | . . . 8 3 4 . . . . . . . . . . O . . |
$$ | . . a 1 2 . . . . . . . . . . . . . . |
$$ | . . X . . X . . . . . . . . . . . . . |
$$ | . . . X X O O . . , . . . . . , X . . |
$$ | . X X O O . . O . . . . . . X . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

I'm talking about the "idea" variation from the book, Knotwilg had this variation in his LZ analysis. Black plays b next and black is much better according to LZ.

The variation doesn't look like something that could happen. That is what I mean by cooperating. It is not clear why both players would play this way, in fact black did not play this. Why does black not respond with the shape move at a? Why does white play the hane ? It is really hard to play yourself because you don't have an opponent.

Bill Spight wrote: The player with the higher winrate estimate usually has more plays to win the game. For instance, if he is ahead by 30 pts. on the board (which should give him a winrate estimate over 95%) and makes a play that loses 15 pts., he is still a favorite to win, but his winrate estimate will normally drop. He will be more likely to lose the game than if he had not made that mistake. That is not the same as saying that a fight favors him.

The second point was that if black was already much better to begin then this looks really good for black but if that was not the case it is maybe just an equal fight. The fight seems to start out in a way that is good for black, first taking some territory and making his other group strong, and now white has to take advantage. Basically, I mean that there doesn't seem to be anything wrong with this fight but it appears that LZ thinks black was already better. If black was already better then white has a bigger task than black in this fight but it looks like an even fight to me which could make the game harder for white.

I don't mean better in terms of points. The strength of groups, potential to expand and erase territory look important in the initial position. That doesn't really seem to change.

Knotwilg wrote: On your instigation that "going from 69% to 80% has a different meaning than flipping the favorite" if that's what you meant, I differ. A 11% change in winning probability surely means someone made a mistake, regardless of the starting percentage?

I meant that 69% is already a lot. I had in mind that if one thinks that it was an even position to begin with it might be a surprise that black is favored in the end. Going to 80% doesn't mean much to me because it doesn't look forced for the most part and the author of the line wasn't thinking of the percentage. I mean anything can happen if someone makes moves for both sides.

kvasir wrote:

Bill Spight wrote: Are you talking about a variation in the book or an LZ variation? The variation just looks dubious, not just because it is good for black instead of white but because it seems to require black and white to cooperate to arrive at the final position.

You quote me as saying something I did not. Please be careful.

$$W The idea from the book
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . X O O O . |
$$ | . . O O X . . . X . . . . . . X O . O |
$$ | . . . X X . . . . , . . . X . X X O . |
$$ | . O . . . . . . . . . . . . . . . X . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . b . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . , . . . . . , O . . |
$$ | . . O 9 7 . . . . . . . . . . . . . . |
$$ | . . . . 6 5 . . . . . . . . . . . . . |
$$ | . . . 8 3 4 . . . . . . . . . . O . . |
$$ | . . a 1 2 . . . . . . . . . . . . . . |
$$ | . . X . . X . . . . . . . . . . . . . |
$$ | . . . X X O O . . , . . . . . , X . . |
$$ | . X X O O . . O . . . . . . X . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------

Click Here To Show Diagram Code

[go]$$W The idea from the book
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . X O O O . |
$$ | . . O O X . . . X . . . . . . X O . O |
$$ | . . . X X . . . . , . . . X . X X O . |
$$ | . O . . . . . . . . . . . . . . . X . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . b . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . , . . . . . , O . . |
$$ | . . O 9 7 . . . . . . . . . . . . . . |
$$ | . . . . 6 5 . . . . . . . . . . . . . |
$$ | . . . 8 3 4 . . . . . . . . . . O . . |
$$ | . . a 1 2 . . . . . . . . . . . . . . |
$$ | . . X . . X . . . . . . . . . . . . . |
$$ | . . . X X O O . . , . . . . . , X . . |
$$ | . X X O O . . O . . . . . . X . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

I'm talking about the "idea" variation from the book, Knotwilg had this variation in his LZ analysis. Black plays b next and black is much better according to LZ.

The variation doesn't look like something that could happen. That is what I mean by cooperating. It is not clear why both players would play this way, in fact black did not play this. Why does black not respond with the shape move at a? Why does white play the hane ? It is really hard to play yourself because you don't have an opponent.

I do not have the book, so I do not know how the book portrays this variation. It looks like a variation where Black "fell for it". That is, White wanted to play at right away, but did not like the result. is an attempt to get a better result. The above line certainly seems like a better result, since White gets not only but and , at the cost of only two stones. (To me, looks quite bad.) The kicker, perhaps, is that LZ says that Black can follow up with at b, and it is White who fell for it.

Bill Spight wrote: You quote me as saying something I did not. Please be careful.

I'm sorry. Have changed it in an edit now.

Bill Spight wrote: I do not have the book, so I do not know how the book portrays this variation. It looks like a variation where Black "fell for it". That is, White wanted to play at right away, but did not like the result. is an attempt to get a better result. The above line certainly seems like a better result, since White gets not only but and , at the cost of only two stones. (To me, looks quite bad.) The kicker, perhaps, is that LZ says that Black can follow up with at b, and it is White who fell for it.

The book has this variation as something the author wanted to lure his opponent into but the opponent dodged it by turning at the top. No other variations are discussed in the book. As I speculated above, perhaps both were indeed thinking along the same lines and there was some psychological warfare going on. Even at my level, I can sometimes see that my opponent sees what I'm seeing" Even at Ishida's level, they can trick each other into both seeing the wrong thing: LZ would have responded to the "trick play" with a keima but not because of this variation. If you force feed the variation to LZ, she thinks Black's chances have gone up (I know you "saw" that already )