It is currently Mon Mar 30, 2020 3:55 pm

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 6 posts ] 
Author Message
Offline
 Post subject: Life and death problem: ko versus unconditional life
Post #1 Posted: Thu Nov 28, 2019 2:36 am 
Lives in gote

Posts: 410
Location: Adelaide, South Australia
Liked others: 162
Was liked: 201
Rank: Australian 2 dan
GD Posts: 200
Lee Changho life and death book 3 problem 53.

Click Here To Show Diagram Code
[go]$$Bc Black to play
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . O . . . .
$$ | . O O O O . O . . .
$$ | . O X X X X O . . .
$$ | X X . X . O O . . .
$$ | O O . X X O . . . .
$$ | . . . . . . . . . .
$$ +--------------------[/go]


What's interesting is not just black's first move, but white's best response.

Spoiler behind the cut: I'll give you black's first move and two options for white.

Click Here To Show Diagram Code
[go]$$Bc Black to play
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . O . . . .
$$ | . O O O O . O . . .
$$ | . O X X X X O . . .
$$ | X X . X . O O . . .
$$ | O O a X X O . . . .
$$ | . . b . 1 . . . . .
$$ +--------------------[/go]

Should white play 'a' or 'b' next? Why?


This post by xela was liked by: Bill Spight
Top
 Profile  
 
Offline
 Post subject: Re: Life and death problem: ko versus unconditional life
Post #2 Posted: Sat Nov 30, 2019 9:36 pm 
Lives in gote

Posts: 410
Location: Adelaide, South Australia
Liked others: 162
Was liked: 201
Rank: Australian 2 dan
GD Posts: 200
Book solution
Click Here To Show Diagram Code
[go]$$Bc Black to play
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . O . . . .
$$ | . O O O O 8 O . . .
$$ | 4 O X X X X O . . .
$$ | X X 5 X . O O . . .
$$ | O O 3 X X O . . . .
$$ | 9 6 2 7 1 . . . . .
$$ +--------------------[/go]

and there's a second diagram showing how white can play under the stones, but black has enough outside liberties to squeeze, so that black lives unconditionally (I'll let you work out the details).

Variation diagram (also from the book)
Click Here To Show Diagram Code
[go]$$Bc Black to play
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . O . . . .
$$ | . O O O O . O . . .
$$ | . O X X X X O . . .
$$ | X X 3 X . O O . . .
$$ | O O 2 X X O . . . .
$$ | . 4 5 6 1 . . . . .
$$ +--------------------[/go]

White can make a ko this way.

Discussion: Why is the first diagram given as the solution, and not the second?
Why would you let black live unconditionally, when you can force a ko?

The answer is that it's an approach-move ko (aka "yose ko"): white would have to ignore four black ko threats to capture the group, which is unlikely to be profitable. But why not start the ko anyway, what's to lose?
Click Here To Show Diagram Code
[go]$$Bc Black to play
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . O . . . .
$$ | . O O O O . O . . .
$$ | a O X X X X O . . .
$$ | X X . X . O O . . .
$$ | O O . X X O . . . .
$$ | . . . . 1 b . . . .
$$ +--------------------[/go]

In the solution diagram, both 'a' and 'b' end up being sente for white. But if white starts and loses the ko, then black can push out at both those points in sente. So the solution diagram is two points better for white than the variation, unless white has a truly massive supply of ko threats.

This is a nice illustration of the saying A three-move approach ko is no ko!

Top
 Profile  
 
Offline
 Post subject: Re: Life and death problem: ko versus unconditional life
Post #3 Posted: Sun Dec 01, 2019 1:39 am 
Honinbo

Posts: 9517
Liked others: 3003
Was liked: 3166
Very nice, xela. :)

I set this up as a difference game, which might not have been a great idea, but the file does get the point across.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
----
Many are cold, but few are frozen.


Last edited by Bill Spight on Sun Dec 01, 2019 6:16 am, edited 1 time in total.
Top
 Profile  
 
Offline
 Post subject: Re: Life and death problem: ko versus unconditional life
Post #4 Posted: Sun Dec 01, 2019 3:46 am 
Lives in gote

Posts: 410
Location: Adelaide, South Australia
Liked others: 162
Was liked: 201
Rank: Australian 2 dan
GD Posts: 200
I've heard the phrase "mast value" before, but not paid a great deal of attention. The page at https://senseis.xmp.net/?Mast needs a bit more context to make sense. Is there something written down to link this to kos?

Top
 Profile  
 
Offline
 Post subject: Re: Life and death problem: ko versus unconditional life
Post #5 Posted: Sun Dec 01, 2019 6:05 am 
Honinbo

Posts: 9517
Liked others: 3003
Was liked: 3166
xela wrote:
I've heard the phrase "mast value" before, but not paid a great deal of attention. The page at https://senseis.xmp.net/?Mast needs a bit more context to make sense. Is there something written down to link this to kos?


Mast value is a term coined by Berlekamp in "The economist's view of games" in Games of No Chance, edited by Richard Nowakowski (Cambridge University Press, 1996), precisely to deal with hyperactive ko positions, such as approach kos, and 10,000 year kos. Up to this point thermographs had masts which represented their mean value when they were not big enough to play. But it is impossible (or nearly so) to derive mean values for hyperactive ko positions. I go into this in "Evaluating kos: A review of the research" in the proceedings of ICOB 2006 (Myongji University).

So what is mean value? Suppose that we have two simple gote such that Black to play can move to a local score of 6 pts. and White to play can move to a local score of 0. The combination has a value of 6 pts. (for Black). They are miai. If Black plays first in one of them, White can play first in the other, to guarantee a result of only 6 pts. If White plays first in one of them, Black can reply in the other, to guarantee a result of at least 6 pts. Those two guarantees establish a game theoretical value for the pair of 6 pts. The mean value for each is thus 6/2 = 3 pts. The value of the mast of the thermograph of each is the same as the mean value.

Now consider a simple ko which Black may win for a local score of 6 and White may win for a local score of 0. There are two such kos, depending upon who can win the ko in one move. Let's call the one that Black can win in one play, K, and the one that White can win in one play, L. Consider the combination, K + L. It is possible to show that Black can guarantee a local score of at least 6 and White can guarantee a local score of at most 6, assuming that play ends. So we may take the value of the combination as 6. OC, there is no mean value here, because they are two different kos. It is easy to see that ko threats do not matter in this case, although they may prolong the play.

Now suppose that we have 3 of K on the board. Black to play can win one, for a local score of 6, and then White can capture in one of the other two, for a local score of 6 in the pair. The result is a local score of 12. By the same token, White to play can capture in one of them and then Black can win one of the others, for a local score of 12. The game theoretical value of 3 Ks is 12, and the mean value of each is 12/3 = 4. Again, ko threats do not matter.

You can't do that for hyperactive ko positions, because for them ko threats do matter. In the early 1990s Berlekamp discovered a way to take ko threats into account, at least abstractly, with the idea of komaster. We can draw thermographs of such positions, but the value at the mast does not equal the mean value, because there is no mean value. So we call it simply the mast value. :)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
----
Many are cold, but few are frozen.


This post by Bill Spight was liked by: xela
Top
 Profile  
 
Offline
 Post subject: Re: Life and death problem: ko versus unconditional life
Post #6 Posted: Sun Dec 01, 2019 3:41 pm 
Lives in gote

Posts: 410
Location: Adelaide, South Australia
Liked others: 162
Was liked: 201
Rank: Australian 2 dan
GD Posts: 200
OK, so every time I see "mast value", I can just say "mean value" to myself and it will be close enough approximation :-)

Seriously, thanks for the very clear explanation. I really do mean to learn CGT properly some time. I did get half way through ONAG and found it very enjoyable. But I want to improve my practical go skills a bit more first. (Doing a fair bit of tsumego at the moment, in case you hadn't noticed, and also would like to finish Genjo-Chitoku before starting something else big).

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 6 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group