Life In 19x19

After watching the London New Year Tournament board 1 broadcasts in YouTube, I still wonder how these endgame discussions and the real games are so far apart.

Cheers,
Vesa

Wow, that got more interesting than I expected!

Next up, part 1 problem 53. I think this is easier, but I'm posting it because of a cute variation. And of course I could be wrong, there might be hidden depth here too.

Click Here To Show Diagram Code

[go]$$B Black to play
$$-------------------------+
$$ . . . . . . O . . . . . |
$$ . . . . . O X O O O . . |
$$ . . O O . . X X X X O O |
$$ . . . X O O X . . . X . |
$$ . . . X X X . . . X . . |
$$ . . . . . . . . . . . . |[/go]

Book solution: Question 1: how can white make ko?
Question 2: instead of in the solution diagram, what move might black try to avoid ko, and how does it fail?

Ko:
Obviously many moves other than fail, but here is an unsuccessful attempt:

Yes, that's the I was thinking of. For a moment I thought I'd improved on the book solution, but alas, no.

White can also resist this way, perhaps not the best reply but it amuses me that the throw-in no longer works:

Next, problems 49 and 50 of part 1 are a nice pair for practising your counting.

Click Here To Show Diagram Code

[go]$$Bc Black to play: 'a' or 'b'?
$$---------------------+
$$ . . . . . a b . . . |
$$ . . . . . . . . X . |
$$ . . O . O . O O X . |
$$ . . . . . O . X . . |
$$ . . . . . . . . X . |
$$ . . . . . . X . . . |
$$ . . . . . . . . . . |[/go]

Click Here To Show Diagram Code

[go]$$Bc Black to play: 'a' or 'b'?
$$---------------------+
$$ . . . . . a b . . . |
$$ . . . . . . . . X . |
$$ . O . O O O O O X . |
$$ . . . . X X . X . . |
$$ . . . . . . . . X . |
$$ . . . . . . . X . . |
$$ . . . . . . . . . . |[/go]

In both cases:

• What's the best play for both sides after black a?
• What's the best play for both sides after black b?
• Which of a and b is best and by how many points?

Remember that the monkey jump is supposed to be sente. If you're looking at a gote sequence, think again about the choices for .
The book tells us which is best but doesn't do the counting. And the book also has this cute story about the name "monkey jump".

It is often complex to verify whether a monkey jump is sente, ambiguous or gote, and so to verify whether calculated move values and counts are correct. In these examples, assumptions about the left outside need to be made.

Probably for the problem to make sense, we need to assume that both moves a and b are sente. However I don't see the difference between problems 50 and 51 so I must be misunderstanding something.

jlt wrote:Probably for the problem to make sense, we need to assume that both moves a and b are sente. However I don't see the difference between problems 50 and 51 so I must be misunderstanding something.

The difference is the presence or absence of a white stone at P17. I was surprised to find that makes a difference to some of the follow-up moves, which is why I find it an interesting pair of problems.

No, you don't assume something is sente, you figure out which continuation makes it into a sente sequence!

RobertJasiek wrote:It is often complex to verify whether a monkey jump is sente, ambiguous or gote, and so to verify whether calculated move values and counts are correct. In these examples, assumptions about the left outside need to be made.

Thanks Robert, this hadn't occurred to me. Now I'm thinking of things like this:

Click Here To Show Diagram Code

[go]$$Bc What's on the left?
$$---------------------+
$$ . . . . 3 1 4 . . . |
$$ . a . . . 2 . . X . |
$$ . . O . O . O O X . |
$$ . . . . . O . X . . |
$$ . . . . . . . . X . |
$$ . . . . . . X . . . |
$$ . . . . . . . . . . |[/go]

If there are black stones near a, then perhaps white shouldn't play this . Even if there are black stones further away to the left, the potential for black to play around a may affect the local endgame moves. For this problem, we'll assume that black doesn't have any stones that are near enough and strong enough to cause trouble, so that in this diagram, and are captured with no bad aji left behind.

Is that the sort of thing you had in mind?

It is very much more demanding than that. For each interesting first move, study whether it in its alternating sequence is sente, ambiguous or gote and how long it is worth playing successively. Then compare different first moves to determine which should usually be played in the local endgame. Each of these aspects must be assessed by value conditions.

To assess values, we must make assumptions on the left side (Is there white territory? Where does White's territory end? Can Black escape?) and the life of the white group (Which attacking moves must be answered because sacrifice of a tail is considered too expensive?). The problems would be somewhat simpler if at least assumptions would not have to be made in simplified positions, such as:

Click Here To Show Diagram Code

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

There may exist whole board situations where a and b are not sente, for instance if there is a sufficiently big point elsewhere on the board, White doesn't need to respond to or . That's why I think we should assume that is sente for the problem to make sense.

Click Here To Show Diagram Code

[go]$$Bc
$$+------------------------------+
$$|. O O . . . . . . . . . 3 . 1 . . . . |
$$|O . O . . . . . . . . . . . . . . X . |
$$|O O O O O O O O O . O . O O O O O X . |
$$|. . . . . . . . . . . . . X X . X . . |
$$|. . . . . . . . . . . . . . . . . X . |
$$|. . . . . . . . . . . . . . . . X . . |
$$|. . . . . . . . . . . . . . . . . . . |[/go]

Now, even with this assumption, I am missing something.

jlt wrote:There may exist whole board situations where a and b are not sente, for instance if there is a sufficiently big point elsewhere on the board, White doesn't need to respond to or . That's why I think we should assume that is sente for the problem to make sense.

OK, I see the point. Yes, we assume that black won't play until such time as is actually a good move. Most of the time that means it will be sente. The book Endgame 2 - Values discusses this sort of thing in more detail using the terms "local sente" and "global sente".

jlt wrote:Now, even with this assumption, I am missing something.

I agree with your analysis of the second position. (Again, no guarantee that my counting is correct, but at least it's reassuring that we both get the same answer!) Where the first one differs is: For what it's worth, the book's assessment of relative values is

Here is a real-game endgame problem with relevance to the current discussion from a recent game of mine (white).

Click Here To Show Diagram Code

[go]$$Wc
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . a . . . . . . |
$$ | . X O O . O . . . . O O . X . X . . . |
$$ | . . X O . . . . O O X X X . X , X X . |
$$ | . . O . . . . . O X X . . X . X O . . |
$$ | . . . . . . . . . . . O O . O X O . . |
$$ | . . . . . . . . . . . . . . . O X . . |
$$ | . X . . . . O . O O . O . . . O X X . |
$$ | . O O O O . . . . X . . . . . O X O . |
$$ | . O X X X X . X . , . . . . . O O X . |
$$ | . O O X . . . . . X . X . . . O X . X |
$$ | . X X X O O . . . . . . . . . O X X . |
$$ | . X O O X . . . O 1 O . X . X O X . . |
$$ | . X X O O O O . . X . . . X O . O . . |
$$ | . O X O . O X O . . . . X O O O O . . |
$$ | . 2 O O O X X X X , . . . X X O . . . |
$$ | X . X X O O X . . . . . . . . X O . . |
$$ | . X . . X X . X . . . . . . X . X . . |
$$ | . . . . . . . . . . . . . . . X . . . |
$$ ---------------------------------------[/go]

n18 is a classic "double sente" endgame kosumi on the 2nd line, but I decided it was not sente enough for me to play now as I feared a black tenuki to close the lower side at l6. If I followed up with a monkey jump, I thought another black tenuki was plausible kiai as to answer would only preserve another half-dozen or so points in gote and black's group was already super strong. So I decided that it was better for me to poke into the lower side in what I thought was sente for some benefit there and then play n18. I wasn't sure what the best spot was, but went for L5. My opponent resisted my plan and took n18 for himself, but then I broke into the lower side which was pretty big too but gote, so he got the monkey jump in sente on the top. I didn't make any precise value calculations, I wonder what they would indicate as best play. n10 is also a very big move for white with reverse sente characteristics as black moves there have some attacking potential on the white group so maybe white taking n18, allowing black l6 and then n10 is ok to get tedomari (but is s3 much smaller?).

Thanks Uberdude, that's a really interesting example, and points to a problem that I have when trying to apply endgame theory.

I'm confident that, given enough time and willpower, I could count the value of a move at N18. But counting the move value for L5 is much harder for me, not because of the counting itself, but because I'd have a hard time actually trying to decide whether L5 is indeed the best local move for white, or whether I want to start at M6 or M5 instead, or whether the cute-looking K5 is playable, let alone black's correct responses to each of those moves, whether the sequence finishes in sente or gote, and the right move if black gets to go first in that area...

So counting works OK (if I've got enough time left on the clock) once all the groups are settled and there are no "vague-looking" areas, but in a position like your diagram I'm much more in "middlegame mode" rather than "endgame mode", even though I know the endgame is supposed to start a bit earlier.

(Trying to do the calculations in my head -- and falling back on traditional endgame theory, because there's no such thing as sente in modern endgame theory -- I get a swing value of 6 points for N18. Slightly smaller than I would have guessed because N17 turns into a sente follow-up for whoever goes second, which cancels out some of the gain. I look forward to being told where I went wrong.)

In Uberdude's game I count White ahead by a couple of points (I don't know how many prisoners there are). In such a situation I will go for A, the "double sente" at the top, because it has a higher degree of certainty, i.e. it's harder to see how Black can get more points from developing the bottom while allowing White to monkey. If I had counted White behind, I would go for L5, which is the harder value to calculate and brings uncertainty into the game.

Also, after A, White doesn't need to follow-up with a monkey, but can choose to play the next big gote N10 or S3. Here I would probably go for S3 again because I find it easier to gauge its size and be confident it is big enough.

Uberdude's decision to L5 instead, indeed shows fighting spirit, or the lucidity to reflect on Black's kiai but in fact it entices Black to show such kiai and take A for himself, taking the opportunity for turmoil.

These musings have nothing to do with precise value computations, everything with maximizing the chance of winning and/or minimizing the effort to compute the result.

Knotwilg wrote:In Uberdude's game I count White ahead by a couple of points (I don't know how many prisoners there are). In such a situation I will go for A, the "double sente" at the top, because it has a higher degree of certainty, i.e. it's harder to see how Black can get more points from developing the bottom while allowing White to monkey. If I had counted White behind, I would go for L5, which is the harder value to calculate and brings uncertainty into the game.

Black has 5 captures to white's 1, and it's 6.5 komi. I thought I was a little ahead, but not enough to be slack.