Unsure about this tsumego problem

ditashi · #1

Hello all! I guess this is my first actual question on L19 so go easy on me

I've been using this Android app for tsumego (Tsumego Pro if you're interested) and they presented a problem.
Unfortunately I can't seem to embed it here here's a link to EidoGo

This was Black to play and these were the suggested moves, after the last move it was considered solved (for Black). As I see it if White plays Q19 next he's actually won this since I can't see a variation where Black survives with two eyes or takes the group. Could anyone shed some light on this? Am I missing something or did they just made a mistake (they do claim some problems might contain errors)?

If anyone has any comments about the problem itself, what't the motivation behind Black' moves that would also be great since it took me a bit of time to "solve" this.
Thank you!

Uberdude · #2

There is something strange with move 7 in the sgf: both white and black play a move, but assuming you just meant to have black play t16 to take a liberty (a good move) then white q19 is a good move, but black also has a good reply at o19 which puts white into atari. White can't connect or he dies, so has to take the 1 black stone which is a ko. Black can make a threat and then retake. So white's group is "ko to kill": if white wins the ko he lives (but has to ignore a ko threat so black gains elsewhere), if he loses the ko he dies (but gains elsewhere as black ignored white's threat to kill off the group).

oca · #3

ditashi wrote:

Unfortunately I can't seem to embed it here here's a link to EidoGo

Hi ditashi, and Welcome !

You can use [sgf]tags to use EidoGo on the forum. you just need to put the text content of the sgf file in between.

Code:

[sgf]
(;GM[1]FF[4]CA[UTF-8]SZ[19]AB[rd][qd][pd][oc][nc][nb]AW[rc][qc][pc][ob][oa];B[qa];W[ra];B[rb];W[qb];B[sa];W[sc];B[se]AW[sd])
[/sgf]

oca · #4

That should show something like that : (but that don't work if sgf tag is in the same post then the code tag... so I did two post...)

oca · #5

Uberdude wrote:

There is something strange with move 7 in the sgf: both white and black play a move, but assuming you just meant to have black play t16 to take a liberty (a good move) then white q19 is a good move, but black also has a good reply at o19 which puts white into atari. White can't connect or he dies, so has to take the 1 black stone which is a ko. Black can make a threat and then retake. So white's group is "ko to kill": if white wins the ko he lives (but has to ignore a ko threat so black gains elsewhere), if he loses the ko he dies (but gains elsewhere as black ignored white's threat to kill off the group).

Just a try a to represent that ko fight... maybe the value of the ko is not representative... but that should at least show the idea...

Uberdude · #6

@oca: mostly right but:
1. If white connects he is in atari so black should capture, not make a T 4 dead shape (why?)
2. If black ignores white's ko threat* he captures 3 stones to kill: connecting at s19 would not kill as q18 is a 2nd eye when white captures the 4.

*(we usually say threat not menace in English though menace is a word too)

oca · #7

Uberdude wrote:

@oca: mostly right but:
1. If white connects he is in atari so black should capture, not make a T 4 dead shape (why?)
2. If black ignores white's ko threat* he captures 3 stones to kill: connecting at s19 would not kill as q18 is a 2nd eye when white captures the 4.

*(we say threat not menace in English though menace is a word too)

Here is my second try

Bill Spight · #8

ditashi · #9

I apologize for the invalid SGF, I'll fix it up on my next posts

So regarding the problem, from what I gather, that solution is considered a "win" only if Black has a valid ko threat to make on the next turn, am I right? Otherwise it's just an advantage for white on the next turn.

The original problem presented only this part of the board so I didn't assume a ko threat was possible.
In the case no Ko threat exists isn't Bill Spight's second suggested variation (Black at O19) the best option for black? In any case he would lose this and White lives but at least he captures some thus making White's territory smaller?

Bill Spight · **#10**

A ko threat on an otherwise empty board is one thing, but normally this ko will be fought in the early endgame, when the best plays elsewhere gain about the same as the ko plays, and there are many plays of various sizes still to play. Then ko threats will also threaten to gain about the same as ko plays, or more.

This file shows a possible ko threat in the normal situation.

For extra credit, calculate how much each play in the ko or the threat gains.

Uberdude · **#11**

ditashi wrote:

So regarding the problem, from what I gather, that solution is considered a "win" only if Black has a valid ko threat to make on the next turn, am I right? Otherwise it's just an advantage for white on the next turn.

No; as far as solving life and death problems is concerned "this sequence produces a ko to kill" is a valid answer and "win" for black. We don't worry about what ko threats exists on the rest of the board when solving the local problem. It's best if you can kill without needing to fight a ko ("unconditional kill") but "ko to kill" is 2nd best and for this problem is the best that black can do. Sometimes the question will clearly say "find a ko to kill", sometimes just "black to play" and then if you find a ko you should keep reading to see if you can improve on that and find an unconditional kill.

Bill Spight · **#12**

ditashi wrote:

So regarding the problem, from what I gather, that solution is considered a "win" only if Black has a valid ko threat to make on the next turn, am I right? Otherwise it's just an advantage for white on the next turn.

No, making the ko is right, whether there are ko threats or not. Black still gets an extra play for making and losing the ko.

(OC, when to make the ko depends on the rest of the board, but that is not the point of the problem.)

Quote:

In the case no Ko threat exists isn't Bill Spight's second suggested variation (Black at O19) the best option for black? In any case he would lose this and White lives but at least he captures some thus making White's territory smaller?

It would be possible to construct a whole board position where that would be the correct play, but normally confining White to the corner and getting the extra play by making the ko are worth more than letting White live without ko and then taking gote to capture a few stones -- or even leaving those stones for later and playing somewhere else.

drmwc · **#13**

There are actually two ways to make ko here. Here's the second way:

This way, white has to make the first threat. This is a gain for black.

However, if Black wins the ko, white still has a lot of liberties so the stones still have aji. Also, if white wins the ko in this variation, she has better endgame than in the other variation. This is a loss for black.

So, in a game, the choice of which of the two ways to play ko is extremely difficult. I suspect players who routinely make the correct decision are close to professional strength (or in other words it's beyond me!).

ez4u · **#14**

Cho Chikun's Life and Death Dictionary gives four different 'right' answers (all ko) in this position:

Click Here To Show Diagram Code: [go]$$Bc $$- - - - - - - - + $$. . . O 7 4 3 . | $$. . X O 6 5 . 2 | $$. . X X O O O 1 | $$. . . . X X X . | $$. . . . . . . . | $$. . . . . . X . | $$[/go]

Click Here To Show Diagram Code: [go]$$Bc $$- - - - - - - - + $$. . . O . 5 6 7 | $$. . X O 4 3 1 2 | $$. . X X O O O . | $$. . . . X X X . | $$. . . . . . . . | $$. . . . . . X . | $$[/go]

Click Here To Show Diagram Code: [go]$$Bc $$- - - - - - - - + $$. . . O 7 6 5 . | $$. . X O 4 3 1 2 | $$. . X X O O O . | $$. . . . X X X . | $$. . . . . . . . | $$. . . . . . X . | $$[/go]

Click Here To Show Diagram Code: [go]$$Bc $$- - - - - - - - + $$. . . O . 1 2 5 | $$. . X O . . 3 4 | $$. . X X O O O . | $$. . . . X X X . | $$. . . . . . . . | $$. . . . . . X . | $$[/go]

The book notes that if White has an extra liberty, the hane at T17 and the play at S18 still lead to ko. However, the placement at R19 allows White to live.

Click Here To Show Diagram Code: [go]$$Bc $$- - - - - - - - + $$. . 7 O . 1 2 5 | $$. . X O . 4 3 8 | $$. . X X O O O 6 | $$. . . X . X X . | $$. . . . X. . . . | $$. . . . . . X . | $$[/go]

In actual practice the R19 placement is the most common by far (100+ examples of in my database).

Unsure about this tsumego problem

Who is online