Life In 19x19

Bill Spight wrote:As my rules indicate, it is possible for play to continue forever if passes lift ko bans and 2 passes do not end play.

Well, it is true only for a few special situations, for example, when there exists a double ko seki on the board. I agree that the Japanese rule, when applying it literally (and assuming a pass lifts ko ban), cannot finish the game. But it does not automatically mean they really meant that a pass cannot lift a ko ban.

Bill Spight wrote:Yes, but they handle such situations by hypothetical play instead of resumption. At the very least you can prove that the unfilled ko in many, if not all of such situations, leaves a dead ko stone.

As you left a reservation, there are situations that not lifting the ko ban really makes things problematic and hypothetical play cannot save all of them.

Here, I brought a figure from my book: Japanese (or Korean) rule, assume a pass does not lift a ko ban, Black's turn, komi W+0.5, what is the best play for each and who wins? My original plan was to explain it, but I just realized that it will be fun to wait for you guys to find the answer.

Gérard TAILLE wrote:I see another issue with the handling of ko in actual play or in hypothetical play.

Komi O,5 point, White to play

$$B
$$ ----------------------------------------
$$ | X X X X X X X X X . O O O O O O O . O|
$$ | X X X X X X X X X O O O O O O O O O .|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | X X X X X X X X X O O O O O O O O O O|
$$ | O O O O O O O O O O O O O O O O O O O|
$$ | X X X X X X X X X X X X X X X X X X X|
$$ | X X X X X X X X X X X X X X X X X X X|
$$ | X X X X X X X X X X X X X X X X X X .|
$$ | X X X X X X X X X X X X X X X X X . X|
$$ | X X X X X X O O O O O . X X X X X X X|
$$ | O O O O O O O O O O O O O O O O O O O|
$$ | O O O O O O O O O O O O O O O O O O O|
$$ | O O O O O O O O O O O O O O O O O O O|
$$ | O O O O O O O O O O O O O O O O O O O|
$$ ----------------------------------------

Click Here To Show Diagram Code

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

3) The players agreed that it is far to complex to decide whether or not the black group is dead under hypothetical play. What happens? I guess the resumption of the game will be decided but both players will obviously pass. What is the result? Here again I guess black will win

Can we conclude that the 1989 rule changes the result of this game ?

I guess this is also a well-known problem. The hidden question here is, who operates the hypothetical play? Do the players have the right to present the best sequence one can imagine? What if the opponent does not agree?

I believe, even though it is not written anywhere, they had "perfect play" or "God's play" in their minds when writing the rule, assuming that a pro level player will not have a trouble finding such a sequence for the problematic localized area. For this situation, no one can confirm such a sequence, and the only fair decision is to let them "play" the hypothetical play to decide whether Black can make a life on the corner.

Anyway, I do not think that they really cared this type of flaw in their rule-writing. (They will simply say "it won't happen, so don't worry.")

One interesting story is this. The Korean rule is more ambiguous in that it is difficult to see whether the hypothetical play is really inside the rule. I at least interprete it that way, and a few Korean pros I asked agreed. However, when I presented this type of problem to them, they were very reluctant to admit that the Black stones at the corner may get a life. (Yeah, I know. They simply think, well... but... whatever the rule says... its DEAD!!!) So, I think KBA will call it dead if it really occurs. Does Nihon Ki-in have the gut to admit that the corner is alive by seki? I doubt it.

jaeup wrote:Here, I brought a figure from my book: Japanese (or Korean) rule, assume a pass does not lift a ko ban, Black's turn, komi W+0.5, what is the best play for each and who wins? My original plan was to explain it, but I just realized that it will be fun to wait for you guys to find the answer.

jpncom49.png

Nice position.

Here is an SGF. I started with no passes as the default and worked from there.

Edit: Silly me.

Click Here To Show Diagram Code

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

jaeup wrote:
I believe, even though it is not written anywhere, they had "perfect play" or "God's play" in their minds when writing the rule, assuming that a pro level player will not have a trouble finding such a sequence for the problematic localized area. For this situation, no one can confirm such a sequence, and the only fair decision is to let them "play" the hypothetical play to decide whether Black can make a life on the corner.

OK jaeup I agree with your view.
Now assume that the God's play tell us that, due to the special ko rule under hypothetical play, the black group in the upper left corner is alive.
In that case, in one hand you can say the black group is dead under normal God's play and in the other hand you can say the black group is alive under hypothetical God's play.

In that case my understanding of the 1989 japonese rule is that black wins the game though it hurts the common sense of go game doesn't it?

Bill Spight wrote: Nice position.

Here is an SGF. I started with no passes as the default and worked from there.

Edit: Silly me.

I only checked the edited version, so I cannot see what was your initial mistake. Anyway, freezing the game is the best strategy for both, but this is a true disaster for the Japanese rulemakers. They never wanted the game to end this way, and expected the hypothetical play to prevent it.

Here is one minor objection: after two passes, is D7 a dead stone? In the hypothetical play, Black will play at F5 and claim that it is a newly created stone. Anyway, it doesn't really matter because all stones will stay on the board regardless of the fate of D7.

Gérard TAILLE wrote: In that case, in one hand you can say the black group is dead under normal God's play and in the other hand you can say the black group is alive under hypothetical God's play.

Yes, it is a true nightmare. The rule asks White to capture the Black stones because "Black can make a life (in the hypothetical play)", but after White's capture, Black cannot make a life (in the real play). Still, because of the one stone addition, White loses the game.

Gérard TAILLE wrote:

Matti wrote:
Yes Matti, of course I realize that I did not mentionned this point, sorry for that.

Anyway, taking your point into account, I looked for this version of the martian problem and I concluded that this new problem is far easier that the original one, though it is quite interesting. Instead of just excluding a move due to a simple ko you can also chose to exclude a move due to a superko.
With this in mind I built the following solution:

White to play

$$W
$$ ------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | . O X X X |
$$ | O X X X . |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ ------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | . O X X X |
$$ | O X X X . |
$$ -------------[/go]

starting from this position white captures all black stones and continue until position:

$$W
$$ ------------
$$ | O O O O O |
$$ | O O O O O |
$$ | O O O O O |
$$ | . O O O O |
$$ | O O O O O |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ ------------
$$ | O O O O O |
$$ | O O O O O |
$$ | O O O O O |
$$ | . O O O O |
$$ | O O O O O |
$$ -------------[/go]

here black captures all white stones and now you can build the following very simple martian position

Black to play

$$W
$$ -------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | X O X X X |
$$ | a X b X . |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ -------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | X O X X X |
$$ | a X b X . |
$$ -------------[/go]

japonese rule : black must pass to avoid giving an additional prisoner
AGA rule : black must play "a" to save its stones and reach soon a seki
chinese rule : black must play "b" to reach an immediat seki because white cannot answer "a" due to superko

Note : for the original martian problem I searched for a solution with a molasses ko but I had to give up. The two solutions I found do not use this molasses ko scheme.

I see. If one has superko restrictions to start with, they may be at arbitrary locations. PSK and SSK might have different restrictions. This could be achieved by capturing all white and/or black stones from the board multiple times. However a single ko is a different thing. If one wants to use molasses ko, a single ko restriction is necessary.

jaeup wrote:

Bill Spight wrote: Nice position.

Here is an SGF. I started with no passes as the default and worked from there.

Edit: Silly me.

I only checked the edited version, so I cannot see what was your initial mistake. Anyway, freezing the game is the best strategy for both, but this is a true disaster for the Japanese rulemakers. They never wanted the game to end this way, and expected the hypothetical play to prevent it.

Who knows what went through the minds of those who came up with the anti-seki?

Here is one minor objection: after two passes, is D7 a dead stone? In the hypothetical play, Black will play at F5 and claim that it is a newly created stone. Anyway, it doesn't really matter because all stones will stay on the board regardless of the fate of D7.

Well, the enabling clause is ambiguous in English. However, since Black could play a living stone at F5 anyway, you can argue that White did not enable that stone to be played by taking the ko. The few times that I have seen 生じうる in ancient Japanese go texts in an online collection seem to me to be consistent with that interpretation.

So yes, I think that the D7 stone is dead under hypothetical play. And, as you say, the stone is not removed.

Under Berlekamp's rules (no pass go with prisoner return) Black wins by 0.5, also under Button Go with 0 komi where taking the button lifts ko and superko bans. takes the button in the mainline, which lifts the ko ban, but costs 0.5 point.

Matti wrote:

Gérard TAILLE wrote:

Matti wrote:
Yes Matti, of course I realize that I did not mentionned this point, sorry for that.

Anyway, taking your point into account, I looked for this version of the martian problem and I concluded that this new problem is far easier that the original one, though it is quite interesting. Instead of just excluding a move due to a simple ko you can also chose to exclude a move due to a superko.
With this in mind I built the following solution:

White to play

$$W
$$ ------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | . O X X X |
$$ | O X X X . |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ ------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | . O X X X |
$$ | O X X X . |
$$ -------------[/go]

starting from this position white captures all black stones and continue until position:

$$W
$$ ------------
$$ | O O O O O |
$$ | O O O O O |
$$ | O O O O O |
$$ | . O O O O |
$$ | O O O O O |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ ------------
$$ | O O O O O |
$$ | O O O O O |
$$ | O O O O O |
$$ | . O O O O |
$$ | O O O O O |
$$ -------------[/go]

here black captures all white stones and now you can build the following very simple martian position

Black to play

$$W
$$ -------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | X O X X X |
$$ | a X b X . |
$$ -------------

Click Here To Show Diagram Code

[go]$$W
$$ -------------
$$ | . . O O O |
$$ | O O O O O |
$$ | O O X X X |
$$ | X O X X X |
$$ | a X b X . |
$$ -------------[/go]

japonese rule : black must pass to avoid giving an additional prisoner
AGA rule : black must play "a" to save its stones and reach soon a seki
chinese rule : black must play "b" to reach an immediat seki because white cannot answer "a" due to superko

Note : for the original martian problem I searched for a solution with a molasses ko but I had to give up. The two solutions I found do not use this molasses ko scheme.

I see. If one has superko restrictions to start with, they may be at arbitrary locations. PSK and SSK might have different restrictions. This could be achieved by capturing all white and/or black stones from the board multiple times. However a single ko is a different thing. If one wants to use molasses ko, a single ko restriction is necessary.

Yes Matti I see also what you mean.
Though my original intention was to look for a position starting a game, I confess I will be very interesting if you find a position (maybe simplier than mine!) which uses only a simple ko to reach this starting position. Sure I will happy to analyse such finding!
Good search!

I have now found a position where all three rule stes have distinct best starting moves, but they are able to prolong the game without gaining points by choosing other moves.

jaeup wrote:

Gérard TAILLE wrote: In that case, in one hand you can say the black group is dead under normal God's play and in the other hand you can say the black group is alive under hypothetical God's play.

Yes, it is a true nightmare. The rule asks White to capture the Black stones because "Black can make a life (in the hypothetical play)", but after White's capture, Black cannot make a life (in the real play). Still, because of the one stone addition, White loses the game.

I see why you are you talking about a white stone addition but keep in mind that after the hypothetical phase, stating the black stones are alive, white claim logically for the resumption of the game but now, God tells white that the best move is not to capture the black stones but to pass!!

Matti wrote:I have now found a position where all three rule stes have distinct best starting moves, but they are able to prolong the game without gaining points by choosing other moves.

I perfectly understand this point because it was also for me a great difficulty to avoid these "non natural" moves giving exactly the same result than the "normal" best move.

Anyway I see you have made great progress doesn't you?

Gérard TAILLE wrote:

Matti wrote:I have now found a position where all three rule stes have distinct best starting moves, but they are able to prolong the game without gaining points by choosing other moves.

I perfectly understand this point because it was also for me a great difficulty to avoid these "non natural" moves giving exactly the same result than the "normal" best move.

Anyway I see you have made great progress doesn't you?

Yes.

I think now I have a solution on 10*10 board.

Matti wrote:

Gérard TAILLE wrote:

Matti wrote:I have now found a position where all three rule stes have distinct best starting moves, but they are able to prolong the game without gaining points by choosing other moves.

I perfectly understand this point because it was also for me a great difficulty to avoid these "non natural" moves giving exactly the same result than the "normal" best move.

Anyway I see you have made great progress doesn't you?

Yes.

I think now I have a solution on 10*10 board.

Fantastic Matti.
I will be very happy to analyse it and appreciate your job!
If you prefer to let the other players search without knowing a solution, maybe we can exchange via private messages. It is up to you to decide.