Life In 19x19

Ok guys.

While I was hoping to leave this subforum and move to amateurs by now -- but considering how badly I get beaten by 18K (lowest possible) players at Tygem I probably still belong here in the beginner forum ...

Had something similar to the following happening to me while counting in a Tygem game:

Click Here To Show Diagram Code

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

I'm black.

The auto-score function says my group is dead (which it is). But I don't want to agree with the scoring since:

1. White has to play it correctly to really kill the group (kill it the right way, and then play a stone in the right spot, rinse and repeat). I want to test white to find the right moves.

2. Even if so, the kill will cost white a lot of stones (20 or so). Something I kinda considered when the group died.

And in both of those cases I win the game. If I accept the group as dead, I lose. So we end up passing, counting, disagreeing, passing, counting, disagreeing ... for many minutes.

What is the "proper" way to count this? If nobody wants to play any stones? How is this supposed to be dealt with?

What is the "proper" way to count this?

In the corner, black cannot approach, and white can atari black with a "rabbity six" dead shape anytime; thus the black group is to be considered as dead.

Now, the counting depends on the state of the outer white group, which I don't know since you showed only the corner position and not the remaining board.

- If the outer white group is alive, white has to do nothing, and all b stones in the corner count as dead.
- If the outer white group is not yet fully alive but e.g. part of a Semeai, all you have to do is to play on and take its liberties from outside. White will then be forced to kill the black corner and to spend additional moves there.

Maybe you could show the whole position, then it would become clear.

Tapani wrote:Ok guys.

While I was hoping to leave this subforum and move to amateurs by now -- but considering how badly I get beaten by 18K (lowest possible) players at Tygem I probably still belong here in the beginner forum ...

Had something similar to the following happening to me while counting in a Tygem game:

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

Click Here To Show Diagram Code

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

I'm black.

The auto-score function says my group is dead (which it is). But I don't want to agree with the scoring since:

1. White has to play it correctly to really kill the group (kill it the right way, and then play a stone in the right spot, rinse and repeat). I want to test white to find the right moves.

2. Even if so, the kill will cost white a lot of stones (20 or so). Something I kinda considered when the group died.

And in both of those cases I win the game. If I accept the group as dead, I lose. So we end up passing, counting, disagreeing, passing, counting, disagreeing ... for many minutes.

What is the "proper" way to count this? If nobody wants to play any stones? How is this supposed to be dealt with?

It is my understanding that if you disagree, and the opponent demonstrates how it is killed, the stones used for that demonstration are not counted in the final score. Therefore your assumption in #2 above is not correct.

#1 is probably just rude (for lack of better word). I don't know what happens if you ask your opponent to demonstrate how it is killed, and he doesn't know how.

I'm assuming the other white stones are alive or connected to a living group.

Then:

1. If you're playing under territory scoring rules ("japanese-style scoring"), where you count empty surrounded points and captured stones, one standard way to resolve disputes at the end of the game under some rulesets is to play them out to determine life and death, and then revert back to the original position after both players passed after you determined what's alive or dead, for counting. So white can show that black is dead here and not pay the cost of playing all the moves to actually kill because he gets to revert back to the original position for the counting., and therefore wins.

2. Another practical way is to play out the position and just score the result, but both players must play the same number of moves on the board, even if that means one of the players has to fill in their own territory and "lose" a point. Additionally, you need to have a special rule that handing back a captured prisoner is a legal move. So if you play out the position here, white does have to pay the cost of killing the black stones, but black has to pay the same cost in handing back prisoners or filling in his own territory because he has to play the same number of moves during the dispute resolution. So white still wins.

3. If you're playing under area scoring rules ("chinese-style scoring") where you count empty surrounded points and stones on the board, you can just play it out. Playing it out doesn't affect the score.

Most servers support neither of the dispute resolution methods in 1 or 2, partly because they are interface nightmares and complex to implement. So they rely on players to mutually understand the rules, and/or moderators to declare game results by fiat when there is a real dispute.

This isn't directly related to your question, but I'll also point out that 18k on Tygem is infamous for having very strong players mixed in with true beginners. You performance at that level may not be the best way to gauge your progress.

The best solution to a situation such as this is to play Chinese rules. That way is costs neither side anything to play on. Under Japanese rule the black group is simply dead because you cannot prevent white from capturing it.

Tapani wrote:Ok guys.

While I was hoping to leave this subforum and move to amateurs by now -- but considering how badly I get beaten by 18K (lowest possible) players at Tygem I probably still belong here in the beginner forum ...

Had something similar to the following happening to me while counting in a Tygem game:

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

Click Here To Show Diagram Code

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

I'm black.

The auto-score function says my group is dead (which it is). But I don't want to agree with the scoring since:

1. White has to play it correctly to really kill the group (kill it the right way, and then play a stone in the right spot, rinse and repeat). I want to test white to find the right moves.

2. Even if so, the kill will cost white a lot of stones (20 or so). Something I kinda considered when the group died.

And in both of those cases I win the game. If I accept the group as dead, I lose. So we end up passing, counting, disagreeing, passing, counting, disagreeing ... for many minutes.

What is the "proper" way to count this? If nobody wants to play any stones? How is this supposed to be dealt with?

The Tygem auto-score function says that the Black stones are dead, and you agree that they are. You don't want to agree that they are, but you do. Therefore, the proper way to count this is to agree that White has 18 points in the corner, and move on.

BTW, as has been pointed out, assumption 2 is incorrect. The kill costs nothing.

Edit: 1. is a good point at your level. White might indeed not know how to kill, even though the auto-score function says that the Black stones are dead. However, you agreed to use Tygem, with its auto-score function, so you are kind of stuck.

Hi Tapani,

Just to reiterate:

1. White has to play it correctly to really kill the group

No. B is just dead; W doesn't have to do anything, under either scoring systems. If this position happens in a tourney and you summon a referee over, they'll just say B is dead, and W needs to do nothing.

2. Even if so, the kill will cost white a lot of stones

Also no, as others have explained -- under either scoring systems.

So both assumptions (1) and (2) are wrong.

EdLee wrote:

1. White has to play it correctly to really kill the group

No. B is just dead; W doesn't have to do anything, under either scoring systems. If this position happens in a tourney and you summon a referee over, they'll just say B is dead, and W needs to do nothing.

Although that is true for this simple position*, the possibly false assumption that the players play correctly in the hypothetical determination of group status was one of the rationales for the British Go Association adopting the BGA (which are modified AGA which are modified Chinese) rules. It is not uncommon in the UK for the referee to be a weaker player than those having a rule dispute, and in the case of some wacky seki/ko/chosei etc they might get it wrong, so letting the players play it out as part of the game (after dame and making defensive moves to remove ko threats (at no cost due to area scoring or pass stones)) is a cleaner solution that doesn't require a strong referee.

Actually in a British tournament it might actually go like this:
Referee: "So what's going on here?"
White: "I think that black group is dead but he doesn't agree".
Black: "I don't agree it is dead, go on show how you capture me!"
White: "I can play here to make a rabbity six, you capture, then vital point and dead shape"
Black: "I'm not in a petting zoo, what are you talking about rabbits for?"
Referee: "Ok, please continue the game, remember that dame are points in BGA rules and we have pass stones, and Mrs White kill the black group if you think you can"
They then fill dame and after all the dame are filled white plays to make rabbity six atari, black captures, white plays vital point, black passes (giving a pass stone so loses a point), white plays another move in the big eye, black passes with another pass stone, white plays another move in the big eye, black passes with another pass stone, etc etc through gradually smaller dead shapes. Through this process black passes loads so loses as many points in pass stones as white gives in captures, and eventually black is captured and white didn't lose any points to do so. Mr Black might realise what is going on and concede they are dead mid-way through this process.
Referee: "Well, now black stones are all captured so there are no more group status disputes and you can count the game".

EdLee wrote:If this position happens in a tourney and you summon a referee over, they'll just say B is dead, and W needs to do nothing.

Hi,
This is indeed the tradition, although not the rule itself. Some rules (AGA, French...) say that it is up to the players to solve the position. Some leave the question open (japanese rules, at least in their available english translation).

Tapani, if all the above seems difficult to grasp, start with the chinese-style counting: each player gets one point per stone of her colour still present on the board, plus one point per empty intersection completely surrounded by her stones only. That's the basic idea of the game of go.

If black lives, then kills white, then Black has 15 points (the 15 intersections in the top right corner).
If White kills black, she has all the points (361 points if the goban is a 19x19 and there is no other groups).

Now, if the players agree, the game may stop in advance, and the score be counted "as if" the relevant group had been killed, the stones removed, and the remaining intersections left with only one colour surrounding them.

Then, switch to the japanese-style counting : each player gets one point per empty intersection completely surrounded by her stones only, plus one point per prisoner.
In the case where both players play the exact same number of stones, it gives exactly the same result. Most of the times, one player may play one or two more stones, and the handicap stones should be taken into account too. Which leads to actual (very minor) differences between the rules (the AGA, BGA and french rules artificially cancel the effect of handicap stones. Chinese rules nearly cancel them).
Most of the time, it changes one or zero points, and doesn't change the outcome. Quite rarely, it does change the outcome.

This definition of score (territory + prisoners) is nothing more than a convenient way to play the original game of go (the winner is the one who possess the most intersections on the board, stones + territory) without the hassle to count hundreds of stones after each game. But it follows that, with this scoring system, dead strings has to be freely removed at the end.
Otherwise, the game would be completely different. It wouldn't be the game of go anymore.

Wow guys, this was overwhelming amount of replies.


Somehow I can understand that stones are "dead" before they are removed from the board, but also I have hard time understanding the concept of "an oracle" referee that can tell if a group can live or die with perfect play.

I had no idea the group can be just declared dead without my opponent having to kill it.

Also someone thought it was rude to "try" my opponent, to see if they are able to perform the kill. More generally -- is it against go tradition to trick or test your opponent when worse off? Or is it only when scoring?
(I have a background in chess, where it is the rule rather than the exception that players test their opponents if they can win / draw an endgame).

Actually I think I like the Chinese scoring so far, and my very irrelevant opinion is that before scoring all killable stones should be killed. If not, no points for those intersections or stones. That way there is no need to have referees adjucating games. It all seems a little .. informal/arbitrary otherwise. I have no idea which scoring Tygem uses.

In many ways the go community is the way the chess community was long time ago, maybe 50+ years ago. With arbiters adjucating games, more informal/diverse rules, ad-hoc player classes instead of statistics based ratings etc. And more friendly and "honourable".

Thank you all. It is an interesting discussion.

jeromie wrote:This isn't directly related to your question, but I'll also point out that 18k on Tygem is infamous for having very strong players mixed in with true beginners. You performance at that level may not be the best way to gauge your progress.

Thanks for telling me!

Been kinda amazed at the varying level of players. Some obviously have never seen a ladder before, while other attack every single stone I place and killing a majority of them. =/

My question is just ... why? Can't they just challenge lower rated players?

Tapani wrote:...but also I have hard time understanding the concept of "an oracle" referee that can tell if a group can live or die with perfect play.

If play is uncertain, then it should be played out. In this case (and probably most cases), it'll be quite clear whether stones are alive or dead, at least to players with a little more experience. As a beginner, there's a lot of shapes and patterns you haven't encountered yet, and you haven't developed your reading strength yet to judge these situations.

Chinese rules definitely simplifies this, so you're not losing points by playing out something that should already be dead or alive.

Tapani wrote:Also someone thought it was rude to "try" my opponent, to see if they are able to perform the kill. More generally -- is it against go tradition to trick or test your opponent when worse off?

No, of course. The fact is that at the beginner level, we are unsure when to end the game and pass. It can happen that a player makes several moves while the opponent just passes... until one of them realizes his/her mistake.

But above 15 kyu, passing during a game becomes very rare.
What would be rude would be to add stones where it is obvious that it can't work. For example, you're playing at the 1 dan level. The game has come to an end, and suddenly, you begin to remove liberties from a chain that is obviously alive (your opponent passes) just to see if your opponent sees the final atari and connects or let himself/herself be captured. That would be completely out of place, as long as your opponent has demonstrated skills 30 stones higher during the whole game.

The question comes from the fact that the game of go allows both players to add a lot of stones after the territories are complete. In chess, you can't play any more moves once you're checkmate.

Myself, I would on the opposite consider rude from my behalf to ask a beginner not to play unuseful moves. The beginner is learning. He/she has to try and see if it works or not.

Tapani wrote:Wow guys, this was overwhelming amount of replies.


Somehow I can understand that stones are "dead" before they are removed from the board, but also I have hard time understanding the concept of "an oracle" referee that can tell if a group can live or die with perfect play.

That is not a feature of go per se. However, beginning players often cannot tell if a group is alive or dead and do not know what to do to resolve disputes at the end of play. In that case a stronger player can show them how a group is alive or dead. A tricky situation can occur where beginners end play prematurely, when the life or death of some stones is unsettled. Then any advice a stronger player may offer could help one beginner at the expense of the other. As DrStraw suggested, if beginners play by area scoring, then they can resolve their questions themselves by play.

I had no idea the group can be just declared dead without my opponent having to kill it.

That's what it means for a group to be dead at the end of play. (Assuming that you mean "to capture it" by "to kill it". )

Also someone thought it was rude to "try" my opponent, to see if they are able to perform the kill. More generally -- is it against go tradition to trick or test your opponent when worse off? Or is it only when scoring?
(I have a background in chess, where it is the rule rather than the exception that players test their opponents if they can win / draw an endgame).

Traditionally, a game of go ended by agreement. Today, most rule sets end play by consecutive passes, which could happen when the players disagree about scoring. They resolve the question of disagreement in various ways. And different online go servers have different procedures. For beginners this can be a mess. Experienced players rarely disagree.

Among experienced players for one of them to extend play unnecessarily or to pretend to disagree about the life or death of a group is a breach of propriety. That said, even experienced players in the SDK range sometimes reach a false agreement, in the sense of obtaining a score different from that which would happen after correct play from the point of consecutive passes. That is perhaps a point of academic interest, but even 4 kyus would do better to test their opponents at the end of play.

Actually I think I like the Chinese scoring so far, and my very irrelevant opinion is that before scoring all killable stones should be killed. If not, no points for those intersections or stones.

That is a good rule for beginners. But you will find as you improve that following it will become boring.

In many ways the go community is the way the chess community was long time ago, maybe 50+ years ago. With arbiters adjucating games, more informal/diverse rules, ad-hoc player classes instead of statistics based ratings etc. And more friendly and "honourable".

Go has a longer history than chess. Maybe someday chess players will catch on and catch up.

Seriously, your view of the go community will change with experience.