Determining who has won is very hard?

So far i have only played go on my computer, which means that after the game is finished, the computer calculates who has won. But if you play over the board, you have to do it yourself, which can get very hard?? Take a look at this example:
image
Both players passed their turn and the game is over. It takes some heavy calculations to determine which groups are dead, alive or unsettled before it is possible to determine the score. I can imagine that on a 19x19 board, it is even more complicated.

I guess my question is, do you have to be a skilled go player just to determine who has won or is there a simple way to determine who has won, that does not require highly developed tactical skills?

Are you allowed to stop the game if both players are unable to determine who has won?

I see unfinished boundaries the game is not finished and then uncountable

6 Likes

What do you do if the players cannot agree on who has won or are unable to determine it? Is the game unaccountable unless both players play it out until the position becomes as basic as you can get?

No it’s not.

If you insist on passing at this point then technically Black has no territory at all in the region I marked ZERO because the boundaries of this would-be territory are not closed:
d6ff17202be2c8b3a41a2ad2f69884da445f8729.jpeg

But really what this means is that players shouldn’t be passing at this point. The game is not over. In fact if you tried to pass at this stage of the game in a face-to-face game with a referee, the referee wouldn’t let you score it and would kindly explain to you that the game isn’t quite over yet.

If you play on a computer and the computer tells you that black has territory in this area at this stage of the game then it is a bug.

Instead you should keep playing a few more moves, until the game looks like this:
Jpeg3.jpeg

And at this point if it’s still unclear to you which groups are alive and which groups are dead, I recommend playing even a few more moves to actually capture the “dead” groups:
d6ff17202be2c8b3a41a2ad2f69884da445f8729.jpeg-1
Jpeg4.jpeg

Now you can pass and count the territory and there shouldn’t be any difficulty to count.

6 Likes

I’m assuming the game is played under Japanese rules.

  1. If the players pass in this position, …

2 … the next step is to remove dead stones. Dead stones are stones of which both players agree that they can’t avoid being captured.
For argument’s sake, I assume that both players agree that the marked stones below are dead and all the other stones are alive (that assessment is not neccessarily correct[*], but that’s OK as long as the players agree):

  1. When all dead stones are removed and added to the prisoners, we arrive at this situation:

  1. Now we determine which regions of the board are fully closed off by stones of one color. In this position, we can find these regions:

  1. The triangles mark white territory and the squares mark black territory. I count 8 points for white and 10 points for black. Besides this, white has an additional 5 points for the prisoners added in step 3. Also, white gets 6.5 points to compensate for black playing the first move of the game (this compensation is called “komi”).

  2. So in total white has 8 + 5 + 6.5 points = 19.5 points, while black has 10 points. That means that white wins by 9.5 points.


That being said, more experienced players would not pass in the original position, because both players still have opportunities to close off more territory than displayed in my diagram of step 4, so they would do that before passing.

[*]The players may not actually agree on the dead stones that I marked in step 2. Experienced players would assess black’s group in the upper right is also dead, and also white’s group in the lower left. And the black group on the right side can be dead or alive, depending on who plays at L6 or N9 first.

If both players pass to stop the game, they can. If they agree on step 2, they can continue the scoring process as I described above to determine the winner.
If they disagree on step 2, they can resume the game to fix those loose ends.

11 Likes

Ok, to summarize:

  • Players can both pass at any time, even if there is plenty of game left to play.
  • Players agree on the dead stones and remove them. If players don’t agree on which stones are dead, the game continues until it becomes clear.
  • The score is being counted regardless if the assessment of the dead stones is correct or wrong, as long as both players agree.
  • If both players pass without establishing any territory or make any prisoners, white will win by komi alone. (i know this is a silly case but i just want to verify that i understand the rules correctly)
10 Likes

I think you understood perfectly.

1 Like

Thanks for the explanations!

2 Likes

Perhaps I should mention that under Japanese rules you should be careful with game resumptions just to prove that a group is dead.

If one player denies that his dead group is actually dead to force his opponent to play moves in their territory to capture the dead group, and the denier just passes while his dead group is being captured, the capturing player loses 1 point per move.

In a game between beginners this can happen due do simple misunderstanding, but in a game between experienced players, denying that a dead group is dead may be a trick by which the denier hopes to win a lost game. That is a reportable offence (score cheating).

3 Likes

Yes, this is the procedure for various area scoring rules (such as Chinese, AGA, New Zealand, ING). The burden of proof falls on the player claiming that something is dead. They have to play out the capture.

Note that under area scoring, it does not change the score to play out this “encore” phase of players making additional moves inside their territory to capture dead stones.

Under Japanese rules, the situation is different. As @gennan mentioned, it costs points to fill in territory, so resuming the game to play out disputes does not quite work. Instead, life and death under Japanese rules is resolved by considering what would happen under hypothetical play (while applying different rules regarding ko) to capture dead stones. For >99% of situations, the Japanese rules life and death determination is the same as under the area scoring rules, however, there are some very rare cases that do behave very differently. I don’t think there is any concise way to fully explain the exact way the Japanese rules work.

8 Likes

The “encore” (game extension) procedure of Japanese rules, to prove that a group is dead, is a bit complicated. Even though it is the procedure prescribed by official Japanese rules, I wouldn’t recommend it to beginners without assistence of an experienced unbiased referee.
Also, that prescribed encore procedure is not actually implemented by any go server AFAIK.

A more practical method to preserve the score while proving a group dead during resumption in an OTB game, is that players give a prisoner to their opponent each time they pass (those are called “pass stones”).
It does require both players to spend the same number of turns in a resumption phase, so 3 consecutive passes may be required to finish the resumption and go to scoring again.

This pass stone method deviates from official Japanese encore rules, but in >99% of cases the scoring result will be the same if both players play correctly.

This pass stone encore method approximates AGA rules.
AGA rules always use pass stones, even before any resumption/encore. Only the AGA rules require white to pass last. This is to ensure the same scoring result when using area scoring [*] versus territory scoring [**], which is a feature specific to AGA rules.

So for beginners who prefer territory scoring, but who regularly find themselves in disagreement about dead stones during scoring, it’s easiest to just play under AGA rules rather than Japanese rules. It avoids many complications.
Then later, when you rarely have scoring disagreements anymore, you may opt to play under Japanese rules if tournament rules require it or if you prefer that tradition.


[*] area scoring = after step 3, count points for territory and stones left on the board
[**] territory scoring = after step 3, count points for territory and prisoners

4 Likes

To answer the title question: basically, yes (maybe I wouldn’t say “very”): determining the winner of a game of Go is harder than other games likes Chess or connect-four. This is a key reason why computers weren’t very good at it for a long time.

4 Likes

AGA rules allow territory counting, but they’re still area scoring regardless

I would recommend something like Lasker-Maas if you want good territory scoring. If you want something implemented online, I’d settle for area scoring and do NZD

2 Likes

One simple option is to use stone scoring. Whoever has more stones on the board wins.

It prolongs the game, which can be boring, and can change the winner compared to other scoring methods, but I think it could work well for beginners (on smaller boards), who don’t know how to score the game in a modern way.

2 Likes

Great but not available on OGS (besides manually counting with a demo board )

If a beginner can understand that an empty point inside his stones will be counted as a point and his stones will count too then you can use area rules (like chinese), the goal being occupy a bigger space (including your stones) as your opponent. Occupy means your opponent cannot take any of your stones.
Game will become accelerated later by understanding better and more precisely what one can capture or not (mostly the 2 eyes concept)

A beginner would assess at times even more like a possible capture of the white stones in the upper left corner (which is easely avoided by a little bit experienced player if cautious). Another game changer is after white play J1 if then black don’t add a move.

2 Likes

Yet

After doing some more research on different go rules it seems that japanese rules are by far the most complicated to score, since proving that a group of stones is dead by capturing them will cost you a lot of points. The opponent can simply claim that his stones are not dead unless you capture them and force you to into ending up with a lower score. To accurately decide who has won a go game under japanese rules requires an engine or a master then, which sounds insane to me.

Using chinese rules, you are free to capture everything you can without losing points, which eliminates the need for discussion on which stones are dead.

Chess was mentioned earlier in the thread. In that game, it is perfectly clear if it is a win, loss or a draw. There is never a need to prove anything by visualizing a complex tactical sequence 10 moves ahead.

5 Likes

Not in real Japanese rules, as that determination of live/dead status is in the dispute resolution phase of the game, and is “rewound” and the original position scored without any loss of points.

But servers (such as OGS or KGS) don’t implement that properly.

5 Likes

A win is clear with checkmate. Draw by threefold repetition or 50-move rule require a long-term memory and are not simple game-end conditions, but much rarer.

3 Likes

Yeah, not that Superko is any better than 3-fold in this respect except for longcycles being far rarer in Go

Yep. AGA and NZD rules also work like that here. Generally speaking Area Scoring Rulesets work like that, and Territory Scoring Rulesets don’t, but Territory Scoring Rulesets can be fixed to work like that, as with Lasker-Maas

3 Likes