Count pass stones as prisoners in AGA rules


It was brought to my attention that when a game is played with AGA rules, pass stones are not accounted for in the prisoner counts. I read a few threads in the forums, and it was made clear that OGS uses chinese scoring for simplicity, so while the final result is correct, it looks wrong is the game is scored manually using japanese scoring, which ironically defeats one of the main purposes of the AGA rules which is to make chinese and japanese scoring consistent with each other (as much as possible, at least).

Even though the final score remains correct, I think that passing stones should be accounted for in the prisoner counts, to avoid any confusion for people who are familiar with the AGA rules (which should be assumed to be the default for a game played using AGA rules).

The only difficulty I can think of, regarding this, is that OGS does not wait for white to pass for last; but that should be easily accounted for by checking who passed last, and increasing black’s prisoner count at the end of the game if it turned out to be black who passes last.


Yes, it’s strange especially since the right procedure (I think) is documented here:

I think it might’ve been a decision that the OGS AGA rules were fine regarding hosting the US go congress online.

It doesn’t seem to make sense to just aesthetically enforce pass stones, if the game isn’t going to be scored using territory counting, in the same way that implementing a method to allow you to rearrange the territories probably doesn’t make sense if you’re not manually counting, and the computer is doing it for you. It would just be an additional aesthetic.

So is the point that one wants to enforce pass stones and white passing last, just in case people want to score the game manually by moving stones around in analysis mode or a review after the game has ended?

It’s only aesthetic in the same way that the prisoners shown are aesthetic, and if the prisoner counts were wrong, (e.g., you capture some stones and the prisoners don’t change) then I’d also want that fixed. Either the prisoners (including pass stones) shown are correct, or, if they are deemed exclusively aesthetically, then they should be removed altogether. I don’t think white passing last needs to be implemented necessarily, but the result stone can easily be accounted for by checking who played last.

In practical terms, this will at the very least avoid confusion, e.g., long debates which have taken place elsewhere on whether OGS implements AGA scoring correctly. In worse cases, albeit rare ones, someone playing a game might rely on an inaccurate prisoner count to determine what to do and misplay as a direct consequence.


Yeah but in another sense the prisoner counts aren’t truly aesthetic since they’re required in Japanese rules which the site is also supporting. Restricted to Chinese or AGA rules etc, sure they could be hidden.

This is also true of the inaccurate score estimator among other things, and could be a reason to just remove the basic score estimator for instance.

I think if one adds in the pass prisoners then one should probably enforce white passing last, because otherwise it’s just another strange half measure?

Then again, the prisoner count is also useful if you look at the board state and wonder why one player has a certain number of less stones than another player - if you can see the difference in the number of prisoners and handicap it makes sense, without needing to replay the full game.

It’s a half measure in a sense, but it’s also a strictly positive improvement, getting strictly closer to a full implementation of AGA rules, and by a lot. I’m also taking into account that I imagine increasing the number of captures is a lot simpler than changing the way the game is declared finished.


But lets say the official AGA rules has a line in it like

12) Counting: There are two methods for counting the score at the end of the game. One is based on territory, the other on area. The players should agree in advance of play which method they will use. If there is no agreement, territory counting shall be used.

the players by choosing to play with the OGS implementation of AGA rules have agreed to use the implementation and hence area scoring. So similarly if two players played over the board with AGA rules using area scoring they could do away with keeping track of prisoners, the pass stones, and white passing last.

So it’s not clear that it’s not a “full” in the sense of fully functioning implementation of AGA rules, it’s just not every implementation of AGA rules that you might want for over the board play.

1 Like

Even so, even if both players agree to use area scoring at the end of the game, they might use territory scoring during the game (including the last decision on whether to pass) for their own benefit.

And even if they didn’t, even if they agreed to area scoring, and privately also used area scoring, the prisoner counts shown would still be wrong, and should be fixed. That is true of any game, even in a non-AGA Chinese-rules game, if the prisoner counts shown were wrong, I would want it to be fixed.

That’s really the bottom line for me, personally: if the prisoner count said “100”, I’d want that fixed even if it were ignored by the server due to Chinese scoring, because it’s not the real prisoner count, whether it’s used or not.


I can partially agree, except that it’ll be wrong, if, someone passes during the game, or otherwise it’ll be wrong only right at the end of the game for most games right?

So you can still use territory counting privately and effectively throughout the game right?

It can also be wrong after the first player passes, and the second player chooses whether to pass or continue playing. If the “white passes last” rule were implemented accurately, then it would also be wrong if white needs to choose whether to pass again or not.

Granted, not a huge impact, I know, so not necessarily a high priority by all means, but I imagine incrementing the prisoner count whenever a pass happens would be a simple fix anyway. If I knew anything/enough about webprogramming and/or react I’d give it a go myself.

In practice, I raised this topic specifically because there was a “big” discusssion in a go-related discord server about whether the results of a half-point game was correct due to the apparent lack of a passing stone, so I imagine avoiding these kinds of discussions and confusions might end up being the more concrete positive impact of this fix, than actual gameplay decisions.