The situation is that you both passed and then realized the black group was dead. So, according to Japanese rules, how should you score it?
I would never play Japanese ruleset,lol
I would actually grant the restart, cause I failed to see how white kills the black left side 
Is the white b13 stone already there before the pass?
1 Like
The game would have been over in real life. People don’t second guess what they don’t know.
The scoring tool changed the outcome of the game. That’s the issue.
Also, it would not allow you to mark the group as alive and grant the territory to black. So had it been material enough points to change the win/loss, a restart would have had to happen and black could have lost.
2 Likes
B13 was not played.
1 Like
Never mind, i got it now.
1 Like
In other topic people arguing now what to do if game ended incorrectly early and territory not sealed, there is a hole. I think instead of rereading old official rules and trying to find interpretation there, its better for OGS to create its own solution. We have analysis and score estimate anyway, why not create one more thing?
Currently with clicking and ctrl+clicking its possible to change status of groups, but that system has too complex mechanics and sometimes returning to game and actually closing the gap is the only way to reach score that both players wish.
Simple territory pen tool is needed. So you can click any coordinate and it becomes color that you need - black, white or neutral. And it should not matter what stones are around. So if players agree that they would place stone somewhere if they didn’t accidentally missed it, they can just draw black or white square here.
1 Like
I don’t see what this issue has to do with territory scoring (Japanese rules) vs area scoring.
Or perhaps it would have been a loss for both by official Japanese rules. Is that what you mean?
Wouldn’t it be easier to simply resume the game ? I’m afraid that this pen tool would go astray from basic Go rules on territory, and create the wrong feeling that closing your borders is not important.
To be clear, there is a genuine difference in my view between:
- situations where borders are not closed and therefore it is a basic rule that there is no territory : I think understanding the proper notion of territory is important and we should not deviate from it, therefore this territory should not be scored and players should resume play; and
- situations where borders are closed, but a stronger player (or AI) could question the living status of a group : this is a matter to be decided solely by the players, and the scoring tool should not interfere in their appreciation.
3 Likes
My view. Disable all suggestions of dead stones in scoring phase, dumb old OGS algorithm or superhuman new KataGo AI one, and force players to mark the dead stones manually. It only takes a few seconds and avoids all problems of old one being wrong and players believing it (one of most common posts on Reddit baduk is someone querying a score from OGS and the answer is invariably “The OGS dead stone marking is just a suggestion from an imperfect simple algorirhm so it can be wrong and you shouldn’t trust it, change it to be correct before confirming”) and new problem of game changing hints from AI. Might even force people to learn how to score, and then OGS wouldn’t be full of forum threads with people who don’t know how to score!
5 Likes
Please God no. Territory is contiguous regions of the board surrounded by stones of only 1 colour (after removing dead stones). This is one of the fundamental rules of Go.
2 Likes
logic of score estimating is: “what score there will be if game will be played until the end?”
and when game is ended, players themselves decide which stones would be dead if they continue to actually capture them.
So fundamental rule of Go in my opinion is prediction of what would be without actually playing it out.
What I propose is to let players to manually paint score themselves because now Kata bot doing it anyway instead of players and players have no way to change it.
Indeed. This is all fine when they agree. But what happens when they disagree or when the borders are not closed?
I’m all for a manual score tool. My point was the unit on which it operates must not be individual empty intersections or stones, but empty regions and chains or groups of stones. Or else it’s not following the rules of Go. That’s how it works on KGS and worked on OGS long ago. Adding auto dead stone marking has 3 purposes:
-
Saving a few seconds work for competent players who could do it correctly themselves but are lazy
-
Helping beginners who don’t understand what is dead to score correctly.
-
Avoiding cheating and trolling from naughty users intentionally mismarking stones (particularly a problem Vs a bot if it can’t disagree).
Point 1 is pretty minor, saving a few seconds is not worth all these troubles. 2 I question the value of beginners blindly trusting a fallable or superhuman algorithm. Better that they do a well designed scoring tutorial and learn how to do it. After all you need to understand the objective of the game to decide what to play during it. 3 call a mod.
4 Likes
when borders are not closed, they may agree to what borders would be if they continue to close it - same as they can agree what stones would be dead if they continue to capture it. I don’t feel big difference from official offline rules here.
I think in not serious tournament players should have right to be lazy and do errors. When they not agree, then Kata mod or human mod would be solution. But before it players always should have way to decide themselves what to do with their own game.
1 Like
I’m not necessarily talking about the “both lose” option; I just don’t know what happens when a dead group is discovered on a border during scoring. Kind of expecting it to go like this:
White: Oh, that group is dead.
Black: Oh, you’re right!
White: So actually almost all of the left side (including three more black points and a lot of white territory) counts as seki.
…both players count carefully…
Black: I think I can do better than that, let’s resume play.
White: kills the group
Black: finishes closing the border
Or if the seki is better for Black:
…
Black: Okay, fine, it’s seki!
White: Wait, I can do better than that, let’s resume play.
Black: saves the group
when borders are not closed, they may agree to what borders would be if they continue to close it - same as they can agree what stones would be dead if they continue to capture it.
Beginner players who don’t understand Go well enough to finish borders and be confident about dead stones before ending the game, might not be helped out by a novel interface that requires them to mark out point-by-point exactly the thing they don’t understand. Asking them to guess “what would the result be if you had finished” when they don’t even understand how to finish is asking too much.
Slightly-less-beginner players who just overlook a gap in the borders would also probably not gain much value from a specialized interface either.
It would probably be much easier for everyone (including OGS devs implementing the interface) if they just used the usual interface to finish the border - i.e. the interface you use for all the rest of the game, where you take turns putting a stone on the board, playing it out.
Also, regarding some of the other discussion earlier in this thread, references to strict Japanese rules like “both players lose” and “anything that touches dame is seki” seem a little beside the point. 
Complex endgame procedures about this, and details like hypothetical play and pass-for-ko and other such things are completely unsuitable for beginners. For casual amateur and beginner games, including in all cases I’ve seen in real life clubs and teaching environments, the rule has always been “resume the game by some reasonable means, e.g. continue with whoever’s turn it was, and play it out”.
6 Likes
Of the solutions mentioned, I lean towards Uberdude’s proposal too. With possibly the exception of allowing players to use AI scoring if they both explicitly opt in to it (as mentioned by someone else in a different thread on this same topic that I’m having trouble finding now).
By the way, I have known beginners who have been confused in KGS by its scoring tool’s insistence on operating on whole groups and regions, when the borders aren’t finished. And anyone who’s played on KGS knows that if you accidentally click in the wrong way during scoring, you might end up marking the whole board or a large region of the board as dead. So this would be a downside of that proposal.
I suspect this could be mitigated though, if the marking of the whole region were slightly animated. In other words, rather than the entire region changing its status instantaneously as it does in KGS, it does the floodfill of status change would be spread out over, say, 0.5 seconds, for visual purposes. Showing the animation of the floodfill would make far more obvious where a small gap in the border is, since you’ll see the wave of territory-ownership-change pass through that gap. The visual animation alone, even without any further rules clarifications or tutorials, would probably help some beginners understand.
Although this would be yet more interface work and dev time (which OGS seems to be constrained by more than anything else).
7 Likes
I don’t see why one can’t have both tools, for marking contiguous regions and for marking individual points if needed.
It’s especially useful for dame and points in seki to mark them individually.
Maybe they’ll start off unmarked anyway and one wouldn’t need to do anything but imagine one player clicked in error (judgement or misclick) and it give a bunch of contiguous dame to that person as points. I’ve seen the old score estimator do strange things when trying to mark stones as dead, so could imagine trying to uncheck those points and then it auto marking the connected black stones as dead so as not to give those points. That could be a group in seki or an entire region etc,
1 Like
Because it is against the rules for one intersection of a continuous region to be assigned differently to white territory / black territory / dame than another in the same region. A small region may be of only one interesction. EDIT: to be clear: you mark a contiguous region of empty intersections, or chain of connected stones, you click on a single intersection that forms part of the region/chain. Just like now. And it automatically affects the whole region/chain because that’s the rules of Go. What I interpreted the alternative as was allowing individual control of intersections so that IN THE CASE OF UNCLOSED BORDERS the players can mark only some of a contiguous region as territory and not those at the unclosed border to work around the fact the players were silly and didn’t close the border.
To make this clear with game image from first post of thread. If T1 is White territory then it is nonsense (and violating the rules of Go) to give tools to allow saying T2 is concurrently black territory. T1 and T2 are connected according the rules of Go so must be scored as the same status. Similarly if you mark p9 as dead then o9 must also be dead. It is nonsense (and violating the rules of Go) to say p9 is dead and o9 is not dead.
With my proposed no automatic dead stones marking proposal, contiguous regions with no dead stones surrounded by stones of only one colour will be automatically marked as that player’s territory (but you can change if needed for example of eye inside a dead group). There’s lots of those in that image T1, t9, R4, q5, n1, M3, d5, c1, k9, h15 To name just a few. So players don’t need to waste their time marking these. All they need to do is mark the dead stones, which are treated as connected chains. So click on g5, M3, n9 (or any of those 5 and whole chain is marked dead), C9 (because players think it’s dead, that KataGo can find a tesuji so disagrees is irrelevant). Doing that to white stones will make the regions adjacent become black territory and vice versa. 4 clicks takes a couple seconds. Easy peasy. No AI Review blue circle to hint at b13 tesuji. Players can discover that when they review the game AFTER scoring.
Hexahedron’s idea of animated flood fill to help find those gaps is nice touch to improve UX.
2 Likes
there is difference between
and