That’s not score cheating in the traditional and widely understood meaning. Expanding the term to include that would be to introduce tremendous confusion in the terminology. It would also hugely increase the time needed to resolve a report unless you required the reporter to explain what was amiss. Of course, most wouldn’t, and CMs would have to go searching the board to see whether another condition, not just a dishonest alteration of the board, were present.
1 Like
Some months ago there were many people, including myself, suggesting improvements to the so-called anti-stalling feature to make it actually an only anti-stalling feature and not an anti-not-resigning-when-you-are-losing feature, but these seem to have been forgotten and no action taken, perhaps with the recent development focus on the automatch changes. Addressing these, rather than telling people off via moderation channels for using the current feature with its problems, would be a good idea.
2 Likes
Was there a way to improve and only trigger it when appropriate ?
Was going to suggest re-coding the feature to only trigger when appropriate as probably the most simple and reasonable solution to it being usable in non-stalling circumstances too, but couldn’t recall if there was a coding issue in why it hadn’t been done, or if it was considered too difficult to have the feature only triggered to only use it e.g. after dame and teire are filled.
(there was something about using KataGo to compare whether the score changes after passing which I seem to recall, though ?
Or if not, it seems possible/simple to use that to verify that after KataGo passing at that point, the score remains the same to it, and only allow anti-stalling feature use if that’s the case ?)
Edit : It seems to work if KataGo analyses the position with Chinese or NZ rules (area scoring), but in games with Japanese rules which I checked, the OGS KataGo score estimate fluctuates a bit after passing, even in completed games with all dame filled and nothing left in endgame to try.
Chinese rules :
End of game :
Score after passing once :
after passing twice :
Japanese rules games :
End of game :
After 1 pass :
After 2 passes :
End of game (the entire board is dead in a 9x9) :
After 1 pass :
After 2 passes :
Edit : Actually, a bit of fluctuation in a NZ rules game I checked, too. Not sure if it’s a bug of expected behaviour by KataGo ?
NZ rules, end of game :
After 1 pass :
After 2 passes :
1 Like
I don’t remember all the gory details, but yes I thought we came to a reasonable consensus that only triggering if there were no more valuable moves left, which could be determined by comparing KataGo scores after passing or not (could allow +/- <1 to deal with examples you showed above, doesn’t need to be exact) would be a significant improvement over what we have now.
5 Likes
I moved the absolute time tangents to another thread.
The proposal to remove it wasn’t necessarily being strongly suggested but it inspired some discussion that can belong in its own thread.
And we can keep discussing here how to improve the existing anti-stalling and anti-escaping features 
5 Likes
Unfortunately (?) we are continuing the solution talk on that thread as well.
1 Like
Yeah, I don’t think it should have been split. The absolute time discussion was always in this context
I would say at least half of the posts were not.
Essentially this covers the only suggestion in the other thread.
If there’s no more useful moves left, then allow the anti-stall, three pass and claim a win, thing to kick in.
The downside of course is that some players don’t just stall when all the dame are filled. It could be that they just lost a group of stones in the midgame and start playing all along the first line in 19x19 to waste time.
E.g. a players last moves all are
The player with black just resigned in that game (one that was easy to find from a good while ago) because the opponent even mentioned in the chat that they were going to fill the board until no moves were left.
How is that stalling? it’s impolite, sure, but the game is clearly not over, as there are open borders, so how is it stalling? Black can’t pass to those moves even if they don’t all require a local response, because there are points to be gained in the open border, and isn’t the definition of stalling being able to pass against it without losing points?
I think that’s overly reductionistic, we also gathered input from @Bunburyist that this would solve the issue from an Absolute Time player’s perspective
Are we looking at the same game?
They’re playing moves that achieving almost nothing, to prolong the game.
Normally both players tidy up the borders in a way that will increase their score over passing.
If you have some definition of “stalling” that you consider different, then this is definitely trolling and the anti-stalling will help prevent it.
That’s not even the definition in the draft guidelines here
- Stalling definition. Stalling usually takes three forms:
- repeatedly playing moves (whether dead stones or passes) that serve no purpose, infilling one’s own territory for no reason,
- repeatedly rejecting the correct score, or
- clicking the autoscore repeatedly when no change has occurred.
What White is doing is preventing the end of the game, by playing moves that barely matter in order to troll the opponent. Literally “stalling” the end of the game. They even described their intention to play until no legal moves were left.
I don’t see why or how you can say this is merely “impolite”, as if they just played in the “wrong” corner of the opening or something.
2 Likes
It’s really not. A lot of the discussion was about in person absolute time games, Ing rules in comparison to Absolute time, @benjito joking about removing absolute time.
If you’re talking about 1-2 posts you made and the response that said “that might be helpful”, we can merge that back in here sure.
The discussion of Absolute Time games was an effort to understand them that that use case might be integrated into the Anti-Stalling feature improvement. I’m concerned that the split might obscure the rationale for the solution to fix the Antistalling feature
That’s exactly why it deserves its own thread.
You can discuss and understand it over there, and then when we have a good idea of how it relates to the topic of this thread, incorporate that into a suggestion to improve the feature.
Rather than having 40-50 off topic posts separating Uberdudes suggestion, and your suggestion, which are very similar in nature.
It’s not really a different rationale in essence.
One player wants to be able to play all the dame, because maybe their opponent might not notice an Atari or something late in the game even though they know they’re losing.
One player wants to play right to the last dame in case their opponent times out in absolute time, especially when they’re losing on the board, as that’s the nature of absolute time games (though it’s possibly not the most popular time control).
Have we obscured the rationale?
1 Like
I suspect an absolute time player would find your framing biased, but in the sense that they’re edge cases but still legitimate play which the Antistalling feature is interrupting, yes, they are in agreement
We can call it biased, but in the case brought up, it is the literal reason.
There is, let’s say, an end to the game that a large fraction of experienced players can agree to is the “end” of the game. Those players would just pass and score the game.
The absolute time player wants to play to that end, no matter what, because it’s only that end that is reasonable to accept, because time is a resource. It doesn’t matter who is winning until that end is reached and both players have time remaining on their clock.
The anti-stalling feature can end this kind of game prematurely, as it can end other games prematurely - see
That’s the drawback, games that are “decided” can be ended early.
We would like to reduce that if possible, but we also want to be able to keep the purpose of the whole feature, to allow players that are winning to end the game when their opponent refuses to play sensibly anymore. That could mean, filling in territory, refusing scoring, or, and I would argue it is stalling in the literal sense of the word “stalling”, changing how you play to troll the opponent and prevent ending the game by not resigning, scoring, passing or filling the borders.
It doesn’t even have to be efficiently filling the borders.
1 Like
OK – if I could make a truly fair system for absolute time that had a very minimal chance of being leveraged to unfair advantage it would be this: (without regard to implementation)
Each player gets their time on the clock and plays within that. Obvs, if you run out of time, you lose.
When the board is closed to the point that it can be scored, and no further move will change the score*, then the game is scored and the winning score wins. Ideally this is done by each player passing, but if one continues to make moves – the server takes over and scores it.
If someone passes before the game is settled, and the opponent keeps playing, then the server does not intervene, both players lose time accordingly, and it plays until the board can be scored.
For me – that’s enough. If I close up, and the opponent keeps playing random spots, I tend to try to handle that by having more time and making sure they run out of time before I do. I would not object, however, to someone doing that leading to either a forfeit or forcing a server score (should be the same result, unless they’re just silly).
So for me, and I seem to be the one who plays absolute here, it would be enough to simply turn off server decisions when playing absolute time. Alternatively, if someone wants to make the server come in and make those more complicated decisions – I’m fine with that. But turning it off is certainly better for me than the current implementation that creates an opportunity for cheating that did not previously exist.
footnote * I say that the repeated stalling moves do not change the score. However, in certain scoring systems, filling in the blank spots actually decreases the opponents score each time. So that would have to be factored in.
QUESTION:
Stalling seems to only truly be a problem in an absolute time system, as it can force a loss on time when the board is already settled. It’s annoying in a game with byo time, but it can’t actually change a result unless the annoyed player quits, is that correct? Or am I missing something.
2 Likes
It is still truly a problem in other time settings, even if it cannot force a loss. It is not acceptable that players purposefully stall the game in the hope that their opponent will resign out of frustration.
So for me, and I seem to be the one who plays absolute here, it would be enough to simply turn off server decisions when playing absolute time.
I don’t have an issue with that solution though, and I imagine it would be simple enough to implement.
@Bunburyist wouldn’t it be more optimal to just fix the feature rather than disabling it where it causes the most egregious harm? Far better to highlight where it causes the most harm to show the need for a fix, right?
Let’s consider:
A. Continuing to play although there remain no available moves that are better than passing. For instance
A.1. Play dead stones inside the opponent’s territory
A.2. Infilling one’s own territory
A.3. Playing dame under Japanese rules.
B. Continuing to play although the opponent could pass three times in a row while the leading player keeps a lead of at least 10 points. For instance, continuing to play 1 pt yose moves when you are behind by 50 points.
Moves like A.1, A.2 are generally considered as stalling, except that A.1 is ambiguous because one player may try to kill and not be sure whether his move works.
A.3 (filling dame) can be considered as stalling under absolute time if Japanese rules are used. On the other hand, filling dame may reduce liberties, thus forcing the opponent to add a move inside his territory to protect a group.
Moves like B can be considered as stalling but not always. Even if you are 50 points behind, it’s generally acceptable to go to the counting phase, especially if the game is close to completion.
So ideally the server should declare a winner when moves like A are detected. However detecting such moves automatically is not easy so OGS’ current implementation detects B instead.
For comparison, EGF rules say this
Finite thinking time under Territory Scoring
The following applies under Territory Scoring. It is not compatible with the Ing Overtime System.
The alternation consists of the competitive phase followed by the neutral phase. During the competitive phase, one or both players moving next can make a play to improve the score or to fill a basic endgame ko. During the neutral phase, neither player can do so because only dame and teire, if any, are left.
If the first two successive passes occur prematurely during the competitive phase, then the clock is neutralised, each player’s time is set to exactly 1 minute, and the clock is restarted for resumed alternation.
Until two successive passes during the neutral phase, every legal play is considered sportsmanlike.
During the neutral phase, a player has to pass if his opponent has just passed. Then on neutralised time, more dame and teire may be filled quickly in continued alternation.
4 Likes