I have been in plenty of games where the opponent, or I, have a server-determined 99% chance to win, only to lose. Because the it’s 99% chance to win if you know how to close it all out like the server does. And if you make errors because you’re not that good, because you’re under time pressure, or because you misread something, you can easily still lose.
3 Likes
I fully agree. A 99% win in the eyes of AI, when starting the endgame can be far from being the same in human eyes.
More points of interest:
endgame popularity
The problem is that endgame is not a so popular part for us, westerners, well at least according to some asian pros comments, and I fear that some times we will find this shorcut quite convenient 
.
implementing bargain
The bargain of 3 passes may even change some way we play the endgame, like you know some difficulty could arise somewhere so instead of playing it you bet and pass 3 times. That makes the game of go quite different, in which you should assume everything from the beginning to the last move anyway.
automatic detection?
In fact even it’s difficult to detect automatically stalling. How can you affirm you detect a negative attitude when there could be nothing intended with bad spirit but just some (ridiculous) hope with unawareness? How you put the limit? It needs some time to see your stones in atari… Of course with more indicators like your rating, the position on the board itself…, stalling attitude can be guessed, not saying it doesn’t exist. But false positive can be quite offensive.
The idea of anti stalling i do understand (and appreciate the effort) it but the criteria are yet far too primitive and can’t guaranty we are still playing go.
2 Likes
Intention is irrelevant to the issue of stalling. A stalled car blocks an intersection whether or not the driver intended it, and stalling actions in go block the end of the game whether or not someone intends that. Beginners might try to retake a simple ko immediately, but the system legitimately blocks them from making such a move—their intention or level of knowledge is irrelevant.
That is mostly untrue. There are three kinds of stalling in the technical sense of a reportable violation: (1) repeated, consecutive resumptions (pass, pass, resume, pass, pass, resume); (2) repeated infilling of one’s own territory; and (3) self-atari moves for no purpose. The first two are easy to detect. #1 is the most annoying because there is no way for a player to stop it, and it theoretically could go on forever (I have seen two instances of players resuming more than 100 times). However, an alternative way to stop #1 would be to place a limit on the number of resumptions (or consecutive resumptions); that would not even require a pop-up. Similarly for #2, a limit could be placed on consecutive infilling; beyond the limit, such a move could be blocked as illegal.
#3 is more problematic, because certain tactics may require self-atari. I can’t think of any way to handle it that is better than the new feature. Of course, this thread has already made clear that some tweaking may be required—perhaps by increasing the number of passes required—in order to avoid the rare unintended consequences that have been enumerated.
BTW, ending stalling has an excellent ancillary benefit in that it will also reduce the number of score-cheating reports. Many score cheaters first attempt to win by stalling, and cheat only when that fails.
2 Likes
I want to add that “some go servers” include OGS as this thread i found may be well deserved by this inconsistency in scoring
1 Like
I think the bug I reported is resolved by clicking the “Autoscore” button or marking stones dead. No need to keep playing to capture.
Well, good news. Still @Conrad_Melville mentioned a collection of erratic scoring, but hopefully these are solved now too.
1 Like
No, the autoscore bug isn’t solved. Indeed, the autoscore fails to work right in more than one way, as has been demonstrated in two different threads.
1 Like
I agree that all of these are bad.
But shouldn’t filling neutral points be excluded from this feature?
These are NOT “useless moves”—yes, some neutral points are plain Dame, but others are Teire and can thus force the opponent to play inside their territory (—> -1 pt).
I was taught that at the end of a game all neutral points should be filled, and I’ve not yet been convinced that I was taught to play useless moves.
Now, some of my opponents are (IMNVHO) sloppy and pass instead, repeatedly, while I go on and fill those neutral points, and BAM, there comes this:
(I stole your screenshot, @anoek, forgive me 
)
7 Likes
Nobody disputes that. Several people, including me, have pointed out that one needed way to tweak the new feature is to allow the filling of dame before the feature is triggered.
3 Likes
I’m not disagreeing that one in theory should be able to fill dame if one wants to, but it does feel like there’s a fine balance between just doing it for the sake of it and tradition and so on, and doing it because there’s a decent chance you can win because of it.
One reason we’re taught (usually by someone in person) to fill in dame, is because that’s what we have to do in order to score the game over the board. It’s part of the counting procedure.
However that isn’t actually necessary online, and I can understand if people don’t want to fill dame, it’s not something unique to OGS, as you often see KGS games or pandanet and so on leave dame unfilled (although there they have manual scoring, marking dead stones manually etc much more prominently).
I do think the 10 point margin after 3 passes probably works in a lot of cases. It may not appeal to our idealised idea of what online Go might be, and there might be cases where it can be abused, like many other things, but my guess would be that it helps in a lot of cases also.
Similarly, if the player passing isn’t the winning player, one might wonder why they’re passing while the opponent isn’t, in this special case when we’re debating about why you should see the button if you’re not passing.
I think there’s a bunch of situations like:
- someone passing instead of resigning, where it’s not clear if the board will be score-able sensibly (they should resign but they don’t);
- the passing player has tried a bunch of failed invasions near the very end of the game, and now the non-passing player is just killing dead stones, since the opponent wouldn’t pass up to now - game is probably over and so why not offer the option to just end it - it prevents players that would stall and when stalling doesn’t work then score cheat of which I’ve seen some.
I think it would be more unfortunate if for some reason the 10 point margin estimate included something being counted as dead “incorrectly” for the players level, but I’m not sure if I buy into the idealism or an appeal to the idea of being deprived of scoring as a counter if it just allows a lot of bad actors more opportunity to ruin the experience of players.
No-one is forced to click this button at the end of the day.
2 Likes
But as there is not a cancel button, they may be misled into thinking they must. “What’s this weird message that appeared instead of the usual game interface, I want it to go away, click”. Hence my call for better messaging and a cancel button.
7 Likes
Normally on real board you fill the dames in order to count the score correctly.
With area scoring you usually fill the dames since they are worth 1 point each.
With japanese rules you usually fill the dames in order to make sure that all necessary teires and connections will be played
I think the only exception for not filling dames is when you’re playing online and using japanese rules and you are absolutely sure that none of the dames are worth one point.
But generally speaking, you should fill dames before passing.
5 Likes
Right so having an option to dismiss it could help. Although maybe in some cases you dismiss it, but then wish it would pop up again a handful of moves later 
That’s not true, it’s even recommended to be sure the score will be well calculated.
I don’t care, i need a working software in all cases. Easy to imagine difficult Life and death (or just unusual hard to see) that will involve more as 10pts/3 passes.
But you buy an unfinished tool which we even don’t know how to make it better, a tool which is at times depriving players from some part of a game of go.
3 Likes
I mean it is true, you don’t even have to use the autoscore, you can still manually score if you want to. You can recommend it but it’s not necessary.
I don’t really buy into this myself. In the extreme you could also forbid people from resigning, because they’re missing some part of a game of go (i.e. scoring).
1 Like
Yes
Also yes
My solution would be to to have s “wait” button which dismisses the message for another 3 passes.
5 Likes
That seems like a very straightforward middle ground.
I think we could still label it “Dismiss”, since “Wait” sounds like it might just pop up again later (regardless of passes).
2 Likes
Playing through a game that should have finished by resignation deprives you of your next game of go 
Do you play other servers online?
Autoscore is sometimes the only scoring available.
Besides if you recommend it, it’s a bit weird to pretend it’s not necessary. It’s because yes, the autoscore can be wrong if you don’t give him a little help.
That’s why we see people taking time to ensure that autoscore will run well, and why the 3 passes can then leave you without scoring the game.
That happened already so let’s consider a solution: get a warning after the 2d pass that a third pass could lead to a victory decided by AI. Not very esthetic but honest. At least you can resume playing to get scoring. Honest but still unilateral anyway.
2 Likes
Then why wait 99%+10 ? Some games are already clearly lost in the opening. You could play even more games 
3 Likes