Fox uses this method to prevent scoretrolling. They make it so you can force counting after 350 moves. It requires you to limit the amount of times you can pass in order to work with Japanese rules though.
It might be too late for OGS to introduce a game feature like this but having to wait for a moderator or have a win annulled because of a troll is frustrating and it seems a lot more common than it should be.
IMO when you need to use a force count feature, your opponent is behaving so badly that it’s better to call the moderator. You will have to wait a bit longer, but the moderator will talk with the player and hopefully get him to stop this behavior, or maybe ban the troll. Then future players will not have to face the same experience.
Your example of someone trolling 100% of games is exactly the situation we most badly need players to call moderators to deal with the troll, for the good of all of us.
Our moderator team is exclusively volunteer. We also have other commitments in our lives. Maintaining 24/7 coverage is not possible and anoek would never ask us to commit to such shifts. We do this because OGS is our passion not our job.
We strive to be as responsive to incoming issues as we can be, but no service guarantees instant responses at any time of day.
For Japanese rules, except in exceptional cases (double-ko seki), a trolling opponent has to eventually run out of ko threats, right?
So you could make it so that forcing scoring is possible if both of the conditions are met:
Your own last N moves in a row have been passes for some reasonably large N.
Simulate the removal of all stones that are considered dead by the normal automated scoring from the board, replacing them with empty points. Now, consider all maximal connected empty regions that touch groups of BOTH colors. The total number of board locations belonging such empty regions in total is less than C, for some reasonably small C.
The idea behind 1 is that if the opponent has so thoroughly run out of things to do that you’ve been able to ignore them the last 6 turns, or whatever, then you should be able to force scoring.
The idea behind 2 is to prevent ending the game prematurely when the game is not over but the opponent cannot catch up within that many moves in a row while you pass. For example, the starting position of a 9 stone handicap game - you should not be allowed to pass your first 5 or 7 moves in a 9 stone handicap game, then force scoring, and still be ahead by the automated scoring. So this condition basically says the total number of “dame” in the game must be very very small - if there are any incomplete borders, likely the number of “dame” will breach the limit “C” because the entire incompletely bordered region should be touching stones of both colors.
Under Japanese rules, technically both players should be filling the dame anyways, and you could educate players that forcing scoring requires you to fill most of the dame. The only unfillable dame would be those that occur in sekis, so C can be chosen so that even if a couple of sekis occur on the board, the number of unfillable dame will still in 99.999% of cases be less than C.
Any sort of forcing scoring relies on the automated scoring’s life and death detection being good enough. Both for dead stone removal as part of condition 2, and just for the scoring of the game itself. OGS’s could definitely be better, but might be okay in most cases now, and potentially one could try hooking into the KataGo engine used for analysis for an improved version (although it’s not trivial even with katago, given the possiblity of japanese rules needing future protective moves, and that katago’s estimates can sometimes be a little jittery).
Is the post you link specifically to do with blitz games or?
I can’t imagine losing by timing out if all my opponent is doing is passing. It’d be annoying sure because it’ll keep going in and out of counting.
I don’t think a force count is going to solve trolling. If I really wanted to troll someone or stall a tournament I could play on the first line every move for 350 moves even if the game force counts at that point. That and I’ll use my available time ~30 secs per move say in some settings.
I don’t think trying to get rid of one niche type of trolling by ruining other perfectly good games is acceptable. (Maybe the word forced in forced counting is the problem, presumably you can’t just resume the game if agreed by both players?)
I like @hexahedron’s proposal a lot, especially if we let C equal the number of points in seki, and use KataGo to spot those (scoring with KataGo is on the planning anyways). In other words, a scoring can be forced if a player passes multiple consecutive times (say 5 times) and all dame are filled up.
Endgame trolling is not very niche… It happens a lot, often with people who are playing normally all game long, but can’t accept their loss at the end, or became annoyed for some reason or another. Most of these are not the kind of trolls that would play only first line moves for the entire game, but just sore losers.
And naturally force counting is meant to reduce the number of troll games, not eliminate them entirely.
It doesn’t even solve endgame trolling, since an opponent can just play a ko threat every couple of moves in between the troll plays, breaking the consecutive passing.
It could also be considered to put forced games automatically on a review list, so the trolls or users abusing the feature will remain on the radar.
I’m not sure you can let C “equal” the number of points in seki, that defeats the whole point. If in a 9 stone handicap white plays one stone, and then the game ends and neither player disputes anything, then technically at that point the entire board is seki, so there will be 351 points of dame in the seki. (There are 351 points in regions that border groups of both colors, following dead stone removal).
The idea is precisely to say that the number of dame points in “sekis” is small - it seems like you need to actually pick a number though, you don’t get anywhere by saying that the number of dame points in sekis must be <= the number of dame points in sekis.
Thinking about it, with the second condition in place, if it’s working well, you could reduce the number of consecutive passes needed by the player forcing scoring to just 2 or 3 perhaps, if you also specify that there must have been no captures during that time either. It’s hard for me to image a border-complete situation where you can pass several times in a row that shouldn’t be scored, except for certain kinds of consecutive ko chains (the opponent is capturing into some ko chains, and you have no threats or anything else to do, so you pass, then they take the next one, and you can only pass, etc). Or certain kinds of seki cleanup (capture a stone in involved in a seki before passing, to score it as a capture). Requiring no captures during that time takes care of those cases.
This is very categorical. I’m saying dragging a game out beyond 350 moves which is what the original proposal was supposed to address is niche. I don’t think having a cap on the number of moves in a cap really solves as many types of trolling as is claimed and does have the potential to disrupt or ruin perfectly reasonable games.
So I don’t think forced counting should be based solely on something arbitrary like move number. I also agree @hexahedron’s post/idea could have something to it. Some reasonable conditions are met that would allow someone to force score a game that could only be prolonged by trolling
We could define seki in such a way that any move played on any of the points in seki by either player will result in a large drop of points for them, that rules out your first example of something that is obviously not a seki.
You might have to propose an actual definition and how it would be practical. I gave a simple programmatic definition: “maximal connected empty regions that touch groups of BOTH colors” after the already-OGS-implemented dead stone removal, but it sounds like you want a different one?
Keep in mind that “all locations that drop more than X points if played” is not obviously a practical criterion - you have no guarantee that KataGo or any other analysis engine even sends playouts down that move in an analysis to say how many points it may or may not lose, so you might need to do many extra queries in the worst case, and you’re also at the mercy of the fact that the score estimates are not reliable if KataGo wants to try a sequence that is predicted to create some chances to win at the cost of some points (e.g., a last ditch attempt to exploit unclear endgame aji).
Think of the normal users, you guys. For normal users this functionality should have an obvious place and an obvious way to use it.
Call moderator is very good in this sense. User has a very clear instruction: don’t ignore this behavior, call moderator and wait in game or resign the game and have moderator annul it.
Your force count should also be very easy to find and have intuitive description. Imagine someone asking how do they force count and we’re like: “you see, what you need to do is”
That’s overcomplicated. And relying on automated scoring for anything is not a good idea in the first place.
For example, with 350 move rule the intuitive implementation can be easy. As soon as the game reached the mark (and one of the players passed a few times maybe?) you get a pop up “Would you like to force counting?”. I can understand that.
Ummmm… the part you quoted is not something the user needs to understand or do, that’s the algorithm in the background. As soon as both criteria are satisfied, you could show a pop up to the user exactly the same way you propose at move 350.
Indeed, if you wanted, you could combine the two - have the popup at move 350 + a few consecutive passes, and have the popup whenever the two criteria I mentioned are satisfied, whichever happens first. The user experience would be exactly the same as in your move 350 proposal, except that almost all of the time that the game is well and truly done much before move 350, then the popup would show much before move 350 as well, allowing the user to end a game that the opponent refuses to finish, but without having to continue dragging out the game many more moves all the way to move 350 before the popup finally appears.
(The one caveat being final dame filling - maybe it’s possible to eliminate that requirement while still catching the case where an actual border is incomplete. Requires a bit of thought for how to do it right, but probably the condition can be refined to do this).
First of all, this means you’re gonna run the algorithm each move, that’s bad.
And I don’t trust the scoring algorithm to correctly see the game is ended. I can’t know if it’s good or bad yet, of course. But considering how it scores games sometimes, I’d like it to be moved past 350 moves mark (to be safe from it otherwise).
Also, sometimes people pass and resume multiple times when disagreeing about the score in seki. You definitely wouldn’t want to auto-score that.
I think we should follow foxy.
Once 350 mark is passed and the victim passed once already. If they’re gonna pass again the pass is sent to the server but they get a pop-up offering to do something about it.
If they agree there’re three cases:
Auto-score by terrible scoring method.
Auto-score by Katago.
Freeze the game and call moderator automatically letting humans decide what happens.