The opponent also needs to be aware of the 3-pass rule, and I think the vast majority of players on OGS will not be aware of it. So in most cases the opponent will just be wondering “Why is my opponent passing early?” and won’t look for ko threats.
3 Likes
Yes, your example in the other thread of the ignored first line hane seems very plausible.
Raising awareness of this 3 pass strategy/exploit was one reason for making this post, so I reject Leira’s characterisation of it as insincere clickbait spam.
I think it is an interesting variant, but I am critical that it is now the only game you can play on OGS rather than proper go.
4 Likes
Has this been rule been tested properly with high handicap games? If it’s fixed to be move 120, then for a 9 stone game, if white has only caught up only about half of the disadvantage by that point (because it’s only half the game), then it could be that 3 further moves still doesn’t catch up.
Maybe black would need lot of boldness to actually try such a thing, but it seems like a possible exploit if that’s the way it works.
9 Likes
I guess its easy to level up to a dan level just by playing handicap games only, by black only and passing after move 121
1 Like
This is a pretty interesting post, I didn’t realize you had to be 3 moves ahead to get it to work. The automod at fox seems to be superior. Of course, at fox it’s absolutely necessary, because the percentage of players that lose their mind when they lose and decide to waste as much of your time as they can is very high.
Stalling on OGS, especially among beginners, is extremely common. This new feature should greatly reduce that, although as this thread shows, the feature may need a tweak or two.
The difference with Fox is that Fox has a “request to count” button, i.e. you can tell the opponent “let’s finish the game” without actually passing.
I’m not sure I get the joke: for this to be a new Go variant, I think there would have to be a position which is losing in the old rule set but winning in the new rule set. But if both players understand the rules fully then all positions should have the same status either way.
Instead, I think it would be fair to characterize this as a new score evaluation system. That’s a bit less provocative because there are already half a dozen different score evaluation systems, several of which have been invented relatively recently, and at least one of which is pretty similar to this one. It is admittedly a bit annoying that these systems seem to be proliferating rather than stabilizing, but it’s less annoying than the stalling / score trolling problem that we had to put up with before.
3 Likes
This is true only IF KataGo is correct in its evaluation (usually, but not always, it can get complicated semeais etc wrong, particularly with the low playouts OGS presumably uses because doing 100,000 playouts for this feature would cost too much compute resource) AND IF both players play don’t make mistakes according to KataGo’s optimal line (much more likely with the fallible humans playing on OGS).
To take frolag’s example from the other post in this new variant passing 3 times made frolag lose. However, had they been playing Go and play continued then there is a non-zero chance his opponent failed to see the necessary teire once the dame were filled, and frolag could have won. I don’t like letting KataGo decide who wins unfinished games to prevent the human players making mistakes and getting a different result from who KataGo predicts would win if it took over and played both colours.
And it’s not just a different score evaluation system, it changes how you should play during the game: are you ahead enough and your opponent has small enough ko threats that you can afford to pass 3 times to try to get the win now and avoid yourself making mistakes later? What are the relative risks of you making that mistake later, vs missing some large ko threats so then you are losing after the passes? It’s a bit like backgammon’s double-or-quits, but on steroids and acid.
6 Likes
However, had they been playing Go and play continued then there is a non-zero chance his opponent failed to see the necessary teire once the dame were filled, and frolag could have won.
In Japanese rules, there is a nonzero chance that after all points are settled one of the players will accidentally fill in one of the eyes of a two-eyed group, or eliminate a key liberty in a seki. Nevertheless there must be an end condition for the game; in Japanese rules it’s when both players pass, in OGS rules it is sometimes when KataGo judges that the game is lost for one player. Reasonable people can disagree about which of these end conditions is superior, but obviously OGS’s system is no more a variant of Go than AGA rules, or Ing rules, and so on.
So which scoring system is better? In an ideal system we would use one of the existing rulesets verbatim, and as in a real tournament a human moderator would be available to resolve disputes. Evidently OGS is unable to keep enough human moderators on staff to achieve this, so an AI moderator is needed, and it is not clear that there is a substantially better approach to AI moderation than what OGS has chosen.
And it’s not just a different score evaluation system, it changes how you should play during the game
By this standard, Japanese rules is a Go variant: it disagrees with other rulesets on the status of certain groups, and unlike some other rulesets it allows for the possibility of a draw.
are you ahead enough and your opponent has small enough ko threats that you can afford to pass 3 times to try to get the win now and avoid yourself making mistakes later? What are the relative risks of you making that mistake later, vs missing some large ko threats so then you are losing after the passes?
Time controls have a much larger impact on this sort of risk analysis than this: with fast time controls you might be inclined to avoid complex fights if you think you’re ahead. Is blitz a Go variant?
6 Likes
we could probably make Tromp-Taylor the only ruleset, to prevent the non-zero chance that players could score incorrectly before all legal moves are played.
3 Likes
The 2 nonzeros here are quite different. 10% is a lot more than 0.0001%. Even dan players missing tiere actually happens. I’ve won and lost games because of it. Filling in your own eyes doesn’t beyond trolls and super noobs.
The idea this system cannot be improved is also wrong. Anoek had already made adjustments, and there are suggestions in other threads for further improvements like only applying when there are no valuable moves left to play to ensure it is only an anti stalling feature and not an anti not resigning when you are behind feature.
4 Likes
Who is arguing that the system cannot be improved?
10% is a lot more than 0.0001%. Even dan players missing tiere actually happens. I’ve won and lost games because of it.
I suppose I’ve seen examples of this too, but 1 in 10 games? That seems at least an order of magnitude too high to me. At any rate, my point remains: it’s just a different scoring system, not a different game.
The idea this system cannot be improved is also wrong.
I’m sure it can and will be improved, but it is likely that the existing system will be equivalent to any improved system in 99.9% of games, so I doubt there is room for a substantial improvement.
2 Likes
Not 1 in 10 of all games, 1 in 10 games where there are teire and the player who needs to defend is passing rather than defending immediately. Passing in this case could have the internal dialogue “I know I need to defend, but I am winning, why aren’t you resigning? passing is my way of asking you to resign” (kinda rude) or “I don’t see that I need to defend, why aren’t you passing so we can score” and then once the dame are filled they might or might not see the problem inside, or “I know I need to defend, but I’m not going to until you force me to”.
2 Likes
Yeah, i think teires and determining whats actually going on pose the biggest problems with the current, or any, implementation of this thing.
The main focus of that anti-stall feature are the situations where game is done, player A passes but player B plays a random stone which has neutral or negative effect on their score, A passes again and B plays another random some, and so on. That unfortunately happens quite a lot, most often in beginner-level games where players might be unsure whether game is done or not, but also some within the higher ranks when someone just refuses to end the match :<
The problem is making a clear distinction between that kind of kind of stalling, and the situations where player A passes too early and player B plays a normal move to continue the game. Espesially if the “normal move” is some sort of mistake according to ai’s analysis, which usually is the case since most moves played by humans are sub-optimal
Teires are extra tricky, because those create situations where player A can pass multiple times without losing any points, but in order to count the score correctly, player B should keep playing moves which have neutral value until A is forced to add a move inside their territory. If theres multiple dames which need to be filled before the aji in teire starts working, then filling those dames is the right thing to do for B.
But that anti-stalling system is still really new feature, i expect it to be tweaked more than once in the near future ^^
It’s often easier to find fault with any change and reject it than to give constructive criticism. Are we arguing for good features, or just for the sake of it? In the spirit of the former, let me identify one problematic aspect that so far noone has described:
The new feature does not self-identify as “anti-stalling”. Although announced and discussed as such, if you were just using the site and not the forums, you are not being told what it’s for. It is an empowering tool to help players stop their opponents stalling the game, that’s what it’s for.
Once you keep that in mind, the challenging questions resolve naturally.
“Has this been properly tested?” – There is no test because there is no proper scenario in which one player keeps playing while the other passes. At least one of the players is confused or malicious.
“Is this a new way to win?” – No. It is a new way to handle unsportsmanlike stalling.
“Can the feature be improved by triggering only when dame are filled?” – Depends. If stalling only occurs after all dame have been filled, then yes.
If there is anything to be improved, it would be the button text “Accept predicted winner and end game“. It is already very descriptive about the function of the button, but not the purpose.
All that in mind, is it that big of a deal? Curious or disaffected users can read the forums or ask around and find out.
5 Likes
I guess the internal dialogue that I’m more worried about is “I lost, I’m angry about it, and so I’m going to use whatever tools are at my disposal to take it out on my opponent by wasting their time and swindling them out of the win.” I encounter players with this internal dialogue on a weekly basis, and I’m grateful that the stewards of OGS are addressing it.
3 Likes
Link to the relevant thread
As I discuss in the other thread, two of the three forms of illegal stalling could be effectively addressed by directly blocking the stalling action, The remaining, problematic form of stalling is by self-atari moves. The new feature addresses that, but, as already noted, could be improved by delaying its triggering until all dame are filled, and by increasing the number of passes required.
2 Likes