I would hope so, but there should still be options to consider resuming, chat, call moderators, etc., which could take time.
Really? Wouldn’t this mean that OGS scoring system remains in this gray zone where users get hints? (even more than now, because the hints would always be correct)
I’d think that if OGS moves to full autoscoring, 2 consecutive passes (or perhaps 3 in some cases) is guaranteed to end the game then and there, and the algorithm determines the result. No resumption option should be given.
Is that how other servers do it?
I don’t see how “hints” would be a problem with a more conservative autoscore.
The current problem is that they might reveal a weakness by marking a few stones dead in a border, that a player could use to start a large invasion.
But if it’s just marking whole groups dead or alive in accordance with the rules of go, I think there’s nothing to be gained by resuming unless you want try to prove the autoscore wrong, which should be allowed.
Especially when the players leave an open border, I really think they should be allowed to fix it rather than end up with a giant dame permanently in the game record.
I can only say that I disagree with this statement. This is exactly the kind of gray zone behaviour that should be avoided IMO.
If you want to allow players to spot open borders before finializing the game, I think you should opt for a fully manual scoring system. No hints about group status, but display correct scores and territory markings while the users are marking dead stones (which would require the system to properly detect seki). Basically this comes down to the KGS scoring system.
Sounds good to me
If you’re comfortable with the idea of autoscore being an instant final ending to games, we don’t really need it to be any better. It’s already right about 99% of the time.
And in your manual scoring system, wouldn’t you get “hints” from the computer showing you how the border is still open, and have a chance to resume? I don’t see how the use of “AI” is changing that fact.
I think that fully manual scoring requires players to mark dead stones. The system should at least not give hints on group status.
I personally feel that showing the score while the players are updating their markings of dead stones is not giving too much of a hint, but if the platform disagrees and decides it is too much of a hint, they can decide to hide those numbers until both players accept the markings and finish the game. That means players can be surprised by an open border. I’d be OK with that too and remind myself to be extra careful on that platform (just as careful as I would need to be with fully automatic scoring).
AI will tell you the status of a group: alive, dead or not fixed.
Right, but I’m talking for the moment about how players should at least have the option to resume when there’s an open border. I thought gennan said
If we’re talking about AI revealing a group status, suppose for example it shows some black group as dead, but it’s some really subtle thing that the players don’t understand. So Black resumes and adds a stone to try to save it, and after a few moves White gives up fighting and it’s alive.
Sure, compared to manual scoring, Black maybe had to give up 1 point that ideally they wouldn’t have. But in comparison with forced autoscore where you don’t get to resume and Black loses the whole group, how is it worse?
Or is there some other scenario for “hints” that I’m not getting?
Good IMO:
- Beginner ends game prematurely with open border
- Game is scored with a load of dame, so different score to if they continued playing and closed, but correct score according to the rules of go at that terminal position
- Beginner feels sad
- Beginner learns about need to close borders
- Beginner plays another game
- Beginner closes border and is not such a beginner
- Next game beginner closes border and is not such a beginner
- Next game beginner closes border and is not such a beginner
- Next game beginner closes border and is not such a beginner
- Next game beginner closes border and is not such a beginner
- Next game beginner closes border and is not such a beginner
- Beginner is happy they now understand scoring.
- Ignorance in the world decreases, someone learnt how to play go, they could teach others
Bad IMO:
- Beginner ends game prematurely with open border
- Score estimator is used as a scoring tool so assumes continued skilled play and scores as though border closed, so not following the scoring rules of go, but what beginner thought it would kinda be like
- Beginner is happy
- Beginner thinks you don’t need to close borders
- Beginner plays another game.
- Beginner ends game prematurely with open border
- Score estimator is used as a scoring tool so assumes continued skilled play and scores as though border closed, so not following the scoring rules of go, but what beginner thought it would kinda be like
- Beginner is happy
- Beginner plays another game.
- Beginner ends game prematurely with open border
- Score estimator is used as a scoring tool so assumes continued skilled play and scores as though border closed, so not following the scoring rules of go, but what beginner thought it would kinda be like
- Beginner is happy
- Beginner continues in the delusion you don’t need to close territory
- Ignorance persists
- Beginner gets to 20 kyu
- Uberdude meets 20k who doesn’t know you need to close borders and is sad.
- Beginner teaches his friend you don’t need to close borders.
EDIT: I forgot the whole calling a mod who wastes their time annulling games in the bad story.
I think so. It’s not only about adding a stone to be alive. There are much more cases (for ex both players may think a group alive when it’s dead)
That was exactly my scenario above. Both players think it’s alive, but it’s dead in some subtle way. Or can you spell it out more? How would they use this “hint” other than disagreeing with it and trying to prove it wrong, arriving at a result more consistent with their skill levels in the process?
Well at times you can’t even add a stone to make it alive like in the scenario you suggest.
It’s more the situations somewhere in between dead/alive (or situations where both players are potentially wrong about a group’s status) where hints about the situation from the scoring system are undesirable, because it would provide outside assistence of the kind that’s not actually allowed by OGS (at least not in a ranked game).
Can you explain more? Assuming autoscore is just marking whole groups dead or alive properly as I think everyone wants, what kind of resumption would be undesirable?
- Players pass thinking a group is dead (alive)
- AI autoscore marks it alive (dead)
- Someone resumes to try to kill (save) it
- ???
- …result that is somehow worse than the AI result that both players disagreed with?
What is wrong with that scenario, is that the players resume the game based on new information provided by a 3rd party. The wrong thing about it is that it’s against the rules of the game that others intervene in the gameplay.
I don’t think dead or alive is so controversial (once the players are past the beginner stage). The issue is with small surrounded groups where an invasion might result in seki. Do you need to add a stone inside to defend against this, or can you leave it? If the autoscore shows a black group alive but with no territory, then you resume play, that means white can choose the next move knowing that an invasion is possible, and so they steal 8 points by finding a sequence of moves that they wouldn’t have found without the hint.
That’s why you want the autoscore to just mark the obvious stuff (any bunch of stones with spaces inside it is alive, and the spaces are territory), and not have it reading to determine the actual correct status. (And yes, I know that defining “the obvious stuff” in code ends up being not obvious at all…)
Obviously not obvious at all.
So two opposite choices only.