I agree. On the one hand, we recognize our responsibility to teach those who ask to learn from us, and on the other hand we recognize the value of software to take care of fools who, instead of wanting to learn, place stone after stone into atari just to see each one being captured, or insist on placing stones inside connected living groups where they have no room to create even one eye, much less two.
I doubt it’s always feasible to find people to help out with scoring. The only thing like that I could imagine would be a “let engine finish our game” option for free matches.
That said, I have observed newbies accepting completely wrong auto-scores at the end of a match because they didn’t finish the borders… that kind of check could be feasible. Does the autoscore-result match a (hidden) engine estimate? If no, … do something. 
@Vsotvep had, I think, a very good specific suggestion for an algorithm to clean up most discrepancies. I’d love to see someone try to implement that as a replacement for whatever we have now!
But I don’t know how you’d use AI to determine that people are playing useless moves in a lost game. How would you avoid cases like this?
3 Likes
I mean, black should resign right?
2 Likes
With 6.5 Komi, clearly one of the players should resign at move 1 
Mh, no, the ONLY automatic ‘you should resign’ feature I can envision would enforce a variation of something that exists in chess (quoting some guy on YouTube):
If you run out of time, and your opponent could’ve checkmated you in ANY possible way, you lose… however if your opponent couldn’t checkmate you in any possible way (he doesn’t have enough chess pieces to do so like a king and a knight) it’s a draw.
I’d say… if there is no sequence of legal moves (barring suicide / filling the 2nd to last eye?) that could possibly lead to player X winning, declare Y the victor.
Not sure if that’s easily doable, just saying that’s the only automated judgement tool I’d employ.
1 Like
I think finding the line between “legal” and “legal except suicidal” is difficult.
Example from a random game, if we consider the blue area on the right:
There are plenty of legal sequences that lead to that territory becoming Black’s, but IMO the only scenario where that happens is if white loses desire to live and commits suicide.
2 Likes
Again - if that can’t be done, I simply wouldn’t allow engine judgements.
Defending against nonsense is part of the game; as much as it’s annoying, it’s defensible when newbies are playing.
2 Likes
Stalling by self-atari moves is reportable. A mod will end the game if that is happening.
1 Like
Are we still talking about full beginners?
I have some compassion when this happens. One cannot be guilty when understanding almost nothing, and still have a try to win or to feel deeper how lost he is.
Then i will try myself to communicate, won’t call a mod for that, sorry.
2 Likes
Same here. Nothing wrong with calling a mod if one prefers, though; they’re not going to ban someone for being new to the game.
4 Likes
Groin, I agree with you fully. I would never call a moderator because of the ignorance of a beginner. That might make them give up go completely instead of want to learn it better.
Computers can help us get things done. So we should use them. In the case of beginners who do not understand how a game ends, we should have OGS help explain this to them. This helps both players.
1 Like
The purpose of calling a mod is (1) to decide the game, and (2) kindly to instruct the beginner on game ethics and site rules. The mods are not truncheon-bearing secret police.
4 Likes
Sure, but you know you are crawling in a cloud, asking yourself what are you doing here playing an obscure game, and suddenly, even with a big smile, a moderator come to check what is happening… Not to say they are doing a bad job ofc. I admire our mods team.
1 Like
I think the discussion is getting mixed into multiple topics, but my two cents:
I agree with @stone.defender general principles for the SE:
- it should provide mechanical counting to eliminate manual tabulation
- it should provide an intuitive visualization so we understand how the result comes about
- it should not leave vast swaths of neutral territory
- it should not apply Go AI which could give hints as to life/death
- Easiest way to accomplish the above is make users mark any and all dead stones (with obvious instructions), the algorithm either recomputes or continues from there
- (my opinion) it should account for the rules set when tabulating
- To @GreenAsJade point regarding “placing a few stones” in the current SE; I find that in the mid game it can take a lot (20+) stones to close off territory, which is very tedious
Regarding @david265 AI resignation suggestion… I’m not a fan myself. Any level of AI that can handle the ladder examples given above will likely give away knowledge of life/death of groups that DDKs might not see. Also unpopular opinion, but I think ‘useless’ end game invasions are not always useless at DDK levels. Obviously there is trolling potential, but what might be obviously futile to stronger players and AI may not be obvious at DDK, and DDKs must also learn how to kill an invasion. I’ve certainly seen invasions the should not work, survive, and that’s a valid victory for the attacker.
7 Likes
You typically don’t have to place more than a few to persuade it to fill the way you picture.
1 Like
yes, but looks like there still should be limit of distance of influence of stone
on AI estimation its w+42,5
without limit, its b+8,5 on influence estimator
with (max diameter=11) , it shows correct winner w+1,5
it doesn’t makes much sense to paint entire territory as black when there is so huge distance between black stones
1 Like
Soo @stone.defender when you putting your estimator on OGS? 
Manhattan score would crush the current estimate on boards like this 
2 Likes
2 Likes