It’s not wrong of you, but it really illustrates the issue caused by having a “smart” scoring tool…
We should not be in a situation where players are afraid of an AI intervention.
It’s not wrong of you, but it really illustrates the issue caused by having a “smart” scoring tool…
We should not be in a situation where players are afraid of an AI intervention.
Right. That is what the discussion is here: A compendium of OGS's terrible scoring system confusing beginners - #109 by Conrad_Melville
Note that the situation was that the AI was not helping, and it was scored according to the players’ expectations. So I don’t think that complaining about AI helps answer the question of whether it was right for me to leave the bottom scored as alive, given that I had some doubt.
How bout we all play Tromp Taylor, nobody has to worry about ending prematurely
Note that the situation was that the AI was not helping, and it was scored according to the players’ expectations.
Yes there was no issue in this instance, but my point remains that you were worried about it and have even developed a “tip” for players to avoid AI intervention. I’m not blaming you, but this is a serious flaw of the system in my view, we should not have to resort to tips…
whether it was right for me to leave the bottom scored as alive, given that I had some doubt.
Setting AI aside: I completely believe it was right of you. If both players agree the invasion failed then it did fail. You have no obligation to help your opponent by defending further or marking it as alive.
Perhaps a stronger player would have played differently, but this opponent did not see a continuation and it’s not your job to change their mind.
Another example of AI revealing weaknesses:
The scoring tool marked two seki points in the upper left, but that’s an invasion that I (6k) can’t read so I cleared them and accepted the loss by 0.5.
I guess it would have been pretty easy to figure out if I had taken the hint and resumed.
Bravo for your honesty! We had a long thread that I started on this subject last year (I don’t feel like looking it up). Basically, players are playing against KataGo (the autoscore) as well as against their opponent. There are cases of multi-moves tesujis used as the basis of scoring. In many cases players can use the info to change the result. I have collected additional examples since last year, but why bother. Nothing has been done.