Auto-score improvements

[elsantodel90]

I know that there has been ongoing discussion for years and so the community must still be split on “what is the best thing to do”, ranging from “zero AI, no intervention at all before players end the game completely by themselves” to “maximum AI intervention, game is immediately scored after 2 passes” but there surely is some definition of “currently intended implementation that should be running now”. I have not been keeping track of the discussions, What does OGS currently attempt to do regarding auto-scoring?

This capture is from a very recently played official game from one of our Argentine online tournaments. The auto-scoring is showing something quite confusing, but is clearly advising the player to resume the game because something is going on / they passed early. The key detail going on seems to be that the group is unsettled and N19 would kill. Is it intended that this kind of thing is flagged to the players this way before the game has ended, or is this currently a bug @anoek ?

[anoek]

Yeah that’s a bit too much of a hint, I’ll see about working out how to address that.

[shinuito]

Also if you were to play out a variation and click estimate score, it seems like the estimator retains the stone markings and doesn’t take into account new dead groups unless they’re removed from the board,

It’ll still say W+9.6 approx.

and you can see the dead stone from the game still be marked in a variation

[square.defender]

@shinuito

estimator

"Estimate Score" counts score incorrectly on final board showing scoring

[Feijoa]

@anoek

too much of a hint

Seems worse than a “hint” - it’s seriously confused about something if it thinks that territory completely surrounded by live black stones is unsealed.

Are those decisions not being made based on the dead/alive states shown?