Please see this game: damisul vs. 이승배
In final position we have a seki, where white has one eye, and black has a false-eye, but black cannot do approach move due to damedzumari. As far as I know, in japanese rules there are no points in seki, but OGS counts one point for eye for white, and doesn’t count black’s false eye.
In this case it does not affects result of the game, but it still looks like a bug to me.
I forked the game (though I didn’t adjust komi for captures - I should try that next) and the autoscore did this
But when I estimate the score it sometimes looks like this
The first image is ok, but it would be better if neither of the eyes were marked as territory.
it’s debatable, is it a false eye if it can’t be falsified 
As in if it can never be forced to be filled in, is it false?
But yeah aside from that it would be better to have no points in seki.
I’ll try fork the game adjusting komi, in case that changes anything.
It’s the same result with the site autoscore with adjusted komi, A1 is a point for black.
It’s possible one of the players deselected A1, by accident or by some misunderstanding, since E6 should also have been deselected.
I think the result would still be correct with the current autoscore behaviour, but it would be better, especially in other cases with dead stones in a seki, where there is no territory in seki as per Japanese rules.
