In Japanese rules, you should capture dead stones in a seki before scoring, but they are still dead when you don’t. They are also outside territory, because there is no territory in seki under Japanese rules. So they are dead outside territory. Examples in the position below are M19, N19 and S12.
Also there can be a dead stone in a location that eventually needs to be filled when properly arranging the board for scoring IRL (also known as teire), so that location is not territory. An example in the position below is A8.
It is true that dead stones are already marked by transparency. I agree that adding a 2nd mark by reducing the size or adding a cross is redundant, but not adding a 2nd mark was not an option in anoek’s latest poll.
I haven’t tried it in production yet, but on the beta server I think there was no option to score the game anyway when a border was open. IIRC the only option available was to resume the game.
Both players assumed the black stone in the bottom right was dead, however we were given a warning that the pink squares needed to be filled. Even when the black stone was marked dead the pink squares and the warning remained.
It doesn’t seem to be possible to manually mark the pink squares as a particular colour. They did end up counting as white’s territory, but unless you manually counted, this was not clear until after the score was accepted.
I wonder what happens if black resumes the game and ends up winning a lost game because of the pink squares. For this game it seems like black won anyway so it didn’t really matter.
I was impressed with how well these updates had improved the autoscore handling of seki. However, it looks like the autoscore is still far from perfect at scoring seki: vanecho vs. kandukuriadityavikram21. This appears to be an autoscore mistake rather than cheating, because the score estimator also scores it wrongly.
