Counting bug in 9x9 Go

Hi, in this game: https://www.littlegolem.net/jsp/game/game.jsp?gid=2507210 only a single captured stone was counted. Probably the territories were to large for the algorithm to count them.

This game also was also mis-scored by Go Arbiter. The offered score of 2.5 seems to be correct, with a win by 2.5 points. In the chat, Go Arbiter declared the win was by 1.5 points, with no explanation. https://www.littlegolem.net/jsp/game/game.jsp?gid=2529234

#2507210 - It is apparently a software bug. The engine returns the point difference as a decimal value, and it should be rounded to 0.5, 1.5, but instead it’s being rounded to 0.5, 1.0, 1.5. So, for example, when the engine returns a difference of 0.8, it should be rounded to 0.5 and not to 1.0. In any case, we should probably switch to Chinese rules with komi 6 and the possibility of a draw.

#2529234 - The engine that calculates the score works by trying to play out the game and only then computing the score. So in this position, the computer as Black played one more move, g6, and that’s why it came out one point lower. This method doesn’t change the result, but it can change the final score. This issue also wouldn’t occur with Chinese scoring.

I am perfectly fine with Chinese counting as far as I do not have to do it myself. Computer counting is great.

I actually preferred the manual version back then, where we just had to mark the dead groups. I don't trust Go Arbiter. (It would help if at least its work is shown, for example if the dead stones are marked. It would be easier to spot mistakes.)

I think Chinese rules/Scoring (more properly known as "Area Scoring" and is used by the AGA) with a komi of 6 would be great!!!

Footnotes: A komi of 6.0 and with Area/Chinese/AGA**-scoring does give the possibility of a tie HOWEVER that is VERY unlikely/rare since there are an odd number of points on the Go board, and an even number of komi results in an odd number of total points available; and thus the ONLY way to get a tie is for a seki to occur that takes out an odd number of points (NOT even) AND for Black to get EXACTLY six more points of the remaining territory than White (so that when the komi is considered both sides have the same number of points).

** I believe New-Zealand rules also use Area-scoring - and I think they also allow for a tie, so they must allow a whole number of komi points.

oops, sorry about the big text. I think I accidentally made some html command/code by my use of asterisks and dashes... or something like that. Maybe that is what happened to Richard Malaschitz ★ at 2025-11-13 also?

And for those who are newish to the Game of Go, a seki rarely occurs, and when it does about half of the time an even number of points are removed from the game.... Furthermore it is rare for beginner games to end with scores with less than a point of difference. For Pro players it might not be unusual for them to win by 1/2 a point, but that is because they make very few mistakes (compared to non-pros ranked below 5 or 6 Dan) and the mistakes they make are often only one or two point mistakes (or even only a 1/2 point mistake). Most of the Go players here on LG (if not all of us) make many mistakes in every game, and they are at least 4-9 point mistakes (each) and thus for a game to end in a tie here on LG it would require a seki (which is rare) and for the two players to make mistakes that all add up to the same loss for both sides. I'd be willing to bet a cup of sake or tea that if we had a 6 point komi here on LG there would be less than two tie games in every five-thousand Go Games.