Yeah indeed it should be 7.5 under chinese rules o.o
Can you link the game?
For that custom komi warning thing, that happens whenever komi is not set to ‘automatic’ - so if the person who created the challenge had set it up as 7.5 manually there will be a warning. Also the game will be unranked in that case
My challenge was only marked as 5.5 komi while it was an open challenge. Now someone has accepted the challenge and the game has started, with correct 7.5 komi: a midsummer game's dream
Well, it’s possible they forced the komi to 7.5 manually because they had tried letting it on “Automatic” but noticed it was then marked as “5.5” when they created it.
Still, 7.5 is normal for Chinese, so the warning is unwarranted in that case, and frankly when I see an open invitation with a warning sign that says “Warning: Custom komi: 7.5” on a Chinese game, the apparent message is “this server has no idea what the usual rules of go are”.
When the server gives a warning that 7.5 komi is unusual for a game under Chinese rules, what I think is:
“this go server thinks the usual komi is 6.5 komi, and they added a helpful warning message for things like a unusual komi, but they didn’t realise that the correct komi is indeed 7.5 in the case of Chinese rules”.
I don’t think: “it must be a bug related to super short time settings or 100 points of komi”; because “7.5 komi” is not “super short time settings” nor “100 points of komi”, it’s just the normal komi for a game under Chinese rules.
I also don’t think “this server is idiotic”, I just think “this server isn’t aware of the usual rules of go”. Because that’s exactly what it looks like when it gives a warning saying that 7.5 komi is unusual.
But if there is a bug and it can be fixed, does it matter what this bug is actually related to?
I created this thread to signal the bug, and to point out that this bug makes the server look bad. I did not create the thread for you to twist my words or to judge me on my possible interpretations of the bug.
And you don’t think the obviousness means it’s likely a bug, but must somehow be a key misunderstanding of the server, or the developer, or the volunteer people contributing to the code?
It’s good to point out bugs, and certainly saying “X looks bad we should fix this faster” makes sense as an argument.
Still I think it’s a bit dramatic to say
I don’t think I need to twist your words, I’m just interpreting it how it reads to me.
Do you not think a Go server not being aware of the usual rules of go is a bit “idiotic” though? That it might be more likely that someone just didn’t take something into account or made a typo etc, given how many rules sets, time settings, customisable options are on offer?
I thought they said “here’s the reason we picked those values” referenced a Japanese handicapping system, and pointed out it been like that longer than someone could boot up Katago and claim that komi is wrong without really having a good idea why it’s wrong.
Similarly, I believe the reason not to change it so far is to change it for a good reason, backed by data, so that in future it doesn’t have to be changed again. Also to change the whole small board handicap system at the same as changing the non handicap one.
I do not know where that bug comes from. I think only the developers can know that.
I’ve only started using the “preferred settings” recently, and I remember that I used to set a “custom komi” of 7.5 myself too, when I started using OGS, because I had already encountered issues with the komi being wrongly displayed. So I’m guessing there are issues with more than just “preferred settings”.
And there is the additional issue of the “Warning: Custom komi: 7.5” icon and message that shows up when someone selects a 7.5 komi as a result of this first bug. So that’s at least two separate issues.
Similarly, I believe the reason not to change it so far is to change it for a good reason, backed by data, so that in future it doesn’t have to be changed again
I think this argument has been presented, but is a burden of proof fallacy disguised as false empiricism. No amount of raw data will be able to determine what the perfect komi is, there will always be room for disagreement of around 1.5 points. What is known is that
a. there is no evidence of fair komi decreasing with smaller boards (to the contrary, there is evidence it increases at 7x7 or 5x5, forgot which one)
b. area and territory scored games use different komis
c. AI suggests that fair komi is 7 for area scoring and 6 for territory scoring.
An intellectually honest person would accept that perfect komi is undetermined and would not impose a single komi standard across all rulesets. This is literally done “because it felt right at the time” without serious inquiry into other scoring systems and potential komis.
The one that should have the burden of evidence is the person forcing 5.5 across all rulesets uniformly while ignoring the different dynamics of counting in each of these rulesets. This honestly suggests stubbornness or fundamental misunderstanding of komi.
Also to change the whole small board handicap system at the same as changing the non handicap one.
This is a valid reason to delay the correction. Despite this, no official acknowledgement of the issue has been made despite repeated calls to fix.
anoek already intends to fix komi on small boards. The hold up and reason it’s not just changing a few numbers is that he wants to fix handicap on small boards too, and that decision is not quite so matter of fact. He wants the time and space to give that decision due consideration to avoid multiple changes and corrections.
Music to my ears – I think this is my first time reading this from an official OGS team member. I understand calibrating the handicap system may be complicated, so happy to have this wait if that is the reason for the hold up. I would love to play ranked 9x9 and 13x13 with New Zealand rules.