So I was scrolling on OGS games one day and saw a 9D player who apparently is 8 years old! And then, I scrolled down to the games list, and saw this! 1300+ of playing against the worst AI with 9 handicap! OGS ranking system has been exploited! I thought in the previous updates, OGS ranked games are only allowed if the 2 players have a difference in ranks less than 9?
Please post the link to the user profile
Sorry, but I would rather not publicly single out someone.
The profile is public. There is nothing wrong about. You are reporting a bug. Not an action of a person.
You have already provide enough information to search the user. High gain of rank for short period is enough to find the user. Please do not waste time to do search next time.
I do not see any illegal actions from the user. They canât beat ordinary kyu players.
I think you are right about strange behavior of rating system.
This is the player, I am sorry if this makes someone feels ashamed.
Why are you repeating after me?
Appologies, your message didnât load until I sent mine.
Please donât worry. It was a joke.
Really? You can do that? How?
Everybody can do it. Just check the history of every user. It is tedious but not impossible.
Like you mean you could find every 9Ds in the game?
Or you can just check ranked 9x9 game history for the bot Amaranthus on Aug. 07.
Ohh! Smart!
Nah, I found another AI issue. If you play against AI Bouvardia in 19x19 and give it 9 handicaps, it just straight up passes. 9x9 doesnât have the issue though.
It does seem like a bug that you can challenge a bot to a ranked game outside the +/- 9 rank difference
However, that aside, exploiting bot weaknesses to gain rank is simply âdisallowed behaviourâ and should be reported.
I agree that it is âdisallowed behaviourâ.
But I disagree about bug âreasonâ.
Rank should not grow fast for playing with weaker opponents.
What about ranked tournaments?
I know what youâre saying - a 5d defeating a 25k should gain essentially no rank.
BUT the game itself is âout of boundsâ. (On face value) it appears to be in a space that the ranking system isnât even designed to handle: a game where the players are more than 9 ranks apart.
I can speculate that itâs treated as if they were 9 ranks apart AND 9 handicap ⌠which results in an âeven matchâ from a ranking perspective.
So arguably if the existence of the game itself were fixed then the ranking system would not be exposed to this âout of bounds caseâ.
That being said, Iâm not an expert of the ranking system, maybe thereâs a reason why this game is allowed, in which case the rank calculation is clearly wrong.
Letâs @anoek at this point
I think it was the opposite, after a previous update this restriction was removed.
I think weâre a good bit on from that proposal now (allowing players to choose their entry level, and removing rank restrictions) that it could be interesting to see if there has been any positive or negative effects generally of those proposals.
It was an intentional change after a long thread of discussions. If that was the simple summary, surely it wouldnât have went ahead.
So it could just be an implementation bug?
^^ relevant GitHub issues, and within the second one
in case itâs useful for debugging.
There are at least two other users who exploited the same weakness. See LES BRETONS