It is possible for a 14k player to be ranked as 23k by OGS. When I challenge that player with automatic handicap, he gets 2 handicap stones, even though he is 2k sronger than I am!!
This is a serious bug, in my opinion.
Even if he sets up a challenge after that, he still gets 2 handicap stones!
Example is player SUNDJER (I am player david265). An example game is https://online-go.com/game/6249237, which is 9x9 with 10 minute timeout.
SUNDJER tells me that other players have told him they have the same problem. He didn’t know where to report it. Perhaps it has never been reported before, but it should get fixed.
This isn’t really a bug the system is working as attended, Ogs has 4 separate ranks Overall, Blitz, Live and correspondence. His rank was being shown as 23kyu because that is his blitz ranking and you two we’re playing a blitz game it seems.
As for the handicap stones its because his blitz rating is a lower rank so if the color is set to automatic the lower rank will get the stones in this case the 23kyu.
I must disagree with you that this is not a bug. If the system works this way, it should be fixed.
As I aleady said, a player with a rank two stones ahead should not get a handicap.
I don’t mind if a dozen ranks are internally maintained, representing a dozen different kind of games. But this design decistion must be hidden from players, because in go one actually has only a single ranking.
If you are the developer here, you do not seem to understand the concept of finding and fixing a bug. Go does not work the way OGS has implemented its ranking system. A 14k player should not be given a 23k rank for one type of game. They should not have two handicap stones when the opponent might need a handicap stone.
The example game is not a blitz game. It is a normal game on a 9x9 board.
It’s not a bug, it works the way it’s intended. And as @matburt mentioned several times, it’s just an experiment that will probably change in the future (although the blitz/live/corr distinction will still be showed in one way or another).
Ranking only makes sense in the context of a ranking system, and as @HowToPlay pointed out, blitz rank is the one considered here for the purpose of handicap stones.
The time control in that game (10 minutes absolute) is in fact considered blitz. I know this is arguable on 9x9 though.
Again, that handicap is correct given the ranks being considered.
If you look at the ranking graph of Sundjer then it’s obvious the ranking system has a big flaw in it.
I think the OP has some point to this complain. Just a bit raw anger shines through
I should repeat, the way this ranking system works is experimental, and this criteria (choosing handicap stones by time control-specific rank instead of overall rank) may change in the future. As for now, this (seemingly buggy) behaviour is caused by players not playing that specific time control for a long time, and not because the handicap stones calculation is wrong.
I don’t believe I am angry about this issue. However, I am always quite firm with people who are not willing to call a bug a bug.
I am retired from a 40 year career in software engineering, having worked in companies big and small as a software engineer and computer scientist, and my professional experience is that letting go of ego and calling a bug a bug is fundamental to supporting any product well.
I believe the primary issue in the delay we see here in fixing this bug is psychological in nature. The early responses to my OP were needlessly defensive, without one shred of evidence that this bug is not a bug. Claiming that a design is experimental has nothing to do with seeing that a product is not performing correctly. Neither are the other excuses that were given above.
If my postings have helped anyone see that focusing on the topic is more helpful than defensiveness, I will not consider my time spent on this thread to have been wasted.
In my own career, I have seen this kind of psychology (sometimes called company politics), when unchecked, ruin projects and companies. The published research on software development also confirms that objective evaluation and review are indispensable to producing quality software. Mistakes are inevitable when developing software as complex as OGS. Trying to cover up mistakes by denying the existence of bugs is counter-productive. I’ve learned that recognizing and fixing bugs efficiently, without ego, and then moving on, is the best policy for any project or company.
I disagree with this statement. For example some players have fighting skills and will perform better on smaller boards or faster games. Others who understand how to use influence will perform better on larger boards.
SanDiego, Well of course my statement is wrong when taken out of context. The context here is a bug in OGS. In that context, the fact that someone could be thought to have “no fighting skills” just because they never played a blitz game while at OGS is a bug. As a first approximation, the rank one seems to have in one type of game can be taken as their initial initial rank in any other type.
That’s rather obvious, isn’t it? Really, why am I under attack here? It makes no sense. I’m simply reporting a bug.
Can we see the results of the experiment in numbers? For example which rating system predicts the game outcomes better (overall rating or time-specific rating)?
I’m pretty certain there’s a misunderstanding here. @mlopezviedma is correct that the system is working as intended. However—and this is a big “however”—the system is built on the admittedly flawed assumption that improvement in one category has absolutely no bearing on one’s skill in other categories. @david265 has drawn attention to a typical case where, due to the character of the ranking system, an adjustment was “necessary,” whatever that means. Now, we can debate whether this deficiency is or is not a “bug,” but that would be rearranging deck chairs on the Titanic. Let’s agree that an ideal ranking system would account for and automate this process. Then let us petition our hard-working devs to implement a code-level correction in the future.
That’s certainly not true for online go, people have different (sometimes significantly different) rankings on different servers in addition to their local federation ranking (if they have one). For example, there is at the moment a 3 stone difference between my OGS and DGS rankings, so It’s quite possible for me to play the same person here and on DGS giving a handicap in one instance and receiving in the other. Would you consider that some kind of world-wide bug?
Sure, it’s a bit unusual to see 3 separate rankings on the same server for the same game, but that’s not without precedents either.
[quote=“mark5000, post:17, topic:9018”]Let’s agree that an ideal ranking system would account for and automate this process. Then let us petition our hard-working devs to implement a code-level correction in the future.
[/quote]
IMO the current system is unnecessarily complex. When Nova and OGS were merged it was probably natural to see correspondence go and live go as completely different games, but I don’t really think they are that different. Unless the devs have some convincing statistics to support their experiment, I’d vote for unifying the ratings.
