Imagine there was some central organization that handles matters of rating. A Federal Reserve of Go or The Bank of Go. We hear them make periodic announcements regarding “interest” to combat inflation / deflation / compression. They could set dan strength to be relative to the top 10% of players in a time frame. If human-like play gets figured or quantitative measures found, bots could be used to sample & analyze set ranks. When it comes to server and federation comparisons, there may be tools to interpret the various ratings & systems. Such a funny notion.
7 Likes
Completely decouple the ratings, like lichess does
1 Like
This small clip comes to mind:
4 Likes
For sure. But I think it’s ok we get some working rating by aggregating all those different kind of games, quite a miracle!
I’m not sure that having multiple ratings would be much of an improvement. At first would we have enough games in each category?
Also we should not forget that by definition a 1 Dan can win games against the strongest players in the world with 9 stones handicap.
2 Likes
In a positive stance, OGS has a very specific base of players originating from correspondence games. Still many many games running here and this is a major characteristic that other servers are missing.
This may make at times the rating a bit more hard at least in the OGS history; I remember some years ago how the rating was perceived by many players
This aspect is in some way a nice ground to fix and stabilize a rating but I suspect that this could have been a factor to get some underrating. Now OGS seems to care about this and is keeping an eye to improve the whole and get the best for all the worlds.
I had this thought last night. For those who do play irl, bringing up OGS rank as a reference point seems like the natural first step. If that’s the sole reference point to go by, that’s “better than nothing”. To me, that may incur a potential problem. If the goal of OGS is to be within range of AGA + EGF, with folks bringing up their OGS rating, would that not in turn affect AGA and EGF?
Whatever issues happen on OGS could warp those “ecosystems”. The goal at the end of the day for AGA is pretty much finding equitable games for fun. If “today” OGS 15k ends up being “yesterday” 5k, that would be to chagrin of everyone involved, past and present. しょうがないですね (Such can not be helped). From coming across old tales, I get the feeling that large go scenes saw more irl tournaments “back in the day” (Hikaru no Go wave) so perhaps the data was better serving the goals of fun & competition.
I do think old dan could be weaker than today dan. I also think any system that is there for fun should not be hard focused on name barriers. Those saying “I want to earn my rank!” may not understand that they could have already broken past barriers, with everything except the designation by the system de jour. I did a reread of the “Korean gup” system page on SL, which I think this thread should also check out.
OGS is a Western server, so it makes sense to align dan level with AGA and EGF, although it would also make sense to align with KGS, which is slightly weaker.
Really? Is a Dan player someone who doesn’t make failure? A supra consistent player?
Ok it seems easy to criticize one’s own attempt to define what is a Dan player but seriously, is this definition even working?
Myself I see it in a more empirical way, like there is a step, a big one, more and less of the size of all what you have done before reaching it. That’s where the boundary between 1k and 1d should be for any ranking system.
It would be written loud and clear what a Dan is if we knew but we don’t and I 'm not expecting anyone to give the answer.
Historically, there was a very clear cut for players to be “dan ranking” during the Japanese Edo period - got admitted into one of the Go houses (not necessarily the 4 great houses, there were other smaller houses). And they had existing “exam” methods that is winning a game against active members of known strength within certain handicap stones. We can still see the surviving diploma indicating which player they won and by how many handicap stones in them. And since Go houses’ strong players would participate in castle games regularly, their strength was also roughly “calibrated” with each other within margins. And in a sense, this is still a legitimate way of getting an amateur ranking in CJKT associations, where a player can get “certified” by recommendation from professional players (just replacing houses with associations, and pros competition internationally to calibrate each other)
However, the majority of the amateur dan rankings in CJKT area are still sort of “different” than the kyu ranking, mostly due to the kyu rankings being generally regulated by each Go school or classes individually, where winning within top % Go tournaments is not really necessary to get you kyu ranks, but getting dan ranks would definitely needed (not going through the recommandations route). And as the dan rankings go, instead of a “guild-like” membership symbol, it becomes a cultural one.
2 Likes
I am saying that OGS could end up subverting the AGA (in terms of playing evaluation). The goal becomes “achieved” because AGA ends up turning into shadowing OGS instead of the other way around. I will highlight with an example.
Say AGA and OGS were restaurants in different countries. The countries shares a similar culture yeah. The goal for OGS is to mimic “the food back home”, going along with the available ingredients. While some ingredients are overlapped, the cooking methods and other ingredients may differ. When the denizens of one place visit another, they bring along their tastes and return home ever enriched too. OGS has the super popular among locals “California Roll” which isn’t a thing in AGA-land. While OGS has the noble goal of becoming more like AGA, AGA ends up becoming more like OGS. Why? Because the AGA comes to adopt the “California Roll” given the new tastes.
The question in mind is “did OGS become more AGA-like or did AGA become more OGS-like”?
I did come across these at the EGC2025 site.
Participants will also have the opportunity to take part in the International Go Qualification Exam (IGQE) on site at the Congress – free of charge .
What is IGQE?
IGQE is a standardized, AI-supported certification system that assesses your Go level across 19 internationally recognized ranks (G19 to G1). Successful participants will receive an official certificate in English.
at International Go Qualification Exam (IGQE) by HJJ - EGC 2025
5 Likes
Seems interesting and fun. I wonder if by next year they go further abroad.
I do think there is some folly in being pegged to the AGA + EGF 1D range. The AGA has minimum time settings that let a game qualify as a ranked game. What are they? A lot longer than OGS! AGA 10 minutes + 20 seconds vs OGS 5 minutes + 7 seconds.
Look at Track & Field. Compare running the 400 meter versus running the 1600 meter. The 400 is a sprint around the track once. The 1600 involves pacing and timing, going around the track four times. At competitions, a runner tends to favor one type of event than the others.
The utilization of time is a skill. Playing with less time favors intuition. Playing with more time favors reading. While 13 seconds may not make much of a difference in one move, that time that the players bank will be greater down the line. That’s a ratio of 2.857, almost triple. At 200 moves, the difference is 1300 seconds per player.
Wow… A real life version of bluetiles’ post!
4 Likes
I do try to be helpful if I can, even if I disagreed with the premise of the OP 
(not the 25kyu plus stuff though, that I agree should genuinely be improved)
The EGF is similar, though there’s different classes A through D modifiers on how much such a time setting contributes to the rating change. You can see it here
While I agree roughly that playing with different time settings requires different skills, like managing the clock etc, they’re all likely to be highly correlated to the persons Go knowledge.
The main difference between offline and online ratings, and you can see it in the chess world too, is that you can play huge numbers more rated games every day online, than there are in person tournament events that you can submit for offline official ratings.
Basically there’s a lot of scope to improve on short timescales that online ratings systems have to deal with, different time settings, different board sizes, but when you improve in one, you likely improve in the others, although maybe not perfectly proportionately.
You might get much better at life and death and endgame from playing a lot of 9x9 Go, but that won’t necessarily be a bad thing for 13x13 or 19x19. It won’t help with the opening sure, but it’s still an improvement, the way drilling Tsumego will also potentially lead to an improvement.
There is some element of a feedback loop between online and offline ranks, whether we peg OGS to AGA or EGF or not.
When people sign up to their first tournament, often the only rank they have is an online rank, and so that creates some correspondence between online and offline ranks as likely either that declared rank itself or a slightly modified version will end up in the database.
Similarly players that take breaks from tournaments but improve from playing a lot online, might declare a new rank, and again that can be based on some online rank like OGS. Beyond a certain level like 1 or 2dan maybe players do migrate to Fox and we have to adjust more for that for offline ranks.
Essentially though there will be some correspondence between them anyway, and if it’s likely a player will use their OGS rank for real life tournaments in a number of situations, why not try to align them at least a little bit better?
2 Likes
Does it really work that way? If, say, a player registers as 5k and loses all their games it is the rank at the end of the tournament that matters, not the one at the start. Also, in databases like the EGD both the declared rank, final rank, and points are recorded; and I suppose after playing their first tournament they will have at least a first indicator. I am not sure if that is only applicable to MacMahon tournaments, and as always, I may have got some or all of it wrong.
I agree. And again I take the opportunity to say that to me it all feels weird. If I stick to playing correspondence for a while for whatever reason my overall rating climbs to 8k. If I practice live skills a lot my rating drops to 12k. Trying to play reasonable even matches becomes more difficult than it already is. And the obvious ‘solution’, creating multiple accounts is just ridiculous, at least that’s how I feel about it. It has taken me enough time and effort to build a solid account with a ‘reputation’ of sorts. One look at my game history suffices to see what kind of player I am. I really think it would be better for everybody to at least split correspondence- and live games ratings.
1 Like
HJJ Schools is promoting their system quite heavily, and this is part of their effort for international recognition.
Here is their “international cooperation” so far
- Taiwan Go Association
- Hong Kong Mind Sports for All Association
- San Francisco Go Club (NPO)
- Kuala Lumpur and Selangor Weiqi Association
- Philippines Go Association
Although Taiwan Go Association (台灣圍棋協會) was sort of a reboot of an old amateur association, and cannot be confused with 台灣棋院, which is for professional players. They only restarted their new diploma using IGQE since 2023, and last month their partner with a local association Taipei Sport Federation (臺北體育總會 which also issued their own diploma, still using the traditional dan/kyu system) that they mutually recognizing diplomas they issued, but currently only tournaments orgnized by these two would recognize their diploma if you want to participate (which currently only consist some tournmants in Taipei and one or two by the 台灣圍棋協會)
To join the majority of tournaments in Taiwan, you would still mostly need the diploma issued by the largest amateur Go associations 中華棋協 (which somethings also confusingly being translated to Taiwan Go Association, since R.O.C. is kinda not that “acceptable” or easily explainable internationally)
4 Likes
So if the strongest player (AI) got 1 stone stronger, people lose a rank? That doesn’t seem like the right definition your rank should be about your understanding of the game, not how far you are from the top.
1 Like
I think it’s about the strongest human player, for whom it is way harder to improve a whole stone (because improving the technology, i.e. the brain, is not an option). But still you’ve got a point.
And yet I think keeping the connection of ranks and handicap stones in mind helps a lot when comparing or designing rating systems.
1 Like