Dan Rank Weights

[Clossius1]

So i’ve been playing more games to prepare for congress. And I have to say OGS 2k 1k and low Dan are all really strong considering. I really think 1k OGS is worthy of Dan ranks. In my opinion, we should add 2 ranks to 2k and stronger i think.

What I consider Dan level is when every move makes a positive value to the game. Maybe not the most efficient but positive.

I think this is also important for OGS to do this for a couple reasons.

1: It will make me happier lol. I am 3D AGA and 6D Fox. I understand not wanting to make me 6D but making me 1-2D OGS is a pretty large difference imo. (I can pull off 2D but i fall to 1D a couple times a month )

2: Looking at Chess https://chess.stackexchange.com/questions/39078/fide-rating-distribution 2000, which i think is close to 1D is no where near the top 10%.So probably can bump the ranks of those close to or at the top 10%

3: I have heard some people will be using their OGS rank at congress because it is the only one they have. I am 3D AGA and have been know to lose to OGS 1k. So an OGS 1k would destroy an AGA 1k.

Would this be difficult to change @GreenAsJade @anoek

On another note, why does OGS not go down the 30k? The numbers go down to 100 but you have to improve 500 points to get out of 25k? I have watched new player streams and there seems to be a large difference between a weak 25k and a strong 25k. I really think a few more kyu ranks would create a better feeling experience for ranking up as a new player instead of feeling a hard grind in the very first rank in Go. They should be able to move 1 rank in their first day of playing if they learn a couple things.

Edit: I read this later and realized my mistakes with the graphs lol. But hopefully my opinion still makes sense.

[AnjaRubik]

I am 1d KGS then 1k OGS then 3-4d Fox.
I got KGS 1d before climbing OGS ladder.
OGS around 1k is very very broad in term of strength.
I often feel that 1d OGS or even 1k OGS could give me 2 stone and be just fine. They know MORE things than me, way more.
Compare that to KGS or Fox, their knowledge is about the same as me, and i very rarely get the feeling that they are clearly better.

[IRM]

I’m 1k on OGS, 4d on KGS and 7d on Fox. I only really play on fox nowadays and I specifically stopped playing on OGS because starting at 2k the player pool becomes really small, so finding a game takes a long time.

I think this is why 2k and above on OGS is so strong, people just don’t play enough to rank up but still play elsewhere and become stronger; I know some EGF 4d who are 1k on OGS.

[MrMTal]

Could there be an effect of AI cheaters sucking points out of the rating system? When an account gets suspended not all games are getting annulled so there’s a minus in rating points from the pool?

But what happens if a new account is created?

I’m sure somebody can chime in and explain how it works

[Dusk_Eagle]

I don’t find these reasons for OGS to change its ranks compelling.

1. If it would make you happier to be OGS 3d, earn it. I disagree that 1-2d in one system is a large difference from 3d in another system. Based on this anecdote, it seems OGS ranks actually align fairly well with EGF ranks at this level.

2. There doesn’t appear to be any basis for the assertion that 2000 in chess is equivalent to 1d in Go. To note, this is also being based on Lichess ratings, which are inflated compared to chess.com and FIDE ratings.

3. In past years at Congress, the AGA has applied an adjustment factor for first-time players based on what server they got their rank at. If OGS arbitrarily bumped their ranks to try and match AGA ranks, then EGF tournaments would have the same problem, but in reverse. It’s also worth mentioning that a similar problem exists with Fox ranks when no adjustment is applied, where players who are a certain dan rank on Fox register at that same level for AGA tournaments and get crushed.

[bluetiles]

I will say, I am sick of the “no dan players play on ogs because there are no dan players there” belief that pervades english speaking go. What if the measures aren’t there? What the heck is 1d anyway?

The need to make a rank have some significance brings along an unfortunate mindset. “A Tygem dan is weak”, “A Fox 3d is a 1d elsewhere”, “EGF 1d is two stones above AGA 1d”, “anyone could test & pay to be 1d Japan”, “KGS and IGS ranks are hard”, “Chinese 5d are six stones above other 5d” are also egregious, no matter the veracity.

FIDE isn’t a good source due to self selection. Those who in chess forums will mention this too. Same with other organizations and servers. The strife between Lichess and Chess dot C•m is pretty funny too.

An another example of a “system that is intended to bring along approximate level” gone awry is found in Magic the Gathering. In the EDH/Commander format (multiplayer, casual), the players had “power level” charts to gauge the strength of their decks before playing. There was not much in terms of specifics. So anyone could oversell, undersell, overestimate, underestimate, find mismatch table, etc when they communicated this info to the other players

The intent was to satisfy rule 0 of the format, that is to say, the idea of keeping fun via social contract. The format itself is ambitious since pretty much all cards (bar a few) from the game (30 years!) are allowed. Of course, this lead to issues and a lot of discussion for the past decade or so. The current paradigm introduced in 2025 by WOTC, the bracket system, is now a simpler stratification with defined list of problematic err gamechanger cards and conditions. While not ideal (the company is stepping in for what was a player led format), this is still an interesting try.

I do think OGS is awesome for having a glicko based system. The idea of being within the spread of AGA and EGF ranks & ratings near their 1k/1d is cool. Issues like rank drift and multi format blend ranks may come up alongside others.

There are some crazy avenues besides changing some formulas, anchors, standards, values, whatever. One could entertain the radical idea of dropping the tie to kyu-dan to be more chess like. This may be heretical to some. One could start using machines made to play go to analyze the players and assign some numbers too. One could also consider reverse komi systems.

[ronin3]

I think OP raises some obviously good points - maybe even too many for a single post!

I would like to add my own insignificant opinion:

• the lack of proper beginner ranks is very bad. Visually improving fast is essential to retaining new and overly-casual players. Not everyone comes in fully committed or with a grind mindset. I think this would deserve its own thread/feature request.
• I think a “true go players” (in the sense of “no true Scotsman”) mindset is still present - and poisoning - much discourse even on this very server. It would seem to me, we are amateurs, not professional or semi professional players, let alone Buddhist monks. Yet, some players have such intolerant and uncompromising stances as to what should constitute the real playing experience, also for others, and shame and ridicule others who dare “stray” from their perfect discipline and grind mindset.
• OP mentions how a Dan rank is so very different between servers, soon we might just call OGS rank Korean “gup” and call it a day. FWIW I would be in favour of the proposal.
• This ties in also with the “elitism” in go, which I’ve tried to put my modest efforts in trying to avoid and lower the gatekeeping barriers - like avoiding overloading beginners with a lot of Japanese nomenclature. It can add to the mystique of the game, for sure, but that’s beside the point I guess.

[bluetiles]

We are at the mercy of what the server has in place. The 2021 rank readjust on OGS was nuts. Many moved up, with ddk becoming sdk by more than five stones up. Imagine having quit go because of crazy the supposed ddk pool was and for how hard the number barriers were. Going to irl play with a rank like that would be hectic without supporting data.

I think KGS (yes K), and the potential consequences that have arisen from there need to be addressed. The perception of ranks there may have done untold damage due to the underlying system being flawed. Seems like some players on there don’t progress (reflected), don’t play elsewhere online (bc dan ranks + english), and don’t start over. If some section is always at XYZ rank, that causes indirect issues for AGA & EGF when they self and others self declare.

If they could shatter their belief, they may be surprised to find out their rank would be different after having found their footing. Something like a whole server reset, a trial experiment with a new system, or a change that drops history beyond a certain point (last few years as an example) could also be interesting.

[Animiral]

From the good old Empty Triangle Webcomic, accurate as ever.

[JethOrensin]

The idea of marking up the ranks to approximate some real-life measurement system like AGA or EGF might seem positive, but I think that it might be “a bit too much trouble, for no real gain”.

If someone has different ranks accross servers then, I’d say, good for them. It seems to me that this is a player that has gathered a diverse experience in different playstyles, opponents and setting, thus, gaining a broader perspective for the game.

If such a player would like to enter a real tournament, they are the ones that know their strenght better and they can make a real approximation on what the tournament might be balanced upon.

For example, in the upcoming tournament in Athens I registered as 6k. This happens to be my EGF rank (since I’ve only played in a few online tournaments), but it also happens to be my own approximation/estimation on my fair rank on a live real life game, even though I haven’t fallen below 4k correspondence games, in OGS or DGS, since 2018.

We can, individually, do those adjustments, so why would we ask for an online server to put in the work to reflect them? It seems that there are too many factors to calculate, to make it worth the trouble.

Last, but not least, on a more subjective level, as a 2k OGS, this suggestion would put me at 1dan.
I’ve never been 1dan before. Not even close. I’d like that accomplishment to be something I - eventually - put in the effort to achieve on my own merit and not by some clerical/logistical alchemy. It would turn a significant achievement to something really cheap and inconsequential.

[Aumpa]

It seems pretty reasonable that OGS, as the most popular western server, could try to split the difference between AGA and EGF ranks, the two western organizations that offer professional certification (AGA through NAGF). And… it seems it’s pretty close? Unless AGA and EGF ranks have drifted further apart in recent years? I’m not sure of the details, but it could be examined and an argument made around that.

The point about 25k being a difficult crab bucket to claw out of is more important. Beginners shouldn’t be stuck there so long while they steadily improve in knowledge and skill but struggle to rank up to 24k.

Why isn’t rank displayed down to 30k?

[shinuito]

@ronin3

• OP mentions how a Dan rank is so very different between servers, soon we might just call OGS rank Korean “gup” and call it a day. FWIW I would be in favour of the proposal.

I think if you make changes arbitrarily, like the “I’d be happier if..” argument in the OP, and there is a problem but you don’t address the root of the problem, this won’t fix it.

With the last rating update everyone got bumped up a few ranks, more so in the high kyu ranges, but that was to align better with EGF and AGA, particularly around 1kyu, but generally.

If you just shift the ranks randomly now, there’s no reason to expect people will be able to maintain their ranks.

@Clossius1

we should add 2 ranks to 2k and stronger i think.

Somebody bordering 2kyu will still occasionally beat clossius (if that’s what’s happening) even if we call him a 3dan, but now we’ve he’ll lose even more rating points.

Basically we’re likely to regress back to the status quo, unless there’s a significant fix that finds and addresses a root cause.

---

If sandbagging or AI cheating as @MrMTal is wondering, is causing issues like a rating drift, then maybe improving detection or policies around annulling games might help.

If it’s something to do with how we changed how we allow players to declare ranks for new accounts, then we could revisit that. I’d expect that to be an issue below dan more so, unless we’re getting a lot of 25kyu accounts beating dans regularly from new account sandbagging.

---

Basically there has to be more of a reason to make a change, to a) get the change done in the first place, or b) have it be an effective change.

Otherwise you might as well ask Fox to subtract 3 ranks from all their ranks below 9dan. Basically has the same effect.

[Groin]

Always the same question, who is first the chicken or the egg?

You can ask AGA to change their over rated rating.
Or ask EGF their under rated.

I find OGS to be pretty well situated between both, let’s keep it like this.

Beginners rating is a different matter, OGS could do better for sure. I’m mildly interested as ranking is always a bit too much a concern especially when you know so few of the game.

[Jon_Ko]

Ranks should reflect the handicap system. One rank difference should make a game even, where black gets one extra stone (and white gets ~7 points komi). The question then is where to anchor the system. Should top bots be around a 13d? Should 1d be close to 1d EGF/AGA?

[shinuito]

@Jon_Ko

Should 1d be close to 1d EGF/AGA?

That’s the current situation.

@anoek

We are attempting to align our low dan ranks to be comparable to the EGF and AGA low dan ranks. Currently we are projecting that a 1d OGS rank will be about .7 stones weaker than an EGF 1d and about 0.8 stones stronger than an AGA 1d. These numbers will be re-evaluated once the dust settles, but we believe we should end up fairly close to the goal of being within a stone of each system, on average.

Though I guess we should be well in the dust settling phase.

There’s also one more external effect that might be worth considering, which is that the average level of playing might be going up.

Nowadays you have easy access to ai reviews for every game you play. If you speak English there’s a wider availability of courses, videos, books than were available before. There’s still probably more than there was I guess in other languages too, with live streams and video content. Availability of lessons from strong players is potentially higher also.

So if we have players that have been a certain rank over a number of years, if you’ve been at 3 or 4 dan for the last 10 years, then maybe that rank is deflating, not because you’ve gotten weaker but because the player base is getting stronger.

I’ve been fairly slowly improving over the years, but I think if I stop actively trying to improve I probably won’t just stay at my current level, I’ll drop down, in EGF rank even, not even thinking about OGS necessarily.

[Jon_Ko]

@shinuito

That’s the current situation.

Is it? Or is it just supposed to be?

[triangle_fuseki]

---

---

---

---

---

---

---

---

---

---

---

---

---

---

Go Rankings Survey (Feb 2024)

That’s a really nice visualization. I’ve generated similar ones for all server/association pairs and uploaded them to the results table. [image]

[shinuito]

That’s the current intention. That was the last big rating tweak I remember. I don’t remember if we got data about how well it aligns since the dust was supposed to settle.

[ronin3]

I am just wondering, at this stage, with a slight amount of curiosity, but a lot more resignation: why are admins and site developers so hellbent on ignoring community feedback? This reply seems reluctant to even discuss the possibility that OP and others in the thread may have a point. People have come back with figures, and yet it seems to me that they are being disregarded almost entirely.

And just to add: one player may not have statistical significance, granted, but OP (if that is his verified account) is a well-regarded teacher in the community, and if that does not merit a little pause, I do not know what will.

[martin3141]

I’m not sure I even understand what is being suggested.

@Clossius1

In my opinion, we should add 2 ranks to 2k and stronger i think.

Sounds like a “increase ratings once” kinda deal, which - if applied to only some players, it is unfair and doesn’t change that much long-term, and - if it applies to all players, not sure if I get the point, since ratings are relative to each other, right?