Dan Rank Weights

Ok, so… First, i wanna clarify that this is just my gut-feeling about ranks, i don’t have any actual data to back me up, or inside knowledge on any of this, but…

I do feel like OGS’s ranks have deflated a bit since that last rating update in 2021. I used to be quite strongly hovering at 1-2k after the big adjustment, but nowdays i’m more like 3-4k. I believe there are a couple of separate reasons on why the ranks have been deflating:

  • strong ai’s which are 9d’s are so much stronger, that even the strongest amateurs cannot beat them with 2 or 3 handicap stones. This means that people who were considered as 6d or 7d are now more like 4d or 5d, which then means that people who are little bit weaker and who were once considered as 4-5d’s are now consider 2-3d etc. It all keeps just escalating down.
  • beginners learn and improve fast, and there is much larger flow of new beginners joining the server than people who have already strong. Once these beginners improve, they start winning games and thus draining the rating points from the rest of the player pool, meaning that players who are not actively improving will start slowly losing ranks to those who are actively getting stronger.
  • The overall level of go skills are increasing due better learning tools. So for us who are mostly stagnant, we move down. Its like the proverb: “if you’re not going forwards, you’re going backwards”, because everyone else around you are getting ahead of you.
  • We have people who are cheating with bots, again draining the rating points from the rest of the player pool.

In general, i feel like the consept of ranks is little bit outdated. The kyu/dan system used to make more sense back in the day when we didnt have computers calculating ratings after each match, so it was an easy/quick way of estimating whether two players were close to eachothers in terms of skill. And whether some sort of handicap might be in order to make the game more enjoyable for both of them.

But now, when we do have the ability to calculate ratings, using ranks is more like a relic of the past. We still use ranks because its a tradition and everyone loves traditions, but it does not mean that ranks are somehow superior than plain rating. In fact, for the purpose of matchmaking and such, we use ratings, which are then translated into ranks via complicated formula because that is what go players expect to see.

Chess players don’t carry the same tradition with ranks as go players, and they seem quite happy to just declare their strenths as ratings. I believe most of european tournaments also nowdays use EGF’s GoR (Go Rating) for matching people in tournaments, and they dont really care about proclaimed ranks either. This seems to work better than using kyu/dan ranks, altought many traditionalists dislike it.

IIRC anoek has said that revisiting the ogs’s ranks is on his to-do list, but i believe that is not the top priority right now as there are multiple more urgent things to do. But i’m sure that a proper analysis on rank deflation is coming once he can tenuki from the things he’s currently working on.

All that being said, i gotta add that i dont think the ranks here are deflated too much yet. While i don’t like to see my own rank dropping from 1-2k to 3-4k, i admit that its in line of what my rank is on EGF’s rating system. I’m currently 3k on ogs and 3k on EGF, so i feel like our system is matching quite well with european ranks.

So while ogs’s ranks maybe are not “between egf and aga” anymore, its still not that hugely off. It just means we are currently just closer to european ranks than north american ones. Some adjustment is probably needed to get us back to what was intented in 2021, but my gut-feeling is that it will only be a stone or two.

8 Likes

Plain rating does not tell how to compute handicap. Ranks are superior for matchmaking.

5 Likes

It is about humans, not about AI. The strongest player at the moment is Shin Jinseo.
If he is one stone stronger than for example Go Seigen, then everyone drops a rank if they don’t improve either. There is no special level of understanding which defines for example a 1 Dan.

I guess a 1 Dan in Europe from 40 years ago would today play as around 2 kyu.

2 Likes

That’s two different worlds. Chess never claimed to be a martial art.

And considering the quality of the relationships between players respectively in both worlds I don’t think chess has anything to teach to weiqi players.

Maybe. Maybe not. Maybe it’s the reverse, a 2k from 40 years ago would play as a 2 Dan now.

2 Likes

This IGEF is dreaming to take the place and authority of the national federations for delivering the grades of their players.

I find always astonishing when a small circle of pros come on the international scene to proclaim that their system is the best. Be some school in Germany, the go magic catalog, the IGEF no yet internationally recognized but pretending to be, the polgote hub of teachers… I know I’m not going to make me a lot of friends by telling this but that’s just how I feel and I may not be alone.

If any international standard could emerge one day, I hope to not be in the way they do but with more collaboration between the existing entities.

2 Likes

(cough) nice to see a thread back, this monkey of a feeling needs data to be corroborated. I do think that today’s kyu would outdo yesterday’s dan, assuming yesterday was over a decade ago.

1 Like

I am surprised about this statement. What makes you think so?

1 Like

Why are you surprised? I’ve no idea, no proof anything reliable to guess the level of a player 40 years ago compared to today. Furthermore that may vary immensely for each player.

It’s not about specific players, it’s about the average. How strong is today’s average 2k compared to the average 2k of 1985? I have no clue.

2 Likes

I played 1985 and I play today. That’s why I have a clue und my clue is the 2k of today can give the 2k of 1985 about 2 stones.

3 Likes

Not my clue. For what I remember myself of a 2k 40 years ago.

2 Likes

What do you suggest doing in the future or establishing a new site that would have ratings similar to chess ratings? I personally think that ranking on different sites doesn’t matter to anyone because it’s a matter of periods of years. For example, two years ago something changed in my approach to the game that I started to feel stronger. I beat a lot of people in my medan and felt 3-5 dan, but after two years I went back to what I was, which was 1-3 dan. Today I play to maintain a level of fitness in the game of Go. What I’m saying is that it doesn’t matter what a player’s level is as long as he plays in a balanced way.

I’m writing this here, but I opened this topic elsewhere in Forayim. I thought there was a site where AI could explain to us the mistakes we make in speech instead of just showing them to us. I’m addressing software developers.

Nah i’m not suggesting anything, that was more like my thoughts about ranks in general, and how ranks on ogs have changed in the last 4 years. And like i said, its only my gut-feeling that our ranks have deflated in recent years, i dont actually have any real data about this >___>

2 Likes