Please (re)instate ranks beyond 25k and 9d

OGS Development
1

So if your rank as converted from your ELO is lower than 25k or higher than 9d, the site will still display your rank as 25k or 9d, respectively. It used to be that people’s and bots’ displayed rank could exceed 9d if their ELO merited it, but not anymore. As far as I know, ranks lower than 25k have never been displayed by the site.

(Automatching and the ranking system are based on ELO rather than the displayed rank as converted from that, so for all practical purposes, the site DOES distinguish between the strengths of these players; it just doesn’t DISPLAY the differences in strength.)

In the case of ranks exceeding 9d, I think this is due to the old convention that 9d was the highest official rank that could be reached, but I don’t see the purpose of sticking to that convention. Why lump all the strongest players into one rank if there are clearly differences, especially since the site already used to do that?

In the case of ranks below 25k, it makes even less sense to me since I don’t think there was ever any convention that stated that 25k is the weakest possible rank. I think absolute novices to the game are probably around 40k, so there is a LOT of variation in ranks that isn’t displayed (but considered for practical purposes anyway in the form of ELO).

We can already see the variations in actual strength if we look at players’ ELO instead of rank, but I think it’s pointless not to also convert the ELO into the appropriate ranks beyond the arbitrary boundaries of 25k and 9d.

Who’s with me? :slight_smile:

8 Likes
2

Especially now that we know that there is a whole universe of skill above 9d - the concept that 9d is “the best you can get” is silly.

And it seems worth being able to aspire to and be recognised for being able to improve visibly further.

I’m not so sure that skills below 25k are differentiable. Unless you are deliberately trying to lose, at some point lack of skill becomes indistinguishable from random play.

I’m not sure it’s worth recognising any difference that could statistically be seen. I could almost make arguments in the opposite direction. I think it’s good for the game if novices are encouraged to actually learn something by being ranked equally all together at the bottom. It hardly takes any effort at all to get to 23-24k.

But on the topic of 9d plus I am really strongly in favour! It could be a small positive differentiator for OGS for those gurus: the opportunity to actually be rewarded for their studies!

6 Likes
3

When I started playing go, I was around 620 ELO (which I guesstimate equates to about 40k). It took me about a month to get to proper 25k, which is approximately 1000 ELO. I would have liked to see my rank progress within that month, but only my ELO progressed, making me wonder when I would ever reach 24k.

5 Likes
4

The formula site uses to calculate rank based on glicko is:

kyu=30-31.25*log(glicko/850)
dan=31.25*log(glicko/850)-29

dan=1-kyu, so 1 dan equals to 0 kyu and 2 dan is minus one kyu.

In this calculations glicko is first bounded between 100 and above 6000, though.

1 Like
5

That’s interesting to know. The converted ranks cover a range from roughly 1000 (25k) to 3000 (9d).

1 Like
6

I have a friend who has just started playing… I think she’s at about 700 elo and regularly gets slaughtered by 25 kyus that are supposedly the same rank as her… I agree it’d be nice if people could more reliably get a game with someone of a similar ability no matter how low or high that may be.

It’s easy to say “I went from 25k to 20k in a week so everyone else should too” but the reality is everyone has their own journey to walk and will find challenges at different places.

6 Likes
7

I’m suprised by the 25k thing. If that’s the case out there, then I withdrawn my argument, and just sit in resounding agreement with the proposal :slight_smile:

5 Likes
8

I used to think the same way as you GaJ since we had a similar experience of quickly moving from 25 to 20 but it’s been really rough watching my friend get absolutely destroyed by fellow 25 kyus (who in reality could all be 200-300 elo stronger which equates to 3-4 stones)

3 Likes
9

Never mind 25 to 20. I would have thought 25 to 24 is really a pretty basic step.

But - happy to take exprience of others closer to what’s going on with beginners than me.

I don’t really understand how the 25ks who are destroying other 25ks aren’t rocketing out of 25k.

I also don’t understand how these people arrived at 25k in the first place given that they start at 13k and move down to their natural level.

But … :man_shrugging:

1 Like
10

a 950 elo player will likely lose to a 1000 elo player and will more often than not destroy a 700 elo player… at the higher ranks you would see a 9k losing to an 8 kyu but destroying a 12 kyu and so you would say “yes, 9 kyu is a good rank for him” and he may well be stable at that rank if he balances out playing up or down…

but the same scenario played at the bottom has the same elo differences all being labelled as "25 kyu"
So basically a 26 kyu doesn’t “rocket” out of 25 kyu because he usually loses to true 25 kyus but at the same time the 26 kyu can comfortably defeat a 29 kyu, even though they all have the 25 kyu label

3 Likes
11

I still don’t understand the point of so called sandbagging, but it might be the case. And some players get to 19-22 kyu in first days of playing.

12

“Real” 25k is around ELO 1000. I think it’s easy to see how people around, say, ELO 850 could be destroying people around ELO 700 without coming even close to rocketing out of 25k.

Not really sure what you’re trying to say here. Their natural level might be 25k, or it might be 40k, but it’s always gonna be displayed as 25k. They all arrive there by losing a lot of games, except that the 40k will lose a lot more games than the true 25k (and has to play for a lot longer in order to reach 24k).

4 Likes
13

I was thinking that if someone is destroying 25ks then they aren’t 25k really.

Therefore I was questioning how their rank came to be 25k in the first place. Because at OGS we all start with 1500 glicko. If you are really better than 25k, you will never drop down that low.

25k is not something you have to “climb out of” at the beginning, it is something you fall down to if you can’t prevent it by winning a game against someone at all.

However, I realised I made one assumption that might be wrong: that once you are at 25k your glicko doesn’t get any less.

If it does, then that does seem unfair and likely to cause trouble in the way that you are saying.

2 Likes
14

Either that, or they’re destroying someone who isn’t really a 25k. :wink:

Unless you’re significantly weaker than ELO 1000, which as an absolute beginner, you are (around ELO 650 based on my personal experience).

It’s not unfair or likely to cause trouble, since for automatch and ranking purposes, the glicko rating is used. It’s just annoying that the true corresponding rank isn’t also displayed.

2 Likes
15

I think you don’t have a handle on how ranks work at OGS these days, @Sarah_Lisa.

When someone signs up, they are give 1500 glicko. Everyone gets that, with an uncertainty of something like +/- 6k if I recall correctly.

This means that absolute beginners never have to “climb up” to their actual rank, they get “beaten down” to it.

This is one of the reasons why I find it surprising that there would be a pool of people at 25k spanning the lower realms of ELO/glicko rating.

1 Like
16

GaJ, you’re the one that doesn’t have a handle on what we’re saying…

AFTER all the 25 kyus fall down to 25 kyu, they then need to start their climb…
for some, they’re right on the borderline and quickly find 24 kyu right around the corner…

but for those who fell to greater depths of “25 kyu” they have a very long climb to get to 24 kyu, after which they may find that extra ranks come a bit easier :wink:

4 Likes
17

I do realize that’s how it works :slight_smile:

Yes, people start at 1500 glicko and then get beaten down. The point is that absolute beginners get beaten down to way below ELO 1000, but still displayed as 25k even though they’re really around 40k. They then have to “climb up” until they ACTUALLY reach 25k (ELO 1000), which can take a long time (around a month in my case).

I surmise that you find this surprising because you joined OGS at a time where your rating was considerably better than that of someone who’s never played go before. :slight_smile: (That, or you’re a very quick learner)

3 Likes
18

Right - now we’re on the same page. As I said, I was assuming that your glicko never goes below the equivalent of 25k.

If it does, then everything you are saying makes sense.

The solutions would be either to not let that happen: make 25k/1000 ELO the absolute minimum, or reflect the lower ELOs correctly in rank.

… And I had a quick look and I see you are right, so I remain totally in favour of this proposal :smiley:

4 Likes
19

Yup… I explained this in the very first post :wink:

I don’t think that option 1 makes sense or is even possible, since there is a broad spectrum of strengths below ELO 1000, and it’s useful (unavoidable?) to consider that when automatching or ranking those players. So yep, thus the basis of my proposal to always reflect the ELO rating with its appropriate corresponding rank :slight_smile:

2 Likes
20

I am torn on reinstating 9d+ ranks… it seems confusing, since you ll find these ranks nowhere else. When the new ranking system came and ranks like 14d popped up, it was a laughing matter to me tbh :slight_smile:, and there never was a question that a 9d cap would be installed. But maybe its for players of that strenght to say, whether theyd see higher dan ranks as an improvement. I ve heard before, that tygem has quite a problem with the wide gap of strength inside the 9d rank.

On the other hand there seems absolutely nothing wrong with expanding the kyu ranks to 30kyu (eg) to make the climbing experience more enjoyable for beginners!

4 Likes