Go server rank survey

General Go Discussion
21

Could someone post a link the results of the survey that doesn’t go through reddit please?
I have that site quite thoroughly blocked, in numerous ways, for productivity purposes :slight_smile:

22

Sure thing.
https://idex.github.io/go-rank-survey/go-survey-results

or just the image:

r5YIHZX.png2399×2379 638 KB

We are glad OGS passed through your diligent defenses :smiley:

2 Likes
23

Thanks Adam, much appreciated :slight_smile:

Haha, yeah, I seem to be able to keep my go playing at healthy levels but reddit has been an endless time sink for me in the past.

24

Yes, yes! Keep lying to yourself :smiley: :stuck_out_tongue:

2 Likes
25

I can’t figure how to read those charts. Could someone please help me?

3 Likes
26

Yes.

1 Like
27

The essence of the analysis is in the tables at the end: https://idex.github.io/go-rank-survey/go-survey-results#A-combined-table-of-mean-and-standard-deviation-of-ranks

2 Likes
28

Nice results!

EGF 6k = Japanese 1d?! The Japanese need to get their s*** together.

Go to https://idex.github.io/go-rank-survey/go-survey-results#A-table-of-rounded-mean-ranks for a straightforward rank table.

1 Like
29

Sorry for OT ;D

I would hazard a guess… that the large discrepancy (apart from being based on very limited data) is due to the fact that Japanese rating is usually club rating, and clubs, in my experience, are basically islands of elderly people playing only amongst themselves, whereas EGF rating is highly competitive in that only tournament games count toward the rating. These tournaments can, but usually aren’t that small so the player pool and therefore rating pool is subject to decent cross-pollination. A Berlin 1k is prooobably going to perform similarly to a Paris 1k, whereas “igo kaijo A’s” 2d might lose horribly to “igo kaijo B’s” 3k.

(if i’m mistaken, @Tokumoto, please correct me!)

4 Likes
30

Here are the factors contributing to the Japanese amateur ranks, especially around 1d, being so inflated.

  1. Nihon Kiin is controlled by professional Go players for the professional players.
  2. As such, credibility of amateur ranks is not a high priority for Nihon Kiin, but pro welfare is.
  3. Every Japanese professional player is entitled by Nihon Kiin (or, Kansai Kiin, etc.) to issue amateur rank certificates for a fee.
  4. As such, rank certs, mainly 1d certificate (because price x demand is the largest there), are source of pro income.

As a natural consequence, inflation and regional rank/strength disparity (the more the number of pros in the region, the weaker 1d’s are in the area) became ridiculously large, and most of the serious Go players in Japan are quite aware of this situation.

So kyu/dan system in Japan mostly has lost its meaning, and people rely on other methods, such as Points Rating System (https://senseis.xmp.net/?PointsRatingSystem) and its variations, to find good opponents to play. I was a bit surprised to find the Go Salon in Nihon Kiin HQ building using this Points Rating System several years ago thinking “Hey, this is like their own admission that kyu/dan system is not working”.

15 Likes
31

That makes so much sense. Thanks a lot for the detailed explanation and background info! :heart:

Next time I’ll play a game at the Nihon Kiin club for sure.

1 Like
32

The system in Japan described by Tokumoto seems a lot like the old papal indulgences in Europe during the middle ages. Done anything heretical lately? A word with the right cleric along with some cash could definitely smooth it over. I wonder if the Japanese pros hear confessions from amateurs. Bless me sensei etcetera.

4 Likes
33

Hi, 4k KGS here. Just played in Japanese amateur tournament held by Nihon-Kiin as shodan, with 2 wins and 3 losses. My impression is that they average 3k KGS, which probably means their ranks differ by 3 stones between online ranks.

34

Thank you for the info. Was the tournament held in Tokyo, or somewhere else?
As I stated, regional disparity is huge, and the info is more meaningful with location.

1 Like
35

Yes, I played in the Takara Shuzo Cup Tokyo preliminaries held on 12-13 May at Nihon-Kiin Headquarters.

1 Like
36

Does anyone know why the OGS-Japan and OGS-Tygem correlation plots look to be higher order fits than the rest of the plots?

37

First of all, correlation relies on linear relationships. Given the logarithmic tendencies of learning curves, it is probably wrong to assume that the difference between, say, 15k and 14k is the same as the difference between 6d and 5d. Second, when you’re comparing different servers’ ranks, you have to deal not only with both server’s internal (differently~) nonlinear rank progression but also circumstantial factors - different population, playstyle, time settings, rating algorithm, rank projection, etc.

2 Likes
38

It is almost two years since the last survey, should we do another survey to update it?

3 Likes
39

I vote yes. What do others think?
I have no clue how to change the name of the poll but,
What should we do?

  • Another Survey
  • Not Another Survey

0 voters

Sorry for the unoriginal names, but I didn’t know what names to use.

1 Like
40

As if this were a question… :smiley:

2 Likes