The history and evolution of kyu rank in Go game

In the discussion of the new OGS ranking adjustment, I dove into the history of how modern kyu rank was perceived in the past, and at the time this was the oldest documents I found in 1931

From this content with its relatively complete description, I was convinced the birth of modern kyu ranking must have an earlier origin. And then I looked into the Japanese wiki for Go ranking

There are some descriptions about the dan rank and kyu rank history but without any sources. It basically says the dan ranking has been established since the Edo period, but no kyu ranking existed at the time, and points the origin of the “kyu ranking” toward the Hoensha (方円社), established at 1879 (明治12年). It is generally considered the first modern Go association, and the precursor of the Nihonkiin.

The wiki description of the Hoensha has these two sentences about its “kyu ranking”

1883年(明治16年)、従来の段位制を廃し、級位制を採用する

また1893年には、級位制から元の段位制に復帰した。

And these create more confusion as to the origin of the modern kyu ranking. The first one says the Hoensha abolished the dan ranking in favor of the kyu ranking system in 1883, and the second one says it reverted back to the dan ranking in 1893, which doesn’t help us understand how kyu ranking was later returned, and they don’t have sources to back them.

So I went on to dig deeper into the history of Hoensha, and this is the first document I found.

It listed all of the members of the Hoensha in 1883 when they switch from dan ranking to a kyu-ranking only system. They mapped the ranks of 9d~1d to 1k~9k (9d to 1k, 8d to 2k, … 1d to 9k), and added the 10k to 12k ranks (which consist of about half of the total 91 members).

rank number of members
1k (9d) 0
2k (8d) 1
3k (7d) 1
4k (6d) 2
5k (5d) 7
6k (4d) 1
7k (3d) 7
8k (2d) 7
9k (1d) 17
10k 5
11k 13
12k 30

At the time, the 4 old Go houses still existed, but at much-reduced capacity because of Meiji Restoration. The next year 1884, the house of Hayashi was merged into the Honinbo House, and we have an updated member list of the Hoensha:


The total members increased to 188 and still used the 1k to 12k system, with some players from the defunct house of Hayashi who didn’t join the House of Honinbo.

rank number of members
1k (9d) 0
2k (8d) 1
3k (7d) 1
4k (6d) 2
5k (5d) 9
6k (4d) 1
7k (3d) 10
8k (2d) 14
9k (1d) 24
10k 21
11k 37
12k 68

And interestingly one of the 11k members was annotated as being 10k, and the locations listed where these players resided had spread across Japan (instead of mostly around Tokyo in the 1883 member list).

The next member list I dug up is in 1891 (right before the returning to dan ranking system in 1893). And the total number of members increased to 285





It still used the 1k to 12k ranking system, and a lot of these members are promoted within the Hoensha, through the precursor of the Insei system - 塾生制度 for young kids, including some famous Go players like Tamura Yasuhisa (later became Honinbo Shusai, 8k on this list) and Hayashi Fumiko (Kita Fumiko, 9k on this list, before she was married, the mother of woman Go players and the teacher of Sugiuchi Kazuko who is still alive and active today).

Next, I will look into what happened after the reversion back to the dan system after 1893

part 2 - the rank strength handicap in the Hoensha
part 3 - dan rank prior to the Hoensha at 1880, and transcribed Hoensha monthly games
part 4 - selected transcribed Hoesha monthly games continue
part 5 - selected 17 Hoesha monthly games from a shodan (Hoensha 9k) player during the 1880s
part 6 - one early known recorded game with a player weaker than shodan (Hoensha 9 kyu) in the 1880s - Hoensha 11 kyu (2k)
part 7 - Hoensha reverted back to the old dan ranking system in 1893 with three kyu levels of Shodan, and the revival of Honinbo house
part 8 - relationship graph of key persons involved in the creation of Hoensha
part 9 - connection to the late 19th century to early 20th century influential historic figures and upper class
part 10 - starting a journal into the late 19th to the early 20th century Go community world
part 11 - how local communities build from the bottom up and multitude ranking systems
part 12 - estimation of Go players using local member list compared to period sources

part ?? - the growing popularity of Go to common people and the increase of listed players
part ?? - The inverse of kyu strength order
part ?? - The establishment of Nihonkiin and the originally intended kyu ranking

To be continued…

30 Likes

Sorry to be completely off topic but I just wanted to say thank you for this and all the other amazing contributions you’ve made to the community here @claire_yang. I hope you know how much it is appreciated. Thank you.

20 Likes

Wow, interesting stuff!

Your posts really are great :D

4 Likes

I’d watch a PowerPoint presentation of this. :+1:

2 Likes

@teapoweredrobot @bugcat @Gia Thanks so much for all you guys. I’ll keep digging and get to the bottom of this history and evolution of kyu rank, and I already have some very interesting findings.

First I’d like to add documents describing how the Hoensha determined their 1k to 12k rank

This came from the original version “坐隠談叢 : 囲碁宝典” vol 1, by 安藤如意 (Ando Nyoi, 1866-1914, birth name 安藤豊次 Ando Toyoji), published in 1904. It is the first modern publication of Go history and a dictionary (the first addition has 5 volumes, and republished and organized in 1910 into 3 volumes). And there is a newer updated version base on it published in 1955 by 渡辺英夫 (Watanabe Hideo 1903-1998, Nihonkiin 8d professional and a Go historian)

Here it referenced a memo from the newpaper/magazine issue 46 published by Hoensha in 1883 and how their “kyu-ranking” system worked -

In the old dan ranking system, a Jozu (上手 equal to 7d) player would give a Shodan (初段 1d) 3 handicap stones in games as a qualification for becoming a dan player (from 9d to 1d, each rank difference is half a stone handicap). However, the old system was riddled with problems, such as 8d and 9d were very difficult to achieve and often stay empty, while many players not as strong as they should be were promoted to 1d (there was even a phrase for it 田舍初段, a shodan from the countryside, strong players in remote counties but much weaker than a normal shodan in well-developed regions like Tokyo), where most competitive players clump together between 2d and 5d (the ranking adjustment is always an issue even hundreds of years ago).

So they proposed to put the best possible player Meijin (名人) at the top using kyu 級 as rank name as 1 kyu, and old 1 dan as 9 kyu, and added 10 to 12 kyu ranks with a larger handicap gap of 1 stone per rank. That is a 1k player can give a 9k player 4 stone handicaps, a 10k player 5 handicaps, an 11k player 6 handicaps, and finally a 12k player 7 handicaps. And they would put this to practice only within the Hoensha (they also wanted to test and see how the new system worked) and would give players diplomas based on these new kyu ranks, which is a big deal since at the time only the four great Go houses can issue diplomas as one of their major source of income after the Meiji Restoration (when they lost their government stipends).

The publishing of newspapers and magazines was also a new source of revenue for Hoensha. Even the catalog of Go players would cost money to buy, and we can see on the member list with the price to purchase it, 定價金五錢 5 sen at the 1884 member list, and 定價金六錢 6 sen for the 1891 list (about the price of a book today).

Regarding the relationship to the evolution and history of kyu ranks, the 10k to 12k in the Hoensha indeed feels like the precursor of the modern kyu rank, with 1 stone rank difference and all that, but only have 3 ranks among them. It was probably designed to include and promote Go-playing for a wider audience to join Hoensha to compete with the traditional Go houses, and most likely a form of promotion system for their 塾生制度, the precursor of the Insei system for discovering young talented kids around the same time.

12 Likes

Need the pronunciation for it, intent to use as an insult IRL

2 Likes

Good luck with correctly pronouncing Mandarin, though you may actually achieve an even worse insult than you intended, as you do so :smiley:

Google translate’s version sounds right to me :wink: (I know no Mandarin :slight_smile: )

The translation is curious… if the pronunciation is as curious, you might have a good insult for sure!

Edit: Ah - doh, it’s Japanese of course. I imagine similar risk exists!

いなかしょだん
Inaka Shodan

Although I am not sure it consists as an insult though. More of a description to a fact.

2 Likes

Imagine using it out of context tho:

"- You missed the headshot, noob!
"- Mind your business, InAKa ShOdAn! (aka bumPKin shOdAn!)

Alas, @claire_yang is too civilized for this. :stuck_out_tongue:

3 Likes

I am very serious.

This phrase actually has a very delicate meaning in Japanese, and I believe it changes over time. It might not start with anything other than stating the difference between players of different origins.

2 Likes

From my understanding, the word originated in the era where the Great Go houses started to dominate the Edo Period Japan. They were not the only Go houses, just the most wealthy ones and in term attracted stronger players to join. There was a time when every Go house can issue their own diplomas, even the minor ones, since they were just a piece of paper to describe the in-house rank. But over time, when the Great Houses were the only one left to compete for the old Meijin title, and their strength compare to each other would be easily measured, they came to an agreement that every diploma they issued had to be signed by all the headmasters of the Great houses. However, the minor house still have their own ranks amount themselves not backed by any Great House headmaster. So players of these “houses from the countryside” were called 田舍. The strength of these players were not better or worse, just untested by the Great Houses.

4 Likes

I read a teensy bit about that because of the last “guess the stones” I posted, trying to find out what the Kansai-kiin was. Maybe it’s not related, though. Sorry for the OT.

At a time, probably during this period at late 19th century when the Great Go Houses started to lose their influence, this might start to describe players who have no “formal training” with less “elegant” play styles (fashionable joseki, fuseki, popular for the era). Since publications, newspapers, magazines containing up-to-date games would cost money to buy and more accessible for players living in the cities, the difference of strength start to be noticeable for players from the countryside (and before the widespread cheap printing press in the modern era, each great house would have its own different joseki and tesuji as trade secrets limited to members only)

5 Likes

About Go Association (Kansai is describing western part of the Japan main island, mostly around Osaka), the current Kansai kiin was just a branch of the Nihonkiin split from it. It issues their own diplomas, have their own pro players, and sort of different rank strength, measured amount themselves, play more games amount themselves, and generally speaking fewer stronger players compared to Nihon kiin.

However, the Kansai Go club (関西囲碁会) is a whole different matter. The book “坐隠談叢 : 囲碁宝典” I listed above - the first modern Go history book, was published by the original Kansai Go club (established in 1905 by this very author 安藤豊次), is a rich resource for historic records since the region was kept more intact compare to Tokyo region (which got bombed heavily during the end of WW2, and lots of valuable records were lost).

6 Likes

I’m fascinated to know/speculate about the relative strength of a given rank over time. I don’t suppose there are (m)any records of matches between, say, two 11kyus from that time.

I suppose a contemporary shodan would thrash their counterparts of that time. Which implies that the 11, 12kyus of that time are playing like, what 11, 12kuus now?!

3 Likes

Why do you suppose that?

2 Likes

Well the accumulation of knowledge, not least AI insights, that has occurred on the last 150 years. Just speculating really. And I don’t know if 1kyu in this historical system is equivalent to 9p now or 9d.

I had in mind that 1k then is like 9d now and so 9k/1d then is equivalent to 1d now. But 1d now knows all the AI moves etc so would surely beat a 1d from those times.

But let’s say it’s actually 9p and so shodan (9k in the Hoensha system) is about 1p or 7d (if I’ve understood the pro/amateur rank approximate equivalence).
So a 1d now is 6 stones weaker than that. So if humanity is more 6 stones stronger then we were 150 years ago then I’d still expect the contemporary (amateur) shodan to win.

Finally if we are comparing 1k Hoensha with 9d now then it’s something in between.

Trying to wrap my inexpert head round it but fundamentally assuming that the top players in the world have improved over time and so since ranks are tied to that as a standard in some way then any given numbered rank should be stronger than that same numbered rank in the past. The difficulty is in understanding which numbers are equivalent.

2 Likes

Well, sorry, but according to AI analysis of ancient games, they haven’t. There may be some advances in theory, but what really matters is strength in reading and positional judgement. If there is a difference, it is certainly not worth 6 stones. I think that most of even the top pros today doubt that they had a chance against the likes of Dosaku or Shuei.

9 Likes

That in itself is fascinating. Thank you.

3 Likes

I think as well, imagine the comparing average western 1dan or an amatuer 1dan in JCK that probably has a job or school and isn’t totally devoted to go vs a player studying in one of houses like Honinbo. I think at the very least one would have to compare insei and not just an average player.

I’m not sure if current insei play regularly in amateur events, so I’m not sure how rating equivalents would work.

Edit: I also am not sure myself what the equivalents between then and now are, but

that could be a decent idea.

3 Likes