Go Playing Population by Rating

[nazgand]

I am wondering how many Go Players on Earth are better than me at Go, but I have been unable to find data on how many players are of each rank. The closest thing I could find is that about 24 million people play Go.

Edit: I found the KGS rank histogram on Sensei's Library http://senseis.xmp.net/?KGSRankHistogram%2FData which says that 22965 out of 31063 are better than 9kyu KGS.

Japan's 2001 census https://web.archive.org/web/20021217041220/http://www.fin.ne.jp/~igo/census.htm says there are 24141920 people who play Go.

22965*24141920/31063=17848218 estimated people who are better than me.

I have a lot of work to do.

This list assumes that all the pro rankings are better than amateur rankings.

• Better than 9 kyu: 17848218
• Better than 8 kyu: 16630358
• Better than 7 kyu: 15155247
• Better than 6 kyu: 13534801
• Better than 5 kyu: 11899589
• Better than 4 kyu: 10139249
• Better than 3 kyu: 8299635
• Better than 2 kyu: 6390073
• Better than 1 kyu: 4621961
• Better than 1 dan: 3045816
• Better than 2 dan: 1874587
• Better than 3 dan: 1119156
• Better than 4 dan: 589888
• Better than 5 dan: 275903
• Better than 6 dan: 124350
• Better than 7 dan: 60620
• Better than 8 dan: 38859
• Better than 9 dan: 24870

[Bill Spight]

Based on your statistics, I guess that about 12 million people are better than you.

You're welcome.

[Chariot]

http://senseis.xmp.net/?RatingHistogramComparisons

There are many issues with this kind of data, but I'm sure it's the closest you're going to get.

[Bill Spight]

Edit: I found the KGS rank histogram on Sensei's Library http://senseis.xmp.net/?KGSRankHistogram%2FData which says that 22965 out of 31063 are better than 9kyu KGS.

Japan's 2001 census https://web.archive.org/web/20021217041220/http://www.fin.ne.jp/~igo/census.htm says there are 24141920 people who play Go.

22965*24141920/31063=17848218 estimated people who are better than me.

I have a lot of work to do.

Maybe so, but the real message is that there are a lot of people to play go with.

As for the number, 24141920, it is spuriously accurate. Besides that, taking a look at their estimate for America, 120,000, that doesn't sound like too bad an estimate to me. But how many of them play regularly online? Maybe 10,000, give or take? Most of that 120,000 play occasionally with friends and family. The average KGS ranking is around 2 kyu, but online players are much stronger, as a group, than casual players. If you are in the 25th - 30th percentile of online players, you are in a much higher percentile of all players.

[EdLee]

Hi nazgand,

Thank you.

It'd be nice to visualize those numbers.
( I'm unable to render charts and graphs at this very moment. )

[calantir]

The results of OGS's go server survey so far also provide some data on how rankings are distributed across different servers:

https://www.surveymonkey.com/results/SM-FQV9CTPC/

For example, if you're a KGS 9k, you're behind about 75% of KGS players.

[gowan]

I remember two people whose go rating put them at about 15k. They were enthusiastic, frequent tournament players whose ratings stayed at that level for decades without their becoming discouraged. They are good examples to show that, unless you are a professional playing go to make an income, your rating is of no importance. At any level you can find people to enjoy the game with. I know another enthusiastic sdk go player who said, seriously, that he didn't want to get stronger because if he became a dan level player he would not have as much fun playing his kyu level friends.

[ez4u]

The results of OGS's go server survey so far also provide some data on how rankings are distributed across different servers:

https://www.surveymonkey.com/results/SM-FQV9CTPC/

For example, if you're a KGS 9k, you're behind about 75% of KGS players.

...you're behind about 75% of KGS players who answer OGS surveys on surveymonkey.

Sorry, I couldn't resist.

[Uberdude]

The results of OGS's go server survey so far also provide some data on how rankings are distributed across different servers:

https://www.surveymonkey.com/results/SM-FQV9CTPC/

For example, if you're a KGS 9k, you're behind about 75% of KGS players.

...you're behind about 75% of KGS players who answer OGS surveys on surveymonkey.

Sorry, I couldn't resist.

I too had that temptation, but as the median kgs rank was 5k in 2010 (see http://senseis.xmp.net/?KGSRankHistogram), this survey result is probably not actually too far off the population result.

[HermanHiddema]

I am wondering how many Go Players on Earth are better than me at Go, but I have been unable to find data on how many players are of each rank. The closest thing I could find is that about 24 million people play Go.

Japan's 2001 census https://web.archive.org/web/20021217041220/http://www.fin.ne.jp/~igo/census.htm says there are 24141920 people who play Go.

This source does not list a methodology, so lets look at some of the actual numbers.

The census claims there are 40,000 players in Germany. The European Go Database (EGD) lists some 5,700 German players total, which means 5,700 people attended at least 1 rated event in the past 20 years. Other countries have similar discrepancies. E.g:

Russia: census 80,000 EGD 4,400
France: census 20,000 EGD 5,300
The UK: census 35,000 EGD 2,000
Holland: census 30,000 EGD 1,300
Ukraine: census 20,000 EGD 1,300
Romania: cencus 2,000 EGD 2,700

And those EGD numbers are from 20 years of gathering data. There's people in there that played 1 tournament as 20k in 1997 and never played again.

So most of the time, the census wildly overestimates the number of players, as compared to the EGD. Now I know that there are players who play only online, or only at their local club, but I have a hard time believing they represent over 80% of the active playing population. Rather, I suspect the census estimates the number of people that "knows how to play". That includes anyone who's ever sat through an explanation and played a few 9x9 games.

Suppose someone asked you: "Do you know how to play chess?" Would your answer be different from the one to the question "Do you play chess?" or the question "Are you an active chess player?"

Without knowing the methodology, it is hard to know how many of those 24 million are actually active go players with some basic grasp of tactics and strategy. It is entirely possible that the majority of those 24 million are weaker than 25 kyu! (and therefore weaker than you! )

[Kirby]

I remember two people whose go rating put them at about 15k. They were enthusiastic, frequent tournament players whose ratings stayed at that level for decades without their becoming discouraged. They are good examples to show that, unless you are a professional playing go to make an income, your rating is of no importance. At any level you can find people to enjoy the game with. I know another enthusiastic sdk go player who said, seriously, that he didn't want to get stronger because if he became a dan level player he would not have as much fun playing his kyu level friends.

I generally agree with this sentiment. However, I do feel that some satisfaction can be attained by working at something and showing noticeable improvement (e.g. through game results). You can still find enjoyment from the game when you're stuck at the same level, but I have to say it was a happy time when I used to see my rank go up.

I'm pretty much stuck at my current level. I still enjoy playing go, and it's fun. Studying is fun, too. But the enjoyment of ranking up, sadly, has been lost for some time now.

[Uberdude]

So most of the time, the census wildly overestimates the number of players, as compared to the EGD. Now I know that there are players who play only online, or only at their local club, but I have a hard time believing they represent over 80% of the active playing population. Rather, I suspect the census estimates the number of people that "knows how to play".

I don't actually find it so hard to believe over 80% of players in the UK aren't in the EGD, though obviously there is a wide spectrum of very active player to learnt the rules once. Just taking some personal anecdotes, I am in EGD and am what I would call a highly active player (78 tournaments in 10 years), though there are more active (e.g. Francis Roads with 268 tournaments in 20 years). I was introduced to go by a school friend who isn't in EGD. He goes through Go phases and sometimes goes to a club or plays on KGS, maybe 0 times one year and 10 times the next; he's about 10k. Another school friend is in EGD as I persuaded him to go to one tournament years ago, about 7k, used to play online and in local clubs, probably not played for a few years. Another is about 25k and only played with us, not at a club or online. Another about 25k and similar. So of those 5 who I would count in 'how many go players in the world? only 2 are in EGD. There are a few more friends who probably know the rules but I wouldn't call players. Also I taught my sister years ago and we played a fair few games.

Then at the university club each year maybe 50 people turn up to learn the rules. As the weeks pass numbers drop down to maybe a dozen regulars who turn up say every other week or more. Of these maybe half will go to the local tournament, and maybe half of those get more adventurous and travel to other tournaments and join the national go scene. Also a bunch of work colleagues played some go amongst themselves (the catalyst was me and a colleague who was previously one of those university club members who never went to a tournament, he's maybe 10k), so that's about 5 more players not in EGD.

So whilst the plural of anecdote is not data, for me being one active player in EGD and one friend in there with one game, there are about a dozen players who one could count as being 'affiliated' with me but not in EGD. I don't know how typical my situation is, and apportioning up those rolling university players is a bit tricky as they know other EGD players and you don't want to double count, but multiplying the EGD count by 5 or 10 doesn't seem too far-fetched to me (Herman's cited number is 17 times).

[HermanHiddema]

I don't actually find it so hard to believe over 80% of players in the UK aren't in the EGD

I don't either, but I'm more specifically talking about 80% of the "active playing population". I think we agree that for every player in the EGD there's probably at least ten that "learned the rules". I just wouldn't count most of those as active.

[sybob]

Interesting topic, many good posts.
At around 8k - 9k, I think you are in the top 15% (somewhere about).

I feel with you Nazgand, I sometimes also want to know where I am. I am a casual club player, (now) 8k on KGS and also around 8k at my local club.
I think it depends on the population, definitions and methodology, as already mentioned.

I played 9x9 using the GoQuest app lately. Do not think lightly of this seemingly simple app: many players, and many (very) strong players, most of them from Asian countries. It has a nice little feature too: you can see how you rank in their total population. I am somewhere around 15%. So, in my view, someone at 8k-9k ranks around 15% of total active players (otherwise they wouldn't have GoQuest installed). But then again: it does not allow 19x19 games. Still, I like to think I rank in the upper 15% of total both casual and experienced players. KGS is less 'reliable' in my view, because of their user base/population: many older and more experienced players, more sandbagging, multiple accounts per user, less popular among the casual and younger players, less players from Asia (population is skewed towards the higher regions, so to speak). FWIW. Just enjoy go.

[wineandgolover]

sybob,

I have played just a few games on GoQuest. I agree with you, and I like it a lot. But I wouldn't rely its ranking stats. After a few games, I am ranked somewhere below you, but still in the top third of its population, which demonstrates the problem. I don't yet have a stable rank, so I am ranked well below SDK. It seems that anybody who just tries the game gets included in the stats. But they could be 30k or 9d, and their rating will be relatively low. Until they play the recommended 40 games, their rating will be unreliable. Yet the stats are based on all accounts, not just those who've played 40 games.