Reliable way to start estimating no. of go players?

[John Fairbairn]

Many go players have an abiding interest in the number of fellow players. Estimates vary wildly and often have little or dubious factual basis, and then there is the knotty, gnarly and niggly problem of even defining a go player.

But chess has had the same problem - estimates there mostly vary between 200 million and 600 million, but it's also unclear whether these include xiangqi and shogi. However, as a possibly more reliable factual starting point, the chess.com website is now claiming over 20 million members (members, not accounts), with over a million active on a daily basis.

That's nowhere near even 200 million, of course, but it still seems impressive and offers opportunities for humans to continue to make money out of the game. And, as they say of their own site:

Of these, well over a million are active on a daily basis. While Chess.com is the most popular chess website and has the most popular chess app in the world (and the best in our opinion!), there are also many more millions who play chess regularly on other websites, apps, and over the board.

Does anybody have an equivalent way to assess the number of go players active online?

[Gomoto]

Just add the active online players from the major go servers

[hyperpape]

This may be because I have less familiarity with the asian go servers, but I think chess.com is way ahead in ways that draw in people. I tried to play chess a more this past year[0], and chess.com offers a lot of content outside of playing games. It was actually a bit annoying to get a daily email with interviews from grandmasters, offers of training, articles about strategy, but I also suspect it helps raise the number of daily users. Even if you can't make time for a game, maybe you'll read an article on your lunch/smoke break.

[0] Results: I am perhaps 1000 ELO: probably equivalent to sub 20 kyu in go. I am truly mediocre at games.

[gowan]

Of course we can't just add up the number of active players on online sites because people play on more than one server, e. g. KGC and Tygem. Not to mention the fact that people have multiple accounts on the same server, e.g. one for blitz and one for serious games. One meaningful number would be the total number of members of national go associations but, of course, many more people actually play go without participating in any organized aspects. Obviously, defining who players are, is another issue. Do you count someone who learned some version of the rules, perhaps played a few games with other similar beginners and hasn't played since. This vagueness applies to the traditional go playing cultures in China, Japan, and Korea. So why do people want to know with some sort of precision how many players there are? Is it for targeting selling of things?

[Kirby]

Number of go players is a meaningless stat unless it’s qualified - as the OP points out, what IS a go player?

The qualification specified here is still a bit tricky - I play games only for the AYD these days. Compared to other times, I wouldn’t consider myself “active”.

The people with the greatest insight are the go server admins, who know ip addresses of the users. This, combined with email stats, could help a bit in finding unique number of users, along with a distribution of their playing frequency.

All in all, it’s a pretty tricky stat to get accurate, unless you add constraints on the filter that make it less meaningful (e.g. number of distinct accounts on KGS that play games two or more times a week).

I’m reminded of a quote that is something to the effect of, “All models are false, but some models are useful”... Which begs the question: what utility do we aim to achieve with such a stat? Just give a number to an interested player? If we better understand our goal in estimating a number of Go players, perhaps we can be better equipped to come up with an estimate.

[Bill Spight]

To paraphrase an old joke:

Count the number of go stones and divide by 400.

[Gomoto]

Holy moley!
Do you want to be philosophers?

Let us just count!

There are 4666 active online users at tygem for example right now,
914 active online users at GoPanda,
please continue with KGS, Fox, OGS, ... if you like

But instead of counting you can discuss the issue as well if you prefer

I estimate there are more than 10000 and less than 1000000 players active online at any given time:

Therefor as a way to start estimating no. of go players I propose 100000 as a good starting point

[hyperpape]

I’m reminded of a quote that is something to the effect of, “All models are false, but some models are useful”... Which begs the question: what utility do we aim to achieve with such a stat? Just give a number to an interested player? If we better understand our goal in estimating a number of Go players, perhaps we can be better equipped to come up with an estimate.

It's easy to think of questions.
1. Is the number of go players increasing or decreasing (worldwide or in a particular country).
2. Is the number/proportion of go players larger or smaller in one place relative to another?
3. How popular is go relative to other pass-times?

[Bill Spight]

Bozulich once told me that before launching Ishi Press he did some research and discovered that in the last half of 1968 people on the east coast of the US bought 100,000 go sets. This was at a time when the AGA had about 600 members.

[Kirby]

I’m reminded of a quote that is something to the effect of, “All models are false, but some models are useful”... Which begs the question: what utility do we aim to achieve with such a stat? Just give a number to an interested player? If we better understand our goal in estimating a number of Go players, perhaps we can be better equipped to come up with an estimate.

It's easy to think of questions.
1. Is the number of go players increasing or decreasing (worldwide or in a particular country).
2. Is the number/proportion of go players larger or smaller in one place relative to another?
3. How popular is go relative to other pass-times?

For numbers 1 and 2, I’d think that the stats you get from Go server accounts provide decent estimates, even considering users with multiple accounts.

[Nyanjilla]

The people with the greatest insight are the go server admins, who know ip addresses of the users. This, combined with email stats, could help a bit in finding unique number of users, along with a distribution of their playing frequency.

Sorry, no. Nowadays, IP and email addresses are disposable. I know of users who get through them like boxes of tissues when they have bad colds....

[Kirby]

The people with the greatest insight are the go server admins, who know ip addresses of the users. This, combined with email stats, could help a bit in finding unique number of users, along with a distribution of their playing frequency.

Sorry, no. Nowadays, IP and email addresses are disposable. I know of users who get through them like boxes of tissues when they have bad colds....

Indeed, I agree. But you still have more information than the general public.