Talk:Go (game)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Former featured article Go (game) is a former featured article. Please see the links under Article milestones below for its original nomination page (for older articles, check the nomination archive) and why it was removed.
Main Page trophy This article appeared on Wikipedia's Main Page as Today's featured article on June 13, 2004.
Wikipedia Version 1.0 Editorial Team / v0.5 / Vital (Rated C-class, Mid-importance)
WikiProject icon This article has been reviewed by the Version 1.0 Editorial Team.
C-Class article C  Quality: C-Class
Checklist icon
 Mid  Importance: Mid

Women players[edit]

I have a copy of Shotwell's book which is referenced in the sectionTop Players. I have looked through the book twice, and I was unable to find any page specifically referencing women players or their special tournaments, as represented by this paragraph:

  • Historically, as with most sports and games, more men than women have played Go. Special tournaments for women exist, but until recently, men and women did not compete together at the highest levels; however, the creation of new, open tournaments and the rise of strong female players, most notably Rui Naiwei, have in recent years highlighted the strength and competitiveness of emerging female players.[1]

I also noticed that the author of this reference did not list a specific page number... TheGarnet (talk) 18:05, 27 November 2013 (UTC)


  1. ^ Shotwell, Peter (2003), Go! More Than a Game, Tuttle Publishing, ISBN 0-8048-3475-X 

Date of Invention[edit]

This article claims that “Go … originated in China more than 2,500 years ago.” However, Go and mathematics claims that “Chinese scholars of the 11th century already published work … based on the go board.” So, when was Go invented? -- (talk) 22:29, 15 December 2013 (UTC)

  • That "11th century" means 11th century AD; this article claims Go originated in about 400BC. There is no contradiction here. Reyk YO! 23:48, 15 December 2013 (UTC)
And evidently the traditional date is more like 4000 years ago... Til Eulenspiegel /talk/ 00:10, 16 December 2013 (UTC)

Capitalization of "go"[edit]


I've noticed that this article always capitalizes the "g" in "go" even in cases where it is not being used as part of a proper name or the first word of a sentence. Is this based upon something in the MOS or is it simply common convention? Other game articles, such those for chess, backgammon, checkers, card games, etc., do not capitalize the first letter when the word is being used as a common noun and I think this is consistent with commonly accepted rules for capitalization. So, in my opinion, "I play go" is correct while "I play Go" is not for exactly the same reasons that "I play poker" is correct and "I play Poker" is not.

Is "go" being capitalized because it is a foreign word or for emphasis? If that's the case, then I think it's better to use italics instead such as "I play go."

Is there concern that people will mistake "go" (the game) for "go" (the verb)? I guess that's possible if no context was provided at all, but since the article is about "go" (the game) that seems a little bit unlikely.

I am interested in this because I am currently working on some articles about shogi. "Shogi" is also a foreign word and is also a board game. "Shogi" is capitalized when it is part of a proper name, e.g., the Japan Shogi Association, or used to begin a sentence, but otherwise it is typically not. So, I am trying to understand how it is different from "go". Thanks in advance. - Marchjuly (talk) 01:13, 4 March 2014 (UTC)

Ok, i found some things on this both on the WikiProject Go's Talk Page and in this talk page's archives here and here, but it doesn't seem to me that a solution was found that was acceptable to all. Was a consensus eventually reached? - Marchjuly (talk) 01:51, 4 March 2014 (UTC)
I personally prefer the capitalized version, to distinguish the name of the game from the rather common English word. Reyk YO! 10:32, 4 March 2014 (UTC)
Hey thanks for the reply Reyk. I can see the merits of that approach. On the other hand, the card game "bridge" is not capitalized each time it is used even though the word is also rather common. - Marchjuly (talk) 10:54, 4 March 2014 (UTC)
Good example. Another (less good though) is "solitaire". Ihardlythinkso (talk) 11:24, 4 March 2014 (UTC) p.s. And I think there might be a lot of them. (E.g. "fencing" is both a sport, and a material for fences.)
Sure, but I think people would be more likely to use the verb "go" when talking about the board game than use "bridge" as in "thing for crossing the water" when talking about the card game. That said, I don't feel particularly strongly about it. Just a slight preference for the capitalised version. Reyk YO! 11:25, 4 March 2014 (UTC)
"Bridge", "patience" and "fencing" are nouns, so they do not interfere with grammatical parsing of a sentence. Context lets the reader immediately know which sense of the noun is being referred to. But "go" is commonly a verb and can cause difficulty in reading and parsing a sentence for those not familiar with it as a noun naming a game. Sure it functions as a noun in specific phrases like "have a go". I think guiding the reader to immediately read "Go, the game" justifies the capitalisation. Tayste (edits) 19:25, 9 March 2016 (UTC)

Seibu Museum[edit]


I only noticed this because it showed up in this diff for an edit made by somebody else to the same sentence.

I believe that seibu in the sentence He said in the French book Arman, un entretien d'artiste[125] that the picture of him published in the Go newspaper was more important for him than his exhibition at the seibu museum. should be capitalized, especially if this is the Japanese word seibu. The Japanese word can be used as a common noun, but in this case it seems to be referring to the name of an actual museum [1] in Tokyo. I don't read French and don't have access to the source this references so I may be wrong. However, if the two seibu museum are the same then, not only seibu but also museum should be capitalized. Furthermore, the actual name of the museum should be probably used since this is the first and only time it is referenced in the article. - Marchjuly (talk) 01:05, 10 March 2014 (UTC)

9x9 Ko Fight picture[edit] In the article, E is described as a ko threat. It may be a ko threat, however it doesn't work - white will just connect the ko. — Preceding unsigned comment added by Fafas (talkcontribs) 20:57, 2 April 2014 (UTC)

Actually it says "possible" ko threat. I think the picture is fine other than the point E being suggested as a ko threat. Fafas (talk) 21:06, 2 April 2014 (UTC)

I think Fafas is right, it is not a ko threat. I think a new image should be made which shows a clearer threat for black. --– sampi (talkcontribemail) 05:13, 26 July 2014 (UTC)

E is indeed not a ko threat for white. However, it is one for black. NathanWubs (talk) 06:28, 26 July 2014 (UTC)
No, it is not a ko threat for black. I think we need a picture were it is abundantly clear to beginners why it's a threat. --– sampi (talkcontribemail) 09:49, 26 July 2014 (UTC)
How is it not, it threatens to kill the left white group. But if you are not satisfied with it. The easiest thing you can do is upload an image to wikipedia creative commons and replace this one. As I do not think anyone would object to a more clearer example. NathanWubs (talk) 11:57, 26 July 2014 (UTC)
I might make a better image and upload it later. In any case, I can assure you E is not a threat (if it helps, I am 3 dan).--– sampi (talkcontribemail) 21:31, 26 July 2014 (UTC)
I played it out, you are right in the end its not beneficial even if black gets the left side. Also being 3 dan, does not help a lot as there are way to many different rating systems these days. Depended on country, or depended on client you use. NathanWubs (talk) 22:24, 26 July 2014 (UTC)
I just mentioned that I was 3 dan (on KGS, for example) to clarify that I am not beginner, and that I know what I'm talking about. No, if black plays E as a threat he does not get the left side (he doesn't get anything, really). I have created an SGF with some example variations here. It is indeed more complicated than I though, but in any case I think a clearer example is better. I am making another image now to replace the one in the article.--– sampi (talkcontribemail) 06:26, 27 July 2014 (UTC)

OK, I replaced the file with the new image. I also restored the comment about E previously removed by Fafas. let me know if you have any comments.--– sampi (talkcontribemail) 06:47, 27 July 2014 (UTC)


Go was seen in the 60s in the British television series, The Man in Room 17, ( about two men who, rather like Mycroft Holmes, solved crimes without leaving their office, but preferred to spend their time competing against each other at Go. — Preceding unsigned comment added by Schollard (talkcontribs) 00:16, 16 April 2014 (UTC)

Good Article nomination[edit]

I see someone has nominated this for good article status. This game definitely deserves a GA article. But in my opinion this article is still quite a way off. It is disorganized, and the sourcing could be much better. We should develop a plan for a more coherent article. What do you all think? Reyk YO! 11:53, 5 May 2014 (UTC)

I agree with you, @Reyk: I am currently working to fix that a little, so what do you suggest?TheQ Editor (Talk) 14:01, 5 May 2014 (UTC)
Also, I noticed in the middle of the article, there is a random{{GoBoardGame}} template. It is kind of irrelevant to the subject and I think it should be moved to the bottom and make it something more similar to the {{Chess}} template.TheQ Editor (Talk) 14:37, 5 May 2014 (UTC)

I think this article is more than a little off. Some terms aren't used consistently throughout the article. But, I'll try my best to help improve it. HowardEzW (talk) 17:57, 5 May 2014 (UTC)

  • Thanks for your responses @TheQ Editor: and @HowardEzW:. I think it will take more than just adding a source here, fixing some terminology there, to get this article up to a reasonable standard. You'll see in the edit history that I have made some incremental changes of this kind too, but IMO the article will need an extensive rewrite starting with a coherent plan regarding the structure and layout. The last time I worked on a major rewrite of an article, we started with a subpage of the talk page where we first decided on sections and listed every source on the topic that we could get our hands on. Then when we had everything together we started writing. This seemed to work well. I suggest starting a subpage Talk:Go (game)/Rewrite where we can do the same thing. What do you think? Reyk YO! 22:19, 5 May 2014 (UTC)
Sure, why not?@Reyk:. So do we just copy the whole article or section by section?TheQ Editor (Talk) 22:37, 5 May 2014 (UTC)
OK, I've started a basic outline. Feel free to tweak it. Now I'll start collecting all the sources I can get my hands on. Reyk YO! 00:55, 7 May 2014 (UTC)
I'll also help on the outline. HowardEzW (talk) 14:56, 7 May 2014 (UTC)
@Reyk:, @HowardEzW:. Should we remove the Strategy and Tactics section. Wikipedia is not a guide. And in the article, it contains a lot of original research, mostly from Go players themselves. TheQ Editor (Talk) 15:28, 11 August 2014 (UTC)
I agree the strategy and tactics section reads like a how-to guide, and has too much original research as it currently is. But I think this is an important part of the game that should be covered in some form, though not the way it currently is. Reyk YO! 00:34, 12 August 2014 (UTC)

Go in Vietnamese[edit]

What is the significance of Go in Vietnam? In the opening paragraph, the name of go in Vietnamese is given. Go is significant in China, Japan and Korea; and often go terms in English come from Chinese, Japanese and Korean. Also it's quite common to refer to go as "weiqi" and "baduk". So it makes sense to put the name in the game in these languages. But, as far as I know, go isn't that popular or culturally significant in Vietnam. If there isn't a reason, I'll remove it. HowardEzW (talk) 17:54, 5 May 2014 (UTC)

Since there was no response for over a week, I've removed the name of go in Vietnamese. HowardEzW (talk) 22:50, 14 May 2014 (UTC)

Modern and low-cost alternatives[edit]

This paragraph contains no citations all and I can't find anything on the web about this. If no references are provided, I would have to delete it. TheQ Editor (Talk) 19:43, 6 May 2014 (UTC)

' "Wei-chi" redirects here. For the Chinese word, see Chinese word for "crisis".'[edit]

What? I can see how Go might be romanized as "Wei-chi", but the Chinese word for crisis has nothing to do with it. (talk) 16:06, 13 May 2014 (UTC)

I get what you mean. The Chinese word for Crisis is more like WeiJi. But on the other hand, Go in chinese is weiqi not weichiTheQ Editor (Talk) 00:15, 14 May 2014 (UTC)
Both are wei-chi in Wade-Giles. Turned the link into a dab per WP:TWODABS and deleted the hat note. — kwami (talk) 02:09, 14 May 2014 (UTC)

improving reasons why go software is not so advanced[edit]

It would seem of fundamental importance to discuss how much money has been invested in developing good quality go-playing software. For example, IBM dedicated massive resources to developing chess playing software. I presume no comparable amount of money or resources has been dedicated to developing go-playing software (and hardware support) I believe this section would be greatly improved if this topic were addressed — Preceding unsigned comment added by (talk) 19:38, 16 May 2014 (UTC)

If you find reliable sources go for it. NathanWubs (talk) 19:59, 16 May 2014 (UTC)

Why this article would fail Good Article review as of now[edit]

@TheQ Editor: This article clearly fails Wikipedia:Good Article criteria 2b/2c due to numerous unreferenced paragraphs and sentences. For example, the entire section "Komi" is unreferenced. All four paragraphs in "Capturing tactics" are unreferenced save the first sentence. Almost the entire section "Strategy" is unreferenced. "Phases of the game" is unreferenced... I think you get the point. "In popular culture and science", in addition to being not fully referenced, is rather disjointed, and needs to be rewritten from the current list of trivia facts into something that flows logically (failures of 1b, and probably 3a and 3b). Virtually all notes are unreferenced. Finally, reference section is a in need of a cleanup, (2a), at least one book is missing page number entries. This article is not ready for a GA by a long shot, through if anyone wants to do a major rewrite of this, I can wait a few days. I'll check here in a week, and quickfail this if no editing has taken place, unless there's a reason for me to delay further. PS. Most of those issues were pointed out almost five years ago: Talk:Go (game)/GA1. --Piotr Konieczny aka Prokonsul Piotrus| reply here 05:48, 30 May 2014 (UTC)

We started a major re-write from the ground up, but it will be a long time before that is done. Reyk YO! 06:03, 30 May 2014 (UTC)
I am glad to hear this article will be improved, but the GA nom should take place when the work is finished (or at least, almost finished), not before. Perhaps rather then me failing the article you'd like to have it withdrawn from the review queue? --Piotr Konieczny aka Prokonsul Piotrus| reply here 06:15, 30 May 2014 (UTC)

Incorrect statement, in the first paragraph[edit]

Baduk does not mean 'encircling game'. I currently do not have my Korean etymology dictionary but it has a completely different etymology based on native roots. 'Encircling game' would be 'Wigi' or 'Duleossagi Noli', it should be obvious to a native speaker like me. It seems the editor knew 'weiqi' and 'igo' meant 'encircling game' but didn't know what 'baduk' meant and so inaccurately assumed that 'baduk' also meant 'encircling game'. Also, because of errors like this I seriously doubt that this article is ready for GA status.--Seonookim (What I've done so far) (I'm busy here) (Talk with me) 13:34, 30 May 2014 (UTC)

Increase in IQ[edit]

At the moment this section is dubious at best. Its gives two sources, a user discussion at sensei library and, a blog post at rail spikes. the first sections says. "Studies show" while in the source that is given no studies are shown. With one link to a study about internet addiction and how go maybe could prevent that from happening. The second statement comes from a programmer and is a guess nothing substantial but anecdotes are offered, which all once again lead to further speculation that does not seem relevant. While it would be lovely that the wonderful game of Go would increase our IQ. There would need to be at least some WP:RS in a relevant field making that claim, to add it to the article. So I suggest for now at least that this section is removed. Or at least to remove the first sentence if consensus is that the speculation of the programmer should remain. NathanWubs (talk) 16:27, 6 June 2014 (UTC)

recent results in computer go[edit]

I was surprised to hear recently on the radio that Go software have started beating professionnal players on 19x19 boards. I was expecting this to take decades more. Yet this article still says only amateurs can be beat, while Computer_Go has a fairly extensive records of professionnal level players being beat (although often with handicap) by go software.. maybe this should be clarified? --TheAnarcat (talk) 13:41, 22 October 2014 (UTC)

Those recent wins are from 4 or 5 stone handicap games, NOT from even games. Go software are certainly capable of beating professional go players -- with handicap stones. Even 20 years ago, go software could beat professional go players as long as there was enough handicap. The only area where go software is as strong as professional go players is on smaller boards such as the 9x9, where professionals have occasionally lost in even games. HowardEzW (talk) 12:40, 2 December 2014 (UTC)

Go and Philosophy[edit]

The article says that " There is significant strategy and philosophy[2][3][4] involved in the game...". The source cited for this is Sensei's Library, which is a wiki for go. Also, there is no mention of go and philosophy throughout the rest of the article. I think this bit should be removed. HowardEzW (talk) 03:28, 21 December 2014 (UTC)


This article has a long-standing template that says it is being rewritten, but Talk:Go (game)/Rewrite hasn't been touched in six months. I would like to remove the template, since it scares off other people from editing, and have asked the two editors involved in that project (judging from history) their opinion. I have been editing this article lately, because it really needs pruning and tightening. - DavidWBrooks (talk) 22:51, 13 April 2015 (UTC)

I have removed the template, which points to Talk:Go (game)/Rewrite = DavidWBrooks (talk) 00:38, 18 May 2015 (UTC)


I have removed a section titled "pedagogy" which had unsourced opinions about the best way to teach or the "common method" of instruction. [[2]] notes that we aren't an instuction manual. - DavidWBrooks (talk) 00:07, 12 June 2015 (UTC)

Page doesn't work in Google Chrome[edit]

The main page works fine (text only) in lynx, but comes up blank in Google Chrome and then issues repeated pop-ups that the page is unresponsive. (talk) 21:30, 21 June 2015 (UTC) Twitter.Com/CalRobert (Robert Maas)

Number of possible games[edit]

According to this video, there are far more possible games of chess than 10^120. 10^10^50 is the most recent number somebody came up with, although it is still a very rough estimation.

If you have reliable sources, free feel to add the information. A video like this is hardly a reliable source. A “number somebody came up with” is irrelevant unless he has a sound augment; blind guesses don't qualify. Regards. Mario Castelán Castro (talk) 16:48, 24 July 2015 (UTC).
There are have been several considerations of the number of possible Go games. (E.g., see NumberOfPossibleGoGames at Sensi's Library.) The size of some these numbers – e.g, on the order of 10^(10^170) – is mind boggling. That makes the more realistic estimate of practical games of "only" 10^800 seem rather mundane.
For a more comprehensible notion of the complexity of game play consider how quickly the game trees expand. Chess has 20 possible first moves, and for each the second player has 20 possible replies. So after the second move there are 400 possibilities. A game of Go usually starts with a "move" in a corner, and usually on one of about eight locations. To which the second player usually responds on one of the similar positions in one of the other three corners, which is roughly 500 possibilities. At this stage the game tree has about 4,000 branches (and those are just the "reasonable" moves). Any time symmetry is broken the rate of game tree expansion can go up by a factor of over 300. All this is why the very notion of "number of possible games" seems incomprehensibly transcendent. ~ J. Johnson (JJ) (talk) 22:33, 7 September 2015 (UTC)

An online little tool for editing goban[edit] (talk) 19:29, 27 December 2015 (UTC)

Chinese or Japanese?[edit]

It's interesting that this game is referred to as go (the Japanese name for it) and identified as a fundamentally Chinese game (which the Chinese call wéiqí). Article should have some comment on that; if it's here, I didn't see it. Obviously, the origins are in China, but the English word for it is go.

Kortoso (talk) 23:31, 27 January 2016 (UTC)

Software Solution to Go.[edit]

Google recently built an AI which beat an expert go player. This article should be updated to reflect that. (talk) 01:16, 28 January 2016 (UTC)

Yes. But note that Wikipedia is an encyclopedia, not a newspaper. So while the event is significant, we can probably wait a week or two until the Nature is more available (it's paywalled) and people have a chance to look it over. Till then the best commentary seems to be Google's official blog at ~ J. Johnson (JJ) (talk) 21:51, 28 January 2016 (UTC)
P.S. For anyone not familiar with AlphaGo (the program referred to above), it is a neural network that uses pattern recognition. It was trained (in part) by having it play thousands of games against copies of itself. An interesting development to watch for is whether the trained network can be examined for heuristics on how to play. ~ J. Johnson (JJ) (talk) 22:02, 28 January 2016 (UTC)

How long do games last ?[edit]

At social, serious amateur, professional level ? Good to be in article. Rcbutcher (talk) 04:26, 28 January 2016 (UTC)

Most tournaments give each player 45 minutes (plus some overtime), allowing three, sometimes four, rounds in a day. In social play, where players are not trying to work things out to the utmost, or may be trying some new approach that quickly hits a wall, a game may last only a half hour. At the top professional levels games can be carried over into a second day. ~ J. Johnson (JJ) (talk) 21:43, 28 January 2016 (UTC)

Oldest board game??[edit]

The article states go is the oldest board game. I think backgammon is olderPlozort (talk) 22:28, 9 February 2016 (UTC).

Is it "Go" or "go" mid-sentence?[edit]

Is the title of the game a proper noun? There seems to be a mixture in Wikipedia articles of capitalised and uncapitalised usages. The Rambling Man (talk) 15:22, 9 March 2016 (UTC)

Since "go" is such a common English word, the convention is to use "Go" for the game, despite this not strictly adhering to the usual rules around proper nouns in English. Tayste (edits) 19:12, 9 March 2016 (UTC)

Recent results in computer go (AlphaGo)[edit]

As WP is not a newspaper I believe we do not need to day-by-day coverage of "breaking news" regarding the AlphaGo versus Lee Sedol match (currently being played out). Nearly all observers are greatly impressed by AlphaGo's play, and this appears to be a significant advance in AI, which may eventually have profound effects on Go. But, again, this article does not need day-by-day coverage of current events. ~ J. Johnson (JJ) (talk) 20:21, 10 March 2016 (UTC)

"Complexity" of Go vs chess[edit]

I removed this paragraph from the lede:

There is significant strategy involved in the game, and the number of possible games is vast (10761 compared, for example, to the estimated 10120 possible in chess), displaying its complexity despite relatively simple rules.

First, these numbers are probably meaningless. According to Go and mathematics, the number of Go games depends enormously on what restrictions you place on game length. 10120, the "Shannon number", is a lower bound (not an estimate) for chess based on a game length of 40 moves. A real estimate would be dominated by long games and would probably be vastly larger.

Second, game tree complexity isn't complexity in any ordinary English sense. In the Go variant where there is no capturing of territory and you just place pieces until the board is full (at which point black wins 181–180), there are still 361! ≈ 10768 possible games. That's dwarfed by the number of possible games of Super Mario Bros. (which is finite if you add a no-repeat rule), and that's in turn dwarfed by the number of possible games of tag (which might be finite if you take quantum gravity into account).

There's clearly some sense in which tic-tac-toe is easier than chess or Go. I think that sense is that if there were tic-tac-toe rankings, a very large fraction of players would be top-ranked and consistently play each other to a draw. I see no way to argue that Go is harder/more sophisticated than chess where human players are concerned. They are both hard.

Computers tend to do badly at games with large branching factors, but that's because we don't know how to do AI. Even average humans easily outperform computers at, say, reading printed text. You could react by being proud to be human, or by being embarrassed that we can't even figure out how to replicate the functions of our own brains. -- BenRG (talk) 21:49, 11 March 2016 (UTC)

No, those numbers are not meaningless, though they are so wild and crazy they seem to verge on fantasy. (See for details.) I don't know what you mean that a "real estimate" on would be "dominated" by longer games; I believe that 40 was taken as a representative game length for chess. But even you doubled that, Go games are still longer (typically around 150–200 moves). And Go has many more options per move, so the relationship is (crudely) something like 80≈35 << 150≈250 35≈80 << 250≈150. If you (or anyone else) can come up with a better illustration of the relative complexity of these games, fine, but until then let's stick with what was there.
The reason computers "tend to do badly at games with large branching factors" is because when those "branching factors" (game trees) become freakingly LARGE it becomes computationally infeasible to search them. The "branching factors" in chess are flat enough that Deep Blue could use brute-force search of the game tree to beat Kasparov 27 years ago, but that won't work in Go. The awesome thing about the new AlphaGo program is it uses a different approach (neural networks) that does pattern recognition. And already it has beaten a world champion three times running (see AlphaGo versus Lee Sedol) with moves that are hardly short of stunning. ~ J. Johnson (JJ) (talk) 19:12, 12 March 2016 (UTC)
I don't know how to respond except with what I wrote above. You don't seem to have understood any of it. You could read the Go and mathematics article, which has a table of the number of possible Go games as a function of maximum game length, with values ranging from 10128 to 1010171. The same thing happens in chess, which is why the total count is dominated by long games. -- BenRG (talk) 19:53, 12 March 2016 (UTC)
Funny thing, from what you have wrote I get a feeling that you don't understand any of this. Particularly, you seem to have missed that the sentence you cut was not about the longest possible game of Go (for which this universe has not time enough to play), but of the comparative size of the game trees (as a measure of complexity) for any reasonable game of chess and Go, however you want to define "reasonable".
BTW, your Go game analysis above, and particularly that black wins, seems faulty. The way I see it, after the 358th move each player has 179 stones on the board, with three three points left. Black plays, then white passes, in order to avoid self-atari. Ditto for black. At that point, by Chinese scoring black has 180 points, and white has 179 points. By Japanese scoring, each player has zero points. By AGA scoring each passed stone is a point, so the score is black 2, white 1. However, under all systems white gets komi of any where from four and a half to eleven and a half points. So white wins. ~ J. Johnson (JJ) (talk) 19:28, 13 March 2016 (UTC)
You guys are off to a bad start so let me jump in. The paragraph that BenRG removed and that J. Johnson restored is factually incorrect (10761 and 10120 are not the number of possible Go/Chess games), so the paragraph must be fixed or removed. BenRG removed it claiming that the hardness of both games cannot be compared, but I tend to disagree with that. I think that Go is more complex than Chess by any reasonable measure, so it looks reasonable to me to make that claim in the article and to briefly support it with numbers. Now if we decide to fix the paragraph, then changing the figures to the correct number of possible games is not what we want to do because the number of possible games depends more on details of the termination rules than on game complexity. Indeed, a minor chess rule change made in 2014 decreased the number of possible chess games from infinite to about 1030,000 according to [3]. So to fix the paragraph I would recommend any of these 3 possibilities:
  • compare the number of possible board positions: 10^170 vs 10^47 (easily understandable by people),
  • compare game-tree complexity: 10^360 vs 10^123, or
  • compare game-tree complexity, but call it "the number of possible games of average length" (less scary than "game-tree complexity" and still factually correct). Egnau (talk) 17:22, 15 March 2016 (UTC)
I concur that that sentence could use some work, and particularly that "the number of possible games" is not simply "vast", but rather an absurd concept with little connection to ordinary experience, and offers little insight into the character of these games. Neither the number of possible board positions, nor the number of possible games of average length" does much better. I think the relation I gave above (80^~35 << 150^~250 35^~80 << 250^~150) would serve for "wow", and is resonable accurate estimate that can be sourced (e.g., in the AlphaGo article in Nature). For the edification of readers who are not familiar with either Go or complexity theory, a simple description of how quickly the game tree expands should be sufficient for purposes of comparison. This could be done in the "Nature of the game" section, but it seems to me this is of interest more generally, so ought to be mentioned in some way in the "Game theory" section. Either way, this could support a summary statement in the lede that Go is deemed more complex than chess. ~ J. Johnson (JJ) (talk) 23:40, 15 March 2016 (UTC)
Well, 3580 = 10123 and 250150 = 10360 are the number of possible games of average length, so if you think that the calculation is absurd then you'll have to find some other calculation, but if you change your mind about the absurdity, then I agree with the calculation since it was one my 3 possibilities. Note that the AlphaGo article isn't the original source: the AlphaGo article cites Allis's Ph.D. thesis. The game complexity article contains the same figures in the table and also cites Allis's Ph.D. thesis. Egnau (talk) 02:44, 16 March 2016 (UTC)
I have no problem with going back to original sources. And I see that Allis' thesis is on-line, so it is more accessible than the AlphaGo article. ~ J. Johnson (JJ) (talk) 20:21, 16 March 2016 (UTC)
The number of possible games or possible game positions is nearly irrelevant, as even a mediocre player can discard 99.99999...% of these as irrelevant to realistic play. What convinces me Go is inherently more complex and more human-friendly/computer-hostile than chess is this (but of course we need sources, not my opinion):
  1. A game of Go typically has 4 different openings (one in each corner) that interact with each others throughout the game - a game of chess has only one opening. Arguably, each opening in Go is simpler and shorter than most chess openings, but the possibilities stemming from the combination of four openings (and play switching back and forth between them) result in greater complexity.
  2. In Go, the influence of one game piece is mostly local, decreasing as one looks further away on the board, though there are notable exceptions such as ladders and ko's, and other concepts involving sente and gote. By contrast, in chess, many pieces can move quickly across the board, or threaten to do so. In my mind, chess is like a Go board folded up like puff pastry, which is confusing for humans but no problem for a computer that excels at combinatorics. Thus, though chess is probably the less complex for a computer (and thus in an objective sense), Go is more accessible for human intelligence and (geometric) intuition - i.e., in Go much more than in chess, human intelligence (and now neural networks) can take some shortcuts through the complete combinatorial complexity to arrive at nearly perfect play.-- (talk) 09:02, 16 March 2016 (UTC)
The numerical comparisons aren't as bad as you think. The number of possible games or possible game positions works just fine for comparing two games if the fraction 99.99999...% that you have to discard as irrelevant to realistic play is the same fraction for both games, or the same power of the base number for both games. Shannon and Allis discard games above average length which discards a lot of unrealistic play and does a lot towards making a fair comparison. Then as long as the number of realistic moves at each turn is some power of the number of legal moves at each turn, the comparison works. It's not perfect and if the numbers were 10123 vs 10124 then it would be unwise to call Go more complex, but it turns out that Go's exponent is about 3x as large which is enough to withstand inaccuracies. Egnau (talk) 15:58, 16 March 2016 (UTC)
I don't think there is any basis for assuming that the percentages are even remotely similar. Is it 99.999999999999 % ? Is it 99.99999999999999999999999999999999999999 % ? I've got no idea (but probably still more 9's)! The length of the game is of course a factor, but (maybe especially in Go) the vast majority of the possible average-length games will be completely silly. Just try placing stones of alternating colours at random allowed points; it will not even remotely resemble a real game of Go. I suspect the percentage may be smaller (fewer 9's) for chess than for Go, but I really don't know.-- (talk) 16:24, 16 March 2016 (UTC)
I think we can all agree that the number of possible games is "nearly irrelevant". I am inclined to think that length is of game is rather irrelevant, especially for purposes of describing Go to a general readership, that a useful comparison can be made in just looking at the first several moves for each game. E.g., out of 20 possible first moves in chess perhaps only ten are generally seen, and likewise for the second move. In Go there are 55 unique possible first moves (disallowing symmetrical duplications), of which only about four are "realistic". But as symmetries get broken the game tree starts to go wild: typically around 30 responses for move 2, and I estimate a hundred or more for move 3. Taking this out another step or two I think it would be more useful in demonstrating the relative complexity of these two games than a bunch of really big numbers. ~ J. Johnson (JJ) (talk) 20:24, 16 March 2016 (UTC)
You're having a hard time estimating the number of silly games, but can you try to estimate the number of realistic games instead? It looks easier. Egnau (talk) 21:12, 16 March 2016 (UTC)
If Egnaus comment was for me, that's exactly my point: It's virtually impossible to estimate the distribution between silly and meaningful games among the astronomic numbers of possible games ind the tree. 0.000000000000000000000000 % ? 0.0000000000000000000000000000000000000 % ? I like J. Johnson's approach - but even more, I'd like reliable sources.-- (talk) 06:25, 17 March 2016 (UTC)
Yes the comment was intended for you, please don't change my indentation level which conveyed that. You find it virtually impossible because you're thinking about the fraction and it feels hopeless. What I was suggesting was to directly estimate the number of realistic games which is the number we're really after. The total number of games is usually estimated by (number_of_legal_moves_per_turn)(length_of_the_longest_game), so the number of realistic games would be (number_of_realistic_moves_per_turn)(length_of_a_realistic_game). Egnau (talk) 15:16, 17 March 2016 (UTC)
"Number of legal moves per turn" is a very, very non-trivial number that varies widely as the game proceeds - and "length of longest game" is theoretically infinite with ko fights. - DavidWBrooks (talk) 15:58, 17 March 2016 (UTC)
Indeed. Allis comments that the "average branching factor" (see my comment below) could be taken as an average, or varying (generally increasing) as the game proceeds. It is because of all these complications of trying to quantify realistic games that these numbers are so wildly inaccurate. But, as long as the estimate for each game is calculated with roughly equal "wildness" (i.e., on comparable bases) the inequality should still hold. ~ J. Johnson (JJ) (talk) 22:17, 17 March 2016 (UTC)
Hypothetically, comparing estimates of the sizes of the trees of realistic games of Chess and Go would be relevant, but I think it is immensely difficult to create such realistic estimates, and we'd need a valid source for them (and for comparing them) anyway. Unless someone has a good source, I think all these numbers are moot.-- (talk) 16:06, 17 March 2016 (UTC)

The above discussion got a little deep so let's step back and talk about reliable sources. Allis's thesis discusses game complexity, and he uses these two measures:

  1. the number of legal board positions (technical term: state-space complexity), and
  2. the number of games of average length (technical term: game-tree complexity).

In the AlphaGo Nature paper, the authors chose the second measure to compare the complexity of Chess and Go. That article passed a stringent review process, so measure (2) is a clear winner, and backed by authoritative sources. J. Johnson called the measure "not much better than absurd" and its numerical outcome "reasonably accurate" in the same breath, so I'm not sure what it means. Nø thinks that the measure is meaningless and that counting the number of realistic games is the only meaningful thing to do. My personal view is that Nø's suggestion would order board games in roughly the same way and I wanted Nø to make some estimates in order to reach that conclusion too, but we didn't get that far. Definition (2) has the benefit that it doesn't require a definition of "realistic" which is somewhat subjective. This might be why definition (2) is preferred by the game complexity specialists. So please, let's trust these people when they claim that (2) is a reasonable way of comparing the complexity of two games, even if it might not appear to be the case at first glance. Egnau (talk) 18:41, 17 March 2016 (UTC)

Egnau, you are incorrect. Allis uses average game length (in plies, or pairs of moves) AND the "average branching factor" (p. 160). Silver et al. express this as bd where b is the game's breadth (number of legal moves per position) and d is its depth (game length)." From Allis (pp. 171, 174) we get the estimates used by Silver et al. for 8035 << 150250 35≈80 << 250≈150. I think this suffices for an authoritative comparison of game complexity at the technical level. (See also Allis' figure 6.1, which shows that, for chess and Go, that state-space and game-tree complexity are comparable.)
However, I think we need to provide a non-technical comparison that works for general readers, such as I outlined in a prior comment (20:24 15 March). Unfortunately, I am not aware of any sources for that. ~ J. Johnson (JJ) (talk) 22:22, 17 March 2016 (UTC)
I'm not incorrect. Allis discusses state-space complexity (1) on p158. Your two items (call them 2a and 2b) are simply ingredients Allis uses to compute game-tree complexity (2). Figure 6.1 is a plot of the two measures I mentioned. I agree that definition (2) is a bit unsettling the first time you see it, and might require a footnote to preemptively answer the questions "why not simply the number of games?" and "why not the number of realistic games?". Definition (1) might be a reasonable choice after all because it is also sourced and it feels less technical. By the way, if you plug the correct b and d in bd you should get 3580 and 250150. It's the third time in the thread that you get it wrong. Egnau (talk) 15:05, 18 March 2016 (UTC)
You're right in that I've been getting b and d reversed. (Corrected. Dealing with such immensities made me bit dizzy.) And I am glad you recognize that BOTH are "ingredients" for Alliss' measure of "game complexity". Which is not shown in your #2, where you have left one out.
You seem to be arguing that state-space complexity is a better measure than game-tree complexity. That could be argued, but why? Is there any reason we can't mention both? That one "feels less technical" (because it doesn't use exponents?) seems irrelevant: it's still a number so big as to lie outside of human experience. Which is why I have been suggesting that (in addition to any technical measures of complexity) we need a simpler, more intuitive, non-technical statement. ~ J. Johnson (JJ) (talk) 20:57, 18 March 2016 (UTC)
I mean "better" for the purpose of the third paragraph of the article where the goal is simply to support the claim that Go is a complex game. Conciseness is a big factor, so I assumed that we would use only one measure and pick the simplest one that does the job correctly. If you think that we should mention both of Allis's measures and go as far as describing the ingredients and the calculations behind (2), then go ahead and write up something, but I think there's a risk someone will revert your change, claiming that it's too detailed for the article's introduction. Egnau (talk) 00:52, 20 March 2016 (UTC)
Not just too detailed, but too technical for the introduction. Which gets back to what I keep saying: for the introduction we need a simpler, more intuitive, non-technical statement. A more detailed explanation can be put into a more appropriate section. ~ J. Johnson (JJ) (talk) 23:38, 21 March 2016 (UTC)
So if we change the 3rd paragraph to say "the number of legal board positions is vast (10170 compared, for example, to the estimated 1047 in chess)", then are you satisfied? Egnau (talk) 15:42, 24 March 2016 (UTC)

How about a simple descriptive statement like this, with no math to scare people away:

Go is considered more complex than chess, having both a larger board with more scope for play and longer games, and, on average, more alternatives to consider per move.[1]

That concisely covers both kinds of complexity, and the gory supporting details can be put into a note:


  1. ^ Estimates of game complexity can only be crudely estimated. The number of legal positions (state-space complexity) for chess has been estimated at anywhere between 1043 to 1050, comparable to 10172 for Go. Alternately, a measure of all the alternatives to be considered at each stage of the game (game-tree complexity) can be estimated with bd, where b is the game's breadth (number of legal moves per position) and d is its depth (number of moves [plies] per game). For chess and Go the proportion is very roughly ≈35≈80 ≪ ≈250≈150, or 10≈123 ≪ 10≈360. (Allis 1993, pp. 158-161, 171, 174, §§6.2.4, 6.3.9, 6.3.12)

I think that covers all of the points and considerations raised so far. ~ J. Johnson (JJ) (talk) 22:06, 24 March 2016 (UTC)

These estimates of "complexity" should be explicitly related to the assumption that game-tree exploration is the method of solving. It is perfectly conceivable (though it doesn't seem likely) that there is some computationally efficient algorithm for solving the game that does not require game-tree search. AlphaGo, in particular, is clearly not exploring the game tree in the straightforward way. It was said that AlphaGo estimated that one of its surprising moves had 1/10,000 chance of being played by a professional. I wonder what the fanout of moves with > 1/10,000 chance of being played is? --Macrakis (talk) 22:25, 24 March 2016 (UTC)
Why? That AlphaGo doesn't depend on game-tree search in no way makes the game less complex. The issue that we have been exploring here is how to present the relative complexity of Go and chess to the general reader. How this complexity is to be tackled (and solved??) is independent of the existence of the complexity. ~ J. Johnson (JJ) (talk) 22:45, 24 March 2016 (UTC)
Looks good to me. I can think of a few minor fixes and improvements to the supporting details (for example, "plies" are half-moves in chess, not pairs of moves), but I can make these minor changes in a later edit. Egnau (talk) 01:58, 25 March 2016 (UTC)
Glad you like it. But "ply" needs to be resolved. My understanding was as pairs of moves, what I have also heard called "rounds", which I believe I picked up from a game theory book without reference to Go. (I'll check that source tonight.) I see that Alliss (p. 5) defines it as "a move by one player", without reference to any specific game. Ply (game theory) is interesting: it explains the chess "half-move" on the basis of one move being a turn by each player. All that seems to come together, so I'll probably revise accordingly. ~ J. Johnson (JJ) (talk) 21:27, 25 March 2016 (UTC)
Yup. Don't know where I got that double-move idea, so I am revising accordingly. ~ J. Johnson (JJ) (talk) 19:31, 26 March 2016 (UTC)
And the fix is in. ~ J. Johnson (JJ) (talk) 21:13, 27 March 2016 (UTC)

A break[edit]

@Egnau: Regarding your replacement of "comparable to" with "much smaller than": what I was trying to communicate is not the magnitude of the relationship (which is certainly evident as it stands), but that it was a fair comparison. That is, these estimates are derived on the same basis for each game. The "much smaller than" is only the result, not the basis. If this wasn't clear, perhaps we could try some alternatives? ~ J. Johnson (JJ) (talk) 23:16, 29 March 2016 (UTC) I didn't notice the ping was incorrect because your link is always red. ~ J. Johnson (JJ) (talk) 21:44, 3 April 2016 (UTC)

I see what you intended now and you can try some alternatives, but I don't think that it will work well. I think that if you don't say anything then by default people already assume that it is a fair comparison, so anything you add to make that explicit will be unnecessary or confusing. The "much smaller than" might be evident, but it makes the sentence flow well.
If you want to improve something, then my main worry right now is that you use Allis as your reference for everything, but Allis's thesis is from 1994 and it doesn't have the latest numbers (for example it has 10172 instead of 10170 for Go). I know you spent time finding all the page numbers, but I think it would have been better to simply use Game_complexity#Complexities_of_some_well-known_games as the reference because that table has the latest numbers and is sourced. What do you think? Egnau (talk) 11:02, 7 April 2016 (UTC)
In the first place, we do not cite Wikipedia. If you mean to suggest using the refereces used in that table, I note that (for Go) the first reference is (!) Allis, 1994 (same as I used here), the fourth reference is from 1983 (oops), and the third is Tromp's own web page (and not peer-reviewed). So perhaps what you really meant to suggest is using the inadequately cited second reference to Tromp and Gunnar Farnebäck (2015). (And why couldn't you just say so directly?) The problem is, they examined only Go; there is no comparable examination of chess. They may indeed have a more exact figure fo Go, but the point of the comparison is how Go and chess match-up. They don't address chess, and your table provides no basis for a fair comparison, done on similar assumptions, etc. Allis' numbers were good enough for Silver et al. just a few months ago, and your desire for "the latest numbers" serves no purpose. ~ J. Johnson (JJ) (talk) 22:21, 7 April 2016 (UTC)
What I meant was linking to the Wikipedia table to benefit from its references without having to copy-paste them. This is done sometimes in Wikipedia. For example in Cosmic Calendar the phrase "13.8 billion year lifetime" is a link to the Age of the universe article, not to the original source of the number. In other articles the references are copy-pasted, so that's certainly also an option.
The number of legal positions is a self-contained question for each game, so I don't understand why you'd accept an estimate for Go only if the same person also did it for chess. Your concern might be valid for something subjective like the number of realistic games, but here we're talking about a thoroughly objective definition.
Since you brought up the topic of "fair comparison", let me point out that Allis calls 10172 a bound (not an estimate), so right now you're the one being unfair by using an estimate for chess and an upper bound for Go.
If you have issues with the references given in the Game complexity article concerning 10170 then I suggest that you first bring it up in that article's talk page and we'll see what happens. Allis' numbers were good enough for Silver et al. simply because they do not discuss state-space complexity. Egnau (talk) 04:25, 8 April 2016 (UTC)
Egnau, are you trying to pick a fight? It sure sounds like it. If you and everyone else working on Game complexity are happy with the sources you have there, then I am happy for you. Where I do have an issue is with your petty, niggling objections of no merit. E.g., you say that Allis's state-space complexity numbers are "unfair [for] using an estimate for chess and an upper bound for Go", apparently because Allis explicitly says "upper bound" for Go, but only "estimate" for chess. What you seem to have missed is that state-space complexity is intrinsically an upper bound to game complexity, and that both figures are estimates of an upper bound. That someone has trimmed the figure the figure for Go by two magnitudes is quite inconsequential, in that the difference between the respective figures is on the order of 120 magnitudes.
As linking to other parts of Wikipedia "to benefit from its references without having to copy-paste them": that it is sometimes done shows only that guidance is imperfect. I commend to your attention the Citing sources#Say where you read it section, which says: "Don't cite a source unless you've seen it for yourself." Note also that if you take a figure from one source, and another figure from another source, and then compare them, you violate the WP:SYNTHESIS rule. ~ J. Johnson (JJ) (talk) 22:19, 8 April 2016 (UTC)
State-space complexity is not "an upper bound to game complexity". We called it a measure of game complexity earlier and that's what it is. See 6.2.4 from Allis for a confirmation. The reason you know that 10172 is only two orders of magnitude off is precisely because of the 10170 estimate. Without it, you wouldn't be able to tell how accurate 10172 is. My argument is that since you need 10170 anyway, you might as well just mention 10170. This is not a case of WP:SYNTHESIS. Allis is a single source that says that it's ok to compare the state-space complexity of chess and Go. Once that's established, we own it to our readers to plug the most accurate numbers that we know of into the comparison, just like nowadays we plug modern estimates of the masses of planets into Newton's equations, and not Newton's values from 1687. I'm sorry if pointing out flaws in your arguments feels like a personal attack. Double-check your facts and your logic and I won't be able to do it. Egnau (talk) 08:13, 9 April 2016 (UTC)
Allis, §6.2.4: "The main application of the state-space complexity of a game is that it provides a bound to the complexity of games that can be solved through complete enumeration." But, gee, he doesn't say upper bound, so you must be right, those numbers are NOT upper bounds. But wait, he does say (§6.3.9): "In our calculation of the state-space complexity of chess .... an upper bound of 5 . 1052 was calculated." [Emphasis added.] So what is unfair about the comparison?
And no, I do not "know" that "10172 is only two orders of magnitude off" because someone else came up with a different estimate, calculated on some slightly different basis. What you seem to be consistently missing is that the comparison of these figures for chess and Go does not depend on their absolute accuracy, but on their relative accuracy, that they are calculated on similar ("comparable") bases. That Tromp and Farnebäck came up with a slightly different figure for Go is irrelevant, as they did not produce a similar figure for chess. BTW, Allis does not say that it is okay for us to plug in someone else's figures, and especialliy not when it is done for only one game. You should mind the flaws in your own arguments. ~ J. Johnson (JJ) (talk) 21:47, 9 April 2016 (UTC)
Ok I admit that state-space complexity can be seen as an upper bound on game complexity, but mathematically that doesn't give you permission to add upper bounds on top of upper bounds, as you seem to think. The statement "upper bounds on state-space complexity are 1052 for chess and 10172 for Go" would be fair, but that's not what the article citation is currently saying. I already showed that for state-space complexity there is no concept of "comparable base" to worry about, so stop bringing that up. The article citation currently contains the statement "state-space complexity [...] has been estimated at [...] 10172 for Go". All I'm saying is that a more accurate estimate is known, so the number should be updated. The fact that it doesn't change the result of the comparison is not a justification for rejecting improvements. Allis does not say that it is wrong for us to plug in someone else's figures either, so obviously this is a completely fruitless line of investigation. Let's look at what is done elsewhere. Do you think that the numbers in List of highest mountains or List of tallest structures in the world were measured by the same person? There are tons of Wikipedia articles comparing estimates from different sources. Doing this is completely reasonable and simply not a problem. Egnau (talk) 13:12, 11 April 2016 (UTC)
The whole point of the text in question is to give a general reader some idea of how complex Go is relative to chess. You keep wanting to polish up an absolute, stand-alone measure of complexity, but for the average reader such numbers are effectively (as BenRG said way back at the start) meaningless. So we compare it with chess, which (to be fair) requires comparable figures. I keep bringing it up because that is the essential point you don't get. What you also do not get is that substituting 10^170 for Allis' figure would be a misquotation of Allis.
Let's try an example. Suppose that a thousand years ago someone measured the lengths of two boats using a ruler that has since been lost, and found that one boat was a hundred times longer than the other. Suppose also that recently someone found a historical record of one of the boat's length in terms of hand spans. Note that these are two entirely different datums. The first gives us the length of one boat relative to the other; it is a comparison, based on comparable measurements. The other gives us (assuming some notion of a hand span) the actual (but approximate) length of one boat. That you can get a better estimate of the length of a hand span, and thus of the one boat, says nothing about which boat is longer. Is that clearer?
That we accept numbers from different sources for (say) highest mountains is because everyone is using the same measuring stick. The problem with comparing game complexity figures is that there is no standard ruler: each author has a different basis for estimating (e.g.) what proportion of all possible moves are actually legal moves. ~ J. Johnson (JJ) (talk) 22:33, 11 April 2016 (UTC)

Promote "Pebble game" as a generic name for wei-go-duk[edit]

Pebble game has a kind of working, phonotactically. I think. -Anam.saion.ocuana (talk) 04:19, 15 March 2016 (UTC)

As media coverage shows quite clearly, this game is generally called "Go" in English - and hence on I've even seen it described as "the Chinese game Go" (which actually isn't entirely wrong). If someone succeeds in changing the commonly used name in English into "The Pebble Game" or something (starting perhaps with the Go associations?), of course it should and will change here too.-- (talk) 06:44, 15 March 2016 (UTC)
Yes. This is an encylopedia. We follow; we don't promote. ~ J. Johnson (JJ) (talk) 20:20, 16 March 2016 (UTC)

'Computers and go' section out of date[edit]

This section is out of date now, especially in the light of the recent outcome between Lee Sedol and AlphaGo. A Miller (talk) 11:48, 15 March 2016 (UTC)

As I mentioned above, Wikipedia is not a newspaper; we don't need to follow "breaking news". At any rate, it will take days, perhaps even months, before the significance of the AlphaGo is threshed out. And the challenges are still the same, and the history of past attempts unchanged, so it's not a matter of "out of date" so much as there are new developments (still unfolding) which will eventually get folded in. Anything late-breaking and really, really important can be linked to via an external link. Meanwhile, let's not get ahead of the assessments of the experts. ~ J. Johnson (JJ) (talk) 23:51, 15 March 2016 (UTC)

Chinese characters solo?[edit]

In an early paragraph, we have:

A basic principle of Go is that stones must have at least one "liberty" (Chinese: )

Here, and again several places later in the article, we have a commonly used English concept followed by a Chinese character, but no pinyin or other indication of Chinese pronunciation. This seems odd to me; we are not writing for scholars of Chinese. Perhaps (I don't know) a reason for this is that the character applies in Chinese, Japanese and Korean, which would require three pronunciations/translitteration. Still, I think that if the Chinese character is relevant at all, it should follow or be followed by at least one translitteration.-- (talk) 10:25, 16 March 2016 (UTC)

I suspect that someone was trying to be helpful in adding more information, but that seems rather excessive, and it clutters the text. But it has occurred to me that perhaps we should have "Terms" ("Glossary"?) section where we could list the common terms, and there Japanese/Chinese/Korean equivalents. These could be wikilinked from where they first occur in the text. ~ J. Johnson (JJ) (talk) 21:02, 19 March 2016 (UTC)
I see that we already have a List of Go terms article, which provides Japanese equivalent terms (as that language is the primary source of the terms used in the West). I could see wikilinking to the pertinent sections there, though I don't know if adding Chinese and Korean equivalents there would be entirely satisfactory. We could also have a small section in this article, dealing only with the terms needed here, with a brief description and the various transliterations. ~ J. Johnson (JJ) (talk) 21:30, 19 March 2016 (UTC)

Number of players[edit]

The last paragraph of the lead section states "As of mid-2008, there were well over 40 million Go players worldwide." This is rather out-of-date, and the reference does not seem to be working. This source indicates about 60 million. Axl ¤ [Talk] 14:25, 18 March 2016 (UTC)

That source (the British Go Association) doesn't provide any source or basis for that statement, so I would be cautious in using it. But if can find a good source feel free to make an adjustment. ~ J. Johnson (JJ) (talk) 21:06, 19 March 2016 (UTC)

External links?[edit]

Seems to me several of the external links go against wikipedia policy. I haven't read up on policy recently, but I believe we should link sources of information, not online games.-- (talk) 12:32, 19 March 2016 (UTC)

Could you be more specific as to which links you question? I just looked, and they all seemed reasonable. Perhaps you could clarify what you mean by "online games" – there are several websites ("servers") where people can play online, but these are not what I would consider an "online game". ~ J. Johnson (JJ) (talk) 21:12, 19 March 2016 (UTC)
Well I reacted to the recent addition of one such link, and a brief glance suggested there was at least one more.
I now clicked all the links, and here are my brief impressions:
  1. - not obviously useful, but would require closer study to be sure
  2. - looks OK (meaning: I have not really tried to asses the quality of the page, but - well, looks OK!)
  3. - looks OK
  4. - seems to be a paywalled Go problem page; looks irrelevant (meaning: according to wikipedia policy, I don't think we should include this link)
  5. - looks OK
  6. - tutorial; we don't generally link "how to" pages (I believe), so looks irrelevant
  7. - online game; looks irrelevant
  8. - go playing program; looks irrelevant
  9. - Smart Game Format related software; looks irrelevant
  10. - not sure!
  11. - looks OK
-- (talk) 11:07, 20 March 2016 (UTC)
Thank you for your patience. While you initially said that several links seemed to "go against wikipedia policy", you didn't specify which policy, so I can't comment in that regard. However, upon examining the links I find the first and fourth ( and to be dubious; I would concur with their removal. The eighth (gnugo) is a program, which I think would be more appropriate at Go software. The last (Man vs. Machine, at is basically a blurb about AlphaGo; I think it is not appropriate here. I am going to boldly remove the last two mentioned, but will hold off on removing the first two as there might be a question regarding them.
You questioned (as "irrelevant") the tutorial #6), (#7), and the SGF link (#9). Not so! The tutorial is good, cosumi is an on-line go server, similar to kiseido, and the SGF software is a standard tool for accessing and sharing games digitally. You were "not sure" about #10 (the book); that's a classic introduction to the game, and quite properly present. ~ J. Johnson (JJ) (talk)

Concerning the presentation of categories a page belongs to when the list is humongous[edit]

It might not be related to the article's content, but look at the messy list of categories at the end of Go_(game) and Talk:Go_(game). It would look much better if it were presented in columns, don't you agree? — TentaclesTalk or mailto:Tentacles 18:20, 23 March 2016 (UTC)

P.S. I also added this comment to Wikipedia talk:WikiProject Categories#Concerning the presentation of categories a page belongs to when the list is humongous. — TentaclesTalk or mailto:Tentacles 18:20, 23 March 2016 (UTC)

 ??? I see only six categories at the bottom of the main Go article. - DavidWBrooks (talk) 18:28, 23 March 2016 (UTC)
OK, I see 1 line of categories at the end of Go_(game), but then there are 7 lines of hidden categories. Now, I see 8 lines of categories (not counting the 1 line of hidden categories) at the end of Talk:Go_(game). — TentaclesTalk or mailto:Tentacles 18:41, 23 March 2016 (UTC)
OK. Keep them hidden. Or kill them - the main accomplishment of categories, IMHO, is spurring edit wars; I've seen no sign that they're useful to readers. - DavidWBrooks (talk) 18:57, 23 March 2016 (UTC)
Even if I don't show "hidden categories" there are still 8 lines of categories at the end of Talk:Go_(game). I wonder whether some of those could be hidden. Anyway, I just thought that presenting categories into columns would be nifty. — TentaclesTalk or mailto:Tentacles 19:15, 23 March 2016 (UTC)
Since most of the categories of Talk:Go_(game) are added via templates, the presentation into columns would have to be automatized... if ever the idea was considered. (I guess the idea must already have been pondered.) Thanks. — TentaclesTalk or mailto:Tentacles —Preceding undated comment added 19:26, 23 March 2016 (UTC)


Why does "Goe" redirect here instead of GOE (disambiguation? That seems non-intuitive to me, and certainly not worthy of a hatnote, but maybe I'm over-reacting. Any thoughts? - DavidWBrooks (talk) 20:19, 13 April 2016 (UTC)

I agree, and I've been bold and done it now.-- (talk) 21:59, 13 April 2016 (UTC)