Game Of Go

Go is not the best game. It is the only game.


Go came from the Japanese word "igo", and in China it is known as the Game Of Weiqi, and in Korea the name is Game Of Baduk.


Go Resources On The Web refers to the Senseis Library and you can find the incredibly simple rules of Go: senseis.xmp.net .

Senseis Library is a wiki site and it is superb, definitely worth a detour.


Go's legendary history is as a teaching tool. An ancient Chinese emperor designs the game for his son, who he thinks is a little bit dense, and needs to learn something of discipline, concentration, and balance. The son goes on to be the first great player, and a good emperor, and a balanced human being...

Ultimately, all comparisons to chess are irrelevant and the beginning go player learns to let go of any comparisons. Chess is ultimately low-level binary thinking, of patterned build-ups to some somewhat unique problem for solving, and that's it. Once you learn what go is about, chess seems antiquated, like colonial British warfare vs American revolutionary guerilla (terrorist - i.e. 'outside conventional' - or [American foreign policy] - 'low intensity warfare') tactics. Go is deeper, as playing against any stronger player will show. As you rise to each new rank you gain a deeper appreciation for the subtleties involved, and the depth of stronger players. Games can provide a scaffold for the developing personality. Issues such as patience, and observance, are quickly developed with go.

Beginners always start by randomly placing stones on the board, as if it were a game of chance - and they certainly get destroyed. Soon, an understanding of how stones connect to form strength develops, shortly afterward, a few basic joseki (corner sequences) are understood. Further experience yields understanding of the board, the importance of the edges, then the efficiency of developing in the corners first, then walls, then center... a deeper kind of strength - influence, and its radiant effect on play in that vicinity. Soon, the advanced beginner understands that the balance between territory and influence are somewhat interchangeable, sorta like matter and energy. (E=mc2) But there needs to be a balance, i.e., develop more or less at the same rate as your opponent.

Thus the game is highly dynamic. As opposed to chess, which is regimented and stiff, uninspiring to the mind in need of challenge, and like I said before, antiquated, though younger than go. I might even go further and comment that chess is inspired to represent historically weak hierarchical structures (get the king) which are highly undemocratic, while go is fundamentally democratic.

-- Stevertigo


I think the notion that Go is superior to Chess, is like saying Baseball is superior to Basketball. Chess is not binary for humans, it is grasped intuitively and then calculated, Go leads to more wholistic intuition but it has a different feeling to it entirely. I do not see them as being related.

-- D


Go is one of the few games that has a sensible and practical handicap system that allows players of even substantially different standards to have a close game. Games are scored, too, and not simply win/lose, allowing near objective measurement of progress. I'm a weak player, but I wish I had more time to spend on this. It's utterly absorbing, and has a noticeable effect for the better on my programming thinking.

Re handicap: Nine stones is a lot. Someone who's only been playing a few months can have a competitive game with the strongest available computer opponents in a 9 stone handicap game. After less than a year, they'll easily beat the same computer opponent in an even game and will be able to have challenging games with a handicap against a 1-dan. That dan probably spent a year just moving up from 1-kyu (only one level). With 9 stones, they could have an interesting game against a top amateur player, or maybe even a pro. This makes being a beginner online a very friendly experience. If you show a willingness to learn, some stronger player is likely to take you under their wing. They can have challenging teaching games against you with a handicap, so they won't be bored. Likewise, they'll likely be someone else's pupil. Before long, you'll move up in strength and have lots of opponents at your level of play.


Great asian board game. More strategic/less tactical than chess (see below), but equally hard to play well. Very simple rules though - reminds me of Smalltalk :-) Oh, and computers are not very good at it. Good for the ego. :-) -- Falk Bruegmann

Less tactical than chess? I think I need to play you, Falk. One of us is going at this all wrong. Who are the other Go Players On Wiki? -- Peter Merel

On a 9x9 board, it's purely a (tactical) fight; on 13x13, some strategy comes into it. On 19x19, one thing to remember is that there are two strategies. One for black and one for white. They don't mix well. -- Hans Walthaus


WAY less tactical than chess. The first 4 years I played it, I played it from a tactical perspective. I won every battle I fought, and lost more games than I care to remember. It took me a long time to undo my chess heritage. -- William Grosso

That's interesting, William. I've only been playing a few years (and on-and-off, at that), and suffer the opposite weakness. Dan players who have kindly commented on my games say that my strategic play is unusually good for someone so new to the game, but I tend to throw that advantage away with clumsy tactical play. I never was any good at Chess.

The difficulty I've found in London is finding enough other weak players to learn with. Playing against Dan players with a large minority of the board covered in handicap stones is fine, but stops making sense after a while.

I saw no mention of computer Go programs here. As someone pointed out, they are all weak, but as Many Faces Of Go can still humble me at anything but 9x9, I still find it enjoyable (for definitions of enjoyable which include being made to feel small and stupid by an 800MHz laptop). -- Chris Hyser

Gnu Go is a notable open-source program of slightly greater strength than Many Faces Of Go v11. However, recent advances in Monte Carlo Tree Search have brought a huge leap in strength to other modern programs. Many Faces Of Go v12 won the 2008 Computer Olympiad Go competition, and is estimated at 1-kyu strength. Other strong MC programs are Mo Go, Crazy Stone, Leela, Aya, and open-source Fuego.


I don't think I agree that Go is less tactical than Chess, it's just that the strategy and tactics are differently distributed. When I lived in Edinburgh, a lot of my friends were gamers of one kind or another, and I tried to interest some of them in Go. Being soaked in a more "occidental" pattern of game structure, they found Go very difficult to get to grips with. One major complaint was that you have to keep swapping between (or, rather, combining together) strategic and tactical modes. -- Keith Braithwaite

I would tend to say (as someone who has spent a similar amount of life playing each Chess and go: low club std ..3 kyu) that Go is far more than Chess. That's what makes it hard for a computer running a min-max on some evaluation of forward positions. A computer cannot easily work out who is winning and by how much in a given position; even when it is clear to a fairly amateur human -- Andrew Cates

Learning when to play local or board is rather difficult. One thing to try is not to disengage the link between strategic and tactical thinking, but to 'embed' the tactical in the strategic. And since strategy is fluid in Go, try to make sure every stone you put down serves at least two purposes. -- Hans Walthaus


The main reason chess is less strategic than Go for me is that, at chess, my strategies almost never work out, because I'm just not at that level of play. There's always a blunder or opportunity a few moves down the road that I've overlooked, which invalidates all my strategic plans.

I assume that, in chess, the higher the level of the game, the more strategic it is. I don't know about Go, which I'm even worse at. -- Falk Bruegmann

Comments of high kyu players about a dan game:

'What the hell are they doing???'

'I dunno, you?'

'Well... It looks interesting.

Comments of dan players about a high kyu game:

'What the hell are they doing??'

'Well... It looks interesting.'


Go has a greater emphasis on sacrifice than Chess. I remember seeing a book which compared Japanese military campaigns with the game of Go, even trying to show situations in history and an equivalent position on the Go board. -- Peter Mielke


I'm reminded of an old aphorism about guitar playing: "It's easy to learn to play the guitar. It's nearly impossible to learn to play the guitar well."

Falk Bruegmann observed that the Game Of Go is reminiscent of Smalltalk - I agree with him. Also Lisp, Postscript, and Forth. In Go, the game is in the play, instead of the rules - in the same way that in Lisp, Postscript, Forth, and Smalltalk, the semantics are in the environment, not the syntax. Thus, the rules of Go are easy to learn - readily accessible to young children. The implications of the rules have been explored for several thousand years now with no end in sight.

I think the Strategic/Tactical dichotomy might be misleading, though. I think Chess has a very "hard" strategy/tactics division; strategic decisions are made rarely and are (at least to my eye) difficult to discern from a given board configuration. I think that Go, on the other hand, has a marvelous recursive quality - moves that are "strategic" at one scale (such as choosing a double-wing opening) are also tactical, at a finer grain (such as when they are played close to another stone). Stones played early, for strategic motivations, become crucial later as they become part of a "tactical" formation.

For me, there is a delicious softness and fluidity about the time and space dimensions of a Game Of Go that I find utterly absent from Chess.

Tom, that's bang on the nail's head, I think. You've re-stated much more clearly what I was trying to say when I wrote "One major complaint was that you have to keep swapping between (or, rather, combining together) strategic and tactical modes." -- Keith

Go proverb: In the beginning, have no plan. -- Hans Walthaus


Started playing a couple of times per week a month ago, and now I'm totally hooked. Why didn't I do this sooner??? -- Anthony Lander


At Need To Know's Extreme Computing festival on Sunday 9th of June, 2002 there was a representative of the British Go Association. Great bloke. He helped me finally begin to grok Go. I'd always played it like chess and it never really made intuitive sense to me. The point about Go is that the level of indirection is much greater than chess. Good players can look at a position and realize that certain pieces are effectively dead without having to go to all the trouble of capturing them. There can also be multiple separate tactical battles going on on the board, unlike chess where there is really only the one strategic war.

For people who are having problems with LifeAndDeathInGO, try to practice some Tsumego, which are small Go-puzzles. A lot of patterns can be learned and instinctively applied to the game. -- Hans Walthaus


sente:
the opportunity to initiate action.

gote:
the obligation to respond.

On average, gaining Sen Te is worth about 9 points. That's a lot - don't give it up willingly!

In the Game Of Go, gote is the condition of finding oneself obliged to respond rather than to initiate action. Initiating action is sente. An ancient but still well regarded Go Proverb states that Walls (aka distinctions) built in Go Te are never advantageous.

Not quite correct. You don't find yourself obliged, you oblige yourself. Typically, a player will have sente (reasonable freedom of action) and have a choice of moves, each of which will lead to a fairly predictable sequence of more or less necessary moves on the part of each player. If the last such move in such a sequence is that of your opponent, then you are playing the sequence in sente. If the last such move in the sequence is yours, you are playing the sequence in gote. Accordingly, one would generally like to play all the sente sequences first, then finish with a gote sequence.

But you don't necessarily want to play all your sente moves. If you play one that only costs a small number of points to ignore, the opponent may well ignore it and move on to a high-point sente move of their own. Sometimes, foregoing a piddly sente move for a large gote move makes sense.

It can also be useful to save a few sente plays for use as ko threats.


Something interesting about Go - all computer programs are weak. A person who's only played for less than a year can beat the strongest computer opponents in an even game. By comparison, computers bested the world's best checkers players more than a decade ago, and they're now starting to surpass the best grandmasters of chess. A traditional Game Tree search is futile - the number of legal or potentially good movies is far too large. Compared to computer chess, computer go is a much more Interesting Problem.

Also interesting, but unrelated... Go supports many different styles of play. If you are strong tactically, but less so strategically, you can play an aggressive "in your face" style (within reason). You can gain territory by attacking the opponent, killing their groups, destroying their moyos (regions of influence). Someone who prefers the more strategic aspects can take a more passive approach, marking out huge moyos in the beginning, forcing the opponent to attack, and thereby transforming the moyo into bona-fide territory. It is also interesting to see these styles at all levels of play; neither is "better" (though certainly, high-level players have found a healthy balance). The aftermath of different strategies can be read on a board after play. There might be large, curved elegant shapes, or a chaotic jumble of dragons and tiny barely living groups.


A quick question: Is it allowed to link Go terms to the pages at Sensei's Library? Or would people like to see them in this Wiki too? -- Hans Walthaus

Yes it's allowed, yes it could be useful and interesting.


I have never played Go, but from what little I know it seems more visual than Chess, and thus I might find it more appealing, being a visual thinker. Chess's rules often have little or no visual correspondence to the board because you have weird "hopping rules", but Go seems less so.


I find it helpful to play Go on a physical set, so I must think through the results of my actions, instead of impulsively clicking at a program and looking at the results.

From my experience, learning to play Go is about learning to see the relationships between pieces (and groups). The farther ahead (and wider) you can read the game, the more structure you can see, and the deeper the apparent relationships become. Go also tends to display fractal-like relationships across different scales of development that occur in the game.

The learning process is naturally recursive (or iterative), since learning to (automatically) see deeper lets you (manually) evaluate the next deeper iteration, and so on.

... and deeper, and deeper, until ...

Sometimes I have this problem, where I've been very focused and become deeply involved in a game of Go, which then refuses to vacate my mind even when I'm done playing. My brain just keeps processing potential permutations of the game, and they continue to appear right before my eyes. It often carries on into my sleep and dreams, and seeing Go patterns in other things, sometimes up to a day or so after playing. Frankly, it's disruptive until it goes away, but I suspect I must be a better player for it.

I've heard of such phenomena in players of other pattern-heavy games (Tetris for example), but I never experienced it outside of playing Go. It's awfully distracting, not being able to shake ones prolonged accidental focus. It seems to happen most often when you've been In The Zone. Note, however, that since Go does not admit ready mastery, being In The Zone is more akin to an accelerated state of learning than to knowing perfectly what will happen before it does (as is possible with certain games).

Go is like a game of cellular automata-type rules that can run on the human brain. Once you learn enough to understand it, you become a human Go evaluator, exploring the life and death of evolving groups, playing them out one game at a time.


The thing I enjoy most about Go is its application in day-to-day life. The skills (or, more accurately, the attitudes and approaches) necessary to play the game, as well as the Way that these skills(/whatever) are applied (i.e. patience, knowing when to strike, knowing when to give in and change your strategy, etc) are very analogous of the skills necessary to understand(/enjoy/succeed) in just about anything in one's own life.

Hence the proverb:

Student: Master, what is the most complex game that man has invented?

Master: Chess!

Student: What about Go?

Master: Go has always existed.

That is a good one, put it on Senseis Library if not there already


I think that exchange doesn't emphasize the "origin" enough. And "complex"? Eh...

Student: Master, what is the greatest game invented by Man?

Master: Chess!

Student: What about Go?

Master: Go existed before Man did.

Well, perhaps, but that's not the proverb. That's a different saying, and it's saying something different (IYSWIM)


Q Is there an inexpensive but useful Go client that works under Pocket Pc, one that can connect to the likes of IGS and LGS?

A ?


I got really into Go for a while, probably was about 6-8 kyu. Then I noticed that after a few games I started imposing a square grid on the world and seeing everything in terms of Go moves... I don't play much any more.

My favourite Go proverb: "Thinking twice is enough."

Go on a square 2n+1 x 2n+1 grid is one thing, but has anyone here played alternative versions of Go? for example...

Go on boards with other topologies (sphere, torus, klein bottle... depending on how the edges wrap around)

Go on a board without a grid (where the exact location of the stone is crucially important... if you can't squeeze a stone in between two of your opponent's stones, then they must be connected. No moving stones!)

As above, but played with natural shaped stones instead of round ones (I made a set like this once)

Go with three players (black, white, red)

Any more...

Zen Go: You can play a regular game with three players, so that each player alternately places white and black stones. Play proceeds 1-B, 2-W, 3-B, 1-W, 2-B, 3-W, etc. The object of the game is to approach a score of zero. This is a nice way to learn to play beautifully since it takes your ego out of the equation. It is kind of like Pair Programming, since you try to work within your co-players' flow of moves.


I'd like to play a game of Wiki Go with you. -- Go No Hikaru


I never liked Go as much as chess. It is less exciting than chess. A beautiful chess game is appreciated at each move and as a whole, whereas in Go, it is only the whole. Go is like the forest (a bigger forest), whereas chess is the forest and the trees. The immediacy and importance of EVERY SINGLE move is more exciting than the abstract loose moves of go. I can't think of any go games where every move in the entire game is profound and amazing and beautiful, there are fewer 'best moves' in go, usually you have several possibilities. In chess, there is often that elusive beautiful single move that is so challenging to find. As for difficulty, the comparison is useless since each is so difficult than no one except a handful in history have come close to the point of exhausting most of it. Go is a lot more refreshing, in that simplicity transforms into complexity, and is a welcome break from chess. Also, the freeform openings, unlike the fixed ones in chess, make Go less daunting, which is nice. But Chess definitely wins when you examine an amazing game (and all the commentary on it). A sacrifice that extends for 20+ moves with no short-term gain and then wins - Wow!

Believe me, everything you like here about chess can be found in Go, but there are almost a magnitude more levels of expertise in the game. At the 9-dan pro (grandmaster) level, every move is important and profound, but often for reasons not apparent to a novice. Try some Life And Death problems if you like those 20-deep combinations.

There are two things I don't really like about Go. One of them is similar to what's up there, which is that Go is not nearly as suspenseful and can never be nearly as suspenseful, because it is a score-based game(get more territory) and not a goal-based game like Chess(capture/checkmate the king). And don't even try to make Go a goal-based game since that would ruin the thing I like most about it(the incredibly flexible and elegant handicap system). The second problem is that to a Mere Human like me the moves of the go masters often don't make sense. It's a bit too deep, and with games that are less deep it's easier to jump in without drowning in "WTF?"s. Now the problem with Chess is that computers have solved 90% of it. My current favorite Abstract Strategy Game right now is Connect Six which has very simple rules(the rules of Connect6 on Wikipedia are about half as long as the basic rules of Go on Wikipedia's page "Go", and that's not even counting the very long "Rules of Go" page with all the subtle variations!), is more suspenseful(and more understandable, though the best players' first moves often still don't make sense, but usually only the first moves) than Go, has a much larger state-space complexity than Chess(actually, when played on a Go board, state-space complexity approximately equals Go's), and can be played with much less equipment than Go or Chess.(It's usually played on a Go board but I've been mostly playing it at school with graph paper and pencils making Xs and Os.)


Why is it that some go players seem to routinely go out of their way to badmouth chess? Perhaps it's just a few (albeit a noisy few), but this page seems to be full of criticisms of chess - the vast majority of which seem to be elements of personal preference. Chess players don't go out of their way to bash go; and many of us enjoy both games, without feeling the need to elevate on above the other, or criticize either game (or those who play it).

It's understandable that go players may want to share the love of their game. But y'all act like a bunch of Amiga fans sometimes. :)

As someone who appreciates both Chess and Go, and was a former Amiga hardcore (and, for that matter, still a Lisp fan), I resent that last comment, while agreeing with your original point.

Go fans, both here and elsewhere, insult Chess far more than is deserved, out of a kind of "my team is better than your team" sort of loyalty. In truth, both Chess and Go are beautiful, elegant games with extreme depth that has not been fully explored, by any means, in both cases.

That said, there are a few things, but only certain things, about Go, the original Amiga in its era, and Lisp, that are in fact quite amazing compared with everything else. That's not to say that the alternatives (Chess, other computers of that era, other languages) suck, which is the logical fallacy that the most emotional extreme fans make. Certain things are indeed objectively amazing by certain criteria; that doesn't mean that the alternatives have no virtues.

Regarding Chess and Go, they both involve strong tactics and strong strategy, but in different ways, and of the two, Chess has been better analyzed to date, which is why there are stronger Chess programs than there are Go programs. It is interesting that Go is resistant to the same kind of analysis, considering that there are humans who are extremely good at it, but it's important to keep in mind that expert performance involves different brain areas than expert explanation of performance, and that the former has universally greatly preceded the latter.

And there are very, very few humans who are deep experts at both Chess and at Go, yet such people are the only ones who have much worthwhile to say on the topic of comparisons - we can safely ignore rantings by e.g. 4th dan Go players.

In other words, once Go is understood analytically as well as Chess is currently understood analytically (and there's no reason to assume that will never happen), we will be in a position to, in theory, write computer Go programs that are as relatively good as are our best computer Chess programs (other than O(n) issues which we currently have absolutely no basis for analyzing).

P.S. A good way to understand the very most fundamental difference between Chess and Go is to implement computer programs for their most trivial variants, respectively Tic-Tac-Toe (Naughts and Crosses) versus Gomoku (or its Western variant, Pente). It is trivial to write a perfect Tic-Tac-Toe algorithm, and the difficulty of implementing a perfect Gomoku/Pente algorithm is similar, with a larger multiplicative constant - yet the former is easy for programmers, while the latter requires new ways of thinking; it's not that Gomoku is a vastly deeper game - it's not, within that multiplicative constant, it's that, to do it well requires a different approach, more like image recognition (a known difficult soft AI problem, at best); doing it the same way that one might implement Tic-Tac-Toe is simply inappropriate.

Using Tic-Tac-Toe-like rules regarding position for Gomoku results in a combinatorial explosion that is not well handled by tree pruning algorithms. It's not clear what is best for Gomoku, since that might shed light on what works better for computer Go programs, but it absolutely works better to go directly for higher-level pattern recognition - which interestingly enough, was the subject of recent discussion on a Chess-related page.

Contrast this situation with that of Checkers. The best computer Checkers programs have, for quite some years, been better than the best human players, unsurprisingly considering Checkers is a simpler game than Chess. But approximately the same algorithms that work for Chess work even better for Checkers. This is different than the situation with Gomoku. It is possible for a non-expert to write a Gomoku/Pente program that will always beat/tie humans, if the right pattern-recognition approach is taken, with a fairly small and simple program. I'm not at all sure that any similarly excellent Gomoku/Pente programs have been written using the kind of algorithms that Checkers programs use. I doubt it.

The point of which is that we're not just talking about simpler kinds of games, we're talking about games in rather different categories.

Grok that low-level difference before you even begin to think of comparing Chess and Go.

Oh...and for the most extreme of hardcore Go bigots: if Go is just better and harder, rather than merely different, why is it that the very best of Go players don't win every Chess championship? I mean, if Chess is so trivial by comparison, surely that would be a no-brainer. (That was rhetorical; the actual answer should be obvious, and it's not that the best Go players would sneer at being crowned the reigning Chess champion of the world.) -- Doug Merritt

One fascinating aspect of Go is the handicap system - it allows someone with 3 years' experience to play a reasonable game with someone of six months' experience. This is "the area" where I do not find close parallels with other strategy games.

Yes, certainly an important difference. But there is a similar thing in Chess; I have played against very weak players by giving them, for example, a queen and bishop handicap at the start of a game. It's not by any means identical to the handicap system in Go, but neither are the two approaches beyond all comparison. The advantage of the Go handicap system is that it is more linear, and more accepted by players, while the Chess handicapping approach is much more non-linear and much less readily accepted.

Please see talk about Handicap system in the beginning of page. Handicaps in Go can be offered in terms of "komi (score points)". And the "more linear" nature of Go in effect, allow meaningful "teaching games", using the handicapped game itself. In your case where queen is sacrificed, it will not be useful to get the other party to learn to play better. I know only one case of strong chess players converting back to playing chess, the reverse example is much more abundant. Go players are not superior, Go game is "more distinguished", if that is a more acceptable word to you.

Later... there is a variant of Go playing called PairGo. It is a way to make enemies quicker than playing the Game Of Bridge, unless you enjoy getting tormented :). Ask people about Rengo room on KGS.

And later again, I now have my "fix" of Go stuff (much longer than planned), and all that time that has taken me away from my passion about Computing and Patterns and ....


We should probably split this into Chess And Go Comparisons. Another thing to note is that while Go is more complex, that doesn't mean it is more difficult. I won a pawn in the opening from a computer rated 900 points higher than me (it was 2600) on its most difficult setting (hours per move). It made a blindingly obvious bad move that even a beginner would reject, yet it led to a pawn loss >10 ply later. It developed the wrong rook to the center which intuitively 'looks wrong' to a human but only had a positional (not tactical disadvantage - BUT that weak position eventually led to a tactical exploitation.

Also I don't think you can say chess is more tactical and go is more strategic. You can say that in chess your strategies are only as effective as your tactics. In otherwords someone with a 4/10 in tactics and 1/10 in strategy will usually beat someone with a 2/10 in tactics and 10/10 in strategy in chess.


[Stuff about the rules of Go removed; try a search engine or the link at the top.]



See original on c2.com