The game of go
Two basic rules
The goal is territory
I almost forgot
Computers and go
(Dutch) Organizations and Internet resources
A game of Go often is a dazzling experience. Small-scale fights slowly develop until their respective battlegrounds begin to overlap. Then anything can happen. The terminology of Go often reminds of warfare. How about these?
This terminology is tainted by the horrors of real warfare, and it must be hoped that this will cease to be the case someday. All the same Go is a fascinating game which seems particularly suited to the human's mind and intuition.
Below I give an overall impression of what the game of Go can be like. I do not supply a complete instruction set for learning the game - at the bottom of this page you will find pointers to Go instruction pages. Eventually the stuff below might evolve into an instruction set of its own. However, I firmly believe that the best way to learn go is have one person who knows it explain the game to at least two people who don't. Then the fresh people should begin playing each other, the experienced player only interfering to assist them if they are unsure or unable to proceed.
The game is played on a square grid, which can be of different dimensions. The customary sizes are 19x19, 13x13, and 9x9. The larger the board (grid), the longer it takes to complete a game, so beginners are best of playing the first few games on a 5x5 or 7x7 board. Standard tournament practice is to play on the large 19x19 board. For some reason the dimensions are always odd, never even. Nothing in the rules stops you from playing on a 10x10, 20x20, or 5x16 board, but it is probably best to save your creativity for other aspects of the game. The game is played using black and white stones, and there are very many of both. In theory you can use as many stones as you like; in practice a supply of 180 stones for both colours is guaranteed to suffice; being ten or twenty stones short of 180 won't harm at all. The number 180 is for the largest board, which has 19x19=361 intersections. For the 9x9=81 board 40 stones of each color are more than enough.
The game starts with an empty board, one of the two players taking the black stones, the other taking the white stones. Black always starts to play. Both players take alternative turns in placing exactly one stone on the board. Stones are put on the intersection of two lines; this can also be on the border of the grid (where only three lines extend from the intersection) and in one of the four corners (where only two lines extend from the `intersection'). Stones do not move once they are placed, unless they happen to be captured later.
One of the appealing aspects of Go is that there are just two basic rules, apart from the organisatory stuff. One says when two friendly stones (i.e. a number of stones played by the same player) are connected, and the other says on which occasion a group of connected stones is captured by the other player. Moreover, both rules are quite intuitive; Once you know them, or even before, they are both clearly the natural thing. The game of Go can make a deep impresssion, as from simple and elegant rules an intuitive game full of beautiful patterns unfolds.
These rules define the context, as the goal of the game is to surround territory. This must be done by putting your stones in such a manner on the board that they form the outline of walls that will eventually become connected and surround territory. Now the problem is that you would like to build solid walls, because that is safe; However, you use the same amount of stones as your opponent, and using solid walls you do not get so much territory - this is really the same as building a house: The more solid the walls the safer, but the less space you get to live in (if you have a given number of bricks). This means that you have to compromise between ambition (aiming for lots of territory) and safety (being able to hold on to your territory). Note: a wall is not something in Go for which there is a rule; it is just a pattern that arises, and which can be given a name. The best thing is to be a superb bricklayer; to know how to build strong walls with only a few stones, and to know which spots are weak and need inforcement, and which spots can do without. And to know the weak spots of your opponent's walls :)
So that is one issue. Another one is that you compete for territory with your opponent. It is possible that both of you lay claim to the same piece of bare land, each of you trying to make it into your own territory (or terra as it is sometimes abbreviated on the Internet Go Club). In this case it often happens that rather than mapping out territory, you find yourself at war with the opponent, each of you trying to cut the walls of the opponent into little pieces and trying to capture some of them. If you succeed that is usually good, because both of you need pieces of wall to build territory. But watch out! Beware of your opponent feeding you the doghouse while he is out building a mansion. Such situations lead to all the tactical and strategical finesses listed at the top of this page.
To mention that one of the nice things of Go is that it has a very natural mechanism in which players of different strength can play an evenly matched game. Normally the game starts with an empty board, and taking alternating turns, the players start staking out territory by putting stones on the board (one at a time; stones do not move). Now if there is a difference of strength this can be balanced by letting the weaker player put a few stones on the board in advance. He gets a little headstart in the claims to territory and influence. These advance stones are called handicap stones. This really makes a difference in that the stronger player will have to try very hard in catching up; On the other hand, the nature of the game is not affected. The ensueing game is as normal a game of Go as it would have been without the handicap stones. The number of handicap stones can be tuned such that the odds are equal for both players. This is in fact used in creating a classification system of strength, the difference in ranks being equal to the number of handicap stones needed to equalize the odds.
It should be noted that so far the best go-playing computer programs are crushed by the average club player. Humans are capable of assembling intuitively (the more experience the better) the patterns arising on the board into meaningful subpatterns, and are able to see subtle exchanges and shifts in the relationships between these patterns. Go poses a great challenge for Artificial Intelligence, because the usual mindsport paradigm in computer programming (using a game-tree) does not work well at all. It could work, if the branching steps were formulated in terms of higher-level patterns, and if the evalutation function for the leaf nodes could also make use of higher-level patterns (simpler evaluation functions are doomed to fail). The fluid nature of patterns in Go makes this a very tough challenge though.
Find more insights on computer Go in this New York Times article , or in the Mindzine Computer Go special. At Science News you'll find another account of computers and games, including computer Go.
For a start, I occasionally create some PostScript tools I find useful. A script that outputs a highly customizable range of Go boards and some of the boards that it spawned can be found here. In the PostScript section you'll find a Perl script which converts ascii tsume-go problems to pretty-printable PostScript output. It is not feature rich and the code is unfortunately not very maintainable. There is probably not enough abstraction in neither the Perl code nor the PostScript code.
is de pagina van de Nederlandse Go bond, en
vindt u compact vele verwijzingen (go.pagina.nl).
The European Go Cultural Centre is located in Amstelveen, the Netherlands. Their site is here.
Well known, comprehensive, and of high quality is Jan van der Steen's site. It has very much to offer on news, games, problems, proverbs, software, statistics, and what not. A must-see if you haven't. There is also an introduction to Go.
The Mindzine Go News index used to be a nice site for recent news on go, but it seems that financing dried up and the site is no longer updated.
Andries Brouwer offers a tarred gzipped collection of 1111 games. Given the stern nature of this number, let us hope that his next release will contain 2222 games (or perhaps 11111, or 1339).
Jan van Rongen has a large collection of Cho Chikun games. This site is definitely worth visiting.
An impressive amount of go-songs was collected by David King. The British go-players are famous for their creative outbursts.
At Jean-loup's Go page Unix and Linux users can download the highly recommended Internet Go Club client named xgospel, maintained by Jean-loup.
The Internet Go-school of Guo Juan.
De gopagina van Andre Engels.
A very good initiative is the European Rating List. Go either to the site where it is computed or to the official EGF ratings page. The latter used to be slow, but lately the connection seems to have been improved.
Sensei's library is a collaboration web site around the game of Go, started and maintained by Arno Hollosi and Morten Pahle. Anyone can add content to existing sections and threads and start new threads. No passwords or user accounts required. Features include on-the-fly creation (by you!) of cool-looking Go diagrams using a very easy ASCII format. Radical technology and an inspired initiative.
Playing Go in Amsterdam.
The Amsterdam Go tournament.
E. told me about this very nice page: A Kid's Guide to Playing Go. Thanks!
A place cramped with go-related links is The Web Go Page Index , though they claim to have only a small list of web pages. There are many sections, and they also have a section with links to rule explanations and tutorials. Please beware! Going there, you might note that there is mention of different rule sets, and you might mistakenly think that there are several kinds of Go. It is true that there are slightly different rule sets in use around the world, but this is inessential. The way in which the game is played is exactly the same in each case, the differences only affect the final stages of some games that are freak cases occurring only once every 100.000 games or so. If you are new to go, then at the link above, concentrate on the tutorials, and skip everything that says something like rule set.
Come to think of it, I include some links to tutorials here. John Tromp's explanation uses a nice small board, but in the example game he uses you will find none of the walls and house building I mention above. This example game does not give a good impression of the goal of Go, which is to make (more) territory than your opponent. It does clearly show the concept of the two basic rules though; when stones are considered connected and when (groups of connected) stones can be captured.
In Jan van der Steen's introduction you will find a nice instruction (example) game, which does indeed nicely illustrate the idea of building walls to make territory. This game does not feature any capturing, probably for a good reason. You see, if you start learning to play the game, it can be confusing at first because there are so many moves you can consider. Then there is the somewhat vague goal of making territory (how in earth do I accomplish this?) and this complicated thing about capturing.
An example game that illustrates both concepts simultaneously is likely to be confusing for some. All that it takes however is to play a few games, preferably with another beginner, and preferably while someone who knows the game is watching so that you can ask all the questions you like. That is the best way to learn the game! Learning Go is a lot like learning how to ride a bike. At first it seems a bit impregnable, but after a short while you get the idea. The analogy stops there however, because one can always learn more about the game of Go (no offence meant, bike enthousiasts), while it stays a highly enjoyable game at all levels of play.
Back to Stijn's index page.