Go Rule Dialects

Remark: the Game of Go is also known under the name Igo (Japan), Wei-qi (China) and Baduk (Korea).


A short, concise set of Go rules

Trivial knowledge preasumptions


  1. Two players, called Black and White
  2. Sufficient black and white stones
  3. A square (Cartesian) grid of size 19x19 is used as board which at each of its 361 intersection points can have the state of either be empty or occupied by either a black or a white stone. Each of the grid line intersection points is referred as a position or point of the board. The 'whole-board-state' is denoted as configuration.
  4. The players like to play a Game of Go and agree mutually of a komi which is an integer or half of an integer. If this agreement doesn't take place the komi is defined to be zero.
  5. Further the player must agree whether they like to start with an empty board or put some (black) stones, called the set-up (handicap), in agreement on the board before playing starts.

Playing procedure

  1. Black starts playing as far as there is no handicap --- which is considered as Black's first move --- else White is to start.
  2. Playing means: the players make their moves on alternate turns.
  3. Each move of a player is either passing --- the current board configuration is not changed --- or put a stone of his color on an empty position of the board. The last option can result in a removal of one or more stones of a color. This removal of stone(s) occurs if and only if:
    1. there are either opponent stone(s) which have after placing the just played stones no liberty, these must be removed,
      or, if this is not the case,
    2. if own stone(s) (necessarily including the just placed stone) still have no liberty, then those must be removed.
    A liberty of a stone is an empty position on the board which is connected to this stone by a vertical and horizontal path of stones all of the same color as the considered stone.
  4. A move is forbidden if it changes the configuration and recreates a previous configuration.
  5. If both players pass in a row the playing procedure ends.
If any player violates the playing rules, the Game immediatly ends, he loses and his opponent is declared the winner.

Determining the Score and the Winner

  1. If the playing ends with two passes, calculate the difference of white stones minus black stones now on the board and add the komi. This is called the score. If the score is positive White is the winner, if it is negative Black is the winner and if it is zero a tie has occured.


Variations and Time of Playing

Variations which doesn't effectively change the score, but shorten the playing or the scoring procedure

  1. Instead of violating the rules during play, a player may 'resign' which indicates that he immediatly lose the game.
  2. If the players agree in common for each empty point which belongs white, black or noone --- as it would belong if they would continue the playing --- they can stop playing and count all empty points as if they were colored by the agreed color.
  3. Further, if the players agree in common for all stones which of these can be removed by further play and their positions are considered to be occupied by the opposite color, their removal during play is not necessary and the playing can stop. These stones will be removed immediately after the playing has been ended.

Variations which can change the score and need mutual agreement before playing starts

  1. Putting a stone such that only stone(s) of its color has/have to be removed is forbidden.
  2. Recreation of the some configuration is allowed as long as this configuration is more than 3 moves ago. If this occurs and both players dislike to leave the board configuration cycle by playing the same positions, the game ends immediately and is either to be declared to be a draw or as to be not effective played. Which possibility is to be chosen, has to be agreed upon playing starts.
  3. Passing is not allowed. At each move a player must put a stone on the board. The end of the playing procedure must be declared by the players' agreement. A deeper analysis shows, that in this case one needs to modify this restriction such that passing is only allowed if the player to move returns one (earlier) captured opponent stone to avoid the possibility that the 'true' winner is defrauted of his 'natural' win --- which else of course would be very inpolite.
  4. Instead of own colored points, only empty, but by own stones controlled points count for the finally score. In this case, also the removed and removable stones of the opponent add to a players gain.
  5. Special groups or configuration of the board can be assign a value number in the scoring procedure instead of counting their belonging stones.
  6. Instead of a 19x19 grid, a rectangle Cartesian grid of any size can be used as board.

Time restrictions for play

  1. For each player a sufficient clock / time control mechanism must be in place to indicate his actual time supply.
  2. A fixed base time for each player might be given for his playing procedure.
  3. Repeatingly, a fixed at most time might be given each player for a fixed number of moves which repeatingly follow his moves made in his base time.
  4. A fixed time bonus might be given each player for each of his moves, immediatly after the move is completed.
  5. The real time a player spends on his move is subtracted continiously from his time supply. If his time supply becomes negative, he immediately loose the game.


Variation overview

Principle Variations

  1. Set-up (pre-turn phase)
    1. initial board configuration
      • handicap
        • free handicap points to choose
        • fixed handicap pattern
      • even game
        • empty board
        • fair preset (symmetric if colors are exchanged)
    2. komi (value to balance the first move advantage)
      • no komi
      • 4.5 - 8 komi given white in case of even game
      • 0.5 komi given white in case of handicap game
  2. Turn Phase
    1. clearify whether passing is allowed (putting a stone on the board is a right or an obligation?)
      • passing allowed
        • nothing to change in the whole game status
        • give the opponent one of your own stones
        • give back the opponent one of your captured stones
      • passing disallowed (player must change the board configuration)
    2. clearify whether a suicide move is allowed
      • generally forbidden (player is any move denied which only takes away the last liberty of own stone(s)
      • single (only just played) stone suicide disallowed, but multi stone suicide granted
    3. clearify whether a move recreating a previous board configuration is forbidden
      • always forbidden (positional super-ko-rule)
      • only forbidden if the same player as previous to move and the same basic ko-plays are identical (situational super-ko-rule)
      • only forbidden if it is two moves ago (ko-rule) else the game result is to be judged as a tie or 'no-result' (game to be replayed)
  3. End of the Turn Phase
    1. both players agree to finish turn phase
    2. if passing is allowed consecutive passes finish turn phase
      • 3 consecutive passes started by the start-player (black) finish
      • 3 consecutive passes always finish
      • 2 consecutive passes always finish
    3. no legal move for the player-to-move exists (thus, passing disallowed) and then he is defined to be the looser
    4. a player resigns (polite alternative to violate the rules) and then he is the looser (no scoring)
  4. Board Scoring
    1. only colored points count (called stone scoring)
    2. colored and empty points may count (called area scoring)
      consequence: removal of opponent stones in turn phase doesn't change later score
      • some opposite colored stones are allowed to be emptied by agreement (if disagreement then turn phase restarts)
      • the rule set defines explicitly who controls each empty point by
        • empty points surrounded by both colors count for both players equally (modern chinese rules)
        • empty points surrounded by both colors count weighted by the topology (SST rules)
    3. only empty points may count (called territory scoring)
      consequence: removal of opponent stones in turning phase changes later score
      [further consequence: a captured or 'death' stone has the same value as one empty point]
      • some own colored stones are allowed to be emptied (Sun-chang Go rules)
      • some opposite colored stones are allowed to be emptied (Japanese Go rules)
      • the rule set defines who controls each empty point by
        • status depends only on the color(s) surrounding topologically the empty point
        • status depends on the local concept of live and death (explicite Japanese ruling)
Remark: the conditions 1.2, 2.2, and 2.3, as well as the alternative scoring principles --- effectively the slightly different aims of the game --- 4.1, 4.2, and 4.3 might result in different strategies, tactics and results!

Rule References

The excellent, english writen reference I known is:
Mathematical Go / Chilling Gets the Last Point by E. Berlekamp and D. Wolfe,
1994, A. K. Peters, Wellesley, Mass. USA, pp. 113-168.

(also published as Mathematical Go endgames by Ishi Press International, San Jose, London, Tokyo, 1994 in Soft-cover) which gives an exact, concise and detailed overview of more than 6 dialects of Go, including: Ancient Chinese, Modern Chinese, Japanese, Mathematical, and North-American rule set.
For further refinements or rules like the SST or New Zealand rule set see:
R. Bozulich / The Go Players's ALMANAC,
ISHI Press, San Jose, Tokyo , 1991

The abstracter but less ambiguous Tromp-Tayler rule set and Robert Jasiek's collection of links to the topic Go rules is also noteworthy.
Also note, that exact stated Go rules are an achievment of the 20th century. Before about 1930, there was spread only the traditional Chinese, the old Korean and the Japanese Go rules, except same minor local variations --- the only excpetion of this seems to be the Tibetan Go rules --- , and they all were given through oral delivery.

First we list these Go rules which has been played by very many (> 100 000) players:

subject / dialect Japanese Ancient II Ancient I Korean China SST Taiw NA NZ
Set up stones no 4 4 16+1 no no no no no
handicap placement fixed non non non free free ? fixed free
Passing allowed yes no no no yes yes yes yes yes
Suicide allowed no no no no no yes yes no yes
Configuration repeation cycle length >2 >2 >2 >2 >2 no ? no no
board scoring type territory territory stone territory* area area area area/territory area
worst case of scoring procedure difficult+ difficult simple simple modest simple simple depends simple
average time the tedious phase lasts 10 10 20 12 15 19 11 10/15 11
Let the 'tedious phase' start when only 1 point gote moves exist, continuing the scoring procedure and after this, lasts until the stones are again arranged in two heaps of black and white stones as at the beginning of the game.
*certain stones of the own boundary were removed.
+here the japanese rules variates with the history

Secondly we list these Go rules which has been played less players which is due mostly to the fact that they are a modern invention (the earlier SST-rule variations count probably also into this category). Of cource this list can't be complete because Go rule theoreticans (mostly self-named amateurs) consider more and more variations. Hopefully here are given the more important ones:

Tibetan Rules
Very old, probably the oldest delivered, but nearly out died (destruction of tibetan cultur respectively life foundation of monks by chinese politics and military since the occupation of Tibet in 1950).
Firstly, a special fixed 12 stone set up on the 3rd line is placed and a 13th (black's first move) stone in the center, secondly putting a stone on an empty point which was just emptied (by the opponent's previous move) is forbidden (no snap-back possible), thirdly a restriction where to place the next move in distance consideration to your previous placed stone, fourth having colored special points at the end gives raise to extra values (and maybe some more we have no knowledge anymore).
Mathematical Japanese/Chinese/Ancient/universal go rules
In the early 1990 Elwyn Berlekamp, a profound mathematician game theorist, created these rules to approximate the Japanese, Chinese, Ancient I Go rules. The last named seems to be almost perfect approximated, say equivalent. In contrast for the japanese there this seems to be totaly out of reach because of many illogical scoring cases which needs interpretation by authorities.
One of these Mathematical Go rule variations considered by Elwyn Berlekamp seems to be the shortest and most elegant: here scoring is never necessary to determine the winner but needs the 'if one passes then one must give one captured stone back' clause.
In his high level mathematical research on best (end-game) play he needs to consider the effects of various Go rules on his theory and therefore created also the above mentioned mathematical variations of the go rules. He observed (and stated) that in the mathematisation of go rules, when it goes to scoring evaluation, the "2 point group tax" occurs naturally which indicates that stone scoring seems to be the simplest (and logical conistst) form of scoring.

Historical Development in Scoring Philosophy

In the philosophic development of taking control as much as possible of the board during playing, it is notworthy that in ancient times, the vacant points counting which is territory scoring stood at the (historical delivered) beginning of Go. This was changed in the early Ming-Dynasty to the stone scoring philosophy but effectively used as area counting with 'group-tax' to shorten the endgame play. This usage assumed the knowledge of the term 'group', 'eye', and 'seki'. Technically, it preassumes special knowledge about removable and non removable stones, but in practice each Go player has such knowledge. In this the 20th cnetury this 'group-tax' was abolished by the Chinese in an attempt to make the score similar to that of the Japanese and they introduced area scoring.
The Japanese, whose Go-theory and Go-player strength was undoubtfully the far highest developed upto the eighty-years of the 20th century, kept their territory --- empty, but controled points --- scoring since the introduction of the Game from China, via Korea to Japan. Effectively, it is mostly still quicker than area counting, but it can be only equivalently to area scoring if the passing player is forced to give a stone to his opponent (because there can be (many) empty dame points occupyable by only one player) and if the second player puts the last stone onto the board --- both is demanded by the NA rule set. Also, the japanese rules count only territory which belongs to an immortal group (what is the status of bent-for-in-the-corner?) in contrast to points in a seki. All this indicates (and needs) a far higher level of abstraction and historical influences than necessary for a simple Go rule set. They (japanese Go authorities) changed e.g several times the status of moonshine-life by current definition, which reflects the controverse view on decisions based on local considerations versus global board configuration. See also Robert Jasiek's detailed Scoring classification of different Go dialects.

It become evident in the 19th century by the Japanese Go players that black has a significant advantage by moving first in an even game. But only in the 20th century komi was invented by the Japanese players to give white a fair chance to win an even game. The Nihon Ki-in started with 4.5 komi, increased to 5.5 komi(around 1970?) and defined in the beginning of the 21th century a 6.5 komi for professional tournaments.
Here is a nice pointer to Ancient Chinese rules and philosophy

Time scale of Go rule changes

In ancient times there were only the different chinese culturs, the korean and the japanese cultur playing Go in a noticable number. Undoubtfully, the Game of Go has its origin about 2000 B.C. (perhaps even earlier) somewhere in the mainland of China. The first records of this Game describe it as 17x17 board as well as a 19x19 board! Sadly, first game representations are only from about 200 A.D. Some times about 500 A.D. the Game of Go must come via Korea to Japan. From the very first game records in China upto the middle of the 20th century, a preset of 2 black and 2 white stones at the starpoints diagonal opposed was common sense in China. In Japan this preset must have been abolished already in the Middle Ages.

The eldest writen complete rule set seems to be the Japanese of the beginning 20th century. These rules were first restructured and simplified 1949, because the until then valid rules has been slowly, but steadily increased. They contained more then fourty special configurations (e.g bend-four-in-the-corner) which needed special score evaluation in the scoring if they arose. These rules were adopted after world war II by the Koreans and the Taiwanese Wei-Qi Professional Association from 1952-1975. They were again reformulated by the Nihon Ki-in (the eastern Japanese Professional Go Association) 1989, but sadly, they still have logical defects (e.g. three-points-without-capturing).
Likewise was the development of all descents of the traditional chinese rules: They were officially writen down by the new founded Wei-Qi Association in the People's Republic of China about 1955 and in the seventy-years the modern Chinese rules were invented. The now valid Go rules in the mainland of China, propagated by the official Wei-Qi Association, were fixed 1988. The North-American Go rules by the AGA 1991, the New Zealand Go rules of 1978 and all other Go rules not used in Easter Asia are similar new. Remarkable are the SST rules of 1986, redesigned 1991, which were invented to convince up the world's Go players by a very rich taiwanese business man and Go fanatican, Mr. Ing. His main aim is to avoid the 'whole-board-repetation prohibition rule', but to make the Go rules nevertheless logical consistant. As similar as he, respectivaly his foundated GOE institutes, tries to spread his rules and philosophy of Go into the world with the support of his money behind them, each community favours their rather young, after 1950, --- except the current japanese --- fashion of Go rules.
Starting around 1970, computer scientists, who try to write Go playing programs and mathematicians who are interested in combinatorial game theory, enlightened the logical foundations of the many Go dialects which now exist. Especially the mathematicians try to simplify the Go rules and make them consistantly.

Chronologic overview of philosophic opinions which results in Go rule changes

Tibetan Go, before A.D. or probably earlier upto now
Go board is 17x17 sized grid
Immediate recapturing is forbidden (no snap back allowed)
A preset of 12 stones, alternatingly black and white, on the third line, spaced 3 apart.
Ancient Go, before A.D. or probably earlier
Go board is 19x19 sized grid
Passing seems ever disallowed based on philosopic reasons (Yin/Yang).
Cyclic whole board repetation which is not an immediate Ko Recapture is allowed and leads to 'games with no result'
Always a preset of 2 white and 2 black stones diagonal opposite each other on the four star points to simplify the opening strategy
Very old representations of Go board configuration were dug out in China, which show stones placed on the physical edge of a 17x17 board, hence effectively a 19x19 grid seemed to be used! This could indicate a change from a 17x17 grid to a 19x19 grid.
In the Sui Dynasty (581-618) it was common to use territory scoring rules like now-a-days in Japan. This lasts into the Tang and Song Dynasties (618-1279) and seemed to changed during the Yuan Dynasty (1279-1368) into the traditional Go --- of cource in some local playes the ancient Go rules may be played longer.
Traditional Go from the early Ming-Dynasty (1368-1644) into the 20th century
Usage of a 2 point 'group-tax' to accomplish the scoring procedure quicker, but not effect the final score in comparision to stone counting.
Japanese Middle Age Go rule changes
Abolish the preset of the 4 star point stones at the beginning. Probably to get a more variation-rich opening.
Interpretation of the 'empty area' as the 'to-strive-for territory', survived from about 500 upto now. This forces to define certain constallations for the scoring as death or alive. The number of such configuration increases slowly, but steadily, upto 1949.
Old Korean Go rules upto 1949 (?) but at least 1937.
Territory rules like the Japanese, but not only death opposite stones were removed in the scoring procedure, but also superfluious own stones to enlarge the own territory, as long as the enlarged territory remained safely in respect to capturing stones directly by the opponent (no stones in atari allowed).
This indicates a mixture of Chinese Area counting and Japanese territory counting. Hence, it was costless to prove life-and-death status, but it leads to a major tactic factor and a difference in strategy and of course result.
Revisions and Restructure of Japanese Go rules 1949
To get rid the need to define the status of many special configurations (37 until then) for the scoring procedure.
And have the first writen defines Go rules (it is said, that it was upto then not even clear, whether putting a stone on the board each move, was right or an obligation).
The Republic of China Go Association defines it's Go rules 1952
These Taiwanese Go rules are made in view of the Traditional Chinese rules and the dominating Japanese rules. They abolish the 4 stone preset.
To have a writen set of Go rules and to indicate their independence from the People's Republic of China.
The modern Chinese Rules arises 1975
After the Chinese Culture Revolution the Wei-Qi Association was founded in China to have an official control of Go playing activities.
These rules were invented as a compromise of the traditional Chinese rules and the dominating Japanese rules. The 4 stones preset as well as the group-tax was abolished. But they declared the 'super-ko'-rule. Since 1980 Go was promoted stronly officially, even by inventing many Japanese professional players. The modern Go rules were reformulated in 1988.
The SST rules become the official rules of the Taiwanese Players 1975.
A rich Go fanatician, Mr. Ing, dislike the 'Super-Ko-rule' and invented his ING-rules for resolving Ko-desputes consistantly. Sadly, this results in a very long rule text. They were redesigned by Mr. Ing 1986 and at last 1991. After his death his Goe foundation tries to popularize these Go rules.
The New Zealand Go Society introduced a short, consistant set of rules 1978
They allows explicit suicide moves and forbide any repetation of whole board configuration.
Their rules, except for area scoring instead of stone scoring, are the Go rules here recommanded.
After 1980
With the spread out of personal computers, many amateurs set up Go programs which manage Go games or even play, of cource very weak, Go. This causes a wider scientific research of computer scientists of the existing dialects of Go rules. Also new dialects, like the IGS-Go rules, were invented.
Combinatorical Go rules around 1988
This results in a further simplification in comparision to ancient Go rules.
In the sense of the established combinatorial game theory which is used by mathematicians:
Passing is not allowed, but instead of this a player can return a removed stone. So no scoring takes place, and the player loses who neither can put a stone on the board nor have any more opposite captured stones.
New AGA rules for players in the USA 1991
To abolish the still then used Japanese rules, because of their unsatisfactory logical consistence. They look similar like the NZ-rules, but forbide suicide moves.

Remark: The abbreviation AGA in Go context can mean either American Go Association, Australian Go Association or Austrian Go Association. The first name is the organization of the USA-amateur Go players --- excluding Canadian and Mexican players not to mention all the south american Go players :-).

Explanation of the Go rule development

Somewhere in the Chinese mainland Go was invented before Anno Domini. It is uncertain when in the range from about 2500 B.C. until 500 B.C. the earliest Go game was played. Also the first rule sets and the sense of this game are unknown and gives rise to different speculations. Reconstruction of unearthed boards and historical data indicate that as well as the Tibetan rule set as also the Ancient III rule may the oldest delivered Go rules. But it is sure that sometimes in the early middle age (about 500 A.D.) it became more and more common to play on a 19x19 sized board in contrast to the earlier spread 17x17 (and even 15x15) size. Shortly afterwards Go was brought to Japan from the mainland (and probably also the korean Go cultur branched off). The Japanese abandoned early the 4 stone preset and started with an empty board. Later, between 1300-1400, there took place a change in scoring philosophy in the chinese mainland, which didn't spread into the far apart countries of Korea and Japan. Also in the remote Tibetan area, the local Go variante could survive for many centuries without change. This situation continued until the Japanese expansion began in the late 19th century.

Starting with the occupation of Korea they also brought their cultur to parts of China and because the Professional Japanese Go players were indisputable the strongest of these three Go culturs there was a lot of pressure to adapt the Japanese rules as well as in Korea as in China. In Korea the original Sun-chang Baduk died out completely until 1945 and in China mixtures of Japanese and Ancient Chinese Go rule exist in local variations. Then in 1945 the Japanese empire falls and next in 1949 the successful communistic revolution took part which let Taiwan break away politically from the Chinese mainland. What followed was a definition of the Taiwanese Go rules in 1952 and a decline of Go in the mainland China until the end of the Chinese cultur revolution from 1966-1969. After this Go was supported politically in China and the modern Chinese rules were invented/defined by a new founded Wei-Qi Association in the beginings of 70th. About the same time, the small Taiwanese Go player community accepts the GOE-rules as their new Go-rules sponsered many financially by a Go interested business man Ing Chang-ki (he changed these 2 further times because of his unsatisfactory with the rules). In Europa as well as in Northern America the Japanese rules were adapted since early reports and cultur research journeys from the begining of the 20th century.

With the start of computer Go (around 1970) Mathematician and Go theorists considered different Go rules. As a consequence new rule sets in New-Zealand (1978) as well as the AGA-rules of 1991 was stated. Also the Combinatorial Game Theory mathematized the different Go dialects geting principle insights in the principle variation possibilities.

Collection of Internet resoruces

created 1998-02-18 12:30 UTC+01
updated 2003-10-21 17:32 UTC+02
Achim Flammenkamp