Other Rules

return to Variations on Go

In battle, there are not more than two methods of attack - the direct and the indirect;
yet these two in combination give rise to an endless series of maneuvers.
[Sun Tzu, The Art of War - V.10]


Li Ch'eng (China 11th Century)The rules of Go are so simple and deep, that a game inventor like myself hesitate to propose some variations on them. However, the goal is not to enhance a (nearly) perfect game, but just to present some different ideas based on Go in order to people enjoy the strategic and tactical endless ways of abstract games. I'm not talking about different interpretations of the rules of Go like what happens in Japan, China or New Zealand (!) (to read about this subject, check the British Go Association in here). I'm presenting rules that change the game producing a specific Go variant  (other variants and lot of ideas can be found at Gregory Patten's website and at Andre Engels website). Don't forget to check this page with many Go proverbs.

I will divide these rules by sections, depending on how they affect game play in tactical or strategic planning. This page starts with some regional variations of the old game.

Tibetan Go is a two player game played on a 17x17 square board, with the following setup:

An example

After Black's first move, White may choose to play around one of his initial stones, or play near the dropped black stone, where the possible options are given by the green dots.

Old Korean Rules - Sun-Chang Go (also called sun-chang pa-tuk)

From an article of John Fairbairn:[...] the game was played on a specially marked traditional board, of which several examples survive. The 17 starting-stone points are marked. It appears that both players placed eight stones each on the points shown in the game below, and Black then played first. But as he was obliged to play his first move on the center point, we can effectively regard this as a starting stone too. These stones have no special status, unlike their equivalents in Tibetan go.

The basic rules are the same as in modern Japanese go, which no doubt encouraged players to abandon the old form. [...] The real difference in sun-chang go, apart from the starting position, is in the counting rules. Ko and seki are treated exactly as in Japan (no points are counted in a seki). There was traditionally no komi, so clearly Black had a big advantage - yet another factor favoring adoption of Japanese rules.

These variants are based on board/rule restrictions. The soldier deployment must obey a set of predefined rules.

Density (by Joćo Pedro Neto)

Mutator Density(N,M) is a rule stating that for each NxN cell (N odd) inside the board should have at most M pieces on it. With no restrictions, i.e., Density(N,N*N) we get standard Go. I estimate that M > N*sqrt(N) to create an interesting game. In the example, the game has Density(3,5), so White can protect is soldier by playing at [2] for example, since Black cannot play [1] or else the density would be 6 on that 3x3 square.

Capture GO (described by Andre Engels)

Mutator Capture(N) applied to Go creates a game where the winner is the first player who captures N stones from his opponent. Capture(1) is like Go, but with some remarkable differences. For example, two connected internal liberties count as two eyes instead of one. In the picture there are two nice handicap positions.

Anti-Square (by Joćo Pedro Neto)

Any four stones of the same color cannot form a square. In the example, White cannot play at [1] since it would form a square with other three white stones already in the board. So, the white stone at north of [1] cannot escape.

Simultaneous Capture (by John Tromp)

After placing a stone, remove ALL (i.e. both black and white) stones that have no liberties (read more). If White plays at [1], both center stones are captured.

Impartial Go (described by Jeff Erickson)

Either player can place a black stone or a white stone on each turn, capturing stones of the opposite color. The winner is the person who captures more stones. So, both players may drop a white stone at [1] to capture the left black stone.

In Impartial Othello, both players may drop any stone and flip stones of the opposite color. The winner is determined by the color of the last piece. So if there are more white pieces, and the last player played white, he wins.

Knight Jump Go. (described by Bill Shubert)

Every move must be exactly a Knight's move (two spaces horiz/vert, then one space vert/horiz) from the previous. There was no reference for what happens when all such spaces are occupied, two possibilities: the game ends, or a stalemate is a loss (that means a no pass rule). 

In [1] and [2] there are two possible playing sequences for placing stones.

Dagger Go (described by Henry Segerman

Instead of handicap stones, the weaker player gets a number of 'daggers', which can be played at any time, and give the player a free move. Generally it seemed that 1 dagger is worth at least 2 handicap stones. The stronger player has to play far more solidly - the dagger can very easily kill things which would otherwise be very alive.

The [1] structure (if surrounded by Black pieces) is not safe. Chain [2] is safe since it has 3 eyes.

Balanced Go (described by Eric Osman

The board is balanced on top of a spindle. One must take care to make plays that keep the board physically balanced enough so as not to topple off the spindle. This should not be played by email unless you have a physics simulation.

Let's say, that the board may support a maximum of stone in the edge (or two stones at half a distance between Tengen and the edge, and so on...). If Black drops at [1], White should choose a move around [2].

Life Wall Go (by Gregory K. Van Patten

[Using the author's words] This Go variant is played on a regular Go board with four sides. However, call one particular side the "Life Wall". The rules of this game are exactly like regular Go, except for scoring the game. At the end of the game, the only groups of your stones which give you points are those groups containing stones on the Life Wall, or whose territory contains intersections on the Life Wall. If we call these groups "Life groups", then your score is equal to the number of your stones which form Life groups, plus the number of territory points contained in your Life groups.

In the sample, the top edge is the Life Wall, and so group [1] is alive, group [2] is not.

Progressive 9x9 Go (described by Bill Taylor

The rules are even simpler than normal Go. Black plays 1 stone, white 2, black 3, etc. Captures as normal. Suicide is legal. Ko cannot exist! You may pass any of your turns. When both pass a full move it ends. Wins by usual area scoring. Note that stones are dropped in a sequence, not simultaneous, or else it would be impossible to create living groups! In the sample, Black started at e5, then White played c7, g7, and Black c3, g3 and e7. White then may move into cells [1], for example.

A Scottish variant include the following rule: "Whenever you put any opponent group into Atari, your sequence stops.". This extra rule enables more safe structures like in Go.

Another possible interesting variant is "Choice Progressive":- at each turn, the player *chooses* how many moves he will make, which can be anything from 0 to one more than the opponent's immediately preceding choice.

Onboard Go (by Ed Pegg Jr

Given a 5x5 board. Black has 12 stones, White has 10. There is no pass, no suicide, no repeated positions. First to get all their stones on the board, wins. I assume that prisoners are returned to the owner. In the top example, Black lost the game since he was unable to control a sufficient number of cells, leaving White with too many choices. In the bottom example, Black captured 2 stones, which gave enough time to win the game.

This can be extended to bigger boards. A possible formula could be: for a NxN Go board (odd N), Black has (N^2-1)/2 stones, and White has the same minus 2 stones.

Btw, even if it is a bit out of context, do not forget to visit Ed's fabulous puzzle site at www.mathpuzzle.com.

  Goal Go (by Graham Lamont

Each player is given a list of goals to achieve. The player who first achieves all goals wins. Alternatively points can be awarded for goals, player with most points at the end wins.

Example goals: [(x,y) denotes a value range]

* Capture (x,y) stones
* Create (x,y) groups
* Limit opponent to (x,y) territory
* Create a seki
* Don't create more than n kos
* Don't let opponent create groups bigger than n stones.
* Create a dragon!
* Don't play any stones on the nth line.
* Capture n corners
* Win/lose by (x,y)
* Don't create groups with more than n eyes

A lot of the skill will be in generating sets of goals which are challenging, only marginally exclusive of each other, and appropriately handicapped.

Royal Atari Go (by Jianying Ji

Each player in addition to regular pieces have a King piece. 

In the first move each player drop their King piece. Then the person who dropped second decide whether to take black or white. Once decided the player play the regular pieces. Black go first. In each turn the player drops a piece according to the rules of Go, except no passing.  The winner is the one who captures the opponent's King.

Perhaps the game should be played on a small Go board, since it's very likely that the King will only be captured at the endgame.

Alak (by Alan Baljeu)

This game is played on a 1-D Go board (or line...) 

Black and white alternate in placing single stones on a line of n points. If placing a stone thereby removes all the go-liberties of any group of stones of the opposite color, those stones are immediately removed. It is illegal to place a stone where one was just removed. Placing is compulsory if legal, and the game ends when the player having  the move cannot legally place anywhere. The winner is the player with the most stones on the board at game end.

A sample game follows where black wins 6-5.

Check this site with more about linear Go.

  No Pass Go

All rules like Go, except no passing is allowed. A player loses if it cannot make a valid move.

The game ends when only one player is left on the board, with all his interior spaces being single points. In practice, it will end long before this by resignation, when the Go-like part has finished, then the "cold" (check next paragraph) part has finished, (all territories reduced to solid stones with 2 eyes each), so that finally someone has to fill in one of an only 2-eyed group and his opponent captures it; thus gaining a huge amount of extra territory and free moves, till the loser has to self-destruct another group, and so on till all are gone. The best way of filling in opponent territory and own territory make very intriguing Combinatorial Game Theory (CGT) example puzzles.

A hot position is one where both players really want to move, where no-one would pass even if it were legal. Most positions in chess and Go are such, and all in Hex, Moku, etc etc. A cold position is one where no-one wants to move, and would pass if it were legal. e.g. Zugwang positions in chess, post-dame (fill-eye) moves in Gonnect, and a great many positions in Nim and so forth. There may also be lukewarm positions where no-one cares if they or their opponent moves first, such as filling in dame in (territory scoring) Go. It is a very useful term in CGT.

Sliding Go (by Morten Pahle?)

Morten Pahle says: The placement of stones is carried out by a slide on the board. The stone is initially placed on a free intersection on the edge closest to the player (home edge) or next to a group connected to the home edge, and then slid to its final position. A slide can only follow the gridlines, and only 'turn' at right angles on intersections. Any number of turns are allowed, as long as the stone always follows the gridlines and only moves through unoccupied intersections. As a consequence, kos are impossible, placement tesujis will be out, most big eyes are alive etc.

On the left, Black has no eye at the circles, since White can fill it up without Black responding. (Black cannot 'get in there'.)

  Slidey Go (by Tamsin Jones 1999)

Like Go, except that a player may also move a friendly stone to an orthogonally adjacent empty cell.

  Line Go (by Joćo Pedro Neto)

Black drops the first stone. After that, each stone must be dropped at the same row or column as the previous stone, and without any stone inbetween.

Influence GO (by Alex Waldon)

Same as Go but with this setup. From the author: I didn't know what else to call this. I haven't tried it yet, but my idea was to create a variant to practice using thickness and influence. [...] I suspect that the best opening move is probably at tengen, after which some interesting fighting should break out as both players try to claim some territory and connect to their sides.

  Choice Go (?)

On each turn, each player states two possible moves. The opponent chooses what move he prefers.

Another similar variant, Forced Takeback Go, the player may force his opponent to take back the most recent move and replace it with another, but only once per turn.

  Proselytes Go I (André A’)

There are no prisoners. Stones who's last liberty is taken, are transformed into stones of the enclosing color. At the end of the game the same is done with the stones surrounded on large scale. Chinese counting.

  Mission Go (André A’)

There are no prisoners. Stones who's last liberty is taken, are transformed into stones of the enclosing color and redropped on any empty cell in the board. At the end of the game the same is done with the stones surrounded on large scale. Chinese counting.

  Compensation Go (André A’)

There are no prisoners. Stones who's last liberty is taken, are given to his owner that redropped them on any empty cell in the board. If more stones lose their liberty the process repeats until it stabilizes.

At the end of the game the same is done with the stones surrounded on large scale. Chinese counting.

  One liberty Go (André A’)

Groups are captured when they have just one liberty.

When this happens simultaneously to groups of both colors, then it is only the non-mover that is captured.

Bill Taylor suggests that suicide should be legal, especially if it results in a capture.

  Progressive Liberties Go (André A’)

Groups are captured when they have a number of liberties defined by their own size. The author proposes that a group should be captured when he has less than one tenth rounded down of its size.

Other criteria are possible.

  ZenGo (Bob Sloane)

Players change colors always after N moves.

note: I do not know when the game should end.

  DanGo (Caspar Nijhuis and Robin Upton, 1991)

Stones once captured, are not removed from play but relocated on the board by the capturer. The game starts with two stones of each colour on opposite corners.

  Environment Go (check this webpage)

From the site: Environmental Go is Go played in the normal way, but with a stack of forty cards, numbered [ from bottom to top] ½, 1, 1½, .., 19½, 20 as well as the normal board and stones. For a player's move, they may either play a stone on the board in the usual way, or take one of the cards. At the end of the game, a player's score is the sum of their score on the board and the cards that they have taken.

  Paper and Pencil Go (Luis Bolańos Mures) cf. webpage

[author's words] The rules are exactly those of Go, with the following exceptions:

  • Surrounded stones are not captured, but just marked. Points occupied by marked stones count as territory for the surrounding player, but neither player can play on them for the remainder of the game. This implies that any group which touches a marked stone is unconditionally alive.
  • Suicide is allowed, i.e., you can make a play such that one or more of your own stones, including the one you just played, become marked.
  • Area scoring is used.
  Stoical Go (Luis Bolańos Mures)

[author's words] In this Go variant, standard ko rules don't apply. Instead, it's illegal to make a capture if your opponent made a capture on his previous move.

All other rules are the same as in Go. Suicide of one or more stones is not allowed, and area scoring is used.

All known forced Go cycles (basically those described in the Sensei's Library article plus the funny pinwheel ko) are impossible with this rule. The nature of the rule itself suggests that forced cycles are either impossible or astronomically rarer than they are in Go when the superko rule is not used.

It's interesting that, while ko fights are usually not possible in finite Go variants, here not only are they possible, but also ko threats have an additional use compared to Go: when a player wants to answer a capture with another capture, he has to make ko threat before being able to do so.

A while ago, I posted on LifeIn19x19.com about this rule, which I then called "passive ko", and it generated some discussion. One player illustrated the rule with this simple position:

White to play kills if Black doesn't have any suitable ko threats available. In Go, Black is simply alive.

I've also programmed the game for Zillions of Games. I've implemented all odd-sized boards from 7x7 to 19x19, as well as the option to use neutral blocking pieces to change the shape of the board.

These variants have the following common goal: To achieve a specific pattern in order to obtain some extra advantage. I give some possibilities, but million others are possible:

Cross (by Joćo Pedro Neto)

Each N cell cross is transformed into a N outlined square. All pieces inside the future square are removed (if they belong to the same player) or captured (otherwise). In the example, if White play at [2], he captures 4 black stones.

Line (by Joćo Pedro Neto)

When a player creates a line of 5 soldiers (vertical or horizontal), he can append an extra soldier to one of the extremes, capturing any eventual enemy soldier. In the example, if White plays at [1], he can extend south and capture the black soldier. If he plays at [2], he can extend at [1] or [3] but does not capture the black soldier. In the other example, Black can capture 2 white soldiers if he plays at [4]. If he plays at [5], he can only extends at [4] and does not capture any white soldier.

Pentaminoes (by Joćo Pedro Neto)

When a player creates a pentaminoe, he must append an extra soldier to create an hexaminoe, capturing any eventual enemy soldier. In the left example, White can capture the Black piece by forming a pentaminoe. In the other example, if Black capture the top white soldier, then the white structure is again a pentaminoe and can capture a black stone. Therefore there is a special Ko rule, that forbids those pentaminoes swaps on two sequential moves.


These variants are based on local features, usually at soldier level:

Electric GO or Magnetgo (by Ralf Gering)

If you play a move, in all 4 directions the closest stone is attracted (if your opponent's) or repelled (if your own). That is, it is moved in a straight line away from or to the stone just played, until it reaches the edge or another stone (in the case of attraction always the stone just played). In the example, if a white soldier is dropped at [1], the 4 pieces (2 whites and 2 blacks) go to cells [2].

A variant may define a maximum length of this magnetic effect (only attract pieces closer than N cells and repel pieces at most to N cells).

Atom GO (by Thomas Hillebrand)

If a stone is killed all 4 stones around it also die. So, if a group of stones dies then all surrounding stones are removed, too. In the example, a black soldier dropped at [1] will kill all 5 stones.

Rotating GO (by Joćo Pedro Neto)

When a stone is dropped on a cell l that has a friendly adjacent neighbor (including diagonals) the player may rotate all neighbors (including diagonals) one cell clockwise. A stone cannot be dropped in a cell with no liberties, even if the stone gets some liberty after the rotation. In the example, a white soldier dropped at [1] will produce the rotation of its 3 neighbors. In the other example, the black stone dropped at [2] rotates the 3 stones, and because of that, captures the white stone.

Push GO (by Joćo Pedro Neto)

This one is similar to Abalone's push rules. A line of N stones can push another line of M<N opponent stones, at a distance of 1 cell, if that next cell is empty. In the example, the black line can push the white stone to [1]. The triple black line can push the double white line into [2], thus capturing two white stones. The triple white line cannot push the black stone, because there is another white stone above [3]. There is a KO rule, that forbids to drop a stone on a cell just made vacant by a push move.

The game can be improved by giving a limited number of neutral stones, for each player, that cannot be pushed, but don't count in the score.

Switch GO (by Joćo Pedro Neto)

When a stone is dropped, the player can swap it with one (if any) of the adjacent opponent stones, if after the opponent stone, in the same direction there is a friendly stone. A stone cannot be dropped in a cell with no liberties, even if the stone gets some liberty after the swap. In the example, White played [1], and since the south black stone is between two white stones, it can be swapped with white stone at [1].

Othello GO (by Gregory K. Van Patten)

A group is dead when it has been surrounded by enemy stones, just as in Go. However, rather than removing the dead group, its stones are "flipped" to the opposite color, just as in Reversi. Since no stones are ever removed, the game never repeats. The winner, just as in Reversi, is the player with the most stones of his own color on board at game end. [The author says that this rule was disappointing in practice]

In the upper example, when Black drops at [1], the white pieces is not captured, but changes color. In the other example, after white move at [2], Black may turn all pieces into black, if he plays at the [3] cells.
Persistent GO (by Gregory K. Van Patten)

If a move creates dead groups of both colors, then all these "dead" groups are left on the board, and may be used to capture other groups. As soon as one of these becomes live again (as a result of capturing something), the dead groups of the opposite color die. [The author says that this rule was disappointing in practice]
Step Go (by Joćo Pedro Neto)

Stones may not be dropped in cells having exactly one adjacent friendly stone.

On the top left, only the green dots are valid cells where black can move. On the right, to connect both white stones, White must drop three extra stones (so, knight connections are not that good in step Go). On the other example, Black cannot escape the Atari, because he cannot play at [4].

Caution Go (by Joćo Pedro Neto)

Stones may not be dropped in cells having more adjacent enemy stones than friendly ones.

On the top left, Black cannot drop on the red dots. On the other example, White cannot plat at [1] and Black to drop at [3] must first drop in one of the [2] cells.

Go Life Game (by Ted Drange)

One player construct a position with Black and White stones. The other player decides which side he plays. If he plays White, he must create a safe position, otherwise if he chooses Black, he must prevent any opponent's safe structures.

One possible position (called the corner game) is where Black has a solid wall of black stones all along the 10th rows out from the corner. White must live within one of those corners. According to Ted, White usually starts on 3-3 and Black answers with 4-2.

  Goof (by ??)

Played on a small board, perhaps 9x9.

On each turn, Black drops two stones. Then White swaps one of those into a white stone. The winner is the player who captured more stones.

At first move, Black cannot put his stones onto adjacent cells.

  Tenuki Go (by Chess Whiz)

Same as go but no drops (orthogonally or diagonally) adjacent to the opponent's last move. Basically a "forced tenuki".

Huang Kung Wang (China - ?)