MG Rules Of Go


mgoetze: I really like the [ext] New Zealand Go Society Rules of Go. However, it seems that some people hate the superko rule so much that they will reject this simple and elegant ruleset in it's entirety for that reason alone. This is my attempt to change this one aspect of NZ Rules. I also took the opportunity to address some concerns raised on the New Zealand Rules page here on SL.

Please do not write comments into the rules text itself.

Version: Draft 2

Rules of Go

Go is played on a board with a 19 x 19 grid, by two players, one using a set of (about) 180 white counters, the other a set of (about) 181 black counters. These counters are called stones.

Adjacent intersections are those intersections connected by lines of the grid, with no intervening intersections.

Two stones of the same colour are connected if they are on adjacent intersections or if they are both connected to a third stone.

A liberty of a stone is an unoccupied intersection adjacent to that stone or to any stone connected to that stone.

A play consists of placing a stone (of that player's own colour) on an unoccupied intersection, then removing any of the opponent's stones that then have no liberties (if any), and then removing any of that player's own stones that then have no liberties (if any).

A game of go begins with an empty board, and the players take turns to move (beginning with black).

A move consists of

  1. making a play so that the position just before the other player's last move is not recreated or
  2. saying 'pass' or
  3. resigning, either by placing two stones on the board simultaneously or saying 'I resign'. In this case, no further moves are made, and the game is said to have ended by resignation.

The game is finished when both players agree that there are no more worthwhile moves. 'Dead' stones may then be removed from the board by mutual agreement. In this case the game is said to have ended normally. If they cannot agree which stones are dead they must play on. (The player whose turn it was when they agreed to end the game plays first.)

The game is also finished when a set sequence of plays is repeated at least twice in immediate succession, and the players agree that this repetition will not end. In this case the game is said to have ended by repetition.

Territory of a player (at the end of the game) consists of all points occupied by that player's stones plus all unoccupied points adjacent to that territory.

In an even game, 7 points (komi) are added to the white players territory.

In a handicap game, white passes the first n - 1 moves where n is the size of the handicap. There is no komi.

The result of the game is determined as follows:

  1. If the game ended normally, and both players have the same amount of territory (jigo), the game is a draw, else
  2. the player with more territory wins. Otherwise,
  3. if the game ended by resignation, the player who did not resign wins or
  4. if the game ended by repetition, the game is a draw.


RobertJasiek: What is the purpose of requiring specific numbers of stones in the game (and then to relax that by an "about")?

mgoetze: There is no particular purpose as far as I can see.

RobertJasiek: Since agreeing on no more worthwhile moves provides strategic hints, the players might prefer to never agree. In other words, the move type "pass" has a meaning. - In pretty simple rules the terms "dead" and "alive" are, to say the least, superfluous and already therefore confusing. Say: "stones to be removed".

mgoetze: And I might prefer to call those players silly. And possibly disqualify one or both of them from a tournament for unsportsmanlike conduct, if that is the scene.

Rich: Would it not be simpler to have a two-pass definition? What is the purpose if not?
Also, can you confirm that suicide is legal under these rules?
mgoetze: I'm not entirely sure whether a two-pass definition is simpler. Yes, suicide is legal under these rules.

RobertJasiek: So you invent bad rules so that you can punish players? The standard purpose of pass is to allow a player not to move while not giving the opponent strategic advice during alternation. - The purpose of calling something "dead" in the rules is to let the rules determine what is dead. If you want to let the players determine the stones to be removed, then let the rules speak of "stones that the players agree to be removed". The term "dead" is superfluous and confusing here.

mgoetze: If a player is so silly as to believe that the extra information given by agreeing that there are no more worthwhile moves on the board over passing repeatedly, if any, has any influence whatsoever, then, in my opinion, they deserve such "punishment".

RobertJasiek: Example:

Black to play  

(I use an easy example so that we can concentrate on the rules and not on strategy.)

If passes are not required to end the alternation but if the players can also seek an informal agreement to end the game, then Black (who is a weak player and not sure whether he should defend once more) will try to get strategic advice from his opponent. Therefore he askes White whether White still intends to play another stone. If White answers "Yes.", then he also advises Black that Black needs another defensive move. Go was supposed to be a game of greater strategic skill during alternate moves, but now it has become a game of getting the better strategic advice from the opponent. The influence on the game is great, however "silly" you call it.

mgoetze: First of all, that is the wrong question; it should be, "do you agree that there are no worthwhile moves left?", and it is perfectly valid to consider it worthwhile for each player to pass once. Asking about a specific kind of move (playing a stone) obviously should not be answered. Secondly, your example is silly because with area scoring, it is sufficiently obvious that playing an extra defensive move does not affect your score negatively, and therefore this should be the default action if a player is unsure, quite regardless of which "strategic" (actually, tactical) advice he is or is not able to solicit.

RobertJasiek: At the very least, you also need to specify the third type of cycles, where the players do not agree, that lasts longer than 2 moves, and in that the removals-difference does not equal zero.

mgoetze: I don't see how an extra rule for this is necessary, since removals do no have a direct effect on the score in area scoring rules.

RobertJasiek: You have already specified one extra ko rule in your rules. The second type is necessary to prevent a player from not losing by continuing a cycle forever (like in a sending-2-returning-1). This is not bad behaviour but good strategy in your early rules draft.

mgoetze: Please give some rationale why the rules must behave like this.

RobertJasiek: If they do not but behave like your rules, then the player that is behind uses a cycle in that he loses more stones per cycle than his opponent to continue the game forever. Thereby the player does not lose. Not losing is better than losing. To allow his opponent to win the game instead of not allowing him to win the game, the rules must describe this case.

Rafael: Robert, you don't need to bother about sending-2-returning-1. If a player wants to win that badly, he might as well just keep on passing. Since not even 3 passes end the game, he doesn't lose (unless it's sudden death).

I suspect that Goetze will find this example silly, though.

(Of course it also applies to New Zealand Rules. I had come up with a similar but far-fetched example in KGSPossibleRulesBugDiscussion, I've just realized that this one is much simpler)

RobertJasiek: Rules of play ought to be complete. Then a player is reasonable naturally because he cannot avoid losing by silly play.

Discussion about superko moved to Superko/Discussion

DougRidgway: (unsure where to put this). These rules seem to ban passing after an opponent passes (due to ko). Is this what you want? Otherwise they look like area scoring+simple ko, to use Robert's terminology.

Anonymous: No, Doug, they don't forbid passing (even endlessly :-). See the distinction the rules make between a play (a move with an actual stone) and saying 'pass' (which is a move but no play). The ko rule and the repetition definition both only apply to plays not moves in general.
To get that clearer the No. 1 move rule should perhaps read:### A move consists of
1.making a play so that the position just before the other player's last play is not recreated or ...

DougRidgway: Hm, you're right, not sure what I was thinking. But I'm a little concerned about this endless passing. It looks like these rules allow a losing player to continually pass while disagreeing that the game is finished (arguing that postponing a loss is worthwhile, perhaps). This could conceivably happen in every game, not just obscure ones. I don't see what the winner could do inside of these rules.

(Sebastian:) Please bear with me if I don't understand the repetition rule. What exactly do you mean by "will" in "will not end"? Of course, tenuki is always an option. If you mean: "both agree that they could (if they wanted to) keep playing back and forward ad infinitum", then it applies to any normal ko. So, if I'm losing, all I need to do is play a ko to make it a draw.

mgoetze: I mean that neither player has any intention of deviating from the sequence. It is not sufficient to start just any normal ko, because in a normal ko there is no exact repetition - you can't keep playing the same ko threat. And even if you do manage to start a triple ko or the like, this is still only sufficient to force a draw if the triple ko is so large that your opponent can't afford to break the cycle.

Rafael: Disagreeing that the repetition will end is not the same as committing to deviate from the cycle. :-)

Anonymous: Here's a better definition of territory:
Territory of a player (at the end of the game) consists of all points occupied by that player's stones plus all unoccupied points surrounded by only that player's stones or the edge of the grid.
(Your original rule a) allows territory only to be a 1-ply layer of unoccupied points along/around the stones - see your definition of "adjacent" and b) tells nothing about unoccupied points that are "adjacent" or in the neighborhood of stones of both players.)

Harleqin: Anonymous, the rule text behaves recursively. Note that the rule regarding territory ends in "... adjacent to that territory." Also, unoccupied points that are adjacent or in the neigbourhood of stones of both players will count the same for both players, so it makes no difference. The rule text is precise and complete. By contrast, your proposal needs additional definitions of "surround" and "edge".

MG Rules Of Go last edited by Harleqin on March 28, 2007 - 16:34
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library