The ko rule is a rule in Go that prohibits certain repetitions of an earlier board position. In its simplest form, it prevents a player from immediately recapturing a stone when doing so would restore the position that existed before the opponent’s preceding move. Without this restriction, the players could repeatedly capture and recapture in the same local position without advancing the game. Broader repetition rules, usually called superko, also prohibit specified repetitions occurring over longer sequences.
The basic ko position
Go stones form connected groups, and a group’s liberties are the empty intersections immediately adjacent to it along the board’s lines. When a move removes a group’s last liberty, that group is captured and removed.
A basic ko arises when a player captures one stone and the capturing stone is itself left with just one liberty: the intersection from which the captured stone was removed. If the opponent immediately played there, the capturing stone would be removed and the earlier board position would be restored.
For example:
- Black captures a single white stone at an intersection called A.
- Black’s capturing stone, at an adjacent intersection B, now has A as its only liberty.
- White could otherwise play at A, capture the black stone at B, and reproduce the position before Black’s capture.
- The ko rule makes that immediate recapture illegal.
Not every capture of a single stone creates a ko. The decisive condition is repetition of the board position, not merely the number of stones captured or the fact that the new stone can be captured.
Under simple ko, White may play elsewhere. If Black also makes a move elsewhere, White can ordinarily return and recapture. The intervening moves mean that the recapture no longer reproduces the position immediately preceding Black’s latest move. Restrictions under a superko rule may nevertheless make a later recapture illegal.
Ko threats and ko fights
The prohibition on immediate recapture creates the tactical contest called a ko fight. The player temporarily barred from recapturing generally seeks a move elsewhere that demands an answer. Such a move is a ko threat.
A typical sequence is:
- Black captures the ko.
- White plays a threat elsewhere.
- Black answers the threat.
- White recaptures the ko.
- Black must now find a threat before recapturing in turn.
A threat is effective when ignoring it would allow a sufficiently important gain. Examples include threatening to capture a group, break into territory, or change the outcome of a local life-and-death problem. A threat need not be answered merely because it is described as a ko threat: the opponent may decide that securing the ko is more valuable than preventing the threatened loss.
This creates a choice between answering the threat and continuing the ko fight, or ignoring it and resolving the ko. Resolution often involves filling the ko point or connecting the endangered stone, although the necessary moves depend on the surrounding position.
Assessing a ko fight
The value of a ko fight is not simply the value of the single stone being exchanged. Its outcome may determine the survival of a large group, ownership of a corner, or the connection between groups.
Important considerations include:
- The consequences of winning or losing: the total difference between the possible local outcomes.
- Available threats: their number, size, and whether the opponent must respond.
- Move requirements: some ko positions require additional preparatory captures or several unanswered moves to settle.
- Whole-board exchanges: a player may concede the ko while gaining compensation elsewhere.
A small ko can therefore become strategically important if it affects the supply of threats for a larger fight. Ko considerations also occur in joseki and the endgame, not only in large capturing battles.
Simple ko and superko
Different Go rulesets do not use identical repetition rules. The distinction between simple ko and superko is particularly important.
Simple ko
Simple ko forbids a move that recreates the board position immediately before the opponent’s preceding move. It prevents the ordinary two-move capture–recapture cycle but does not, by itself, prohibit longer cycles.
Its practical effect in a basic ko is that the player who has just lost the ko must first use a turn elsewhere, or pass, rather than immediately recapture. Whether a later recapture is legal must still be assessed under the complete ruleset, including any special provisions.
Positional superko
Positional superko generally prohibits a stone placement that recreates any earlier arrangement of black and white stones on the board, regardless of whose turn it was in the earlier position.
The relevant comparison is the whole board, not merely the local ko shape. A local pattern may recur legally if stones elsewhere make the overall position different.
Situational superko
Situational superko generally prohibits recreating an earlier combination of board position and player to move. The same arrangement of stones with the other player to move is therefore treated differently from an exact repetition of the full situation.
Exact formulations vary between rulesets, including how the initial position, passes, and the history of play are treated. Passes require explicit treatment because a pass leaves the stones unchanged; a blanket ban on all unchanged positions would interfere with the normal procedure of ending play by passing.
Longer cycles and ruleset differences
Several interacting ko positions can produce a cycle longer than the immediate exchange that simple ko forbids. A familiar example is triple ko, in which captures at three ko locations can restore an earlier position through a sequence of moves.
Other exceptional cyclic configurations include the eternal life pattern, also known by the Japanese term chōsei. Such cases show why a rule against immediate recapture is not a complete rule against repetition.
Rulesets address these situations in different ways:
- Japanese rules use a simple-ko prohibition during ordinary play and provide for a game to be declared no result when a whole-board repetition occurs and both players agree. A no-result game is distinct from a tied score.
- Rulesets using superko instead prohibit specified moves that would complete a repetition. This can force a player to deviate from an otherwise repeatable sequence.
- Other rulesets combine repetition provisions with particular adjudication procedures, so their treatment cannot always be reduced to a single label.
Japanese rules also contain special provisions concerning ko during confirmation of life and death after play has stopped. These are distinct from the ordinary procedure of playing ko threats during the game. Consequently, a position’s legal continuation or final status can depend on the ruleset being used.
Significance in Go
Ko connects local tactical problems with decisions across the whole board. A fight over one intersection may require players to compare threats in distant regions, assess exchanges between territory and captures, and decide when to abandon a local objective.
It also distinguishes Go from board games in which legality depends only on the current arrangement of pieces. Even with simple ko, recent history matters. Under superko, much earlier positions may affect which moves are legal. Computer Go programs therefore need to preserve or otherwise track the relevant position history, rather than determine legality solely from the stones currently visible on the board.