The Rules of Go

Page 05
Explanation of the Ko rule

The rule of Ko states that "A player cannot re-take the ko immediately after his opponent took it - an additional move must be done first". Fine, but what exactly does it mean, and why is it supposed to fix the problem? Well - let us look at the following position (Fig 1-1):

Position 1 Position 2 Position 3



























Fig 1-1. Simple ko sequence

First - let us start with Position 1. Assuming this Ko on the left side just developed, black can now play at 'a' and capture one white stone - which produces the situation in Position 2. It is white's turn now - so what can he do? The intuitive thing would be to re-capture the ko by playing at 'b' - but according to the Ko Rule - this is a no no! So what can the poor white do? Well - he can find a threat, a move which he believes black will answer, giving white the chance to re-capture the ko next! One such move is at 'c', placing the two marked black stones in atari (for explanation of this term, go here).

Let us assume that this is exactly what white does and that black answers him at '3' - producing Position 3. Now, and only now, can white re-take the ko - which he does by playing at '4'. Please notice that, in agreement with the general ko rule (or the superko rule) which states that "No Go position can repeat twice!", Position 3 is distinctively different from Position 1 (by the additional stones at '2' and '3').

Now, lets look at Position 3. It is black's turn again - and he would like to re-capture the ko by playing at 'a' again. But he cannot do it!! The ko rule specifically forbids re-taking the ko in the next move! So black has to find himself a ko threat now - a move which he hopes white will answer, so black will be able to subsequently re-take the ko. One such move is to play at 'd', hoping for white to defend at 'e' (or to play at 'e' hoping for white to defend at 'd').

Notice that the problem of black re-taking the ko is also nicely covered by the super-ko rule, since if he did, he would repeat the exact position which was on the board before white played '4' in Position 3.

What happens if black plays at 'd' is examined in Fig 1-2.

Position 4 Position 5 Position 6



























Fig 1-2. Simple ko sequence - continued

Position 4. So black figured it all out - he has to play a ko threat, which white will answer, and then black will re-take the ko with 'a' and the ball will be in white's court again. A good threat is for black is to play at 'd' trying to induce white to answer at 'e', and this is what black does. Well... lets see how it goes - in Position 5. Black plays at '5'. Now - what is white to do? He can obediently defend at 'e' and allow black to re-take the ko, but this will only lead to black having the opportunity to next play at 'f' capturing the marked white stone (explanation of this here) and if white defends at 'f', black will just capture the other white stone with 'e'... seems like whatever white does, wherever he defends, the result s not satisfactory - so white decides to abandon the fight on the right for now, and connect the ko instead - which is shown in Position 6.

The ko is finished - white has connected it with '6'. The price of this connection is the fact that white had to leave black's threat on the upper right unanswered (he ignores the threat) - and so, since it is black's turn now, he can cash in on his threat by playing 'e' or 'f'. This is his compensation for losing the ko. And the game goes on - black can kill the marked white stone with 'e' now, and then white can start a counter-attack with a move at 'g', for example. The ko is resolved.

Some more example of ko situations are given in Fig 1-3. In each of these situations, white can start a ko by playing at 'a'. And in each of these situations, the back-and-forth killing could go on forever, if not for the ko rule.










Fig 1-3. Some more examples of ko.

Some examples of positions which are not ko are given in Fig 1-4. In each of these positions, white can kill some black stones by playing at 'a', after which black can re-capture some white stones by playing at the marked spots, but this is as far as the killing would go without some additional stones. Because of this lack of "endless chain of back-and-forth killing" these positions are not ko!










Fig 1-4. Examples of what ko is not!
Sep.1999 (rb)