British Go Journal No. 54. October 1981. Page 23.
Every go player improves if he keeps on playing. This improvement is never steady and consistent, it occurs in jumps. Each jump shows that a go revelation has born fruit. But between the jumps are the plateaux where you plod along for months or maybe years always at the same depressing level. I remember being stuck at 2-kyu for a whole year; one of the insights that finally jogged me out of my rut was the significance of 'pre-atari'. Most go players, when they see an atari, play it immediately. Their reason seems to be that atari is a forcing move which compels an answer. If you have successfully forced your opponent into some particular response then you can bask in the illusion that you have control over what's happening on the board. However, the fact of the matter is that you may have destroyed all sorts of delicious aji. Why should this be? The explanation is disarmingly simple.
When a group has just two liberties you can put it into atari. We can describe this state as 'pre-atari'. But which liberty do you fill? If you fill one you can force one response, if the other you force the other response. The simple but crucial point is that there are two ways you can force your opponent. Two possibilities. Here we are very close to the Japanese idea of 'aji'. Do you want to hit your opponent on the head or the backside? If you refrain from hitting him either way but probe nearby he may be forced into making a grovelling answer to your probe. You have forced him off balance because he can't just parry your probe but must shelter himeself at the same time from both the blow on the head and the kick in the pants.
Dia 2 |
Let's look at an example: In Dia 1 White has spotted what he thought was aweakness in black's shape and pushed with 1 and cut with 5. Now Black can play atari at either A or B. Whoa! Steady there, what should Black really do? He should think beyond the two ataris, and play 1 in Dia 2. White has to grovel and connect at 2 in Dia 3. Now Black can settle himeself comfortably. If White tries to resist with 2 in Dia 4, Black can unleash one of the options he discretely refrained from playing before and humiliate White.
Dia 3 ||
Dia 4 |
Dia 5 |
Examples of ataris that should be left alone are very common. Dia 5 shows another example. Black has just played a rather flashy move at 1, tempting White into a tricky fight. White decides to simplify the position and sacrifice his three stones by pushing through with 2 and 4. But white should refrain from popping in the extra atari at A. The reason is that he might later be able to play B, which threatens C and after C the two ataris at A and D really come to life. The conclusion is that White can play B almost as a free move because Black can't respond aggressively. Had White carried on gleefully playing atari at A the aji that lets him have such an easy time with B disappears.
Obviously if an atari is double sente then you can't very well avoid playing it straight away. However, if it is just sente for you then leave it alone because there are always TWO ataris. Try to find a sequence in which the other atari might be helpful. Even if nothing comes to mind now, 50 moves later things might be very different. If nothing comes of it you will have an extra ko threat or some yose in sente, or both at the same time.