Sakata pointed out that reading isn't simply the calculation of variations, but the judgement of the outcome positions. (In a tsumego problem that is easy, though.
) Our own Magicwand has pointed out here that it involves knowing what plays
not to make.
It is also more than visualization. It involves building a tree.
Kotov, in
Think Like a Grandmaster, cautioned against the inefficiency of repeating lines that you have already read. Bridge great Terence Reese warned against dithering.
Building a tree organizes your reading.
I have heard that recent research indicates that it is not a bad idea (contra Kotov) to start again from the original position and retrace your steps up to a branch point. My thought is that perhaps that helps keep the whole tree in memory.
One very important part of building a tree that Kotov recommends is to identify candidate moves at each step. OC, as Magicwand says, that involves eliminating other moves.
This means that reading is not, I play here, he plays there, I play there, . . . It is like this:
There is a good point, and there is another one, and there is another, . . . Suppose that I play here. What are good plays by him? . . . Suppose that he plays there. Then what are my possible replies? . . . OK, I have got that line. Let's start over. Here, here, here, here, . . . Now suppose that he plays here instead. . . .
Also, particularly in go, you consider your opponent's possible plays. Like this: If he plays there he will have two eyes, so that's a good spot for me. But if he played there, and then I got two moves, could I take away an eye or make ko? Ah! I could, so either of those points might be good, too.
If you are calculating variations, purposefully build a tree.
