How to read
Reading (Japanese: yomi) is the mental skill of calculating lines of play (also known as variations or branches). Reading is the muscle of Go. (mitdahand of KGS).
Table of contents | Table of diagrams Status Vital point First reading Second reading Main line Maybe here |
Many players, even experienced players experience difficulty reading. It is generally accepted that solving problems and particularly life and death problems are a good way to train reading skill.
So in order to become good at reading in general, let's first look what it takes to properly read a life and death problem, i.e. I'll start posting how I (think I) do it:
Example
This example is taken from Kyu Exercise 27.
In this problem, the status of the Black group is asked.
First, allow your intuition to downsize the effort and establish the direction in which you are looking. I can only speak for myself, but my intuition tells me Black is either dead already or unsettled. So, I'll investigate Black to move first.
After , the reading branches into a couple of answers, (here and for me a, b, c and d). Three sequences: Black a - White d; Black b - White c; Black c - White b, Black d - White e all give life for Black. But the sequence Black d - White a; Black c - White b leads to auto-atari and death. Here, there is (at least for me) no branching. (See Kyu Exercise 27 / Solution has diagrams for these sequences.)
My assessment of this problem grows closer to "Black is dead already" because his best attempt (I can think of) fails. Further reading should be less hard:
After here, White plays at the vital point of an unfinished bulky five and I need no further reading to know Black is dead. I instinctively know
at b will also be answered at a and Black is dead.
The same will go for other attempts, and here I quit reading. I am not a 100% certain, but in Go I am happy with 99% if that allows me to reduce time spent.
Conclusion
The message is not to always have the solution right. For example,
Not everybody will think of all these moves as viable answers, hence miss the solution or possible refutations.
The important thing is to branch, wrap up and allow your intuition to cut off dead ends. See Kyu Exercise 27 / Solution.
How not to read
Although this move is a viable attempt, playing it merely because it looks right then see what happens is not reading.
Comments
JohnAspinall: I think that any attempt at instruction that says "use your intuition to ..." is doomed to failure. Either your pupil has the intuition already, in which case they don't need your instruction, or they don't. Suppose you go to a 9dan player and ask for a lesson, and he says "Use your intuition to pick the best move. Lesson over." I think you would not feel that was a good lesson.
Is intuition real? Sure. But it's an indicator that the process of introspecting (thinking about how you think) is failing at this point. When you intuit something, you're still thinking about it, but it's a familiar, probably practiced, bit of thinking and it had disappeared below the level at which you can observe yourself performing some procedure.
The goal of much learning is to make more things intuitive, that's why we practice life and death problems, for instance. We want to speed up our thinking so that we can solve harder problems. But the goal of teaching is to enable the pupil to turn something into intuition. You have to give them the "something" first. This is why teaching is hard, and why a good e.g. go player may not necessarily be a good go teacher. Raising a procedure up from the intuitive level so it can be taught is a different skill.
Dieter: I was not saying "use your intuition to find the best move, period", but "allow intuition to reduce reading effort". A method declining intuition would process all available points, which may do for a computer program, but is not the way for humans to read, I think.
unkx80: Treat the intuition as a heuristic to get nearer to the solution path, but it does not eliminate reading altogether.
Computers programs don't have intuition, and currently there are no good heuristics for computers. So it is largely enumeration of all possible moves, but the search space is too large, meaning that it takes too long to properly search a sufficiently large problem. This is why current computer Go programs are weak compared to similar chess programs.
I guess, if there is a good "intuitive" heuristic available for computer Go programs, then they can cut down the search space significantly and get much stronger than they are now.
Calvin: Since reading is the weakest part of my game, I am working hard to improve it, not just by doing tsumego but also by just trying to be concious of the process of reading. Often during real games the problem isn't so much that I can't read but that I'm too lazy. I.e., I might see a the potential for a sequence that is good, but rather than read it out, I'll just play a safe alternative. This is particluarly true the more tired I get. So choosing to read is as important is being able to do it.
I've like to mention some books that talk a little bit about reading. The introduction to the Davies Tesuji book gives an excellent example of a mental process, but it may be too rigorous to be psychologically realistic. Lessons in the Fundamentals of Go by Kageyama of course is a must, but it mainly emphasizes ladder reading as an exercise. It's more of a kick in the butt than a tutorial.
But the most interesting book I've read recently is not a Go book but rather a chess book called "The Inner Game of Chess: How to Calculate and Win" by Andrew Soltis. Now, I'm not a chess player, but this is not a typical chess book; it's more about the process of mentally calculating and evaluating variations.