Strategy vs tactics ( aka theory vs reading )

Kirby · Post by **Kirby** » Mon Aug 30, 2010 3:59 pm

CarlJung wrote:
Kirby wrote:The brain is good at chunking. Learn fundamental pieces of information, and you'd be surprised at what it can put together.

I assume we use the word "read" differently. I use it in the tsumego sense, where you know you have the answer to a problem because you have tried all possible combinations, or at least found a path that can't be refuted. I don't believe that kind of reading can be done on global situations. If you could, you'd only have to play through some 30 stones (I just made that figure up) and then either resign or wait for the opponent to resign since you both could read out what the result will be. Not even the strongest pros can do that.

When you practice reading, you begin to internalize shapes. Situations become recognizable because you've practiced a lot. When you are 30k, you might have to think about how to make two eyes in a 1-ply reading problem. This is because you might consider intersections A, B, and C, and think about the implications of each.

When you practice reading repeatedly, you grow to not have to think about moves "A" and "C". It's obvious that "B" is the correct answer, because the situation is internalized to you.

I believe that as you practice reading, you can grow to be able to solve more and more complex problems. Yes, you might not traverse the entire game tree for every decision you make.

But shapes and situations have become internalized to you since you have practiced reading.

Kirby · Post by **Kirby** » Mon Aug 30, 2010 4:05 pm

flOvermind wrote:You seem to think that "strategy" is trivial. Reading is the only thing that's hard, all that thickness, direction, aji and so on is actually too easy to bother. Nothing sophisticated at all. Yet the pros say exactly the opposite...

I'm not saying that "direction of play", etc. is easy to master. I am saying that they can be achieved by practicing your reading. When a pro says that something is "thick", for example, he says so for a reason. His reading ability is fantastic, and he has internalized shapes and situations from practice.

flOvermind wrote:Has it ever occured to you that you actually have no idea about strategy? Couldn't it be that what you think you know about go strategy is actually completely wrong?

Yes, that's true. What I know about go strategy could be completely wrong. But it will improve with reading practice.

flOvermind wrote:There is a reason why the pros can give us kyu players nine stones without breaking a sweat...

This is because they have internalized shapes and situations from hours and hours of reading practice. When a pro was 5k, he might have had to think long and hard about a situation, and read it out completely. But as a pro, he's practiced the situation so much that it's engrained into his being. So the answer is simple to get to.

But it is simple to get to because he first spent the time to read through the complexity in the first place - not because he heard another pro say "this is good" or "this is bad". He grew his own sense of evaluation by studying reading.

When you study reading, at first you have to think of tons of lines of play. But when you practice enough, things become internalized to you and second nature.

Kirby · Post by **Kirby** » Mon Aug 30, 2010 4:06 pm

amnal wrote:...
Strategy is 100% reading too, I'd have said, just expressed in a notably different way to tactics. But it's still fair to draw a line between the different reading types, and 'reading' usually is used to refer to the tactics side.

This is the point. Even with "strategy", reading is the number one way to get good at it.

palapiku · Post by **palapiku** » Mon Aug 30, 2010 4:09 pm

amnal wrote:
palapiku wrote:They do?

(and "aji", at least, is 100% reading...)

Strategy is 100% reading too, I'd have said, just expressed in a notably different way to tactics. But it's still fair to draw a line between the different reading types, and 'reading' usually is used to refer to the tactics side.

Aji is certainly tactics - you're reading out a specific sequence (that happens to need some supporting stones which aren't there yet).

Kirby · Post by **Kirby** » Mon Aug 30, 2010 4:21 pm

Here's an example:

I was playing go once against an old guy in Korea that was one of my wife's uncle's friends. There was actually a group of people that were there. They weren't strong dan players, but maybe SDK level. There was a "life/death" situation on the side of the board, where I thought I could kill the group.

But upon close inspection, I realized that, if I tried to kill the group, the group could connect outside to another living group, and still be alive.

Anyway, we played the game out, and one of the observers - let's call him Mr. X - pointed out, "You must play at 'a' in this situation, because it kills the group.". We discussed it, and he hadn't realized that the group could actually connect out and live.

Anyway, "Mr. X" knew that 'a' was a key point in killing the group without reading out everything fully. He knew this because he had practiced his reading a lot, and came to know that 'a' kills a group of that shape. In this particular case, killing did not work, so he was actually wrong, though.

Nonetheless, though he was wrong, the reading practice that he did told him how to deal with that local situation.

I think that pros can do the same thing with global situations. I'm pretty sure that Lee Sedol reads out the global situation in a game. He might think "after moving here, then the opponent can either jump out with a keima, and then I can cut, etc.".

This is a form of reading, which has been honed with experience. Of course, a pro does not ennumerate every single possibility when they are doing this type of global reading. They might omit a possibility, just like "Mr. X", above, missed a possibility in a life/death problem.

But the point is, after a lot of practice, you can do this type of global reading, even though there are lots of possibilities. The more precisely and accurately you can do this type of global reading, the less things you will miss.

By practicing reading, you can internalize the shapes and situations necessary to make these decisions quickly. You will prune a lot of unnecessary possibilities.

Occasionally, you will prune a possibility that should *not* be pruned, but more experience will help you to prune better.

daal · Post by **daal** » Tue Aug 31, 2010 1:01 am

Some of the books I've read, such as Kageyama's Fundamentals are full of examples of players who have lost games not due to reading errors, but because they are following a losing strategy. For example, a player successfully defends all of his groups in a 9 stone handicap game and loses (p. 50) or the player who successfully uses his thickness to surround territory and loses (p. 100). These are players who can read well but are embarking on the wrong plan. Reading won't necessarily tell you whether your objective is correct.

Of course Kirby is right that even if you can win some games on KGS by occupying strategically better points, this won't get you too far against someone who can read better. Just try a handicap game. You automatically have occupied some of the strategically most important points, yet will you win? Getting stronger at reading is probably what most of us need to do most. It is however also worth mentioning that when we do this by doing tsumego, we are not getting stronger all by ourselves - in fact, we are being taught by whoever it was that devised the problem.

flOvermind · Post by **flOvermind** » Tue Aug 31, 2010 1:36 am

daal wrote:Some of the books I've read, such as Kageyama's Fundamentals are full of examples of players who have lost games not due to reading errors, but because they are following a losing strategy. For example, a player successfully defends all of his groups in a 9 stone handicap game and loses (p. 50) or the player who successfully uses his thickness to surround territory and loses (p. 100). These are players who can read well but are embarking on the wrong plan. Reading won't necessarily tell you whether your objective is correct.

That was exactly my point. I often lose games where I made no reading mistake at all, and get exactly what I wanted, both locally and globally, just to find myself short of territory at the beginning of the endgame. Obviously, my reading was not at fault in these games. It was the judgement of the reading result that was flawed.

That's what several pros tell us: Reading is the easy part, judging the result is hard. Sure, that's coming from someone who has already mastered reading, and we amateurs are certainly not there yet. But it surely is telling that the pros are all very close to perfect reading, but still consider positional judgement hard...

Of course there are also games where I make reading mistakes, that is, I don't get what I want and fall behind because of that. I'm not saying you shouldn't practice reading. You can surely get very far by studying reading alone. I'm just saying that there is more to the game than reading, and you can only get so far without proper strategy. How far is that exactly? I don't know. Perhaps it's EGF 1 dan, perhaps even EGF 3 dan, but I doubt it's much higher. You certainly won't get to pro strength with reading alone, although reading is an important prerequisite to getting there.

quantumf · Post by **quantumf** » Tue Aug 31, 2010 6:25 am

flOvermind wrote:How far is that exactly? I don't know. Perhaps it's EGF 1 dan, perhaps even EGF 3 dan, but I doubt it's much higher.

7d KGS, according to Jeff Chang

daal · Post by **daal** » Tue Aug 31, 2010 6:43 am

To illustrate the OP's point, here's a game I played this morning that I won with almost no reading. I've commented most moves to show what I was thinking. Although it would be a stretch to call what I was thinking "strategy," it certainly wasn't reading, and indeed this seems to be enough to win some games at this level.

Nonetheless, in order to improve, I think I need to put my tsumego practice to better use, and not to read more books about strategy.

Kirby · Post by **Kirby** » Tue Aug 31, 2010 7:16 am

flOvermind wrote:...

That was exactly my point. I often lose games where I made no reading mistake at all, and get exactly what I wanted, both locally and globally, just to find myself short of territory at the beginning of the endgame. ...

You made a mistake in reading, then. Unless "exactly what you wanted" was not to win the game.

fl0vermind wrote:Obviously, my reading was not at fault in these games. It was the judgement of the reading result that was flawed.
....

Why don't you consider "judgement of the reading result" a part of reading? If you're doing a tsumego problem, and read out a sequence that gives you one eye instead of two eyes, you get the problem wrong.

daniel_the_smith · Post by **daniel_the_smith** » Tue Aug 31, 2010 8:01 am

daal wrote:Although it would be a stretch to call what I was thinking "strategy,"

It's called "instinct," I think.

flOvermind · Post by **flOvermind** » Tue Aug 31, 2010 8:06 am

Kirby wrote:Why don't you consider "judgement of the reading result" a part of reading? If you're doing a tsumego problem, and read out a sequence that gives you one eye instead of two eyes, you get the problem wrong.

Because "reading" is "I move there, he answers there, I answer here, ...". Positional judgement is "here we have a global position, that's good/bad because...". That's two separate things.
Tsumego is a bad example. You don't need strategy or positional judgement for tsumego, but you do need it for real games.

In case you're really able to read out whole board positions until you arrive at a definite status in the sense of every group having or not having two eyes and all borders closed enough to be countable, please tell me how. I'm sure lot's of people, including pros, would be interested

Monadology · Post by **Monadology** » Tue Aug 31, 2010 8:09 am

Kirby wrote:
You made a mistake in reading, then. Unless "exactly what you wanted" was not to win the game.

This point is either trivial or non-trivial.

Inasmuch as it is trivial, it relies on the obvious and unhelpful fact that he didn't read out the entire game. However, this is a mistake we always make and will always make regarding reading simply because we are incapable of it.

In its non-trivial form, within the limits of his ability to practically read out sequences, your claim would require a lot more substantiation. Substantiation that you haven't provided. I don't see that this point is immediately obvious.

Furthermore, it's quite possible that what he wanted was actually not to win the game. Or, more specifically, he wanted two things:

A) To win the game
B) To achieve some set of strategic goals within the game.

It was his supposition that B and A were mutually compatible. He wanted B because he thought it would be a means towards A, which he also wanted.

However, he was mistaken that B would be enough to achieve A. So the question is: why was he mistaken regarding this fact? The two competing answers are:

X) He made a mistake in his reading (in the non-trivial sense).
Y) He made a mistake in his strategy.

Considering that in order to realize the disconnect between A and B would probably require reading into the mid and potentially end game, I think that we can exclude X as a reasonable cause for his failure. No one can read that far, given all the alternatives. Unless there is some alternative to X and Y, then I think the failure to realize B did not lead to A was a result of Y, not X.

Now, this does not in itself justify the claim that reading is not enough. It could be the case that reading the proper local responses out far enough is sufficient to guaruntee a satisfactory and winning global result. And that this level of reading is within human grasp. This is equivalent to the claim that there is another means towards the end of A (winning the game) besides B:

C) To read out as far and as accurately as possible the most favorable (local) result and play accordingly.

C would replace B as a sufficient means to the end of A.

PS: Obviously B requires a supplement along the lines of C that would be:

D) To read out as far and as accurately as possible the result that most effectively leads to the fulfillment of B, and play accordingly.

Because strategy requires reading.

Kirby · Post by **Kirby** » Tue Aug 31, 2010 8:19 am

flOvermind wrote:
Kirby wrote:Why don't you consider "judgement of the reading result" a part of reading? If you're doing a tsumego problem, and read out a sequence that gives you one eye instead of two eyes, you get the problem wrong.

Because "reading" is "I move there, he answers there, I answer here, ...". Positional judgement is "here we have a global position, that's good/bad because...". That's two separate things.
Tsumego is a bad example. You don't need strategy or positional judgement for tsumego, but you do need it for real games.

I don't really agree. I don't think we need to make a distinction. If you think about the example again, for tsumego problems, you need to have "two eyes to live". The phrase, "two eyes to live" can be thought of as "strategy" in the sense that, by making two eyes, you can make a situation in which it is impossible for the opponent to kill you. It is a *fundamental* piece of strategy, but it could still be considered strategy.

Now there are at least two ways that you could come about knowing that you need to have two eyes to live:
1.) You read in a book someplace, or your friend tells you, you need two eyes to live.

2.) Nobody told you anything about two eyes, but you came to the conclusion that two eyes allowed you to be uncapturable, because you read out some basic situations, and stumbled upon that truth on your own.

We can think in the same way for global situations. We can blindly follow what somebody has told us about something being good, or we can, through experience in reading, learn for yourself what is good in a global situation.

flOvermind wrote:In case you're really able to read out whole board positions until you arrive at a definite status in the sense of every group having or not having two eyes and all borders closed enough to be countable, please tell me how. I'm sure lot's of people, including pros, would be interested

Of course, go is too complicated of a game for me to arrive at a "definite status" of the game result. I'm not claiming to have achieved that expertise in reading. People make mistakes and have limits to their reading ability - and the same is true for tsumego problems (some tsumego problems are very difficult).

That's why nobody can play a perfect game of go. But people CAN use reading to try their best to get a "definite status" of a game result. And the player that does a better job of this wins the game (unless they are lucky).

But what I want to point out is that, what seems impossible for you now - global reading, for example - can become more achievable by lots of reading practice. When you do a lot of go problems, the idea of "two eyes" becomes engrained into your ability - even second nature, you might say. But when you first learned the rules of go, this could be a complicated bit of strategy. Some go problems at the 5k level might seem impossible for somebody to read out at the very beginner level.

But as you practice reading, your brain starts grouping patterns together. Your evaluation begins to become enhanced. You can say with some confidence that "this position is good for black". Of course, you may be wrong sometimes. That's still true of tsumego problems too, though.

People make the distinction between local tsumego problems and the situation on the whole go board, simply because they have a high degree of confidence that they are correct with a local tsumego problem. But I don't think we need to make the distinction. Yes, my reading may not be sufficient to tell with CERTAINTY the result of a go game. But as my reading becomes better, my evaluation is also improved, and I can have more and more confidence in the results of my games.

I think that evaluation is *most certainly* a part of reading. If you ignore evaluation, you are just going through meaningless sequences in your head.

Kirby · Post by **Kirby** » Tue Aug 31, 2010 8:24 am

Monadology wrote:...

You cannot guarantee victory for the entire game, yes. Humans have not achieved that level of ability.

But the further you read ahead, the better you can ensure your result.

fl0vermind claimed in his post that he made "no reading mistakes". This cannot be the case if he omitted some possibilities that led him to victory.

So I think that everybody makes reading mistakes in a go game - just some people make fewer than others.

I think the important thing here is that *evaluation* of a position is a part of reading. If you have no evaluation in your reading, then what, exactly, are you reading?

Life In 19x19

Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )

Re: Strategy vs tactics ( aka theory vs reading )