A vague treatise on influence

[daal]

@Bill

Perhaps you could say a bit more about how white uses influence in that game. You mention around move 104 how white cashes in on his influence, but I'm not sure which influence you're talking about. Which stones or groups do you see as exerting powerful influence, and can you say more about how white profits from it?

[Bill Spight]

@Bill

Perhaps you could say a bit more about how white uses influence in that game. You mention around move 104 how white cashes in on his influence, but I'm not sure which influence you're talking about. Which stones or groups do you see as exerting powerful influence, and can you say more about how white profits from it?

White's initial wall was not very strong, but White extended and strengthened it with his counterattack on the top side. Later he initiated an attack on the bottom side. Then when Black starts his counterattack and White hanes in the bottom left, Black does not protect the corner. If he did, his counterattack would fail. White would connect his stones while attacking Black's floating stones. If Black saved those stones, White would wall off the left side, probably on the second line, and make territory there. So Black pushes through, ceding the corner to White.

As White continues the attack with W100 - W110, he makes territory. Not so much, perhaps, but the whole process of the attack has been one of cashing in on White's influence in the center and left side.

[RobertJasiek]

RJ is a go player interested in arcane theoretical matters

I am interested in both ordinary and arcane theoretical matters.

He doesn't write books to get stronger, he writes them so that people wishing to get stronger can do so by reading his books.

While writing the books, I also learn a lot because writing forces me to think yet more about everything than my laziness wishes:) Besides teaching, there are more reasons for writing the books, including a) getting a good reference for myself and b) advancing go theory.

[daal]

Then when Black starts his counterattack and White hanes in the bottom left, Black does not protect the corner. If he did, his counterattack would fail. White would connect his stones while attacking Black's floating stones. If Black saved those stones, White would wall off the left side, probably on the second line, and make territory there. So Black pushes through, ceding the corner to White.

So the stones we're talking about that are exerting influence in this case are G5 and E7?

[daal]

While writing the books, I also learn a lot because writing forces me to think yet more about everything than my laziness wishes:) Besides teaching, there are more reasons for writing the books, including a) getting a good reference for myself and b) advancing go theory.

I'm sure that learning is a nice side effect of writing books and papers, but Kirby's post compares your analytical approach to MW's intuitive one and suggests that your's might not be the best way to improve, and I simply wanted to point out that while you may also be interested in improving, your approach has other functions: it helps your teaching and suits your interest in researching arcane (and ordinary) topics.

While few go players approach go in such a methodical way as you do, I imagine that it's the only way that you can, and whether it's the best way is probably irrelevant.

[Bill Spight]

Then when Black starts his counterattack and White hanes in the bottom left, Black does not protect the corner. If he did, his counterattack would fail. White would connect his stones while attacking Black's floating stones. If Black saved those stones, White would wall off the left side, probably on the second line, and make territory there. So Black pushes through, ceding the corner to White.

So the stones we're talking about that are exerting influence in this case are G5 and E7?

I have edited the SGF file to include a variation for to try to illustrate what I mean. I am talking about outside influence, not influence in the second sense, so the stones I am referring to are the wall in the top left and center and the fifth line wall on the bottom side. There is some possibility of renewing the attack on the large Black dragon, which would involve the wall on the right side, as well.

White played for influence in the opening, and finally made it pay off.

[RobertJasiek]

methodical way as you do, I imagine that it's the only way that you can

Although I do not believe in intuition, I can pretend to decide or learn as if it were intuition-like. For lightning games, it can be useful to save time when reasoning would be significantly slower. For learning it is extraordinarily inefficient by a ratio of roughly 1:10,000 (1 principle with 1 example versus 10,000 random examples of which one conveys some new principle).

[RobertJasiek]

I think that there are two different senses of influence that are in common use. [...] power or outside strength. [...] The second sense comes from computer go (as far as I know), and means the effect that a stone or group of stones has on empty points or stones.

I am not sure if I understand what you are saying. Now I guess you might mean (the influence-generating aspect of, e.g.) thickness when you say "power or outside strength" as influence type I? Do you mean "influence" when you speak of (the influence absorbing) influence type II?

If so, then also type I can have negative connection or life values (it is not thickness then but thinness).

What is your research about, if I may ask?

[lightvector]

I think Jasiek's approach to the game has merit, although I can understand Magicwand's frustration.

For certain tasks, by far the most effective way to become skilled is by simply practicing the task and learning by example, especially if one can get good feedback. Many physical skills and sports fall are like this, as are things such as learning a new language. Almost any amount of abstract rule-based learning cannot teach you how to react effectively and coordinate your muscles to hit the ball in the right way. Similarly, you can try to learn the complex grammatical rules of a language and their numerous exceptions and all the different rules behind sentence structure, adjective order, verb conjugation, etc. But if you want to become fluent, once you have the very basics, it is usually more effective to learn the rest by immersing yourself in the language and interacting with other speakers. Even simple tasks like object recognition are also of this type - you can very easily teach someone to recognize a new type of animal by showing them a few different pictures, whereas describing all the precise characteristics could be difficult (for example, different breeds of dogs can vary enormously in size, shape, and appearance). We take it for granted, but the difficulty of object recognition and classification is evidenced by the fact it's still a major research problem in computer vision!

The unconcious/pattern-matching/instinctive part of the brain has an astonshing ability to learn and generalize in approximately the right way through examples and repeated feedback. Through examples, we can learn incredibly complex rules and interactions, and depending on the task, sometimes even faster than it would be to even write down what all the rules even are (particularly if there are a large number of them and a large number of exceptions to them as well).

Much of this applies to Go. Many players have become experts at understanding "influence" and "aji" and "good shape" and "thickness" by being told very little other than that such concepts exist and being given feedback by stronger players - "yes, that's good shape", "no, that's bad shape", "yes, that group is thick", "no that group isn't thick". As usual, the unconscious part of the brain has worked its magic, enabling these players to learn and skillfully apply these concepts, and yet often leaving them unable to precisely describe what the concepts are! It is plausible that for many players, just like learning a new language, learning by immersion and example could be a much more effective way of becoming fluent than methodically learning all the precise rules and definitions.

But there are disadvantages to this type of unconscious, example-driven learning as well. For one, knowledge acquired this way is difficult to transmit to others, which makes it difficult to teach beginners. Moreover, if you yourself have picked up some misconceptions or converged on something not quite right, it can be extremely difficult to determine what you are doing wrong and how to fix it. In such cases a more scientific approach could indeed be extremely effective. And for some aspects of the game, like capturing races and the microendgame, it is actually possible to enumerate the different cases that can arise and come up with exact rules for evaluating them. And even where it is not possible, it probably helps if you at least start out with approximately the right general concept to begin with, rather than learning purely from trial and error.

As a result, there is no reason why an approach like Jasiek's could not be effective. To go back to the language example, the complexity of human language has not stopped people from developing a useful theory of linguistics and then using the results for useful applications that would otherwise be impossible, such as automated language translation (still poor, but very very gradually improving). Jasiek's research has not been necessary in the past for players to become strong, but it is nonetheless an interesting way to approach the game and could potentially be very useful for many players.

[Kirby]

...

the best indicator of which method is better can only be seen by who wins more games.

There are more indicators. An important other indicator is: Formally proving the truth of a theorem.

I do not see "formally proving the truth of a theorem" as an indicator of a better method. But perhaps this shows a difference in our goals behind the game.

In other words, if having a more intuitive approach leads to more wins than a systematic one, I feel that that approach is probably better for getting stronger.

Why? The alternative explanation is: It has been more popular in the past and this still shows in the present.

Your alternative explanation is possible, but I don't think that it's actually the case when I consider other areas of learning.

One example that comes to mind is language learning. You can devote hours of study to learning grammar for a foreign language precisely. You might even know more grammar detail than a native speaker of that language, if you study for long enough. But in my experience, studying grammar pedantically is no match for practicing speech with a native speaker.

If I compare this to go, the best way to learn the "language" is probably to learn from "native speakers" (i.e. pros). I would much rather study from a native speaker that lacks some methodical knowledge (eg. grammar rules in the language example) than a non-native speaker that has studied grammar more precisely.

I would venture that it's probably more useful to focus on the same things that pros do.

Be patient! Method requires understanding the fundamentals first before the laurels can be collected. BTW, in the meantime also professionals can learn from methodical insight for both playing and teaching.

I agree that studying the fundamentals first is essential for expertise. But if you claim that there are pros that lack methodological knowledge that you have, it is simply evidence that such knowledge is not necessary for expertise.

If pros win more because of experience and intuition, then by golly, I think that experience and intuition is the way to go. Studying other stuff is interesting, but it's probably not the best way to get better if pros don't adhere to those methods.

Wait another 30 to 50 years and you will see that your guess is wrong:)

Well, it's entirely hypothetical, so in another 30 to 50 years, perhaps you'll change your mind, too.

---

I've been thinking about this a bit more, and I think that my opinion has changed slightly. I still believe that experience and learning from those that are stronger than you (eg. pros) is essential. There is also value in learning the fundamentals.

If I have to choose between learning by technical methodology and by intuition and experience, I feel that intuition and experience will get you further PROVIDED you have some grasp on some fundamentals - we can see that pros have that, already.

But where my opinion has adapted is in that I believe it is necessary to achieve a balance between the two approaches. They are not mutually exclusive. You can learn methodology, and can even study proofs of certain aspects of go. I do not find this as useful as playing others by experience, and gaining intuition. But having both your right brain and your left brain work together is bound to achieve good results.

My conclusion: You need to have balance.

[RobertJasiek]

Jasiek's research has not been necessary in the past for players to become strong

Up to 1 kyu, what I needed I found as generalised knowledge in books, except that I had to bite my way through a joseki dictionary because at that time generalised joseki knowledge did not exist; about 50% of the effort was reading that dictionary and trying to understand and learn the variations. From 1 kyu to 3 dan, I needed mostly what I had neglected earlier: opening, life and death and tsumego; no useful generalised knowledge existed at that time in English (not even for opening), so it was hard work but possible.

Things dramatically changed from 3 dan to 5 dan because I had neither: generalised knowledge and specialised dictionary style books were not available for what I needed the most then: fundamentals and thickness / influence. What was not available I had to invent myself to become stronger. So I read what I had: piles of unsorted, totally unstructered example books. All in the hope to occasionally find some interesting diagram (mostly such fake ordinary examples the book author did not highlight) that possibly could tell me something knew if only I would think hard enough what kind of thing that was to grasp as a new general idea.

Therefore I think such research or explanation of its results is necessary for players with a similar experience of needing something not found explicitly in literature before.

[RobertJasiek]

I do not see "formally proving the truth of a theorem" as an indicator of a better method.

A method proven (by means of expressing it as a theorem, which is then proven etc.) to convey the truth is the most correct. No other method can be more correct than the truth. E.g., if your other method is intuition, then it can at best be as good as the proven method. This lets the proven method be the in general better method; always correct instead of only sometimes correct.

But perhaps this shows a difference in our goals behind the game.

No. It shows your lack of understanding of what proven truth implies.

I don't think that it's actually the case when I consider other areas of learning.

(Currently) I do not claim it for other areas. It might well be a speciality of go "learning culture".

If I compare this to go, the best way to learn the "language" is probably to learn from "native speakers" (i.e. pros).

No. My experience is: Professionals mostly teach as weakly as example only books and almost completely failed to assess my major weaknesses (except that they noticed one weakess they were all strong at: local life and death reading), something I wish I would never have to say of any teacher but in go it is the unfortunate norm rather than the exception.

if you claim that there are pros that lack methodological knowledge that you have, it is simply evidence that such knowledge is not necessary for expertise.

Expertise at winning games does not require expertise at all fields because players may make mistakes and can afford to be weaker at some fields for which methodical knowledge would help them easily. They also may be very weak teachers because that is immaterial for have expertise at winning one's (the teacher's) own games.

There is also value in learning the fundamentals.

A great value:)

But where my opinion has adapted is in that I believe it is necessary to achieve a balance between the two approaches. They are not mutually exclusive.

It is not a matter of exclusion but of efficiency. The inefficient method becomes superfluous when the always correct method provides the answer.

My conclusion: You need to have balance.

My conclusion: The still methodically unexplored parts of go theory also need to be explained methodically.

[Go_Japan]

Kirby and LightVector:

There are other examples besides language learning of this type of cognitive intuitive understanding. Two of the most popular ones that have been studied recently involve plane identification and chicken sexing: http://discovermagazine.com/2011/sep/18 ... ou-realize. There are a lot of others, though. Anything written by Gerd Gigerenzer in the smart heuristics research program generally look at such intuitive understandings based on adaptation in our brains through evolution. Other fun examples are included in a book that I think makes a good read by Klein called Sources of Power.

[Kirby]

...

But perhaps this shows a difference in our goals behind the game.

No. It shows your lack of understanding of what proven truth implies.

Um, no. I like "proven truth" as much as the next guy - I was a math major, and that's what we do.

But I don't feel like I can treat go like a closed system the way I do with math.

I think it's easier to use the scientific method: get a hypothesis and test it.

This is the essence of experience. Play games against someone stronger. Play a new move. Improve your hypothesis power.

You can't prove everything in the game. But you can use the scientific method.

[RobertJasiek]

You can't prove everything in the game.

But what is or can obviously be proven can be applied. For example, semeai formulae.