Opening principles for beginners

[jumapari]

Even if some pre-AI opening are now out of fashion (sanrensei, Chinese opening, Kobayashi,...) I think it's still interesting to study them at least at kyu level. First because the ideas you learn may be useful in other situations. And second, because these openings are still played in amateur games, and an opponent who knows these openings well may be able to trick you.

It's like learning reading and writing, before one starts to think about writing his own book
As long as I frequently make huge mistakes in every game (for instance in life and death situations), it not really depents on some small difference out of the opening.
but learning the old school and old common josekis is like learning how to play this game at the opening. If I managed to understand, what is a good shape, and why some josekis are better in a certain situation, I can practice that. Than someday somebody will knock me down with a 3x3 Invasion like Alpha Go, and I will go into that. I think that's a good way.
Thank you very much for your answers.

And thanks for Mr. Jasiek for this great joseki books. Thats exactly what a beginner like me needs to understand them. Even it's a bit strange to read them in english when I know that Mr. Jasiek is a german like me

[Bill Spight]

And thanks for Mr. Jasiek for this great joseki books. Thats exactly what a beginner like me needs to understand them. Even it's a bit strange to read them in english when I know that Mr. Jasiek is a german like me

Cher Robert's mastery of English over the past several years is an example of his persistent and thorough pursuit of excellence.

As long as I frequently make huge mistakes in every game (for instance in life and death situations), it not really depents on some small difference out of the opening.
but learning the old school and old common josekis is like learning how to play this game at the opening. If I managed to understand, what is a good shape, and why some josekis are better in a certain situation, I can practice that. Than someday somebody will knock me down with a 3x3 Invasion like Alpha Go, and I will go into that. I think that's a good way.

As Robert indicated above, the less you have to unlearn, the better. Go is a long game in which small errors tend to accumulate and persist. In terms of the temperature of the whole board, it reduces over time, with fluctuations, such that the swings at the end of play are much smaller, as a rule, than the swings at the beginning. As gennan indicates, in chess the temperature at the end can be greater than at the beginning. For instance, when a pawn promotes there is typically a large increase in temperature in chess. The same is true, as Reti pointed out (without using the term), when a closed position opens up. That's why Reti advised beginners to play openings such as the King's Gambit, to learn how to play open positions, so that they learn to be prepared for when the game opens up.

There is an analogy to that in go, as when the liberties of stones are taken away, it is possible for local situations to heat up considerably, and that can happen at the end of play. Double Digit Kyu players (DDKs) often make make huge mistakes at the end of the game, when liberties are filled in. That's why you should always fill in the liberties before passing. And why you shouldn't resign.

But, as I said, go is a long game. Some people think that a mistake that loses only 2 points in the opening is not worth worrying about. Well, if that were the only mistake in the opening, that would be so. But the difference between a professional and a DDK is on average, about 2 pts. per move. Small losses tend to persist and accumulate.

The thing is, it is relatively easy to avoid 2 pt. losses in the opening. Imitating current professional opening play, even without learning opening principles, per se, is a good start. It's not like Alpha Go and other AI play early 3-3 invasions, but humans don't. Now everybody plays early 3-3 invasions. When reviewing your own games with AI, of course you should learn how to avoid blunders. But also pay attention to the AI's suggestions for the opening, even if they only make small improvements for each play. Go is a long game. Small improvements add up. OC, you don't need to spend much time on each suggestion. And you don't have to slavishly follow them. But take note of them.

[jumapari]

The thing is, it is relatively easy to avoid 2 pt. losses in the opening. Imitating current professional opening play, even without learning opening principles, per se, is a good start. It's not like Alpha Go and other AI play early 3-3 invasions, but humans don't. Now everybody plays early 3-3 invasions. When reviewing your own games with AI, of course you should learn how to avoid blunders. But also pay attention to the AI's suggestions for the opening, even if they only make small improvements for each play. Go is a long game. Small improvements add up. OC, you don't need to spend much time on each suggestion. And you don't have to slavishly follow them. But take note of them.

That's a big difference between go and chess. In chess the computer Innovations start mostly in very complicated variations after something like the 15th move. I Go a 3x3 invasion may start with the 3rd move.

I try to take note of every suggestion, think about it and put it in my kind of thinking. Thanks a lot for that!!!

[John Fairbairn]

As Robert indicated above, the less you have to unlearn, the better.

Is this really true? My experience in many fields, including go, is that it is probably not - at least it is only with very strong qualification.

Leaving aside the usual problem with apparently logical statements, that it all depends where you start - you can get a totally different answer if you start with different assumptions - straight off, I posit two objections.

1. It may work best for people with a certain cast of mind, and not for others with a different cast of mind. It may work better for some subjects than others.

2. It all depends what you mean by learn. Learning from mistakes is often the most powerful way to learn. Deep learning. In fact, the whole process of unlearning can a form of reinforcement learning. You often have to learn that way so that you know what a mistake even is. Being bitten in the bum by losing a group because you filled in a liberty through making a sente and not obviously bad Atari 100 moves earlier is more powerful than reading a list that just says "1. Don't make unnecessary ataris." Even if it shows an example. You usually learn more from your own experience than from others - hence the advice to play lots of games.

I suspect the real best method is a mixture of instruction and experience, and that the proportions in the mix vary according to type of person, subject matter, culture and need, teachers, etc etc. But I am inclined to think experience is almost always the flour in the bread, and that way you always end up with a loaf. Adding sunflower seeds from recipe list might make the load tastier, but without the flour you're up a gum tree. And as I've often said before, it's a long lane that has no loaf on the bread. It sounds less logical than the opening statement, but I think it contains more truth.

[Bill Spight]

As Robert indicated above, the less you have to unlearn, the better.

Is this really true? My experience in many fields, including go, is that it is probably not - at least it is only with very strong qualification.

Learning from mistakes, when promptly pointed out, is one thing. Unlearning bad habits is another story. OC, with the death of neurons and the alteration of synapses, true unlearning is possible, in the sense of erasure. However, what happens as a rule is that the original neural connections remain in place, but are inhibited and replaced by new connections which produce new habits. Under stress, such as may happen during the play of a game, that inhibition may fail and the old bad habit resurface. Akin to what Freud called the return of the repressed.

Edit: That is one reason that overlearning corrections to mistakes is very important. Overlearning reinforces the corrections and also the inhibition of the old mistakes.

[John Fairbairn]

Learning from mistakes, when promptly pointed out, is one thing. Unlearning bad habits is another story. OC, with the death of neurons and the alteration of synapses, true unlearning is possible, in the sense of erasure. However, what happens as a rule is that the original neural connections remain in place, but are inhibited and replaced by new connections which produce new habits. Under stress, such as may happen during the play of a game, that inhibition may fail and the old bad habit resurface. Akin to what Freud called the return of the repressed.

No quibbles with this, except that it's not addressing my main point, which was about learning. This is about unlearning.

I am suggesting there is an important difference between learning from a list, or from what somebody else has told you, which may mean you "know" it, but in a specific and superficial way, and learning by experience, which leads to deeper knowledge (intuition, if you like) which has wider, general application.

If someone teaches you to parrot 2+2=4 at school, you may "know" it and even pass an exam by repeating that. But it doesn't help with 3+3. If, however, your teacher gives you 2 sweets and then 2 more sweets, and asks how many you have got, you may make a mistake initially and say 22 or whatever, but you can be guided to learn from your mistake, as opposed to unlearning it. And once you've grasped that, the sheer delight you feel from doing it yourself, makes it a deep part of your learning - an 3+3 is then a doddle.

I'm certain mistakes are very useful in aiding the learning experience. What I'm unsure about is how much mistakery can be tolerated before it degenerates into bad habits and refusal to listen to alternative opinions. My own experience is: quite a lot at the beginning, and the joy of discovering a mistake is a major element in boosting passion for the subject and wanting to learn more. Things I learn too easily I tire of easily. And vice versa, which is why I've been doing go for over 50 years.

[RobertJasiek]

Learning correct knowledge versus learning and later unlearning (partly) wrong knowledge:

As a kyu, I was taught some correct and some (partly) wrong knowledge. The correct knowledge made me stronger very quickly. The (partly) wrong knowledge made me stronger only a bit and later turned out to be a major obstacle delaying my later strength increment. The net impact of (partly) wrong, later unlearned knowledge might even have been negative.

Correct knowledge has not been available for everything but Bill, I and others enhance it. Therefore, learning from knowledge is becoming easier.

Needless to say, there are also further ways of learning, such as learning from one's mistakes, training a neural net (AI or human brain of a player skilled at subconscious learning), or acquiring skills of reading, judgement or self-controlling one's psychology.

[jumapari]

As I started the thread, and even I´m a weak Go but reasonable strong Chessplayer, I have the impertinence to add my point of view to the words of you experts, because I cannot imagine that this should be so far apart between go and chess.

1. in chess, the doctrine has changed again and again over the last 120 years because the game has constantly evolved and developed. In my youth I had training with the coach of a German national team at that time, and almost everything I learned, did not pass the test of time. Why? Because people have understood - also through the advent of computers - that dogmatism is less useful than concrete solutions in a unique position. If the Soviet school of the former socalled "Patriarch" Michael Botwinnik was primarily marked by dogmatism, it developed towards more and more dynamism and concreteness. Today, even the former world champion Viswasnathan Anand says: "As long as it works, I don't have to understand it completely".

2) Already in my youth my coach told me: The students learn the doctrine and the rules, and the masters look for the exceptions to the rules.

3 What Robert Jasiek writes has also found its way into the chess world: You can learn good habits as well as bad ones. Otherwise it is hard to explain that there are chess and go players who, even after 30 or 40 years of playing practice, barely outgrow the level of the advanced or even weaker. In Chess Elo > 1400 and in Go maybe DDK?!

Thanks again for the many extremely competent answers! I have never experienced anything like this in any other forum. Thanks!

[RobertJasiek]

The most fascinating aspect of AI go is that its play mostly conforms to centuries of human go insight. Yes, there have been minor changes (deeper reading, better judgement than most humans, some new josekis / openings). Overall, however, the game is still the same. That is because 1) the game is balanced between tactics, strategy and evaluation and 2) human go theory has been developed broadly for a long time for most aspects.

[Bill Spight]

No quibbles with this, except that it's not addressing my main point, which was about learning. This is about unlearning.

Well, my point was about unnecessary unlearning. I doubt if there is much difference between us.

I'm certain mistakes are very useful in aiding the learning experience. What I'm unsure about is how much mistakery can be tolerated before it degenerates into bad habits and refusal to listen to alternative opinions. My own experience is: quite a lot at the beginning, and the joy of discovering a mistake is a major element in boosting passion for the subject and wanting to learn more. Things I learn too easily I tire of easily. And vice versa, which is why I've been doing go for over 50 years.

I was fortunate, when learning go, that, with the exception of one or two games, all of my games were with players 5 kyu or better. That limited my ability to form bad habits, because they punished most of my mistakes. There are also degrees of mistakes. Replacing a bad mistake with a not so bad mistake is progress. And there are cases, as we all know, where to get better you have to get worse first. A friend of mine was stuck at 1 kyu for a long time and was frustrated. Since he never played reductions and always play invasions, I suggested that he take a month and reverse that. Always play reductions and never play invasions. He needed to learn to play reductions, and doing so would mean that he would make new mistakes.