Hi again,

I've met the expression "reading depth" here on the forum.

Can someone please explain it?

I guessed that it was like "If I do this, the opponent might do that, then I answer with this, then he might do that...".
At the moment (Capture Go plus the Crazy Sensei Go - who ends up with more stones) I'm happy if I can guess the computers next move.

I've read "reading depth of 10" - does this mean, you guys can predict the next 10 moves of a game?!

I know good chess players can do so, but in Go there are so many options, and (on bigger boards than I play) many territories/battlefields.

If this IS reading depth, how does one learn it?
How can I learn to predict the next 2, 3 moves?
Is this about standard situations or learning pro games by heart?
What if the other player does something else, something unexpected?

And: Should I therefore try to solve all tsumego in my head, or at least try to do so?
(I'm not doing tsumego right now. Stopped after getting your feedback on my other thread in favour of CS/Capture Go).

Thanks!!

Reading depth of 10 moves means that you imagine at least a sequence that is 10 moves long. It doesn't mean that the sequences you imagine are optimal, nor that your opponent will play one of these sequences.

At your level, reading more than 3 moves deep is useless, except in special cases like ladders. Playing games will improve your intuition, so don't spend an excessive amount of time reading for the moment.

Tsumego : what do you mean by "in your head"? Do you still look at the picture?

Remember that's what you have to do for every move in a real game. So you have to start somewhere, and tsumego is good practice. With a real physical set at home, you can experiment and find out which methods work best for you, to learn a shape, a sequence, and variations.

Reading in Go and chess share many common elements, yes.

Yes, I don't mean without looking at the problem, but without using a board to play possible moves, or enter a move on the screen and see what happens.

It is easier to test moves, and if they fail, start over.

But yes, as EdLee says, thinking ahead without having a board as help is closer to a real game.

To read ahead further than just the move that you intend to play, you need to predict your opponent's best response correctly and then your best response to his response etc.

To predict those best reponses correctly, you need to have a good intuition for which moves are good in a particular situation, so you can quickly discard bad moves to avoid wasting time in a wrong branch. But you need to check some alternatives to confirm that your predictions are actually the best response and that you don't overlook something. And in a real game, you have to envision if the end result is actually OK for you, given the rest of the board position.

You need experience to do this well in actual games and to get a feeling for your own reading ability to decide if you go into some tactical path or not.

Reading ahead further and getting stronger go hand in hand (they feed each other). Reading 1 or 2 moves ahead in local tactics is appropriate for novices. Around dan level, players can typically read ahead 20 moves or so in local tactics with few branches. But simple ladders have no branches and even beginners may be able to read those to an arbitrary depth after some practise.

It's easiest to develop this reading skill by practising with small tactical puzzles such as tsumego and tesuji. These puzzles exist for all skill levels.

This is good advice. How do you develop that intuition? Learn good and bad shape. I recommend staring here: https://senseis.xmp.net/?Shape

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

Not hardly.



First, learn what are good moves. Today's bots do that.



Segoe recommended trying that first, then setting up the problem on the board and playing over variations, then looking at the answer.

At your level I would not spend more than 30 sec. trying to solve problems in your head. You do not yet know enough good plays and shapes.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

BTW, "reading" refers to solving some local tactics. It focusses on liberties, cuts and eyes.

When stones are further apart, like in the opening, it's not really possible to read ahead much. There is a lot of freedom in the opening and it involves matters like the latest opening fashion and the style and taste of individual players.

To learn more about the opening, you can read books about this topic, replay pro games or just play and use game databases to look up some of the opening situations you encountered in your own games. From there, you may be able to predict some of your opponent's opening moves, from your experience and/or from knowledge that some opening moves are considered better than others. Or perhaps you just know your opponent and his favourite opening.

@ gennan:


Could you elaborate on this, please?
Does this mean "Ah, yes, that's a ladder." or does it mean "It is a ladder, so stones go there, there, there... oh, there's a white stone she can connect with!! Ladder won't work."?



Those are standard combinations of moves, with joseki being a subgroup played in corners and not "winning" anyone the corner??
I've seen there are websites on Joseki, and an app, looks complex.
Where can I find tesuji for beginners?

@Bill Spight:


Do you mean watching what the bots do (good moves), or getting the replay-SGF?
I've seen that Kozumi has some code for replay, and in the beginning, it calls one move "bad move".
But first, I don't understand the code (how are the lines counted? I was trying from bottom to top, left to right, but that seems to be wrong? Eg 3-4?)
Kozumi never comments a second move.
Do other bots do that?

No, I just meant that today's bots start off playing very badly, worse than human beginners, and gradually (for them) learn what moves are good. Human beginners can actually learn the basic patterns of good moves with less experience than bots. We're pretty smart. IMHO, that's better than learning how to do brute force search, which today's bots don't learn to do, either.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I mean that with some practise, beginners can solve ladders such as the longer ladder example here, to objectively determine if the ladder succeeds for the chaser or the chasee. When there are stones on the path of the ladder, it can becomes harder to solve.


Tesuji and joseki are different things:

Joseki are move sequences aka patterns that occur frequently in a corner in the early stage (=fuseki) of stronger players' games. The results of these patterns are supposed to be fair for both players (=even), ignoring the rest of the board. The joseki at the top of this page is probably the first joseki that people learn, because it's one of the simplest. But there exist very complicated joseki and there exist joseki dictionaries containing tens of thousands of diagrams. It's not recommended to study long and complicated joseki until you are close to dan level, because you're on your own when your opponent deviates and it's easy to collapse by a small mistake or a minor gap in your knowledge.

Tesuji are skillfull moves in local tactics. A tesuji cleverly exploits some issue in your opponent's configuration of stones (=shape), such as cutting points and lack of liberties. There are many types of tesuji. Ladders, nets and snap-backs are usually the first ones beginners learn, also see here. Familiarising yourself with tesuji improves your tactical intuition to detect shape issues and practicing tesuji puzzles helps to improve your reading ability (as does practising tsumego puzzles).

I can recommend the free online beginners course at 321go to start learning about tesuji, tsumego and reading in general. That course has thousands of puzzles organised by different topics, starting with the rules of the game for novices and going up to fuseki and joseki problems for single digit kyu players. The user interface is a bit dated, but the content is really good and well organised.

You will find a few simple tesuji problems here: https://gochild2009.appspot.com

Wow!!

So much to study

I think I'll need to stick with one project for some time, because the resources are so overwhelming.

But thank you very much for all the input!!

I think I'll take "Teach yourself Go" (Matthews) on my vacation next week.
I made a tiny travelling go (some of the plastic chips of my go set, plus a printout of a 9x9 board).

After that, I think, "Learn to play go" might be good?

I also found a great series of Go teaching videos on Youtube, lesson 5 yesterday explained geta.

Watching the videos (and if necessary replaying the situation on a board) is easier than reading a book, so I tend to do that when I'm tired.

(I get up one hour earlier now to play and maybe soon study go. A nice way to start the day, a nice mug of coffee while being slaughtered by a bot )

Thank you for your help - the links are bookmarked.

The great Japanese player Kato Masao wrote a series of books (in Japanese) with titles like "Three Move Tesuji" or "Three Move Life and Death". The idea is you pick a move to study (move 1) then decide what your opponent would do if you played that (move 2) and then decide what your answer would be in that case (move 3). Reading three moves like this for most moves in the game is excellent practice for double digit kyu players and even for single digit kyu players.

For myself I find that I can read better when the problem is set up on a board with stones, rather than just looking at a diagram in a book.