Don't really get how to even begin solving life and death puzzles of any difficulty.

[Disjunctivist]

Currently working through the beginner material, and I notice that now that I’ve got to 4.27-8 life and death puzzles, I don’t really understand anything that is going on at all. It’s not just that I can’t solve them, is that I can’t even begin to see what thought process I could apply that would lead to me solving them. Take this one, for example, no.11 in the 4.28 group: Play Go at online-go.com! | OGS It’s not just that I can’t read a single line all the way through to the end that clearly kills or anything like that. It’s that I don’t understand what I am trying to achieve at any level beyond “eventually all white stones must be surrounded. I mean, yes “play at the vital point”, but that is not helping. At this point I don’t really know how to proceed with “solving the puzzle”. I could try and read out random lines where I start taking away white liberties somewhere. But without some idea of what I’m going for, the tree of reasonable moves quickly expands beyond the point where that is feasible. So presumably I have to instead grok some insight at a higher level. But how do I do that? Stare at it?

How do people actually go about solving puzzles?

[jlt]

Try easier puzzles, like

Tsumego Hero

[JethOrensin]

@Disjunctivist

But how do I do that? Stare at it?

We’ve all been there, so there is no need to be frustrated.

How to find the vital point?
The answer, at least early on, is based on understanding two things:
a) Basic shapes.
b) Basic Tesuji

Unfortunately, those are not things that can be explained in a paragraph or two, so even if you had asked some AI, the result wouldn’t have helped you much.

I am not aware of a free resource that can explain those things fast - checked on sensei’s, didn’t find what I wanted - so, I’ll shamelessly plug this, which is free to download:

A Go book for everyone | A project for a Go book in every language

Pages 141 till 170 are about “life and death” and it explains shapes, tesuji and gives real examples from actual games, not just “problems made in a vacuum”.

If you just want a methodology, here is page 161:

In the problem you mention:

Playing A, and then B, is the solution. Those are the eye making shape points.
A, destroys the main shape point that makes points.
B, is a tesuji where you sacrifice a stone to poke out the potential eye (if White plays there they have an eye - when Black plays there, White nees to play on the 1-1 point to capture the Black stone, thus there is no longer an eye in the 1-1 point).

[Disjunctivist]

Why is it is useful to play at A though, sure, you then can’t make an eye at one above A, but why can’t you just have the eye at A itself instead?

[uzaya]

Ngl I was trying to explain the solution in a long way but then it became too confusing… So to simply put:
White has an eye on above A, if white plays at A then it threatens to create an eye both at the left side of A and right side of A. When you (black) play at A, you prevent white forming an eye on both right side of A and left side of A simultaneously.

[Uberdude]

A useful tip for tsumego is to think where the other colour would like to play: if it’s white to play, where is the most obvious eye-making move that literally makes a single eye point surrounded by 4 white stones? A. So that’s a good idea to start at for black to stop white making eyes.

[JethOrensin]

@Disjunctivist

Why is it is useful to play at A though,

Fair question.
So, let’s think this a bit. If White plays first, then this is the “plan” for White, in the simpliest possible moves and White is alive.

If you visualise this end result as Black, you want to prevent the above image from happening.

Let’s say you try to take out the corner eye out first:

Now White will capture are 4 and get two eyes again, so this doesn’t work.

Let’s try to tackle the main shape then, that anchors both eyes:

This attacks the eye in the middle and with 2, White threatens to make an eye with A, later and by capturing the Black stone, so Black plays 3 at A and kills the group.

Now, why is the group dead?

The space at 3 is no longer an eye, since Black can atari by playing B, and White now is in atari and needs to play 10, which will leave his group with one eye, and thus it will be dead.

I hope that helped.

[jlt]

@JethOrensin

Let’s say you try to take out the corner eye out first:

Now White will capture are 4 and get two eyes again, so this doesn’t work.

Although I think that if Black 3 is at A4 then Black can get a disadvantageous ko.

[JethOrensin]

@jlt

Black can get a disadvantageous ko.

Nice! I hadn’t noticed that.
I just went along with the problem interface which considers this a “wrong solution” and tried to explain why, without looking more into it. Though I most probably wouldn’t have noticed the ko even if I had looked into it, to be honest.

[Gia]

My New Book - Mastering basic corner shapes ( Step by step )

Does this kind of thought process help you? Like see the answer and deconstruct it?

[dokbohm]

Welcome to go— the thought process is beyond belief for some like me —-maybe someone here can help you,—– you got that far in your studies, so that is good – but understanding it thats something else good luck

[Fragile_Pākehā]

@Uberdude

A useful tip for tsumego is to think where the other colour would like to play

^^ This is gold for some puzzles, like the one in the OP.

In the back of my mind for this particular puzzle, I also know “capture three stones to form an eye”. Obviously, there’s not actually three stones in the corner (A2, A1, B1), but if I imagine that there were, I know I’d get an eye for capturing the three (imaginary) stones, plus the eye at C2…if I were White.

---

Inappropriately long musings

fwiw, I’ve often had the exact same question as you “How do people actually go about solving puzzles?” and for the same reason “the tree of reasonable moves quickly expands”. I’m currently working on the AI generated puzzles on Black To Play and, especially if I’m sleepy, it’s like looking at a QR code sometimes. I mean, there’s 19 chains on the board, where do I begin finding the first move? What is even the objective? And after that all the possible responses and before long I’m ranting at the screen about having to read 17,483 branches. “Sadaharu himself couldn’t solve this” I cry then I give up and guess a random move. Sometimes. Other times, like JLT suggests, I swallow my pride and lower the difficulty to something in my range. In the gym there’s a saying “If you lift light weights as if they were heavy, one day you will lift heavy weights like they’re light”. So I’m trying to do the easy puzzles with ‘good form’, actually reading all the way through before playing the first stone.

Developing reading skill is difficult - but it’s difficult for you opponent also.

[Myrsilos]

Disjunctivist, do you understand the basic point that when a group is “killed”, all its liberties can eventually be removed and it can be captured and removed from the board?

That’s the only “conceptual” thing to know about L&D puzzles. If you’re having trouble, start with easier puzzles and do a lot of them before returning to the ones that gave you trouble. (You can even keep playing past the “solution” to actually remove the stones, to convince yourself this is the point.)

You say: “But without some idea of what I’m going for, the tree of reasonable moves quickly expands beyond the point where that is feasible.”

This relates to training your cognition by doing easy L&D problems, not memorizing an algorithm for puzzle-solving. There are rules of thumb that you learn as you go along, but being able to enumerate the rules of thumb doesn’t make the size of the tree you’re referring to any smaller. It’s a lot like reading out ladders: the first time you do it, the sheer length of the stones going all the way across the board makes it seem impossible, but once you’ve practiced a lot you can read out ladders that hit a complex mess of stones, change directions, etc.

[FritzS]

Do you think you have a good enough understanding of life and death and the concept of making two eyes already?
Because this is the “higher level” you are talking about.

[PJTraill]

I shall try to sketch out a thought process that leads you to the answer. If you can follow it, it should help, if not, you need to review some more elementary techniques.

Here is the original puzzle:

We need to trim that to manageable proportions. We think where White must play to complete their outside border, if Black answers any attempts to expand it. In the long run the largest they can make it is occupying A, B, and C. (They could try tricks like jumping to D, but the will not make room for eyes, and in the end will have to come back to C. Similarly they can push at E, but will have to play A in the end.)

Here the blue line is crucial, as it shows those spots where White has a chance to build eyes; the Black line around the outside of White’s wall is not really important.

So we can narrow the area we look at down to this:

We have already reduced our search tree considerably: 5! = 120, which sounds doable, with some effort.

Now we think what White could do to make two eyes in there. If you are familiar with nakade shapes you should see at once that Black could kill by occupying these two spots:

Your can also find these spots by seeing that they are the central ones, that connect the others. They have two and three vacant neighbours respectively; the rest have only one.

So White must get one of those spots to live. The resulting shapes are

and

We now go back to the original shape, thus reduced:

Before going further, let me add that in most Life & Death problems we have to choose between reducing/enlarging eyespace from the outside and playing a vital point in the middle of it. Here we have been given a hint, “Play the vital point”.

We now think what happens if Black takes one of the vital points. The more promising seems to be the 3-1 points, as it has more vacant neighbours:

Most of White’s eyespace has been fused into a single eye, and to get another White may try to block at 2 here:

But then Black can play 3, and White can only make a false eye. If White played 2 at A it would not help, and 2 at 3 would be answered by Black at 2. So this black 1 is a success.

But we should perhaps consider Black’s other option, which might turn out better in some way:

Again, White takes the other of the two central points:

This time White lives: they can not be stopped from occupying A, and x and y are miai, meaning that whichever Black plays, White is fine if they play the other.

[Disjunctivist]

Thanks

[Disjunctivist]

Thanks. I don’t understand why the last bit is true: “x and y are miai, meaning that whichever Black plays, White is fine if they play the other.” Why is white fine if they play at x or y?

[Disjunctivist]

I feel like doing the beginner stuff here should be exactly how you “train your cognition by doing easy L&D problems”, but instead it seems they are mostly extremely easy but with sudden random jumps to extremely hard.

[PJTraill]

@Disjunctivist

Why is white fine if they play at x or y?

Do you not see why White @ x works in response to Black @ y, or not vice versa, or neither?

@Disjunctivist

I feel like doing the beginner stuff here should be exactly how you “train your cognition by doing easy L&D problems”, but instead it seems they are mostly extremely easy but with sudden random jumps to extremely hard.

Have you been doing them properly? That means working the answer out in your head before making any moves, and continually coming back to any you do not understand even when you guess the answer. The jumps may be partly explained by gaps in your understanding, but only partly; a lot of people have the same problem, and there is perhaps too little explanation.

I find the BadukPop app useful for doing lots of easy problems, but the free version may be a bit limited for that. The easiest levels are extremely easy, and the levels of the problems are better calibrated than many other collections. I would also suggest the WeiqiHub app under Train — Topics — Beginner; there should be enough there to keep you going for some time, and the levels are quite well calibrated too (except that they rank nothing weaker than 15 kyu, even when they are really beginner-level, IIRC).

I also suggest you do a mix of lots of easy problems, to sharpen your shape-spotting, and a few hard ones to push the envelope and practice the technique of reading long sequences and variations.

[jlt]

If the one I suggested is too easy, try “Koren problem academy” on

Tsumego Hero

That you can solve easily some problems and fail at others in the same set is normal. The right level of difficulty is a set in which you solve (at the first attempt) about half of the problems.