How to approach a tesuji problem book?

[Javaness2]

I thought that the idea of lie and death or tesuji problems was to read the answers first, then to see if you remembered them. Certainly, when I was looking at opening books, I read the patterns first, then I tried to remember them. The understanding and misunderstanding can come along with practical application.

[Kirby]

Maybe I used the wrong wording, John. I should have said "technique" instead of shape. But technique can also be acquired from solving problems, as opposed to looking at the answers. Staring at a problem for a long time can get you nowhere, I agree. But that just means the problem is too difficult.

I don't know all of the tidbits that pros say, but I speak from experience when I say that I get little value from looking at a solution when I haven't solved the problem. The feeling is, "ok, black's placement here, followed by hane here was the solution, but i just didn't get it."

Solutions do have value to me, and that is to confirm that my solution is correct. When I have carefully "solved" a problem, and then see that the solution includes a line I hadn't considered, THEN it's something meaningful to me.

If I do 100 problems, the techniques used to solve them usually aren't that new. But the practice is useful to really reenforce the "techniques" into my brain.

So to be clear, I believe that the "guided search" to which you refer, can best be acquired through solving such a problem yourself.

[jeromie]

I've actually done better in the problem book than my initial trajectory indicated. I've worked through the first 6 pages (36 problems), and I'm getting about 5/6 on each page completely correct without looking at the solution. This includes the few *** problems I've seen, so I think the book is about the right level for me after all.

I've tried to think in terms of techniques and/or blind spots in strategic goals that I might be missing that caused a few of the problems to give me difficulty.

I think the single biggest concept that I've taken away so far is: whenever possible, leave weaknesses in your opponent's shape even if you can't exploit them immediately and let him/her worry about how/if/when to fix them. I could have told you that before working through these problems, of course, but working through this collection is revealing that that is not always my first instinct.

A lot of the problems are really joseki problems, too. I like having them presented to me as a tesuji problem rather than an attempt to memorize the joseki; I feel I'm more likely to memorize the moves that way.

[schawipp]

One aspect about L&D problems, which might be further helpful, is not only identifying "the" solution but also the refutations for the other "reasonable looking" moves which turn out to be not the solution. In a given L&D problem two first moves may both seem to work on a first glance, however on a second glance one of the two moves has some disadvantage, e.g. the opponent can set up some tricky ko, or one move ends in sente and the other in gote etc. All these tiny nuances can be decisive in a game with the right timing. Especially important: If you look at the solution and find that it's different than yours, you should prove for yourself why your "solution" does not work.

[Knotwilg]

The difficulty (...) is that you need to keep looking at similar problems for reinforcement (i.e. spreading out the brain-wracking work), and we get back to the lack of resources

For people with limited time, such as most of us, the limited resources are not much of a problem. I can regularly return to the same problem and try solving it again. The fact that it is similar to itself is more important than it being equal to itself. In other words: I'm not so much able to solve it again because I've seen the exact position but because I'm reminded of the same technique.

For Life & Death, there is Hitachi's series online, which I've never even finished once and it's still growing.
For tesuji, you will find quite a few with Gogameguru and over time I've been able to return and be baffled multiple times by the same problem.

[John Fairbairn]

For people with limited time, such as most of us, the limited resources are not much of a problem. I can regularly return to the same problem and try solving it again.

My ideal definition of resources is something that is structured - not just a muddled mass of problems.

For tesuji, you will find quite a few with Gogameguru and over time I've been able to return and be baffled multiple times by the same problem.

This suggests to me that you first solved it by brute force once and (naturally) forgot the search tree. If you had learnt the technique the odds are you would have seen it again elsewhere and would remember it still.

I came across something today that may serve as an analogy. I was reading a book about Russian poetry and the author was discussing a very short poem (Я вас любил) by Pushkin. He remarked in passing that it had been learnt by heart by generations of schoolchildren in Russia. I didn't know it, but I do know by heart another schoolchildren's favourite, Черёмуха by Esenin, and so rather sniffily assumed Pushkin's work fell in the same trite class. But in a few short paragraphs the author brilliantly showed how Pushkin's tiny poem contained a stunning array of techniques in addition to the usual iambic tetrameters and rhyme, and these were techniques that you wouldn't normally find in a book of poetry terms (e.g. reinforced rhymes, incremental repetitions, keeping the first person in the nominative case and the second person in an oblique case throughout). And that was just the low-level techniques at the word and syntax level. Overlying all of that, there was a high-level technique of using litotes to give an "optional" interpretation to the whole poem rather than the immediately obvious one, but - again stunningly -the author showed this other interpretation was not really optional. It was inherent along with the usual interpretation in this context (unrequited love) and so reflected a genuine psychological dilemma of the "I" person we all recognise - he wants to keep loving this woman tenderly but at the same time wants to punish her for ignoring him.

I can't do the author's explanation justice here, but none of it is directly relevant here. The point is, I was so dazzled (and I do mean dazzled) that after I read it I found myself going over the poem in my mind to savour it (in other words, I had unconsciously already memorised it by tracing the techniques in my mind) and I found myself desperately eager to look for all these techniques in other poems. Fortunately, on the same train journey, I had a poem by Anna Akhmatova to hand. I had read before that she was a devotee of Pushkin but hadn't thought anything of it. Now I could see instantly what people meant, and I could also see at the same time how Akhmatova added her own techniques, created her own voice, so I was learning even more as I was practising finding Pushkin's techniques.

I suggest that this, in an ideal world, is how L&D should also be studied. There is in fact a good example in the back of one of the books of Kitani's games. Even though it's a book of games they reprint this very, very practical lecture (technique based) on tsumego by him - it's apparently out of place but must have been considered so remarkable it had to be reprinted somewhere. I only wish it could have been in the same class as the Pushkin critic. Perhaps even better for a simple thing like a game would be a Gray's Anatomy for go

The alternative method of brute search and praying for divine inspiration seems to me like the alternative way of memorising poems that, at a guess, we all used at school when we were too young to know about Ricci's "memory palaces" and the like - sheer brute-force repetition combined with fear of detention. Both methods can obviously work in the sense of getting a solution, and they can also teach concentration, discipline and so on. But in terms of teaching techniques applicable elsewhere, a tsumego treated as an opportunity to learn rather than as a puzzle has so much more to offer.

[Kirby]

The alternative method of brute search and praying for divine inspiration

I'm not praying for divine inspiration when I solve a problem. I'm systematically eliminating variations until I find the best one.

Maybe an example is called for.

Here's a go problem that I selected randomly from SmartGo (variations and comments included by smartGo, not me):



I solved the problem by searching through potential responses by black and white.

But what "technique" can I learn from this? Sure, I tried the hane relatively early on, because I "felt" it was a good "technique" to try. But that took about a half second of thought, whereas solving the problem took a couple of minutes.

What technique am I supposed to be learning here? I found value simply from exploring variations, and indeed, I solved it.

[John Fairbairn]

I solved the problem by searching through potential responses by black and white.

From the word "potential" I infer that you did not make a list of all possible initial moves: P19, Q19, R19, etc. You probably selected moves based on your intuition (as confirmed by your mention of looking at hane first). What I think you may have overlooked is that you almost certainly applied other patterns which are so internalised that you weren't aware of using them. For example, this problem potentially includes, as well as space-restricting hane: nakade moves, not answering a cut with obvious connection, threatening two eye-basis stones as a forcing move, a false eye on the edge.

But what "technique" can I learn from this?

The techniques mentioned above are rather simple for a dan player, so I'd expect you not to learn anything from this particular problem. It would just be reinforcement.

whereas solving the problem took a couple of minutes.

But this does surprise me. I took only a second to find the right move and a few seconds to confirm it (using techniques). Normally I wouldn't explicitly list the techniques to myself, but for the purposes of this post I made a special effort, and apart from the techniques mentioned above, my brain also spouted out a few others that were instantly eliminated (things like Black possibly having too few liberties on the edge and falling prey to an oi-otoshi). I don't consider myself a tsumego expert BTW but I likewise learnt nothing new from this problem.

What technique am I supposed to be learning here?

I'm sure you'll understand I don't want to spend precious time digging out an ideal position to illustrate what can be learnt, but here's one that may just fit the bill. White to play and kill.



It's maybe a bit too easy, but I suspect you will not have seen it, and also probably haven't yet internalised the main technique because it is rather rare in practice (it's an "entertainment tsumego by Nakamura Yutaro), so you may learn something. Not quite a clue, but I saw six techniques in about one or two seconds, and the main one instantly, though I have seen this problem before, many years ago. I have found the main technique referred to applicable in quite a few other problems, but I think these were all "entertainment tsumego" rather than really practical problems. This is a curse (or benefit), I suppose, of having worked on books like GTAM.

[Kirby]

Maybe spending a couple of minutes is an exaggeration- I didn't time myself, but the problem does not have an immediate solution to me - I have to verify that some responses don't work. Anyway, I don't disagree at all that I've internalized processes to prune variations, and I think that's due to my practice in doing problems, rather than consciously considering techniques. It's more of a subconscious thing.

My point is that most of the problems that I think are worthwhile at my level don't introduce me to new techniques, but they still take time to solve and are worth doing. I think that practicing visualization and systematically solving a problem is more important than trying to figure out new techniques, since I don't encounter new techniques that often.

New techniques are automatically acquired through practice. Also, learning a new technique is cool, but there's a lot more to being good at reading than techniques (speed, accuracy, etc.).

[schawipp]

I solved the problem by searching through potential responses by black and white.

In your solution, variation 1 leads to a ko. However your variation 5 labelled with "Bad move" would yield the same result if black played P17 instead of Q17. Thus, what is the difference between variation 1 and 5?

[Kirby]

I solved the problem by searching through potential responses by black and white.

In your solution, variation 1 leads to a ko. However your variation 5 labelled with "Bad move" would yield the same result if black played P17 instead of Q17. Thus, what is the difference between variation 1 and 5?

If black plays P17, white plays Q17. By the way, these aren't my variations- it's just a problem I grabbed from SmartGo.

Anyway, my point is not specific to this problem. If I may summarize my argument, I'd say this:

1.) John argues that life & death should be approached not as a puzzle to solve, but as an opportunity to learn new techniques.

2.) I argue that learning brand new techniques is uncommon, and a large part of solving problems is not about finding that new technique, but rather about exercising visualization and systematic breakdown of a problem.

I don't like analogies, but one way to think about it is to compare to long division drills in elementary school. Once you've learned the technique for long division and know how to divide single digit numbers, there isn't much more technique to learn. But there's still a lot of benefit to be achieved by practicing the skill.

[schawipp]

If black plays P17, white plays Q17

That was also my thought, but now that you wrote it I can suddenly see my thinking error. Seems like I had some dust in my optics...

[Kirby]

If black plays P17, white plays Q17

That was also my thought, but now that you wrote it I can suddenly see my thinking error. Seems like I had some dust in my optics...

You knew the technique - you just need a bit more practice to reinforce it

[jeromie]

I don't think John is suggesting that every single problem should introduce a novel technique, but that thinking in terms of technique will help you group problems into reasonable sets and learn to apply the concepts in your game more effectively. I actually think this way of thinking is more important for the authors of a problem collection than for those solving the problems. I can't consciously run through each technique I know in a game situation, but if they have been properly reinforced I will rely upon those techniques automatically when I try to find the important points in a position. As a trivial example, if I do 10 problems in a row where the first move is a hane that lets me kill the group, I'm likely to remember that the hane is frequently a killing move when faced with an in-game life and death situation.

I think good problem book authors do this implicitly, and this is their great advantage over user generated collections like those on goproblems.com (or the SmartGo collection that uses those problems.) Even in the Graded Go Problems for Beginners series, there are often several problems in a row that require a similar approach. They will move a few stones around to show how changes in the position mean you can't play without thinking, but they are thematically connected by the techniques and/or features of the position. I get way more out of problems that are effectively grouped than those presented in a semi-random order, even if the quality of individual problems is comparable.

[Kirby]

I don't think John is suggesting that every single problem should introduce a novel technique, but that thinking in terms of technique will help you group problems into reasonable sets and learn to apply the concepts in your game more effectively.

What I think you should be getting out of a successful L&D problem is not the glow of satisfaction from solving a puzzle but what you get when you can say, "Aha, I have just learned a new technique." The glow of satisfaction should be from adding a tool to your tool chest.

I believe that many "successful L&D problem(s)" don't lead to the satisfaction of having learned a new technique. It's great when this happens, but it's the exception rather than the norm.