Ian Butler's Go Journal

[John Fairbairn]

I've gone through the following books at least twice: Tesuji ... I still get bad shapes in my games.

Seems to me that's where the problem lies: using a screwdriver instead of a spanner. Suji != shape. Presumably you use suji and not shape in solving tsumego problems. Why operate differently in tesuji problems? By all means use shape to identify a starting point, but after that you have to go with the flow.

[jlt]

What I meant is: I have practiced tesuji, hoping to improve my reading during middlegame. That I get bad shape (=the end result of the sequence is a grossly bad position) means that my reading skills didn't improve that much.

[John Fairbairn]

What I meant is: I have practiced tesuji, hoping to improve my reading during middlegame. That I get bad shape (=the end result of the sequence is a grossly bad position) means that my reading skills didn't improve that much.

To which I have to say, of course. Just to make sure we are on the same page here, the context is your own remarks:

Differences between reading tsumegos and reading during a game:

• In life-and-death problems, the group under attack usually can't escape, and outside stones are considered as alive. This is not always the case in a game.
• In a game, it may be better to sacrifice a group than to live while giving the opponent too much profit.
• In a game, the goal is usually not to live or to kill but to get a good position, so it is necessary to ave a good positional judgment as well.

I might quibble about the last bullet point, but the main thing here is that there is (rightly) no mention of tesuji there. So practising tesuji to achieve something that doesn't require tesuji is a bit pointless, no?.

You learn tesujis, you don't read them. You use learned tesujis to speed up your reading. A move inside a nakade shape is a tesuji. Once you learn that you don't need to read out the precise kill. With tesujis you try to achieve a specific tactical goal: e.g. get sente, connect up, separate the opponent, get a squeeze etc. Just like tsumego you know you've succeeded if you achieve your goal - not by whether you've coincidentally got a better (or worse) position. Your overall game goals are assessed by different, strategic criteria.

If you decide strategically that you are behind and need to gain territory while denying the opponent extra territory, you may decide that a good way to do that is to attack him (yoritsuki). In order to attack him profitably you may assess that you need to create two weak groups of his. For that you may need to divide one group into two. For that you need to look for a weakness (the shape element), and then you need to reach into your too chest and pick out the right tesuji for the job. Then you do the job (reading). The more tesujis you have in your chest the easier it is to do the job. Once you've finished the job you may find you have fixed the hot-water tap to the cold-water pipe and vice versa, but at least you got the tap-changing tesuji right.

There are lots of tesujis of course, and some can seem quite complicated (e.g. the gravestone tesuji). But a vast number are built up from simpler tesujis.

Since we are in the Nutcracker season, another way of looking at it is that the various steps learned by a ballerina are like tesujis. She will learn easy stuff pliés, tendus and battements. As she learns these to the extent she has total confidence in them, she can proceed to practise ronds de jambe, jetés and so on, and may even achieve dan level and learn incoprate thse into advanced "tesuji" techniques such as shedding her knickers, i.e. do gargouillades. She might do all this impeccably and beautifully. But it doesn't mean she can then dance the Sugar Plum Fairy. For that she needs to learn the choreography and the music.

One of the major problems with studying go is that people skip the music and start stumbling through the choreography first - playing whole games. They think about learning the steps only later, when they get sick of falling over. This is arsey-versey and if you tried it in real ballet you would probably end up crippled.

Given the social nature of board games it's not easy to recommend a complete solution, but the one point I would make is that if you decide you are going to learn tesujis steps, just learn the steps and don't try yet to mix them up with dancing (the overall game). Just master the steps so well (so they become instinct not reading matter) that when the dance director says, "OK, tendu into bourré, attitude, développé etc etc" you don't ask her to show you the steps, you just rattle them off in sequence. Then you can learn to dance properly - and with the music.

[Bill Spight]

Tesuji (James Davies)
Just a small note on my progress in Tesuji. The first few chapters were good, I solved almost all problems correctly, my reading was good. The last two chapters I did, however, were a lot worse. 'When liberties count' en especially 'linking up groups'. In that last one, I missed more than half the problems at the end of the chapter and this, too, reflects in my games. Given the situations in the book, I'll never find these answers in a game.

You are actually doing quite well. A good level for learning tasks is when you get about half of them right. The last chapter may be a bit more difficult than that, but you are doing well on the book as a whole. Keep reviewing the book and overlearning the material until, as John Fairbairn says, you have total confidence in your knowledge of it. Don't worry about ending up memorizing the problems. You may not run into those exact situations in your games, but your learning will generalize to similar situations.

So it really shows a big lack of understanding/knowledge about tesuji on my part. For some reason tesuji is a very hard subject for me to learn. The reading is not necessarily the problem (i make mistakes but not the main issue), but finding the actual tesujis and seeing how they work out.

The counsel of perfection. Courage, mon vieux! The Tesuji book is right up your alley. You should be able to master it within a few months, at most.

This goes together with the "uninspired" feeling I talked about before. Often I'll settle with an obvious move, missing out on very sharp and strong moves, and throwing away stones that could be linked up.

Join the club.

Not sure how to improve in that, except study the Tesuji book hard

A worthy task. Keep reviewing it until you have mastered it.

[jlt]

@John: yes we are referring to the same paragraph.

Knotwilg said above that the point of tsumegos is not to find the solution but to practice reading life-and-death problems. So I was tempted to generalize as follows: the point of tesuji problems is not to learn tesujis but to practice reading middle-game fights. But your position seems to be opposite: the point of tesuji problems is to learn tesujis so as to use them later with ease.

You are of course right, and indeed I sometimes spot tesujis that I wouldn't have seen 1-2 years ago, so of course the time spent on solving problems didn't go to waste.

On the other hand, most moves in contact fighting don't seem to be in the catalogue of tesuji techniques, so

(a) either I don't have the adequate catalogue in my library,
(b) or these moves cannot be learnt in books but only through practice,
(c) or practicing tesuji and life-and-death problems train "brain muscles" which can later be used during games.

(a), (b), (c) needn't be mutually exclusive.

[John Fairbairn]

jlt

Yes, my view is rather different from that of knotwilg, but not the opposite.

All other things being equal, the player who can read more deeply should always win, in my estimation. So, reading must be valued.

Where I differ (insofar as I correctly interpret knotwilg's view) is that I reject the hair-shirt approach that there is value in reading everything out move by move.

If I want to go and see my friend whole lives in a mountain hut, I can don my cilice and spiked garter and walk all the way there, chat for a few minutes, and then hobble all the back.

Or I can drive my car (i.e. use tesuji) and drive to the bottom of the brae and then walk the last few hundred yards. That way I can spend more time with my friend, or go to visit another friend as well.

The hair-shirters may say, "But look at my bulging calf muscles and see how fit I am." I could retort, "But I can walk up the brae without difficulty, and look at my bulging social life."

In real life, it's a personal choice, of course, and I would hesitate to say which choice is better. But in go I think it is possible to make a very strong case for avoiding the hair-shirt approach. This is especially so for amateurs, who have little time and tend to need to play fast. Because of the relative weakness, I think amateurs would in any case benefit more from the baby-steps approach of learning each tesuji as an amuse-gueule rather than as a four-course meal.

But even for pros there are drawbacks to the far-end-of-a-fart approach. The more time you spend going down the branches of a move tree the less time you can spend on a totally different move. You are likely to end up with blind spots and the nerd view as well as nervous exhaustion. This is a form of déformation professionelle even for "un professionel." A version of this defect is particularly applicable to amateurs. How often do we work through a problem, feel the glow of getting it right and then turn the page to get, for good measure, the author's pat on our backs. "Yes," he says though not in so many words, of course, "Your line was correct, but what would you have done if White had played A?" Prick of balloon - we never even considered that move. This is such a major problem in amateur go, I believe, that hair-shirts could usefully be banned.

On the other hand, most moves in contact fighting don't seem to be in the catalogue of tesuji techniques

I'd need to think hard before accepting that. But in any case, if it's true it gives you a lot of information. It tells you to expect to play ordinary moves most of the time and not try to be too fancy. That alone can limit your reading load. There are also heuristics and other "tricks" to ease the burden, e.g. "five alive", miai, precious liberties.

I also differ (I think) in that I don't believe the way to learn tesujis is to get a problem book and hair-shirt your way through that. I prefer analysing the tesujis in terms of first, and most important, what they are meant to do, and then in tracing the various lines that lead to said end. Only then do you do problems to reinforce what you learned. You don't normally learn things by just doing problems in other disciplines, surely? You don't learn Latin by practising the conjugation - what conjugation? - of amare and then reading Cicero. First you learn amo, amas, amat etc and you analyse what the endings -o, -as, at do, and why, and so on. Go's only different, as I said earlier, because we start actually playing a full game too soon.

I feel confident in my assertions because of personal experience. When I was translating Gateway To All Marvels (i.e. Xuanxuan Qijing) I had to analyse each problem intensively. First I had set myself the task of cataloguing the various types of techniques (I'm too lazy to look it up but I think it may have been about 30-40 at a minimum). That required great focus. I had to prepare many variation diagrams. That required huge focus and also guaranteed that I looked at every variation - no blind spots. I had to compare many different professional opinions as to the solutions. In many cases the pros differed from each other, some were plain wrong and some didn't solve the problem presented - they chose to add an extra stone so they could show off their own line. Imagine the concentration that required of me, tip-toeing round the feet of the gods, scrabbling for crumbs of spilt ambrosia. Then I had to sort out the relevance of the title of each problem to its solution. That bit at least was fun. I had to do that for about 400 problems.

Note that in all that work I never tried to solve any problem in the traditional way. At the end, however, I was able to look at not just these problems and just "know" the solution, but I could turn to other similar Chinese problem classics and say almost instantly, "Oh, that's a double snapback, just like XYZ in GTAM, and you've got to watch out for so-and-so." Interestingly, though, if I turned to a largely dissimilar book such as the Guanzi Pu which deals to a large extent in boundary-play problems, I was lost - because I had never analysed this type of problem. In other words, my move-by-move reading ability had not improved at all. I had just acquired a lot of useful chunks, or tools for my tool chest. I knew that if I had the urge I could - as could everyone else - acquire the necessary chunks for GZP and other books.

This method of analysing and making associations is, for me at least, much more fun, probably sticks in the brain longer, provides a gorgeously fertile mulch in which to plant new chunks, and doesn't preclude at all any extra move-by-move reading that you may want to do.

PS The Fujisawa tesuji book is the one to use if you want to start on tesuji analysis.

[jlt]

I had never thought of analyzing tesuji. Interesting and worth a try.

[Ian Butler]

@John and jlt: Thank you for having that interesting discussion here. You bring up interesting points, John, that I hadn't thought about before. Whether one approach is right and the other is wrong, that's probably impossible to say. It will depend on who is practising. What works and motivates one person, won't necessarily for another.

@Bill: Wow. Thanks a lot for that. Inspiring words, because I really hadn't thought about it that way. I'll stick with the Tesuji book for a while and won't be afraid of memorizing the solutions. Try to 'overlearn' the book, like you said. I'm excited for it!



There are two things I'd like to note down for today. One is how I try to organise my books/planning for Go study, the other is about how I'm doing tsumego and comparing that to how I did it before.

Book Program
After Bill's post about the Tesuji book, I had to add a category. So this is how I think it'll go for now. I divide my Go books into three (or four) categories.

- Category A: tsumego.
- Category B: Tesuji by James Davies.
- Category C: All Go books concerning tactics/strategy (like Attack & Defense, Fundamentals of Go, Opening Theory Made Easy...)
- Extra's: Mostly (pro) games with commentary.

Category A: I have to do daily. Once I finish 1001 L&D problems, I'll either do it again or take another tsumego book. At least 15 minutes a day.
Category B: I do this 3-4 times a week, depending on how busy I am. For Tesuji that means 1 chapter when I do it. So 3-4 chapters a week. I'll likely keep this book in category B for a while, and try to 'overlearn' this one.
Category C: I can only do one book at a time. For me it's a big risk to read too many books at the same time, that's why I need this category system. So in this category I can read through Attack and Defense, for example, but I can't start Fundamentals of Go before I finish Attack and Defense. This category has no time limit and mostly depends on my schedule. When I'm busy with work, I shouldn't feel bad or guilty when I don't touch a category C book for a few weeks. When on holiday, I can go through them faster.
Extra's are extra, obviously. When I wanna replay a pro game, I'll use any book I have. I can do many books at the same time, because it's always just one game.

Tsumego fun!
I love doing tsumego. I'm doing it way better than I used to, before my big break in Go. It's so much fun. I don't know exactly what changed but I think these things changed the way I think of it a bit:
- No pressure. I don't need to get better in Go quickly, there's no time rush. So tsumego becomes the purpose on itself instead of a means to achieve a goal. I like doing the tsumego because they pose Go puzzles, it's fun!
- No rush. I increased my time spent per problem a lot. Right now I'm at like about 2-3 minutes average on a problem, compared to like 20 seconds last year. I try to read out as many variations as I can. I like to challenge myself on the problems: how deep can I read this problem? How many variations can I find for this move? Etc. When I look at an answer that was particularly hard, I go back and try to read it out in my head fully, and compare to how I misread it earlier.
- No frustration. I get a lot of them right. But those that I miss, I don't feel bad about them at all. It interests me why I got it wrong and pulls me back to that problem. I get good vibes from getting a lot right, but no negativity from getting a lot wrong. A year ago I would sometimes be frustrated when I asked myself: why doesn't this work, or why this or this? Now I challenge myself to figure it out. Once I just played it out on a board because I couldn't read it; and than the answer (shortage of liberties + snapback) was so obvious. I could read it easily after putting it on the board.

So for some reason, tsumego is really agreeing with me at the moment. I still have problems reading in game, but man am I enjoying reading in my tsumego book.

[Knotwilg]

Knotwilg said above that the point of tsumegos is not to find the solution but to practice reading life-and-death problems. So I was tempted to generalize as follows: the point of tesuji problems is not to learn tesujis but to practice reading middle-game fights. But your (John's) position seems to be opposite: the point of tesuji problems is to learn tesujis so as to use them later with ease.

My position is that life & death problems can have both complementary but interfering objectives:

- they can be used to train reading, and it's the reading that matters, not so much as finding the solution, while the problem setting does provide a goal to read towards
- and they can be used to enhance the repertoire of techniques (tesuji), and it's spotting the key move that matters, not so much checking all possible variations

In the first case, looking at the solution (too soon) is irrelevant (or counterproductive).
In the second case, looking at the solution will confirm your newly acquired technique or make you discover the key move you did not see.

Most tsumego can be used for both purposes, so I think it's really down to the student on what the material will be used for.

[Boidhre]

What I found was the "read it out fully" approach worked well with easier problems where instead of solving it quickly I was trying to find the best refutations in the paths that did not live/kill. Repurposing easier problems as a different kind of reading training. For problems that weren't easy, I found it better to focus on drill and overlearning the solutions to try and make them part of my mental library. YMMV.

[Ian Butler]

Small update: tsumego still going good. Even on the busiest of days, I manage to do at least 15-20 minutes of tsumego. When I have some more time, it's sometimes up to 40 minutes to even an hour sometimes. I manage 3-4 chapters per week in Tesuji, too. My reading is definitely improving, which is fantastic to notice.

I haven't played any game yet, though. Partially because I've never been as busy as I've been in my life, with my job. Some days I only have an hour or an hour and a half free time. I've made tsumego and Tesuji my main priority of free time, but when I do have more time, I chill a bit with a book, a series or watch a Go video rather than playing games. Maybe over the holidays.
On the other hand, playing games is not something that I'm really craving at the moment, so maybe I don't have to start playing immediately. I liked the idea of improving my reading and Tesuji knowledge a bit more before going back into playing. I feel that I could fall into my bad habits of not reading while playing easily if I started playing too soon, especially on short time settings. If I do start again, I want to do it properly.

Funny discussion: I had a funny discussion a few days ago, too. I taught someone the rules of Go and we played a game. She assumed she was winning and was very disappointed when I pointed out I was winning pretty heavily. Afterwards we got into a discussion where she kept saying the game has such easy rules, anyone can win. Someone intelligent could beat me after just learning the rules. I kept trying to explain Go is very complicated and even someone intelligent has to learn and grow, much like you learn a language before you can speak it well. She kept disagreeing and saying she'll beat me next time. It was pretty funny. Interesting to think about, but I do believe she's just sorely mistaken. Sure, a talent in visualisation will help you grow faster in Go, but even someone very, very intelligent will have to work his way through the ranks and won't beat a SDK or a dan player or whatever after learning the rules and playing 2 or 3 games.

[Knotwilg]

On 5x5 yes. On 7x7 it will already take some time but a win in a couple of weeks is not excluded. On 9x9 it takes 1/2 to 1 year to beat a 1d I think and on 19x19 it takes a year with serious study and a good disposition.

[Ian Butler]

Okay so I've been doing so many tsumego that I wanted to practise my reading in a game. I have no time for 19x19 game right now, so I thought: why not a 9x9? I always used to be scared of 9x9 because my reading was not so good (lazy) and tactics, close fighting, were my worst things at Go while whole board thinking was my strength. So I was afraid to lose ranks, instead of playing a lot of 9x9 to learn, but that's in the past :p
It still scared me to play a little

My objective this game was easy: read. Time settings were 30 seconds per move. A few times I got to the sound of the countdown, so that's good.
AI says is big mistake. It was the move I spend the most time on, though. I thought of good responses to both hanes, but AI would stretch. I think I could live (there's a reading exercise in there) but it'd be small and black would be very thick. So this was the worst move by white. Black's next two moves, however, were the losing moves according to AI.
I was proud of and because they were sente, followed up by which is big.
is a move where the coward in me responds again. when black doesn't answer move 18, I think white can kill, but I'm not sure and 30 seconds were definitely not enough to read all the way through the one point jump into black's group to kill, so I calmly defended. Or cowardly, I don't know
I'm on the fence about. I wanted to press black at the top, but my left was too open so I thought securing this corner would give me enough points to win. MAybe another coward's move, or rather just a good solid move, consolidating the win?
I was happy to get in.
thanks to Tesuji from James Davies I've been reading a lot lately perfect sacrifice play I think. On the other hand, if I push, can't I made a dead 5-space? Can't read it out perfectly, though, so the sacrifice play seems good enough for my reading.
why not just the hane? Here I was already playing too fast again, I stopped thinking because I thought I had won already. Dangerous habit.



Conclusion? I managed to read ingame, which was the point. I made some sharp-ish moves and really tried not to be too passive. I think for me it was very good. So I'm happy with this one

[Boidhre]

why not just the hane? Here I was already playing too fast again, I stopped thinking because I thought I had won already. Dangerous habit.

Descending keeps sente, hane loses it. Getting the next sente as well as in-game is better than the hane here. Nice when fast play still arrived at the better of the two options.

[Ian Butler]

And - because why not? - another 9x9 game to finish the day.
My takeaway from this game... I never know what to do in the opening of a 9x9 game
was probably bad. After that both me and my opponent kind of missed the point of the game, which was the weak group of mine on the right.
not sure if was the best way to harass it from my opponent. 2-2 seems tougher to deal with. I hane'd on the inside but I was trying to read out if I could hane on the outside. I couldn't read well under the "stress" of the game so in the end I hane'd on that side. I thought I was dead when white was about to play 24 but I thought he was going to play F2, pull back, instead of atari. But in the review I saw I was alive at that point. Or maybe I'm still missing something.
I'm oddly proud of . I know it seems like even a 16 kyu would easily play that one, but under the stress of a game, I would so easily take the easy way out and atari on the other side and give up too many points because I won't read it out. But I read it out and captured the stone. It seems very silly but I'm proud of that
Anyway, another neat 9x9 game that I won, which of course I'm happy about. Especially since I used to be so, so bad at them.

It was fun.