For a dan player, doing tsumego problems is certainly better than doing nothing, and if you find them fun and they boost your motivation they may in a way even be better than anything.

But there are drawbacks with them. One is that you know there is a solution. This is quite different from a real game.

Then there is the rara avis problem I've already mentioned. I know lots of dan players who can do fancy under-the-stones problems but who don't know all the lines of the carpenter's square, or the best way to connect in a J group, both of which are very, very common in real games. But if you study real games: (a) you will come across even more tesujis and techniques than you find in books, and (b) you will encounter them and the various related shapes in exactly the right proportions and in the right contexts.

Doing book-style tsumego is rather like learning a language and choosing to learn lists of the names of obscure fishes and vegetables instead of how to order fish and chips. Fine if you find it fun, but don't claim it's very efficient for real life.

In my experience, all dan players are well aware of the danger of simply playing memorised josekis and not looking at how they interact with the rest of the board. In fact, that may even be a marker for dan strength. But the same danger applies for tsumego. They are disembodied capsules - that's what tsume implies. I think dan players should be looking at the whole board and using the most commonly occurring tesujis in that context. But maybe doing that reliably is a marker for high dan strength...

1) I cannot argue with that: to convince yourself there might be something in the game situation is a mindset that is probably not encouraged by doing tsumego but by playing games. However, the "smell" that something might be possible can come from doing lots of tsumego.

2) I agree about the fancy tsumego that come from an encyclopaedic approach. I disagree when looking at my current diet of Hitachi Go Problems. They are very diverse and only partially artificial.

3) There's a difference: when memorizing joseki, you risk narrow the opening into a combination of such set patterns. You cannot similarly force situations into preset tsumego that you know how to solve - although I do sometimes kill a group by forcing it into an L-shape (they are not uncommon).

Overall, I think that fancy under the stones tesuji are not a fair representative of what tsumego practice may have in store for you, even if you're a dan player.

There are problem collections in which each problem has a solution and other problem collections in which some of the problems have a solution.

Obviously I'm using under the stones as exemplia gratia.

When I published Gateway to All Marvels I did a very extensive index and naming of all the themes that occurred. At the same time, I did a fairly extensive survey of other tsumego books on my shelves. I wanted to check things like what proportion of themes GTAM covered, find new themes, etc. I didn't publish that data, but what I can say is that in all such books there was a pretty high proportion of themes that I have never or only rarely seen in actual play.

I'm sure that's partly because many tsumego books have been contrived as works of art, especially the famous ones or ones with titles like "Best of Showa". If you look at books with titles like "Tsumego to get to 2-dan" I'd expect a different pattern, of course, but if you look at the sort of problem books people ask about or carry with them at tournaments, from what I see they do tend to go for the famous works. Maybe that's a little misguided.

It may also be that modern books from Korea and China are rather more functional than the Japanese books or Chinese classics I've been used to.

Just a (perhaps stupid) question, when you speak of tsumego and doing tsumego exercises is it just in the sense of life and death problems or does it also encapsulate tesuji (of the sort found in Get Strong at Tesuji / Tesuji / Segoe)? And if not, how would you say they fit into the whole improvement regime?

I know they are very much related - but I'm thinking of tesuji problems that come in a different format than just local life and death (like cut / connect / endgame / sente / shape).

I'm adding to the discussion about tsumego (non)sense at my own journal:

viewtopic.php?f=48&t=12096&p=210558#p210558

The nice thing about tsumego problems is that they force active learning. You must actively exercise variations in your mind, and you get feedback on whether your analysis was correct. You can do this to some degree with pro games, but you really have to be conscious about it.

It's much easier to click through moves in a pro game without understanding their depth.

I might compare it to watching a lecture on calculus vs. doing some problems. You might get a comprehensive coverage of material by watching a lecture, but you may very well think you understand when you don't since you haven't done any problems.

Anyway, arguing over which is best is somewhat silly. I know we have limited time, but why not do both? I'm planning on studying pro games more partially as a result of this discussion, but not because I think it's more fruitful. I think having a mix of study, in itself, may prove efficient.

_________________
be immersed

I'm referring to life & death only. Tesuji problem collections are very different in that they try to focus on techniques that are very applicable in games and are definitely not trying to wow you - to do so would be almost a contradiction of the meaning of tesuji.

We don't have a huge choice when it comes to tesuji collections. I gave a very gushing recommendation of the Fujisawa book a long time ago on rec.games.go as a result of which Slate & Shell commissioned a translation. Much to our joint surprise it didn't sell very well. To some extent that must have been because it was the time of go publishing world's equivalent of the end of the dinosaurs. But the book still exists and is very substantial. It has everything you need. In my view it's much better than the Segoe one, not just because it's in English but because it explains the range of applications for each tesuji. The very fact that that is possible for tesujis but is largely irrelevant in tsumego problems illustrates the difference in value as regards efficient learning.



From what I've just said you'll note that I don't think they really are much related.

View this through the chess lens:

A book of tesuji problems is the equivalent of the books of combination or mating problems in chess. A book of crafted tsumego problems is the equivalent of a book of problems by the likes of Sam Lloyd in chess where you can have 12 queens on the board and similar absurdities that have no relation to real chess. Very few chess grandmasters give such weird problems more than a passing glance. They are in too much of a hurry to get to shodan and want to acquire tools they can use straightaway.

Amateurs of The Rookie type, however, think this is a bit grubby and prefer to search for the "truth." This is probably often a way to avoid facing up to hard work, but in some case it's probably out of stubbornness. I'm reminded of the stereotypical lost tourist couple. The man turns the map upside down and inside out a dozen times and curses every time the woman says "Why don't we ask someone?"

If we had a lady in this forum I'd expect her to be practical and ask why we don't study the stuff that is immediately useful. Instead we get men who justify learning a hundred obscure knots on the off-chance they might need a highwayman's hitch the next time they need to rescue a penguin that's fallen down a crevasse in the Antarctic.

I say this as someone who did learn a hundred obscure knots and have never got to use any of them. Hang on, is that a penguin I see there...?

Aha, well then I can closer relate to what you say about tsumego, I guess once you know a wide enough repertoire of unsettled / living / dead shapes that there is a diminishing returns on doing the problems - except as a way to drill and keep them memorised. For me however, and I guess for most kyus, learning these shapes (notchers, bulky six, L groups and on and on) is kind of essential. Drilling basics so to speak instead of studying contrived yet exquisitely crafted problems.

I guess I'll have to do some tesuji books next then heh. Also, the books by Fujisawa, would these be the "Dictionary of Basic Tesuji" (4 volumes?)

The taxonomy of chess "puzzlers" (to try to pick a general word not already used by a subcategory) is

Puzzles (this name is not universal but it's the most common): positions that could easily come up in a real game and most often have actually been taken from a real game. The solver is expected to find a continuation that will effectively win the game. There is generally a single best first move but sometimes there are multiple good moves in the variations.
Problems (an unfortunate name since it seems more general than it is): aesthetic but very artificially constructed positions of a form that would never come up in an actual game, usually with similarly artificial goals such as "mate in two" (checkmate Black within 2 moves).
Studies: composed endgame problems that are designed to have a tightrope-like solution with brilliant moves on both sides. These are generally rather artificial by necessity but are usually designed to be as plausible as possible.

Note that by the nature of chess all of these are whole-board problems, which is another difference from go.

I think that tesuji problems are most comparable to chess puzzles, although they cover a small portion of the board and usually result in the player getting a good local result rather than practically winning the whole game outright.

John Fairbairn compares tsumego to chess problems but I think they are more akin to chess studies. Problems are mostly an art form largely irrelevant to the actual game, while studies are often prescribed by strong players as a very effective way of training calculation skill; as with tsumego, it's less about learning common patterns and more about learning to calculate effectively. In fact studies are one of the few kinds of chess tasks that extremely strong grandmasters continue to train with.

Disclaimer: I am most familiar with kyu-level tsumego and I know that the problems become more artificial as their level increases.

In chess I improved from 2000 level to 2200, aka a chess master, by just solving difficult problems. As a 3 kyu I once tried to replay pro go games on go4go but couldn't understand any of their moves. Felt like a completely waste of time. The owner of that website, though, improved a bit by putting all those moves in. I also just recently found out Magnus Carlsen saying in an interview that going over games without engines or any analysis helps him a lot. He said he is trying to find new ideas. Also Vladimir Kramnik famously said that he looks over 10000 games every month to find out new moves.

Yes, four volumes, by Slate & Shell available here: http://www.slateandshell.com/SSFS001.html

Agree. Also, even if some problems are artificial, that's totally fine. Weight lifting dumbbells is artificial, but it still makes you stronger if you do it the right way.

_________________
be immersed

I feel like saying that under the stones doesn't come up much in actual games is kind of missing the point as I always thought of those problems as being about clarity in your reading. It's very easy to see an under the stones situation when you play it out but for whatever reason they're very difficult to see in your head. Getting good at spotting them requires that your reading takes a step forward and I think that pays off even if you don't see the actual pattern all that often.

Also, under the stones situations (and other dan problem situations) do occur even if they're not all that common so I think it's hasty to dismiss them as not being worth studying. It's like learning a language; you get diminishing returns learning less and less common words as you progress to a more advanced level, but in order to be fluent you need to learn these "rare" words. Small steps forward are still steps forward.

There's a point being missed but that's not it.

Something I've found with both go and chess players (and it's not restricted to them) is that some people who have spent a long time doing something wrong can get quite upset, angry even, when someone else suggests there may be a better way. This was encapsulated once when a friend and I who were helping develop the first shogi computer asked a chess grandmaster who accepted that shogi was a better game than chess (no draws, very much reduced opening memorisation, more money) why he didn't switch codes. He wasn't upset, just melancholy, when he explained that he couldn't face admitting that he might have wasted all those thousands of hours on chess.

So maybe that's why some people here are closing their ears.

I have not said that doing traditional tsumego problems will not give you something useful. One thing I've found in life is that nothing you learn is ever totally wasted. As a silly little example, when young at school I learnt the French name for what seemed like every flower under the sun. I thus learnt and have never forgotten that glycine is French for wisteria. The problem was I had no idea what a wisteria looks like. In my world we had bluebells and gowans, buttercups and roses.

I only ever found out what a wisteria is about 30 years later when I had children and we started playing the Japanese flower-card game hanafuda. I had an aha moment. The suit fuji is the wisteria suit, and it was obviously useful to know that to play the game.

Having a slight thrill from discovering at long last what a wisteria was I did a little research, and found that it was name after someone called Carl Wistar and so should be called wisteria, but the spelling wisteria seems close to universal in Britain, and is so pronounced (wis-TEE-ria). I had now acquired a fun fact for dinner parties and go forums. I had no direct use for knowing la glycine but still, I had "improved."

But, in retrospect, because I ended up in my first job as a technical translator doing chemical texts, it would have been much more useful to me to know that glycine is also a chemical compound. And there's the point. I believe that in studying games, where life & death and tesuji occur in abundance, you are more likely to be efficient in your learning of these skills. It is there you learn not just how to read with clarity, but why (context) and in the right proportions for each technique.

There may more than one way to skin a cat, but using a nail file doesn't seem like the best one.

And telling me it's more effective to read a book than click through a game because you speed up when clicking is just telling me you've got a discipline or concentration problem - which is part of the reason some people never reach shodan, apparently.



Kids who want to be baseball stars don't normally use dumbbells. They don't just want to be stronger. They want to be stronger and play better baseball. So they practise pitching and batting and fielding. That makes them stronger and trains the right motor skills and gives them whole-game sense. It's also more fun.

Again, I'm not saying don't do dumb-bells or read tsumego books - a change of pace can be good, and nothing is ever truly wasted. Just don't pretend it's the most efficient way to become stronger for the skill you have chosen - playing a complete game of baseball, not acquiring six-pack abs for prom night; playing a complete game of go, not doing puzzles.

Not the most efficient but maybe very necessary. Like marathon runners very rarely run a marathon.

But they do very long runs. They don't work on their abs.

I think we are going round in circles now, so I think I'll go to the gym instead and let Kirby have his usual potshots at me in peace.

That's a rather unfair argument. We could likewise argue that you think reading a go book is a very efficient way to improve because you've written some, or replaying a pro game is because you compiled GoGoD. We need arguments why L&D is or isn't efficient, not arguments why people cling to tsumego despite proven poor return on investment (or conversely, don't think highly of it because they never invested in it).

I love the above analogy with marathon precisely because it is flawed and thus points out the problem with other analogies, like baseball.

When trying to make a point about tsumego as a tool to improve go, I wouldn't search in other domains. Analogies are like a baby in a cradle: it's hard to fight the imagery but how exactly did that prove my point?

I think Kirby deserves much better than being depicted as a troll for merely disagreeing with you (and only at some points) in the most civilized manner.

Thanks! I never new what a wisteria was, though I have often heard the word, but since my German landlord is always complaining about how the neighbor's glyzine is ruining the facade of the house, I now know what a wisteria looks like.

As to tsumego, I think it gives people the opportunity to practice reading, killing and living, and like you say, that can't hurt.

_________________
Patience, grasshopper.