Reading shortage of liberties

Numsgil · #1

This is problem 127 from "Making good shape":

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O . O X . . | $$ . . . . X O . O X . . | $$ . . . . X . . O X . . | $$ . . . X . . . . X . . | $$ . . . . . . O . . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

The solution involves a shortage of liberties. Which I can read up to, but I find it tricky to read several moves in and then count the liberties of groups after taking into account theoretical stones. Especially captured theoretical stones. For a problem like this I usually read up to a point that I suspect I shortage of liberties will come into play and then just assume it works, which obviously isn't a great strategy

Essentially, I can keep either count liberties or keep stone positions in my mind, but not both at once, which means I have to try and swap stone positions in and out of short term memory, which is error prone and I find myself forced to either re-read up to the target sequence several times or just be lazy and assume it works.

I find myself making mistakes in games involving shortage of liberties that I didn't spot when I was reading out a sequence, but a few moves in it becomes readily apparent to me. I'm wondering if there's any mnemonic devices or the like that stronger players use to help with this. That is, is it just a matter of continued practice or do I need to think of things fundamentally differently?

I'm thinking of Moonwalking with Einstein, (which I haven't read

), where in a year the author won the US memory championship by training with an expert and learning lots of memory techniques. There was a lot of hard work, too, but also specific techniques.

This quote is also relevant, I think:

Quote:

Q: Can you explain the "OK Plateau?"

A: The OK Plateau is that place we all get to where we just stop getting better at something. Take typing, for example. You might type and type and type all day long, but once you reach a certain level, you just never get appreciably faster at it. That's because it's become automatic. You've moved it to the back of your mind's filing cabinet. If you want to become a faster typer, it's possible, of course. But you've got to bring the task back under your conscious control. You've got to push yourself past where you're comfortable. You have to watch yourself fail and learn from your mistakes. That's the way to get better at anything. And it's how I improved my memory.

lightvector · #2

I don't know about others, but for me, I "just see it". That is, when I first looked at the position, the knight's move looked fishy and then after I mentally placed the first move, I sort of saw all the subsequent stones fly down automatically, and the final position appeared. More or less, I can see the liberty shortage the same way you would instinctively see it if it were played out fully, because I actually see the final position on the board from here, albeit a bit more instinct and fuzzy than if it were actually played.

Your quote sort of applies to me as well, but maybe in a different way. For me, the problem happens when the stones that fly down when I see a position are *wrong*. That's always annoying, especially because my conscious reading is much much slower, so I rely on this "automatic reading" a lot. Alternatively, I will see a problem and not have any first instinct - no shape really stands out to me. Then it easily takes me an order of magnitude or more longer to solve the problem, even if it's not a hard problem.

topazg · #3

lightvector wrote:

I don't know about others, but for me, I "just see it". That is, when I first looked at the position, the knight's move looked fishy and then after I mentally placed the first move, I sort of saw all the subsequent stones fly down automatically, and the final position appeared. More or less, I can see the liberty shortage the same way you would instinctively see it if it were played out fully, because I actually see the final position on the board from here, albeit a bit more instinct and fuzzy than if it were actually played.

Ditto, but I'm sure that's because I've spent so long looking at cutting tesujis. It's just the obvious place to cut somehow, and even though the shortage of liberties isn't intuitively obvious (that it works, that is), the sequence is almost completely linear.

topazg · #4

For anyone interested, this is my thought process by move:

1. Cutting Tesuji
2. Attempt to connect (only option)
3. Block (only option)
4. Attempt to capture cutting stone (only option)
5. Extension dies, so make eye false in the hope of shortage of liberties (one of two options)
6. Capture cutting stone (only option)
7. Atari (only option consistent with 5)
8. Connection fails, so choice at 5 works. If it failed, a new "1" would be the only choice, but unless :b1:

@

works, White has connection miai, so the only alternative to read is :b1:

@

... in this case :w2:

@

is the only obvious option that doesn't revert to the main line, and as it happens, seems to save White's bacon.

So, this problem was easy for me as I had to consider a total of 12 moves (including all the moves within a line), therefore not taking very long. I'm a big believer that move pruning is what allows stronger players to find problem answers quickly, and although it makes solutions look trivial, learning the shapes to automatically discard can take a very long time.

I remember some Koreans that came as a goodwill gesture to a tournament I attended in 2007 (3p, 1p, and 7d). I was about 6k at the time, and I had worked with a 5k and a 1k for about 25 minutes on a supposedly dan level tsumego. We thought we'd worked it out in about 10 minutes until someone found a refutation, and spent a whole load of extra time working it out and then double and triple checking it. Using resources available, we thought we'd ask the 1p to look at the problem to verify our answer - he solved it himself just short of 5 seconds. Even though the problem must have been completely obvious to him, the effort it takes to make that problem obvious is very large.

daal · #5

topazg wrote:

lightvector wrote:

I don't know about others, but for me, I "just see it". That is, when I first looked at the position, the knight's move looked fishy and then after I mentally placed the first move, I sort of saw all the subsequent stones fly down automatically, and the final position appeared. More or less, I can see the liberty shortage the same way you would instinctively see it if it were played out fully, because I actually see the final position on the board from here, albeit a bit more instinct and fuzzy than if it were actually played.

Ditto, but I'm sure that's because I've spent so long looking at cutting tesujis. It's just the obvious place to cut somehow, and even though the shortage of liberties isn't intuitively obvious (that is works), the sequence is completely linear.

I've also done countless tesuji problems (though perhaps not quite so many

), but despite often intuitively knowing the first correct move, the rest of the stones never "fly down automatically," and similar to the OP, when they do land, they often just don't stay put. Perhaps it's because so little about the following sequences seem linear to me.

Quote:

1. Cutting Tesuji
2. Attempt to connect (only option)

According to the book, there are two possible ways of attempting to connect - o13 and q13, which to me is enough of a mental roadblock to make the rest of the stones jump about like kids in a back seat.

I once commented on one of your videos, saying that your analysis of certain board situations was way too fast for me to follow. You would say something like "black goes here and white answers and then black white black white black and I'm fine," with a confidence that I never have. Perhaps it's an innate quality to be able to visualize the stones flying onto their postulated positions, but if not I'd also like to get better at it. Perhaps I need to improve my way of looking at and remembering tesuji problems. Or just do a few thousand more.

topazg · #6

daal wrote:

According to the book, there are two possible ways of attempting to connect - o13 and q13, which to me is enough of a mental roadblock to make the rest of the stones jump about like kids in a back seat.

Interesting. I see them as miai, and when you play one, you have to play the other anyway to connect, as Black has to follow through to cut. I suppose I consider the move order irrelevant for those two, and intuitively pick the one where White sticks out further. After the next 3 moves it feels clear to me that starting with the other one would lead to an identical position, and at that point I don't bother returning to read the other one. You are quite right though, they are both options, and it was only in hindsight that I wrote off analysing the other one.

Cassandra · #7

Numsgil wrote:

This is problem 127 from "Making good shape":

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O . O X . . | $$ . . . . X O . O X . . | $$ . . . . X . . O X . . | $$ . . . X . . . . X . . | $$ . . . . . . O . . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

I don't think that "counting liberties" is the main topic here.

To save Black's encircled N16-stones you have to capture White's O16-stones.
To capture these stones, you have to prevent their connection to White's P12 stone.

The only point to start with trying to seperate the white stones above from their allies below is P13 for Black.

The problem with this problem might be that you can fall into the trap that it might be necessary to capture White's Q16-stones, too.
This is because the N16-stones seem to be connected to the Q16-stones. But really they are not, there is only this bamboo-joint.

Numsgil · #8

Thanks for the posts everyone! Sounds like the consensus is that you just practice and you get better at it.

Cassandra wrote:

Numsgil wrote:

This is problem 127 from "Making good shape":

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O . O X . . | $$ . . . . X O . O X . . | $$ . . . . X . . O X . . | $$ . . . X . . . . X . . | $$ . . . . . . O . . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

I don't think that "counting liberties" is the main topic here.

To save Black's encircled N16-stones you have to capture White's O16-stones.
To capture these stones, you have to prevent their connection to White's P12 stone.

The only point to start with trying to seperate the white stones above from their allies below is P13 for Black.

The problem with this problem might be that you can fall into the trap that it might be necessary to capture White's Q16-stones, too.
This is because the N16-stones seem to be connected to the Q16-stones. But really they are not, there is only this bamboo-joint.

That's the point of the problem, yes, but the point of my post here is that deciding if the shortage of liberties at the end of the problem "works" requires a mental switch from the spatial reasoning of placing imaginary stones to the mathematical reasoning of counting liberties. Reading up to the shortage of liberties should be straightforward and manageable for most SDKs. But then you have to make a judgement call and see if the sequence helps you at all. Which in this case requires counting liberties to detect the connect-and-die.

I can't stuff both numbers and stones into my brain's "registers" (borrowing computer jargon), so I get into this uncomfortable situation where I have to try and juggle numbers and stones in my mind at the same time, and I invariably get confused and lose stones or numbers. Which means recounting or re-reading in alternating cycles and verifying the problem piece-by-piece. In this problem's case, I can strongly suspect that the shortage of liberties works, since it's a go problem about a bamboo joint, but I can't verify it with the same certainty that I could verify a straight capture.

...

As far as mnemonic devices, this blog post comes to mind.

Specifically, he quotes A Tree Grows in Brooklyn (which I also haven't actually read

)

Quote:

“She liked numbers and sums. She devised a game in which each number was a family member and the “answer” made a family grouping with a story to it. Naught was a babe in arms. He gave no trouble. Whenever he appeared you just “carried” him. The figure 1 was a pretty baby girl just learning to walk, and easy to handle; 2 was a baby boy who could walk and talk a little. He went into family life (into sums, etc.) with very little trouble. And 3 was an older boy in kindergarten, who had to be watched a little. Then there was 4, a girl of Francie’s age. She was almost as easy to “mind” as 2. The mother was 5, gentle and kind. In large sums, she came along and made everything easy the way a mother should. The father, 6, was harder than the others but very just. But 7 was mean. He was a crotchety old grandfather and not at all accountable for how he came out. The grandmother, 8, was hard too, but easier to understand than 7. Hardest of all was 9. He was company and what a hard time fitting him into family life!

When Francie added a sum, she would fix a little story to go with the result. If the answer was 924, it meant that the little boy and girl were being minded by company while the rest of the family went out. When a number such as 1024 appeared, it meant that all the little children were playing together in the yard. The number 62 meant that papa was taking the little boy for a walk; 50 meant that mama had the baby out in the buggy for an airing and 78 meant grandfather and grandmother sitting home by the fire of a winter’s evening. Each single combination of numbers was a new set-up for the family and no two stories were ever the same.

Francie took the game with her up into algebra. X was the boy’s sweetheart who came into the family life and complicated it. Y was the boy friend who caused trouble. So arithmetic was a warm and human thing to Francie and occupied many lonely hours for her time.”

Instead of numbers, maybe you could do something similar with liberties. And as you work through a problem, you get this little narative going in your mind that makes it much easier to keep track of everything. I've never really tried this sort of thing before, so I was curious if anyone has tried for reading out problems.

John Fairbairn · #9

Quote:

Instead of numbers, maybe you could do something similar with liberties. And as you work through a problem, you get this little narative going in your mind that makes it much easier to keep track of everything. I've never really tried this sort of thing before, so I was curious if anyone has tried for reading out problems.

I don't think it applies to liberties specifically, but there are such approaches for life and death problems in general. One of the best known is hane-then-nakade. It's maybe not what you mean by a narrative, but it's an effective shortcut. You don't have to work out all the moves after the nakade - you just know the group is dead.

As one of the indexes for the edition of the Gateway to All Marvels (Xuanxuan Qijing) I am currently proofreading, I used a list of such motifs. Some will be familiar (e.g. the L group). Some will be partially familiar (e.g. under the stones, well known but do you know there is Type I and Type O?). Some I have made names up for (e.g. the cave tesuji or the centipede tesuji). But the point about all of them is that they represent not single moves but longish sequences which boil down in effect to a single move ("chunks" is the word used by neuroscientists, I believe), or they define useful shapes (e.g. the bamboo joint, as here, is well known as being tricky for liberties (another common one in GATM is the elbow shape), so as soon as you see one, your search acquires immediate and useful focus. Either way, this chunking is a way of drastically pruning the search tree in your head.

Furthermore, you will see eventually that several chunks tend to go together like fish and chips. You can therefore look at a problem in GTAM where the solution is 16 moves long, but if you know it breaks down into two "chunks" of 4 moves and 12 moves, you effectively see it as a 2-move problem (or even 1 move, if it's a fish-and-chips relationship) and can solve it almost instantly. This latter approach is perhaps closer to the narrative idea.

I hasten to add, BTW, that GATM is not as simplistic as it may sound from the above. For example, while it does include the bog-standard type of hane-and-nakade problem (where the hane is always on the edge), it also includes examples of the rather rare but delightful case where the hane is somewhere on the inside, which you tend to find hard because you normally associate such hanes with the edge. Caught looking with a high and inside pitch when expecting low and away.

I would also like to stress that the chunks I have identified for the index are there to help readers find related problems, because there are rather many in the book (almost 500). I have no idea whether they are all genuinely useful for learning life and death. Maybe the book will help us find out.

aurik · **#10**

Totally with you on this one. When presented in problem form, I can quickly narrow it down to cutting the keima, which gives me 2 possible starting points, and one of them fails quite quickly. But the other line, I didn't see the shortage of libs until I mentally played out the entire sequence. And I can tell you for 100% certain, had I encountered this in a real game where nobody was standing over my shoulder saying "you can rescue those" I would have missed the shortage of libs and given them up for dead.

This is probably why one of the ways to improve is to do lots of tsumego. Even if you never see the exact same problems in the game, it helps to train our brain to make these reads effortlessly.

lightvector · **#11**

Numsgil wrote:

That's the point of the problem, yes, but the point of my post here is that deciding if the shortage of liberties at the end of the problem "works" requires a mental switch from the spatial reasoning of placing imaginary stones to the mathematical reasoning of counting liberties. Reading up to the shortage of liberties should be straightforward and manageable for most SDKs. But then you have to make a judgement call and see if the sequence helps you at all. Which in this case requires counting liberties to detect the connect-and-die.

Do you still need conscious liberty counting to recognize that the following position is a connect-and-die, when you actually see it on board like this? Or is it mostly automatic recognition?

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O X O X . . | $$ . . . . X O . O X . . | $$ . . . . X X O O X . . | $$ . . . X . O . O X . . | $$ . . . . . . O X . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

If it requires conscious liberty counting, then how about if we change it to the following position? Is this one automatic?

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X X X X . . | $$ . O . . X O O O X . . | $$ . . . . X O . O X . . | $$ . . . . X X O X X . . | $$ . . . X . O . . . . . | $$ . . . . . . O . X . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

aurik · **#12**

I would think those are both automatic for any SDK. But, at least at my level, at some point the shortage of liberties become obscured even by a "one-way street" sequence of moves.

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . O X . . . . . | $$ . . . O . X . X X . . | $$ . O . O . O . O X . . | $$ . . . . X O . O X . . | $$ . . . . X X . O X . . | $$ . . . X . O . O X . . | $$ . . . . . . O X . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

At what level of reading do you think the answer to this derivative problem becomes reflex / instant?

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . O X . . . . . | $$ . . . O . X X X X . . | $$ . O . . . O O O X . . | $$ . . . . X . . . X . . | $$ . . . . X . O . X . . | $$ . . . X . O . . . . . | $$ . . . . . . O . X . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

What about this one? (I'd think this one would be "easier" because of the overly common use of the crane's nest as a teaching example of tesuji)

lightvector · **#13**

The first variant you posted is not actually reflex for me, in that it doesn't yet trigger my "tesuji alarm". But the move at N16 jumps out, given that it's a cut/connect point, and thereafter, the next move I see for black is P16, which becomes snapback into a connect-and-die the other way. Instead of P16, throwing in at P13 works too but is not intuitive to me. But either way, seeing the liberty situation that results at the end of the sequence is easy.

The second one is pretty automatic.

Anyways, I guess the point I'm trying to make is that ideally you shouldn't need to have to simultaneously read and count liberties. The hope is to get to the point where after you read to the end, you are visualizing the final position with all the stones there, and then you can see the shortage of liberties in that visualized position the same way you would see it in an actual position, with no actual counting needed.

daniel_the_smith · **#14**

The problem took me a bit because I wanted something like the below to work, I thought it was the sort of problem where you throw in a couple times and it becomes a snapback.

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O . O X . . | $$ . . . . X O . O X . . | $$ . . . . X . 5 O X . . | $$ . . . X . 4 1 2 X . . | $$ . . . . . . O 3 . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

Of course that doesn't work here and I did eventually find it.

As far as counting liberties, if I read a position where I think liberties matter and I can't visualize the liberties, I'll mentally replay up to the position I was out and count the liberties as I go. If it's an especially tricky under the stones thing I may have to do this a couple times.

Numsgil · **#15**

lightvector wrote:

Do you still need conscious liberty counting to recognize that the following position is a connect-and-die, when you actually see it on board like this? Or is it mostly automatic recognition?

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O X O X . . | $$ . . . . X O . O X . . | $$ . . . . X X O O X . . | $$ . . . X . O . O X . . | $$ . . . . . . O X . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

Yeah, that's no problem. But if you find a position on a board and a position in your mind to be equally clear, more power to you

Specifically, I always have to remember if there aren't any extra liberties for white somewhere where I wasn't reading out stones. eg:, maybe the actual position is this:

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O X O X . . | $$ . . . . X O . O . . . | $$ . . . . X X O O X . . | $$ . . . X . O . O X . . | $$ . . . . . . O X . . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

[/quote]

Or actually, for this specific problem and my specific brain, I get tripped up this way:

Click Here To Show Diagram Code: [go]$$Bc $$ -----------------------+ $$ . . . . . . . . . . . | $$ . . . . . X . . . . . | $$ . . . O O X . X X . . | $$ . O . . X O X O X . . | $$ . . . . X O . O X . . | $$ . . . . X X O O X . . | $$ . . . X . O . O X . . | $$ . . . . . . O O X . . | $$ . O . O . O . X . . . | $$ . , . . . . . , . . . |[/go]

[/quote]

Stones start drifting. The more stones I try to keep track of the less certain I am of any one single stone (hmm, sort of an uncertainty principle at work here). At a certain reading depth and a certain "local area", it requires all my effort just to keep the relevant stones in place. At best I can usually keep track of the liberties of one or two small groups during reading (I don't think of it in numbers, but in terms of "captured", "atari", "imminent atari" and "safe"), but in this problem black has a throw in that is captured, so right there I have to keep track of that. Then there's the two stone white group, which I can read up to the atari for it. And then I simply haven't (can't, at present) store the liberty information for the larger white group. So I have to go back and re-read for that, and maybe I can do it properly and maybe I can't.

It gets better with practice, but it still feels like dragging a heavy load along the ground where a wheel would make things so much easier.

yoyoma · **#16**

To me these aren't too bad because at the end there are only ~2 liberties left that matter. I have a harder time with large capture races, where there are still tactics left that obscure the actual number of liberties. So you have to read out a few short tactical sequences in different areas of the dragons, and then count liberties of a position that isn't actually on the board.

Bill Spight · **#17**

John Fairbairn wrote:

I would also like to stress that the chunks I have identified for the index are there to help readers find related problems, because there are rather many in the book (almost 500). I have no idea whether they are all genuinely useful for learning life and death. Maybe the book will help us find out.

Chunking has long been known as a way that memory works. It has also been used (with somewhat different meaning) in problem solving. (See SOAR.) I feel sure that the chunks you have identified, John, will help people improve their tesuji and life & death.

Cassandra · **#18**

Clustering the problems may be very useful to the reader.

During the years of editing German Go Journal's Problem Corner, I got the feeling that many solvers are not aware of vital points and vital shape elements, which could give a clue to the solution.

See attachment for a resulting idea of mine.

RobertJasiek · **#19**

Cassandra wrote:

vital points and vital shape elements

In tsumego, what do you consider "vital points and vital shape elements", which types do exist and how in general can they be determined, IYO?

John Fairbairn · **#20**

Quote:

See attachment for a resulting idea of mine.

Some very nice presentational ideas in that.

Reading shortage of liberties

Who is online