Life In 19x19
http://prod.lifein19x19.com/

Memorizing Tsumego via SRS
http://prod.lifein19x19.com/viewtopic.php?f=12&t=6623		 Page 1 of 4
Author:  Lamp [ Fri Aug 31, 2012 10:02 pm ]
Post subject:  Memorizing Tsumego via SRS
The other day I got to thinking. I use SRS pretty seriously for studying foreign languages. And although that's their primary use, it is highly effective for memorizing pretty much anything under the sun. People use it to memorize random facts for tests (med students come to mind), and math equations, among other things.

So my question is how effective do you think it would be to memorize tsumego? Let's say for the sake of argument (extreme example, I know), you memorized all of the tsumego in all 4 volumes of GGPfB, as well as Cho Chikun's Encyclopedia of Life and Death. Would this help, hurt, or have no effect on your skill? There are two ways to go about it:

1) Only add problems you've solved correctly through your own reading skill to your SRS cards.
2) Memorize even problems that are too difficult to you (although at least make sure that after seeing the solution and the refutations of failed moves that you can replay them in your head).

Due to the nature of SRS, eventually you would get to the point where you would be able to answer every tsumego in an instant - not because you read it out every single time, but because you recognize the problem and already know the answer. But I wonder if this would train you to instantly see the vital point in other tsumego with similar shapes when they occur in your own games. This would allow you to only have to check 1 move in a lot of cases, which would be the right move, and not waste a lot of time calculating out failure variations.

Does this sound like a good idea? Bad idea? Idea that could be good but with a few improvements?

What do you guys think?
Author:  Bill Spight [ Fri Aug 31, 2012 10:32 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
Quote:
Due to the nature of SRS, eventually you would get to the point where you would be able to answer every tsumego in an instant - not because you read it out every single time, but because you recognize the problem and already know the answer. But I wonder if this would train you to instantly see the vital point in other tsumego with similar shapes when they occur in your own games. This would allow you to only have to check 1 move in a lot of cases, which would be the right move, and not waste a lot of time calculating out failure variations.


In terms of recognition, what you want is not only the tsumego, but non-tsumego that are very similar, so that you can tell which is which. The thing is, of course, you have to construct the non-tsumego yourself. :)
Author:  speedchase [ Fri Aug 31, 2012 11:42 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
It's a bad idea.
Not because it would hurt you to do it, just because it wouldn't be useful. The point of Tsumego isn't to memorize shapes, but to learn to read.

This might be a good idea with some other type of studying, but I don't really know which. Maybe pro games somehow?
Author:  Lamp [ Fri Aug 31, 2012 11:50 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
speedchase wrote:
It's a bad idea.
Not because it would hurt you to do it, just because it wouldn't be useful. The point of Tsumego isn't to memorize shapes, but to learn to read.

This might be a good idea with some other type of studying, but I don't really know which. Maybe pro games somehow?


I agree it's a little questionable, but the reason I had the idea in the first place was because most strong players can solve in basically an instant tsumego that could take SDKs minutes, if not more, to solve (assuming they can solve it all). So at some level they're eye just instantly fixated on the correct move and they just "knew" it was correct, and needed only to verify it by reading that one sequence. Of course this isn't the case for all life and death problems, but some.

So sure, solving tsumego can help you read deeper and more accurately, but I thought that maybe something like this would be a useful way of training you to be able to read less.

When a pro is analyzing a life and death situation in a real game, do you think he arrives at the correct move strictly by calculation and nothing else, or do you think that similar positions from his past may (if even subconsciously) pop into his mind and guide him toward the correct move? Somehow he has an intuition about which moves have potential for correctness.

FWIW, I have also been thinking of ways to apply SRS to joseki, but it's difficult given the tree-like nature of joseki variations.
Author:  speedchase [ Sat Sep 01, 2012 12:18 am ]
Post subject:  Re: Memorizing Tsumego via SRS
The reason strong players solve tsumego quickly is because they can read fast, not because they can guess the move. Of course intuition is important, but you will not get the type of intuition you need to read quickly from this kind of memorization. It will only help you for problems that are similar to ones you study, and it won't help you at all outside of life and death situations. Besides, even if you can solve a problem and know that your solution works, that doesn't mean that there isn't a better way to live.
Author:  Boidhre [ Sat Sep 01, 2012 2:39 am ]
Post subject:  Re: Memorizing Tsumego via SRS
I started a thread on it a while back, the general consensus seemed to be that it was a bad idea: viewtopic.php?f=18&t=5713

*shrugs*
Author:  LocoRon [ Sat Sep 01, 2012 2:56 am ]
Post subject:  Re: Memorizing Tsumego via SRS
speedchase wrote:
The reason strong players solve tsumego quickly is because they can read fast, not because they can guess the move.


No.

At ~1 kyu, I [sadly] read about the same speed as I did when I was ~20 kyu. I can now solve tsumego I couldn't then, in seconds. The difference is experience, which has honed my intuition.

(Edit: Maybe even stronger players do read faster; barring concrete evidence, I'll just assume they have superior intuition. Not taking into account individual abilities of course; my reading admittedly sucks.)

I recall a post somewhere on this board either about an experiment tracking strong players eyes (I think pros, but maybe high dan amateurs) while solving tsumego, or maybe it was just an anecdote, I don't remember it all. Anyway, basically, their eyes go straight to the vital point, and that's where they start reading.

As for the op; I think of this about the same as I think of memorizing joseki. It can be handy. But it can also be dangerous. Either way, you should always read out the situation to the best of your ability; don't rely entirely on memorization, because chances are pretty good that something in the current game negates what you remember (in the case of tsumego, liberties are of utmost importance, both of the group in question, as well as the surrounding stones; if you just follow the situation you remember, you might either miss a critical liberty, or miss an opportunity to attack the surrounding stones--or to defend them, as the case may be).
Author:  Oroth [ Sat Sep 01, 2012 3:27 am ]
Post subject:  Re: Memorizing Tsumego via SRS
I have actually done this with Anki (a spaced repetition program) and the intermediate problems from Cho's L&D.

Observations:

  • It helped my L&D reading, partly because I was doing a lot of problems every day and worked my way through the entire set.
  • There is a constant danger you will memorise the problem not the shape. That is, you're remembering the answer to problem 447, not this precise shape. Rotating the problem/swapping colours helps a little. Having many problems that are very similiar except for a stone changed here and there is better as then you are forced to recognise the precise pattern of stones and infer the answer. Cho's encylcopedia is good in that there are a lot of variations on the basic patterns (tripod shape, L+1 etc).
  • To further ensure it was the shape I was learning I didn't just try to spot the answer, but to try and recognise why the answer necessarily followed from that particular shape - spotting the reason, not just guessing the point.
  • Finally, what appealed to me about this approach was that it gave me a way to quantify my improvement. There is a business phrase 'if you can't measure it, you can't manage it.' In go, you can play hundreds of games but not seem to improve, which is frustrating. At least doing this I could be assured the effort I was putting in was definitely improving a part of my game.
Author:  Bill Spight [ Sat Sep 01, 2012 3:39 am ]
Post subject:  Re: Memorizing Tsumego via SRS
LocoRon wrote:
speedchase wrote:
The reason strong players solve tsumego quickly is because they can read fast, not because they can guess the move.


No.

At ~1 kyu, I [sadly] read about the same speed as I did when I was ~20 kyu. I can now solve tsumego I couldn't then, in seconds. The difference is experience, which has honed my intuition.


After the first environmental go (AKA coupon go) game between Jiange Jujo and Rui Naiwei, I had discovered a mistake by Jujo and was showing it to him at their house. Rui was going about the house, doing this and that. At one point she stopped by, looked at the go board for about two seconds, and rattled off an eight move combination that neither I nor Jujo had seen. Interestingly, it was an improvement on her own play that she had missed during the game. :)


Quote:
I recall a post somewhere on this board either about an experiment tracking strong players eyes (I think pros, but maybe high dan amateurs) while solving tsumego, or maybe it was just an anecdote, I don't remember it all. Anyway, basically, their eyes go straight to the vital point, and that's where they start reading.


What I remember is a study that showed that weaker solvers looked at where to play, stronger solvers looked at points where eyes would be. :)
Author:  Bill Spight [ Sat Sep 01, 2012 3:49 am ]
Post subject:  Re: Memorizing Tsumego via SRS
Boidhre wrote:
I started a thread on it a while back, the general consensus seemed to be that it was a bad idea: http://lifein19x19.com/forum/viewtopic.php?f=18&t=5713

*shrugs*


I see that on that thread I discussed making similar flashcards with important differences. See the second page. :)

BTW, making such flashcards may be the most valuable part of the exercise. :) See http://calnewport.com/blog/2012/08/10/y ... -learning/
Author:  LocoRon [ Sat Sep 01, 2012 5:54 am ]
Post subject:  Re: Memorizing Tsumego via SRS
Bill Spight wrote:
Quote:
I recall a post somewhere on this board either about an experiment tracking strong players eyes (I think pros, but maybe high dan amateurs) while solving tsumego, or maybe it was just an anecdote, I don't remember it all. Anyway, basically, their eyes go straight to the vital point, and that's where they start reading.


What I remember is a study that showed that weaker solvers looked at where to play, stronger solvers looked at points where eyes would be. :)


Hmm... Maybe that's what I was thinking of as well.
Author:  hyperpape [ Sat Sep 01, 2012 7:27 am ]
Post subject:  Re: Memorizing Tsumego via SRS
Bill Spight wrote:
BTW, making such flashcards may be the most valuable part of the exercise. :) See http://calnewport.com/blog/2012/08/10/y ... -learning/
I believe it: http://www.cscs.umich.edu/~crshalizi/notabene/.

In my small way, I have been trying to imitate that set of notebooks.
Author:  jts [ Sat Sep 01, 2012 8:13 am ]
Post subject:  Re: Memorizing Tsumego via SRS
LocoRon wrote:
Bill Spight wrote:
Quote:
I recall a post somewhere on this board either about an experiment tracking strong players eyes (I think pros, but maybe high dan amateurs) while solving tsumego, or maybe it was just an anecdote, I don't remember it all. Anyway, basically, their eyes go straight to the vital point, and that's where they start reading.


What I remember is a study that showed that weaker solvers looked at where to play, stronger solvers looked at points where eyes would be. :)


Hmm... Maybe that's what I was thinking of as well.

See, what I remember is strong players look at empty intersections, weak players look at stones that have already been played.
Author:  OtakuViking [ Sat Sep 01, 2012 8:31 am ]
Post subject:  Re: Memorizing Tsumego via SRS
Memorizing tsumego defeats the purpose of the tsumego. It is meant to be a workout for your mind to look ahead and find the correct move. But I have thought about the uses of SRS in regards to go before. I think you'd be better off trying with basic joseki and common shapes and opening patterns. Leave tsumego as the exercise in reading that it is meant to be.
Author:  Bartleby [ Sat Sep 01, 2012 9:25 am ]
Post subject:  Re: Memorizing Tsumego via SRS
I think sometimes we have a tendency to overthink these things, and to try to use technology when it is simply not necessary.

Buy a tsumego book with problems at a level that is appropriate for you. Go through it, noting down all the problems you miss. Wait a few days, then go through the problems that you missed, this time making a special effort. Put the book aside for a month, then go through the book again.

After a couple of repetitions you will know how to solve the problems in your book. There is no reason to use SRS for this, even entering the problems into ANKI or whatever you are using is probably an inefficient use of your time.

Problem books actually work very well for the purpose they are meant for, which is to improve your reading while exposing you to important shapes. The only issues are finding a book that is appropriate for your level and putting in the effort to go through it.

Personally, I am not a big fan of SRS even for learning languages. A much better way to learn vocabulary, IMO, is to read a lot and look up words you don't know. In that way you learn how the words are used in actual language, not just their dictionary definition. Also, at least for me, words and phrases tend to stick better when they are placed in a context.
Author:  speedchase [ Sat Sep 01, 2012 12:34 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
LocoRon wrote:

No.

At ~1 kyu, I [sadly] read about the same speed as I did when I was ~20 kyu. I can now solve tsumego I couldn't then, in seconds. The difference is experience, which has honed my intuition.

(Edit: Maybe even stronger players do read faster; barring concrete evidence, I'll just assume they have superior intuition. Not taking into account individual abilities of course; my reading admittedly sucks.)


maybe you don't read faster, but I definitely read faster than I did as 5kyu. How do you think stronger players could read 15 move sequences in two seconds, if they can't read faster than 20kyu'?
Author:  yoyoma [ Sat Sep 01, 2012 12:50 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
I think it's a good idea. No one ever said memorizing rabbity six is dead was bad. This is just extending that concept further. Is it bad that I memorized L group is dead, L+1 needs a move, and L+2 is alive? Or the status of some of the standard side shapes (Door is dead)?
Author:  singular [ Sat Sep 01, 2012 2:05 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
speedchase wrote:
maybe you don't read faster, but I definitely read faster than I did as 5kyu. How do you think stronger players could read 15 move sequences in two seconds, if they can't read faster than 20kyu'?


15 moves in two seconds? I'd better start drinking my vitamin water . . . . I think I can manage four moves per two seconds going at a good clip.
Author:  topazg [ Sat Sep 01, 2012 3:26 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
My experience with professionals is also that they don't read faster, just better.

I'm lucky if I'm reading more than 2 or 3 moves a second, and that's on simple sequences (although my subconscious may be crunching faster in the background).

I remember a fun anecdotal experience when I was about 5k. I had been working on a go problem that was about 1d level with another 5k and a 2k. We spent 20 minutes getting to what looked like the right answer, having broken a couple of wrong answers along the way. We showed it to a Korean 3p, who picked the right move within a second - a second and a half at most. His English was poor, but it was very clear that from a quick analysis of the shape it was the only _intersection_ that could work. No reading involved at all, apart from subsequent confirmation, but he didn't even seem look at any other candidates other than to assess whether they occupied as critical a point in the shape.
Author:  speedchase [ Sat Sep 01, 2012 3:34 pm ]
Post subject:  Re: Memorizing Tsumego via SRS
I'm not saying that professionals don't read more efficiently that mere mortals, I just think that they also have a huge speed advantage.

Ps. take a look at the first footnote on this page: http://senseis.xmp.net/?ReadingDepth
PPs. singular, I don't really understand your point, can you please elaborate?
Page 1 of 4 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/