Spaced Repetition Software for Tsumego

[Bill Spight]

Do not learn what you don't understand. Doesn't that go against what the received wisdom for beginners is? i.e. don't worry about theory, just play games.

I never received that wisdom. I think that the point about playing games is that there are things that come up again and again that you can just pick up. When I was learning there was no go material that I knew about at that level. I suspect that more is available now, I don't know.

This site by Dr. Wozniak, a spaced repetition expert, gives "20 rules of formulating knowledge in learning." http://www.supermemo.com/articles/20rules.htm

Rule 1. "Do not learn if you do not understand."

Rule 2. "Learn before you memorize."

IMO, simply making flash cards (or the computerized equivalent) out of go problems is not such a good idea. It does not follow Wozniak's rules. I suspect that making flash cards based upon go problems is a good idea, and that the main value comes from making them. For instance,

$$ Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | . O X , . . .
$$ | O . O X X . .
$$ | X O O O X . .
$$ | . O . X . . .
$$ --------------

Click Here To Show Diagram Code

[go]$$ Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | . O X , . . .
$$ | O . O X X . .
$$ | X O O O X . .
$$ | . O . X . . .
$$ --------------[/go]

I think that this would make a good flash card based upon this problem, but devising the card would be an important part of the learning.

Edit: In fact, I would suggest making status flash cards from tsumego problems. I. e., zero move tsumego.

Hmm, dead I think since even if white wins the ko they only have one eye and white cannot do better than a ko here I think. I could be wrong though.

You do not mention it, but I suspect that you read this possibility:

$$W Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | 1 O X , . . .
$$ | O 2 O X X . .
$$ | X O O O X . .
$$ | . O . X . . .
$$ --------------

Click Here To Show Diagram Code

[go]$$W Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | 1 O X , . . .
$$ | O 2 O X X . .
$$ | X O O O X . .
$$ | . O . X . . .
$$ --------------[/go]

Your comment pertains to this possibility:

$$W Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | 2 O X , . . .
$$ | O . O X X . .
$$ | X O O O X . .
$$ | 1 O . X . . .
$$ --------------

Click Here To Show Diagram Code

[go]$$W Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | 2 O X , . . .
$$ | O . O X X . .
$$ | X O O O X . .
$$ | 1 O . X . . .
$$ --------------[/go]

You are right. White is dead.

Perhaps this might make a good flash card.

$$ Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | X O X , . . .
$$ | O . O X X . .
$$ | . O O O X . .
$$ | O O . X . . .
$$ --------------

Click Here To Show Diagram Code

[go]$$ Dead or alive?
$$ | . . . . . . .
$$ | . . . . . . .
$$ | X X X . X . .
$$ | X O X , . . .
$$ | O . O X X . .
$$ | . O O O X . .
$$ | O O . X . . .
$$ --------------[/go]

The 2-3 point is a false eye.

[Boidhre]

Hmm, you know Anki has a really useful function where you can upload a deck for free to a central server and then anyone can freely download it. If we had a community effort making images from tsumego we could make quite a few zero move "alive or dead" tsumego relatively quickly and have a free resource for beginners up there. Ditto a more complicated multi move remaining deck for more advanced players.

I'd be worried about making mistakes in my reading though if I was doing it! (i.e. I don't think it's right copying someone else's tsumego en masse for public consumption)

Oh and yeah, I read the self atari if white filled 1,4. I should have mentioned that.

[RBerenguel]

The problem is, it takes more time to do this in Anki than creating the problem in EasyGo or similar (I know because I created a deck for the proverbs in Opening Theory Made Easy... Not that useful)

[Boidhre]

The problem is, it takes more time to do this in Anki than creating the problem in EasyGo or similar (I know because I created a deck for the proverbs in Opening Theory Made Easy... Not that useful)

Sure but Anki is free and cross platform, OSX, Windows, Linux, BSD, Android, iOS, (paid for on mobile devices though), while EasyGo is iOS only.

[RBerenguel]

Yes, I know it's relatively free (you can install the HTML based Anki in jailbroken iOS devices and Android devices for free, but it's not as nice,) but in this kind of things you have to take into account the price in time. Generating the images of 1-move problems takes a lot of time... And is then not that much useful after a while. I'd rather have a cross-platform SGF problem viewer with SRS but...

[Boidhre]

Yes, I know it's relatively free (you can install the HTML based Anki in jailbroken iOS devices and Android devices for free, but it's not as nice,) but in this kind of things you have to take into account the price in time. Generating the images of 1-move problems takes a lot of time... And is then not that much useful after a while. I'd rather have a cross-platform SGF problem viewer with SRS but...

Yeah, that's the ideal.

As a side question: Given the comments above by people, why do you want SRS?

[RBerenguel]

I'd like SRS to be given more frequently problems I can't solve. But I want the full sequence: I may find the first move... This qualifies as a "2". I may read the perfect sequence, this is a "5". But if there are variations I missed? Of course images would also work, but the amount of work of generating the required images is "a lot". I've been "typesetting" a lot of problems, and even that takes a lot of work (around 15 minutes for every 10 problems)! Of course, I can live without SRS for problem practice, repeating solved problems is a way to get a thick "muscle memory".

[yoyoma]

I've fiddled with doing this too. My goal for using Anki/SRS system would be for *memorizing* certain things. Things that would be good to know instantly without reading. Mostly this would be the status of standard corner and side shapes, possibly including memorizing the vital point and/or mainline of the problem. Also certain standard tesujis, such as tombstone. To make good anki cards of this you would make several cards where the tombstone barely works and barely doesn't work. In this case it may be more like be able to read the mainline almost instantly instead of outright have it memorized.

Taking the tombstone example, for a long time I could solve tombstone problems in books with pretty much 100% accuracy. The problem though is that it took me quite a long time to read the sequence. But if you focused on practicing reading that exact sequence over and over, you would be able to read it faster.

Josekis could be included with targeted questions, such asking do you need a ladder for this variation, where the position shown is right at the critical moment that you have to pick a variation before the ladder actually happens. For this case I would be aiming for almost instant recognition that there is a ladder involved, and also knowing which direction it goes.

So for me this is about pattern recognition. This is different from reading practice. I think software could help for that as well, but it would be more like a method Bill described. Trying to feed you problems at a certain rate based on their difficulty to you. Don't bother showing simple problems you get 100%, and don't burden you too much with problems that are too hard. This is to exercise reading, a separate skill from pattern recognition.

[Mike Novack]

b is not, new problems are introduced at a fixed rate per day. At least with the implementations that I've seen anyway.

I am getting even more confused.

I can see a statement like "new problems are introduced as a fixed rate per day" with reference to some specific software (or site or whatever). I cannot even imagine somebody making a claim like that "in general". I cannot even imagine somebody making a claim that all go software even accesses time/date and so wouldn't know that a day had pased(not obvious that it would need to).

Nor did I realize that we were supposedly only talking about "find the first move" problems. In many cases that is a trivial part of the problem (in a problem, black to kill the first move might be obvious whenever there is a point that makes white unconditionally alive. When doing problems beyond the easy ones, I can often find the first move but still not solve the problem correctly.

[Boidhre]

b is not, new problems are introduced at a fixed rate per day. At least with the implementations that I've seen anyway.

I am getting even more confused.

I can see a statement like "new problems are introduced as a fixed rate per day" with reference to some specific software (or site or whatever). I cannot even imagine somebody making a claim like that "in general". I cannot even imagine somebody making a claim that all go software even accesses time/date and so wouldn't know that a day had pased(not obvious that it would need to).

Nor did I realize that we were supposedly only talking about "find the first move" problems. In many cases that is a trivial part of the problem (in a problem, black to kill the first move might be obvious whenever there is a point that makes white unconditionally alive. When doing problems beyond the easy ones, I can often find the first move but still not solve the problem correctly.

My apologies for being confusing and unclear. We're not only talking about first move problems, they're just the ones I'm tending to do at the moment, my question was to others who've more experience than I whether they found this approach useful with tsumego in general, the first move problem element was descriptive of my current practices not prescriptive of what I think people should be doing. For a multi-move problem I'd only count it as a success if every move and reply was correctly read, otherwise I'd mark it as wrong, since I do such problems to practice reading rather than to memorise at the moment (I've no idea whether people memorise multi-move tsumego in the sdk and dan ranks or just use them for reading practice).

[Mike Novack]

Can we return to the beginning?


"I think none exists specific to go?...... I'd very happily pay for such a feature."

I took that as a question about the existence of such software and that we were not restricting the answer to free as in free beer software.

I thought I was giving an answer, that I believe such software exists and that MFOG would be an example. But I never formally tested the behavior of the "solve go problems" part of this program, never kept formal statistics on the frequency of reappearance of problems. Just my "sense" that ones I usually got wrong kept reappearing and those I usually got right seemed to be replaced by new ones over time. Time in this case being number of problems tried, not real time.

I also noted that the free trial version does not allow access to anything beyond the "easy" set of problems (the paid for version also offers an "intermediate" and "hard" set). I suggested that you look at the free trial version and keep statistics on the behavior of the "easy" set of problems if you don't think it is serving problems up based upon your history of getting them right or wrong.

PLEASE -- this is not intended as a push of some specific product. When faced with the claim "none exists" the easiest, most natural response is to give one example that does. People should feel free to add additional examples.

BTW: In terms of exact problems this might be memorization but I think more generalized learning of patterns and possibilities is taking place.

[Boidhre]

Can we return to the beginning?


"I think none exists specific to go?...... I'd very happily pay for such a feature."

I took that as a question about the existence of such software and that we were not restricting the answer to free as in free beer software.

I thought I was giving an answer, that I believe such software exists and that MFOG would be an example. But I never formally tested the behavior of the "solve go problems" part of this program, never kept formal statistics on the frequency of reappearance of problems. Just my "sense" that ones I usually got wrong kept reappearing and those I usually got right seemed to be replaced by new ones over time. Time in this case being number of problems tried, not real time.

I also noted that the free trial version does not allow access to anything beyond the "easy" set of problems (the paid for version also offers an "intermediate" and "hard" set). I suggested that you look at the free trial version and keep statistics on the behavior of the "easy" set of problems if you don't think it is serving problems up based upon your history of getting them right or wrong.

PLEASE -- this is not intended as a push of some specific product. When faced with the claim "none exists" the easiest, most natural response is to give one example that does. People should feel free to add additional examples.

BTW: In terms of exact problems this might be memorization but I think more generalized learning of patterns and possibilities is taking place.

And I appreciated your input. The issue was I was not specific enough in my question in the initial post, I should have specified exactly what I meant by spaced repetition system and so on. I've been considering buying MFOG on and off for a while now as it happens. I'm also being swayed by Bill into thinking that SRS might not be the best method to bring to bear on tsumego.

With regard to your last statement: that is/was my hope with this approach.

[daniel_the_smith]

I'm late to the party, but my website does spaced repitition for joseki positions. I don't have time at the moment to expand that to problems, but it's on my list, which is far too long lately...

[Bill Spight]

I'm also being swayed by Bill into thinking that SRS might not be the best method to bring to bear on tsumego.

I am being swayed to think that SRS might be good for basic shape, tesuji, and status problems.

[Boidhre]

I'm also being swayed by Bill into thinking that SRS might not be the best method to bring to bear on tsumego.

I am being swayed to think that SRS might be good for basic shape, tesuji, and status problems.

Which are essentially what I solve at the moment! Since a one move kill is effectively asking: Show that you know that the status of this group is undetermined. I can see some value in memorising these kinds of shapes, especially corner and side shapes.