Supermemo method doesn't work for me

Johnny -- why don't you try SuperMemo? Mnemosyne and Anki may be simpler to use, but they both use a slightly modified version of the SM-2 algorithm (developed by Piotr Wozniak 22 years ago). See http://www.supermemo.com/english/algsm11.htm for an overview of the algorithm's development. The earliest SM algorithm that I used was SM-6.

Yes, SuperMemo's interface is annoying. You have to learn to work around its quirks. But I have to say, I used it for many years without major problems. And let's face it -- Photoshop is much more complicated. :)

Thanks for the replies, but as you guys can see, my main problem was that I was using the method the wrong way. The answers were not simple enough, so they were difficult to recall and it was hard to grade my guesses. Also, I didn't use it regularly.
Yes, I think Mnemosyne would work well anyway to tell the truth, but since I managed to install the latest Anki, and Anki looks better and has more features, I'll just give it a try.

Tom, I don't use Supermemo because I use Linux. In any case, I don't think there would be a substantial difference: the AJATT guy said any popular SRS is good, as long as you use it correctly (= you must use it regularly and the cards must be good).

There's only one thing that I'm wondering about now. Ok, I understand how the grading system works, but I wonder how I should grade the new cards after I've seen them.
As an esample, Good Night = Buenas Noches, let's suppose I put that on a card. How should I grade it the first time? I think I should take the time to learn it, that is, spend some time on it until I think I kind of "know it"... but how much, and how long will I remember it?
In Anki, the first time it might be like, SOON = about ten minutes or right away, HARD = about 10 hours, GOOD = about 4 days, EASY = about 8 days.
It's difficult to guess how well you'll remember a card, and I've noticed that I need to review a fact within a day (or even better, 12 hours) if I want to be sure I feel like I'm learning. Reviewing a certain fact is often an unconscious process while we are studying a subject (we suddenly think of a fact we just studied or read again about it), but when we use SRS, how do we simulate that?
It really seems to me that I need to choose HARD every time I am trying to learn a totally new card, so that I'll be sure I see it very soon. The second time I'll see it, I can start using the normal grading method (so if I remember it pretty well I can choose GOOD).

"srs" means spaced REPETITION software. it's not a learning/memorization software. this means that you create cards of information that you have learned before, that you already know. so if you don't know a card/information well, learn it first, then start repeating with supermemo/mnemosyne.

if you use it as a learning software the repetitions will be, as you have experienced yourself, very often. so don't forget to use it for what it was built for...

>>the AJATT guy said any popular SRS is good, as long as you use it correctly (= you must use it regularly and the cards must be good).<<

But since SRS is in no way a sufficient condition to mastery of a language… and that there are conditions of using it owing to time factors, there must be some who can’t use it at all. Or rarely.

IIRC, AJATT himself also claimed once that, if you have a better method, or if you have strong preference, whatever, sometimes you can safely get out of flashcards and learn as usual with other things.

I hope I’m still on the topic. The thing is, theoretically, you can learn something as if you were using flashcards, but you use none at all. Let’s say I have a small English grammar book that includes the most important rules, and it’s so small in size that I can carry them wherever I am. If my learning setting is optimal/optimized, I should be able to use the book exactly like a deck of flashcards, so that I can review very often.

Of course you can comment on the design of SRS from time to time, but ultimately it's just a helper. I guess most people would try to modify it to suit their needs.

Johnny--

Look at it this way: You have no idea how well you will remember a new item after the first review, and neither does SuperMemo (or Anki/Mnemosyne). There is no point in obsessing over the first interval. Don't worry about the fact that you cannot recall the item on the second review; your failure is important information for the software. Think of the first few repetitions as a "test run" that helps the software determine the difficulty of the item.

BTW, the GOAL of the optimization in SuperMemo is that you should FAIL to recall 5-20% of the items EVERY DAY (depending on the forgetting index). If you are forgetting less, then your intervals are too short and you are wasting time. If you expect to have 100% recall every day, you are misunderstanding the idea of spaced repetition.

<<If you expect to have 100% recall every day, you are misunderstanding the idea of spaced repetition. >>

Yeah, I remember reading that 100% retention is not something that should be considered when using software like Supermemo. However, the method of spaced repetition in itself would allow such a thing, and that's what I have in mind when I keep worrying that the gaps are too long. The reason why 100% retention is usually not considered is that 90% retention seems to take much less time, and you can use the time you save to learn other new stuff, so in the end you can actually learn "more" if you don't obsess over remembering everything, that is, 100%.

But that is not always the case. There are cases where you might actually want 100% retention. I think you need to change the settings to do that (in Supermemo, but Anki should be customizable in more or less the same way too I think). After all, to increase the probability you remember something, you just need to shorten the gaps, which might also be seen as an increase in the difficulty coefficient of each card.
For example, I expect to see a card with a Chinese character much more often than a card with a new English word. I'd say that remembering a new English word seems much easier than remembering what a Chinese character looks like, what it means, and being able to draw it.

If I have time, I'll take a closer look at the algorithm.

100% recall is not possible because the software has to take guesses. If there were no failed repetitions, it would not know which items are hard and which are easy (all items would be recalled equally well). It would have to assume that every item is as hard as the hardest possible item, which would be ridiculously unoptimal.

I worried about retention, too, when I was starting to learn with SuperMemo. The default forgetting index was 10% (BTW, this is not the same as retention!!!), but I changed it to 5% (the minimum allowed). Today, I think I was just wasting my time. Besides, repetitions are kind of boring when you get 95% of the answers right. :)

Dear Tom*

I am sorry that you have to put up with attacks like this. I will not post my name just in case it sparks off obscenities just when we thought they had stopped.

Being a native speaker of English I have never used Supermemo. However I have read what you have written. I believe that native speakers are constantly reminding themselves of the meaning of words through normal interaction. One of the reasons why I contribute to Antimoon is to keep the whole process of reading and writing going.

I have found that the online keyboard that you have produced is helpful in writing my partner's name, which is Hungarian. (Her husband was Hungarian although she is Polish.)