Rooboy Wrote:Yeah, SRS definitely needs a sexier image (or a simple sexy way to be explained).

Just returned from a nice week in Vegas. Before i went i used Anki to learn Blackjack strategy for the rules of the casinos i planned to play in. Basically if you play perfect strategy for the rules of the table the house advantage is less than 1%. When you couple that with free drink supply (and we're talking regular $15 cocktails) then it's a definite win in my opinion.

I used to own a book called, "comp city" that taught how to take advantage of games that have less that a 1% house advantage to get complimentary rooms, meals, shows, airfare, etc. I lived in Las Vegas for eight years and got comped hotel Suites, gourmet restaurants, show tickets, limo rides to sporting events with box seats, etc.. Pretty much anything I asked for, they gave me. You have to be sly though and trick them into thinking you are losing money. Two or three trips without losing and they will cut you off.
Be careful with tips also. They can add up to more than what you are getting for free.

Grinkers Wrote:In what way is SuperMemo better? I've only used Anki, because it was free and had linux support. It seems most people here are using Anki, but I've never really seen comparisons between different SRS software.

Anki is great and works very well, but it is based on SM-2 algorithm which is 23 years old!

Since 2002, SuperMemo uses the latest algorithm SM-11 which eliminates problems with repetition delay and repetition advancement evident in previous algorithms.

Simple comparison:

http://nihongoperapera.com/spacing-algor...views.html

http://nihongoperapera.com/mnemosyne-anki-review.html


Scientific comparison:

http://www.supermemo.com/english/algsm11.htm

http://supermemo.com/english/ol/sm5.htm

http://supermemo.com/english/ol/sm2.htm


Personally, I use all of them for each one has its unique features.

Newer does not necessarily mean better:

http://ichi2.net/anki/wiki/FrequentlyAsk...Ankiuse.3F

Anki's algorithm is sufficient for most people, but in the long term, the newer algorithm (SM-11) is better, especially if you are consistent and study every day, for many reasons:

http://www.supermemo.com/english/algsm11.htm

I'm pretty happy with anki (and being able to use Linux and review on my iphone, not sure if SuperMemo can do this, but I'm already setup now!).

SRS interest me a lot, so I'll be reading those links when I have more time, thanks!

Are people able to import/export their existing cards to/from Anki/SuperMemo? I'm always open to trying new things, but losing over 6000 cards would be too much at this point.

Edit: It looks like there's no Linux support, and having to use Wine/emulation... bleh!

ahibba Wrote:Anki's algorithm is sufficient for most people, but in the long term, the newer algorithm (SM-11) is better, especially if you are consistent and study every day, for many reasons:

http://www.supermemo.com/english/algsm11.htm

I know next to nothing about math, but I think that's a lot of nonsense that doesn't have any basis in reality except insofar as it relates to selling software and perhaps some of their own non-experiential theorizing about the math of SRSing (when I say non-experiential I mean it lacks empirical validity, whatever they might say about using gathered user info). In my opinion, the only important thing is making sure you've got the 'schedule it sooner/later' automation in a rough algorithmic way, dependent on one's self-grading.

Reviewing resolve's page, if I knew anything about math/programming, that's the sort of thing I would be doing, shaping the SM2 thingy around that perspective. I think the way rigorous mathematical abstraction will work in the future, w/ regards to neuroscience and the greater resolution of imaging technologies and suchlike, I think it will evolve so that it's further reliant on and entwined with the human perspective in this way, actually, but that's a rant for another time...

Don't worry too much about the algorithm being used by whichever SRS you use. No matter which one you're using, it's going to be way more efficient than trying to plan out your studying and reviewing on your own.

It seems that I'm the only one here who knows what SuperMemo really is and who is Piotr Wozniak.

Wozniak does not care if his software sells or not (note: Supermemo 98 is freeware). He is a researcher specializing in the theory of spaced reptition for more than two decades. He wrote the first computer software employing spaced reptition in 1987.

SuperMemo is not just a commercial software like other SRS in the market. 20 years ago, under the supervision of Prof. Kierzkowski, Wozniak wrote his Master's Thesis: Optimization of learning. A new approach and computer application. This 200-pages dissertation was entirely devoted to the theoretical aspects of SuperMemo.

You can take a look at his Ph.D. dissertation which is also about SuperMemo and the theory of spaced repitition:

Economics of learning: New aspects in designing modern computer aided self-instruction systems

ahibba Wrote:It seems that I'm the only one here who knows what SuperMemo really is and who is Piotr Wozniak.

Wozniak does not care if his software sells or not (note: Supermemo 98 is freeware). He is a researcher specializing in the theory of spaced reptition for more than two decades. He wrote the first computer software employing spaced reptition in 1987.

SuperMemo is not just a commercial software like other SRS in the market. 20 years ago, under the supervision of Prof. Kierzkowski, Wozniak wrote his Master's Thesis: Optimization of learning. A new approach and computer application. This 200-pages dissertation was entirely devoted to the theoretical aspects of SuperMemo.

You can take a look at his Ph.D. dissertation which is also about SuperMemo and the theory of spaced repitition:

Economics of learning: New aspects in designing modern computer aided self-instruction systems

Pfft, you actually seem rather late to the party, re: Wozniak/Supermemo, and you also seem to be a bit blinded by the Science™ presented at the site. There's a lot of good ideas and hard information, theoretical as it is, but it's not set in stone, and the stuff about the algorithms is indeed a mix of self-indulgent wankery and software marketing, that's very clear if you get past idolizing or whatever. Trying to precisely match algorithms to subjective learning experiences is rather silly. The emphasis should logically fall on the self-grading, and making tweaks to a basic algorithm is good enough from there. Most of the latter tweaks, I'd say, are best centered around workflow/interface design and suchlike. Anyway.

nest0r Wrote:the stuff about the algorithms is indeed a mix of self-indulgent wankery and software marketing

I totally disagree with you.

If you are comfortable with Anki, stick with it. But don't underestimate SuperMemo.

Remember: SuperMemo is not suitable for all people, especially those who know nothing about math.

Dr. Wozniak has compiled some statistics on Supermemo users, and apparently Supermemo requires a certain kind of personality. Women, artists, and humanities majors tend to hate it; men, scientists and engineers tend to love it. The idea behind Supermemo, just like Ben Graham's "Value Investing'', apparently either hooks you in five minutes--or never hooks you at all.

Stick-to-it-iveness seems to count for something too. About 60% of people who try Supermemo give up within the first two months.

ahibba Wrote:

nest0r Wrote:the stuff about the algorithms is indeed a mix of self-indulgent wankery and software marketing

I totally disagree with you.

If you are comfortable with Anki, stick with it. But don't underestimate SuperMemo.

Remember: SuperMemo is not suitable for all people, especially those who know nothing about math.

Dr. Wozniak has compiled some statistics on Supermemo users, and apparently Supermemo requires a certain kind of personality. Women, artists, and humanities majors tend to hate it; men, scientists and engineers tend to love it. The idea behind Supermemo, just like Ben Graham's "Value Investing'', apparently either hooks you in five minutes--or never hooks you at all.

Stick-to-it-iveness seems to count for something too. About 60% of people who try Supermemo give up within the first two months.

Looks like while I was cutting you some slack, you snatched the rope and hanged yourself again. Oh well. ;p

I retract what I wrote earlier... SuperMemo now sounds really barfy. :lol:

Dr. Wozniak deserves a lot of credit for all the work he's put in over the years - maybe Anki would not even be here if it were not for SuperMemo. However, as much respect as I have for him, I think the later SM algorithms do not adapt so well to breaks or material of varying difficulties. And if you put constraints on the optimal factors matrix to ensure a "hard" grade doesn't ever yield a longer interval than a "good" grade, then you end up with static increments anyway that are no different to SM2.

For a user who never misses a day of study and studies material that varies little in difficulty, then it's quite possible SM can squeeze out a little more efficiency. But I'm willing to sacrifice a small amount of efficiency for an open source program that keeps my data free, runs on multiple platforms, and isn't terribly awkward to use. If Anki didn't exist, I'd be using another open source SM2 implementation like Mnemosyne over SuperMemo.

resolve Wrote:Dr. Wozniak deserves a lot of credit for all the work he's put in over the years - maybe Anki would not even be here if it were not for SuperMemo. However, as much respect as I have for him, I think the later SM algorithms do not adapt so well to breaks or material of varying difficulties. And if you put constraints on the optimal factors matrix to ensure a "hard" grade doesn't ever yield a longer interval than a "good" grade, then you end up with static increments anyway that are no different to SM2.

For a user who never misses a day of study and studies material that varies little in difficulty, then it's quite possible SM can squeeze out a little more efficiency. But I'm willing to sacrifice a small amount of efficiency for an open source program that keeps my data free, runs on multiple platforms, and isn't terribly awkward to use. If Anki didn't exist, I'd be using another open source SM2 implementation like Mnemosyne over SuperMemo.

After reading about SuperMemo, these are my thoughts exactly!

Being a Linux user, Anki (and the free data backups) is just amazing. Once I'm out of my financial pickle, I can't wait to give a donation. It has completely changed the memorization part of my education!

I'm using SM98 for general learning of Japanese (Kanji included) and so far, so good. I only glanced over Anki and got a good impression of it but was already, at that point, set-up with SM, so no change for me.

What I like about SM is:
- voluminous amount of supporting material
- statistically verified efficiency of the method
- rich user options (graphs, searches etc)

On a downside, you can definitely put a shocking UI, to which, nevertheless, you can get accustomed.

Yay - finally another SM vs Anki threat. My thoughts:
1.) Trying to understand the math behind the newer SM algorithm seems like a waste of time.
I trust resolves's judgement on this. (A potentially slightly better Algorithm in SM11 than in SM2/Anki)
2.) Once you use an SRS for a considerable amount of time, you grow to like it. Or maybe it is just your psychology telling you: You did not waste so much time on an less than ideal product, so your product has to be the best.
3.) Most users only used one SRS for a considerable amount of time. So judgment on any other seems misplaced.