We often hear about how important it is to understand, to not just memorize information blindly. But there are situations when long term memorization is an essential part of the study process, for example if you’re trying to learn a new language. A common tool to help in this is flashcards, ie. cards each with a precise question on one side, its answer on the other.
When digging deeper on how to maximize the efficiency of flashcard review, the “when to review such or such card?” question naturally comes to mind. It turns out some people have been researching that very question and have come up with interesting tools and theories on how to best use your time when reviewing.
SuperMemo and spaced repetition
One such person is Piotr Wozniak, a rather eccentric Polish profressor who, since the early 1980s, has been studying and perfecting what he calls “spaced repetition“. His software, SuperMemo, implements this technique.
Essentially, once a flashcard is made and reviewed for the first time, it is then scheduled to be reviewed some time in the future, say in 2 days. Then, depending on how well you remember it in 2 days, it is rescheduled for a next review, but this time further in the future, say in a week. The process is then repeated, spacing repetitions further and further until a point where you won’t forget about it anymore.
The core idea behind spaced repetition in SuperMemo is to present the flashcard just when you’re about to forget about it. Dr Wozniak has developped and refined models of how memorization and forgetting happens in time. The end goal, of course, is to repeat a minimum number of times a given flashcard, so you can repeat more flashcards total.
The SuperMemo website details his findings and tips concerning memorization, and can make for hours of reading. In particular, there are techniques to make better flashcards (how to formulate the question, etc.) and to use the SuperMemo software, of course. There is also an affiliated site where you can find ready-made flashcard sets available for purchase.
EDIT: On the website, you can notably find more detailed information on the mathematical details of the algorithm used by the program and its evolution. As mentioned in a comment below by FlashcardDB creator, the Leitner system is another spaced repetition system which doesn’t rely on computers (it was created in the ’70s), but on moving real cards in real decks, as explained on the FlashcardDB site itself.
Alternative software
Altough it seems Piotr Wozniak has been a pioneer in creating spaced repetition software, other programs have recently appeared that follows a similar model.
If you want something cross-platform (I run Linux so that’s my case), you can go for Mnemosyne. It’s much simpler than SuperMemo (much less options, formatting in flashcards, etc.), but it’s open source and free.
There are also Web versions. SuperMemo itself has an online version. Very recently, websites SpicyElephant, Mind Picnic and Flashcard DB (mentioned above) have appeared that follow the spaced repetition model. Being online communities, they allow you to share flashcards and reuse those made by others.
References
EDIT: Side note: my own experience or lack thereof
Unfortunately, I do not have a lot of personal experience with this. When I was in high school I had programmed something similar for myself (my formula was real simple and very inflexible). That’s why I found it amusing to discover spaced repetitions a few months ago.
But these days I don’t deal with much information that would apply well to spaced repetition (I study software engineering). Encoding the Q/As is too much effort for the gain. Learning through application works better right now.
I’m trying to come back to it with other learning trails, though, as the principle is just so cool 
October 7th, 2008 at 3:49 pm
Hi, this is Jeff from FlashcardDB. Thanks for the mention François! Looking forward to passing through Montréal this long weekend.
With the all the users on FlashcardDB studying with both the SM-2 algorithm and the Leitner System I hope to eventually have a good set of data to statistically compare the effectiveness of both systems. How much does the extra effort of grading your responses gain you?
The Mnemosyne guys are doing something similar with their optional anonymous data collection that tracks how effective changes are to their version of the SM-2 algorithm. See: http://sourceforge.net/mailarchive/forum.php?forum_name=mnemosyne-proj-devel&max_rows=25&style=nested&viewmonth=200805. See also the “Memory Research” section here: http://www.mnemosyne-proj.org/principles.php.
Hopefully we will eventually see more publicly accessible sets of data as the popularity of spaced repetition increases.
October 7th, 2008 at 10:25 pm
Thanks Jeff for your comment! Sorry I went over online versions so fast in my post
I tried to keep it short and stick to the overall principle.
It’s great to see you’re trying to improve the core algorithms yourself. Have you noticed some differences yet? (edit: well, just saw your SM-2 version has only been up since July, but say we set aside statistical relevance? Erm
)
The SuperMemo algorithm seems to have been changed so many times (SM-3, 4… 11!). But then as they mention on the Mnemosyne site: “we are a bit skeptical that the huge complexity of the newer SM algorithms provides for a statistically relevant benefit”.
(edit: moved a part of my comment to the post)October 13th, 2008 at 5:51 am
Thanks for the write up François, it was well timed for the release of the new version of http://spicyelephant.com
Craig.
October 13th, 2008 at 9:05 am
Thanks Craig for passing by.
By the way, I first heard of SpicyElephant through this blog post (was posted on YC News I think):
http://www.neverreadpassively......artup.html
October 15th, 2008 at 10:52 am
Hey François. (I sent you an e-mail with some further rambling thoughts). You are right it will take time to gather enough data to be useful. I do intend to be as open as possible to allow the statistically-gifted out there to do their thing.
November 3rd, 2008 at 1:17 pm
I agree that Mnemosyne is better than Supermemo in terms of usability. I’ve just found a new SRS - BetterMemo (http://www.bettermemo.com) and so far it seems even better than Mnemosyne. Unfortunately, it is paid, so I am torn between the two right now.
As for the web sites for flashcards sharing and learning - I find the clunky at the time, but I hope they will improve in the near future. Desktop software (not only spaced repetition) is far better than the web product, but some day I want to be able to access all my data, no matter where I am.
November 6th, 2008 at 2:32 pm
Thanks John for your comment. BetterMemo seems interesting (compared to Mnemosyne: I esp. like the future reviews lists/graphs). The thing is, though, I’m a features-oriented guy (vs usability-oriented), so I’m thinking: if it’s Windows-only and 30$, my natural inclination would be to go for Supermemo (~40$ IIRC).
As for Web versions, I’m pretty sure they’ll get better with time. Simplicity is usually an imperative when developing Web apps (to send a clear message to users who check your site out for about 5 secs before moving on), so I guess they’re trying to strike a balance with feature-richness.
Concerning access from everywhere, I still need to check out PDA/phone versions of spaced repetition software; it seems like the perfect thing to do when you’re waiting in line for 5 minutes.
December 7th, 2008 at 10:07 am
Hi,
I’ve not used SuperMemo because it is a Windoze (!) only software and I dont have Win. There is another great software for Linux people: FlashQard
and here is the website:
flashqard.sourceforge.net
And of course it is free (as in freedom and beer)
December 8th, 2008 at 9:54 am
Hi Shahab. FlashQard looks real nice, I’m suprised I haven’t heard of it before when I was searching for Linux spaced repetition software (well it’s pretty recent but nonetheless).
I like the way the next reviews are presented, with the number of good answers: it makes the review process more engaging.