Hacker News new | past | comments | ask | show | jobs | submit login
To remember everything you learn, surrender to this algorithm (2008) (wired.com)
215 points by sanmak 36 days ago | hide | past | favorite | 110 comments

You know. I've never been interested in these e-ink notebooks where you can write on a digital screen that acts like paper. I've always felt that actual pen and paper were superior.

However, if you add some good spaced repetition software to one of these e-notebooks then I think you will have finally surpassed pen and paper for note taking.

I'm thinking of something like, you take a note on a page and then save it. A week later the page of notes you wrote is presented to you again and you can draw boxes around parts of the page and blur / hide bits of text, etc; these then become your spaced repetition items. I've heard of studies showing that hand written notes are superior for memory, this is the best of both worlds, hand written and then automatic spaced repetition.

And how much easier is it to surrender your life to a spaced repetition algorithm when it is it's own dedicated device? The idea of being able to memorize math, and other visual information, in addition to the usual written content (which is, again, better remembered if hand written) is so promising I'd happily pay hundreds of dollars for such a device.

I know there are some fairly hackable e-ink tables out there, has anyone done this?

AnkiDroid will support in-app drawing shortly[0]. There's already an e-ink mode which removes most animations, and supports image occlusions via third-party apps which use the AnkiDroid API [1].

Happy to discuss improvements for your workflow if it's an Android-based e-ink tablet via Discord[2]/GitHub issues.

[0] https://github.com/ankidroid/Anki-Android/issues/9363

[1] https://github.com/ankidroid/Anki-Android/wiki/Third-Party-A...

[2] https://discord.gg/qjzcRTx

It's a good idea. I think the problem with being presented the whole page is that we fool ourselves into thinking we know something. I.e recognition feels like remembering.

So it might be better to draw the boxes up front, but I like the concept.

https://www.remnote.io/ tries to do something like this but with textual notes.

I just want taking notes and then chopping those notes into spaced repetition items to be separate.

I didn't mean present the entire page, although a link back to the entire page for context would be nice.

Also, being able to remember things by their spacial layout on the page is an advantage, not a flaw. It allows you to employ your spacial and visual memory to help in memorizing concepts and facts.

A key for when tablets will replace paper is when refresh rates get so fast drawing mimics ink on paper. Until the lag is eliminated it won't be replaced.

We're already there.

The reMarkable (by reports) and Onyx BOOX (by direct experience) both satisfy this requirement.

I picked up the BOOX as an e-book reader, not a note-taking device. It turns out that the note-taking functionality is really, really good.

I still have the objection that notes are stuck in an application-specific interface. Though on balance that's not all bad.

There's also the inherent privacy risk of having an electronic rather than physical record.

I haven't used them myself, but I've heard that the Remarkable tablets have pretty good lag.

I'm using a remarkable 2 and the lag when writing is almost non-existent. The thing traditional paper has over the e-ink tablet is the ease of shuffling through notes. Sure, I can tag, and organize, and search through notebooks on my tablet, but shuffling through a stack of paper is much faster, without the friction of using a (sometimes clunky) e-ink ui.

Well said and I agree. Everything most e-ink tablets do you can already do with paper and pencil, more or less. You can search your notes on the tablet, you can search your notes in a paper notebook, and with the paper notebook you can remember things like "it's towards the back, after X". You can manually tag your e-notes, reality automatically tags your physical notes with things like page wrinkles and spilled coffee stains. So, slightly different but ultimately quite similar for most people.

Again, I think spaced repetition is a killer feature that could finally distinguish an e-ink tablet from paper. I think this idea has more potential for improving learning than almost all the multi-million dollar learning websites out there. It can systematically take you to savant levels of recollection in any topic. And right now this tool is locked behind some clunky text focused UI. (Yes, latex and pictures exist, but they suck when compared to drawing on a page.)

It's 21 milliseconds one reviewer said. And remember, it's a refresh happening under the stylus and under your fingers, so it's not comparable to regular screen refresh times.

The consensus within the Anki community seems to be that yes, the SuperMemo scheduling algorithm (SM-17) is a bit better than Anki's (SM-2 variant), but not by enough to matter, so you shouldn't worry about it.

But when you look into tests people have done to see how much better it is exactly, it's around 30% more efficient.

I still use Anki and am grateful for its existence, but this is a huge difference in efficiency. I wish there was a better open source scheduling algorithm available.

How come Anki cant update it itself? Is SM-17 just a different set of exponential-decay functions from SM-2? Is that even copyrightable?

Also, if Anki is open-source, I'm surprised that there is no unofficial port that uses SM-17. But maybe I'm just ignorant of how much work that would involve.

Woz goes into detail about the SM-17 algorithm here: https://supermemo.guru/wiki/Algorithm_SM-17

... he's built these through decades of thinking of and tinkering and working on this singular problem. SM-2 is from 1987 for SM for DOS 1.0, and is fairly simplistic comparatively.

SM-17 is much more complicated and also closed source. You would have to spend a while reverse-engineering it.

I'm curious about those tests, do you have a link? I'm also curious if there's been any attempts at open-source versions of more recent supermemo algorithms.

Unfortunately there's no single good source for this; 30% is just an approximate aggregate figure I've come up with after seeing various people's posts on different sites.

Some things I could find just now:

Anki vs. SM-8: https://anki.tenderapp.com/discussions/effective-learning/12...

SM-2 vs. SM-17: https://supermemo.guru/wiki/Universal_metric_for_cross-compa...

Anki vs. custom machine learning scheduler: https://old.reddit.com/r/Anki/comments/d0i8uy/improved_algor...

Anki vs SM-17: https://unrelatedwaffle.medium.com/battle-of-the-spaced-repe...

SuperMemo seems to fall behind Anki in the short term (<6 months), but eventually makes up for it. Some of the links reflect that.

As for open-source implementations, there this: https://github.com/slaypni/SM-15

Is there a version of Anki that uses SuperMemo or is there another software that implements it?

Anki is great... however:

... having used it for the last 4 years to try to memorize vocabulary in certain languages (Spanish & French, recently started Chinese), I think it's a base layer... (much like how Bitcoiners like to talk about BTC's Blockchain). Reasons:

  1. I've tired of it. I need a layer that makes me excited to use it.

  2. I've found it far-far less useful for learning other things than rote vocabulary memorization. It needs a layer that makes it easy to store and "pull up" everything I need to know.
That said, I don't want to sound like I'm looking a gift-horse in the mouth. Anki truly is a great piece of software, and my sincere thanks to the devs.

In my opinion people often make the mistake of using Anki to memorize one-to-one translations of foreign words. It is more effective to use it to learn the mapping of whole sentences.

For every word you want to learn, add a few idiomatic and representative sentences in Anki. You can do the same for every new grammatical concept you want to learn.

And don't forget to use Anki's text-to-speech tools to practice listening as well as speaking.

Does it replace other forms of learning entirely? No, but it can be a great component of learning a new language.

That's exactly what I do. From videos I use losslesscut [1] to isolate the sentences I want to practice on and then turn those into anki flashcards with some custom tools I wrote.

[1] https://github.com/mifi/lossless-cut

What else do your tools/scripts do? Does losslesscut work with streaming video or only downloaded files?

Speech to text (via google cloud speech) and dictionary lookups (via a custom gui). Only downloaded files, luckily the sources I use make their content available online for free (and I wrote some python to automatically download new episodes).

If you want to download from YT i recommend NewPipe, better than every dodgy browser extension (i shall install another program on my machine? GTFO) i tried. You can just get the audio too. Android only though .

Any chance you've studied either Spanish or Russian and have decks to share? I've never fully gotten started with Anki, but would love to get back into it to use it to expand my vocab with full sentences.

I'll also be checking online now that you've spurred my curiosity, just curious if you have ones that you like too.

Keep in mind, the full value from Anki really comes from creating your own decks rather than using preconfigured ones.

This was a huge eye opening moment for me. I thought it was about using a deck, and I always found it less than fantastic. Now that I am making my own deck, it is far more useful.

There is this guy who turned the Tatoeba database into Anki decks: https://digitalwords.net/anki/tatoeba-audio/index.eng.html

Totally agree with this.

When learning Japanese grinding vocab was extremely boring but did a fantastic job of bridging the gap between textbook literature and actually being able to read simpler (middle school level) native literature / comics / news stories.

And I didn't even need that many words to bridge that gap. Maybe like 1000 or so over what I'd already picked up from textbooks.

Now? There's no way in heck I'm going back to grinding vocab and instead slowly pick up new words by reading tons and tons of books.

I still need to rote memorize a bunch of kanji since picking them up naturally is pretty brutal, but like with vocabulary I'll only do it to the point where I can stumble through YA-level books without looking something up every other sentence.

It's much more fun to watch native video / read native materials and put in whole sentences that are just on the tip of your foreign language understanding (i + 1 theory from stephen krashen).

Instead of just 1 to 1 vocab mapping, look for chinese tv shows that you enjoy and start to slowly build up vocabulary by mining sentences.

Then when you do reviews, not only will you remember the vocabulary, you will also remember when it occurred in the story and have a cool memory of some of your favorite tv shows / books / etc.

I highly recommend this. A swedish girl told me that the reason for them being good at enhlish is they don't get localized movies. Helps picking up colloquial uses of language too, realized that watching japanese movies.

I've used Anki every day for almost a year now for Spanish vocabulary. My deck has over 4000 cards. Most are one-to-one translations of nouns and verbs, but some are phrases and Spanish idioms.

To your first point, it has basically become a habit for me. The key is to really commit to it for the first several weeks and establish the habit, then you won't feel tired of it as reviewing your cards daily becomes something you "just do".

Definitely agree with point 2. The Anki practice really shines when I combine it with my conversational practice. Memorizing the words alone doesn't make me a better speaker, but it does mean there's a lot fewer "what's the word?" moments in the middle of a conversation.

I believe Anki memorization + regular conversational practice is second only to full immersion as the best learning strategy.

> I believe Anki memorization + regular conversational practice is second only to full immersion as the best learning strategy.

I'd extend this:

  Memory + Synthesis = Useful learning strategy
Blindly learning facts with no intent to synthesize them into something else will help you win at trivia contests. Learning to integrate those facts into something else (like conversation, with languages) is how you effectively learn. Both aspects are needed in order to become effective at whatever subject you're studying.

That said, with Anki or any other tool these rules are incredibly handy:


When I first heard the term "revising your notes" I assumed it meant the kind of synthesis you speak of, or at least finding errors in lecture material. No, apparently it just means looking at them again, at least as a minimum.

I believe "revision" is used in British English to mean reviewing or studying.

Nice man. Yes, why not use both immersion and spaced repetition?

Immersion is great if you can do it. When I traveled up Argentina to visit my wife’s family I wrote down a lot of words and phrases that were new to me for cards.

I'm building a prototype that does that. It's not done yet, but you can check out last month's demo of the "pull up". https://www.conceptionary.app/

(Please don't squish my website HN)

Give Lingvist a shot (commercial, but there is a free trial).

There’s also Mnemosyne which is fully FOSS unlike Anki.



I'm pretty sure Anki is FOSS, though perhaps you just don't like the AGPL?

I think they meant the other parts of the Anki ecosystem that are not FOSS, AnkiWeb and the iOS app.

Anki is just a slice of Piotr Wozniak's ideas, and the algorithm it uses isn't as good as that implemented in his own software, SuperMemo. And besides: it doesn't do incremental reading, which is lights-out amazing: https://super-memo.com/supermemo18.html

I'm interested in incremental-reading, but the video he has really doesn't sell it for me [1]

It seems like you skim through an article, extract a few facts, and add them to your memorization schedule. Would it really help you summarize the article, or help you synthesize the arguments and follow the logic leading to a conclusion?

1. https://www.youtube.com/watch?v=DoQoeK53bP8

Hello! I made a few videos that explain how incremental Reading works:

1 Topics: https://youtu.be/1ZNsn8IL-TM 2 Incremental Reading core analogy: https://youtu.be/Wme2RLm1jWY 3 How Incremental Reading makes micro learning easy: https://youtu.be/piqq1kwYL5s

Spoiler alert: basically incremental reading is like using save states when playing a game with an emulator. As one might save their game state before and after a challenging section in a game, when you find a difficult to understand/explain idea, incremental reading makes it easy to pause and synthesize that idea before moving on. Yes at the end also it’s good to summarize the ideas to yourself and simplify that explanation, and then use SuperMemo to memorize that explanation. Incremental reading is hard to explain why it is so useful, I’ve been using SuperMemo every day for 15 years and I still feel like I am learning new things

Since the early 90s I've used a piano-practice scheme that provides spaced repetition. It greatly reduces the time needed to either (1) learn or (2) memorize pieces. I always feel compelled to cite this poem:

You can get a great deal from rehearsal

If it just has the proper dispersal

You would just be an ass

To do it en masse

Your remembering would turn out much worsal.

I play piano and it takes me forever to memorize pieces. Could you describe the details of your scheme?

Well, the first principle is not about spaced repetition but is more about active learning (wrong term?) vs. passive learning. In other words, don't memorize by playing it a bunch and hoping repetition will beget recall. Instead, play from memory as far as you can (in sections as described below), filling in and approximating whatever you can't remember exactly, but trying to recall parts in that section, even if you can't piece them together in sequence.

The second principle is the spaced repetition. That works for learning the muscle memory as well as the memorization, but let's just address memorization.

First I divide a piece into arbitrary sections, often pages but in information-dense music you've got to go a couple lines at a time. I do not usually divide at a sensible musical stopping point, because we remember the starts and ends of what we do much better than the middles. Choosing weird places to start and end helps get some of those middles into our heads; plus it's really easy to mentally track what you've done if it's at a page or line boundary.

Then, with sections A through, say, D, I'll go in roughly this order, with "A" meaning I play all notes in section A in the proper sequence, even if I have to wait a few seconds sometimes to recall what happens next.

A A A; B B B A A; C C C B B A; D D D C C B A; D D C B; D C; D

...adjusting sometimes to accommodate the different information densities of the different sections. After this you'll need some mental downtime because it sucks a lot out of you to recall so much. Later you can do sections E through H, etc.

By the time you've done that, you're halfway to your goal. You can then start increasing the tempo toward performance quality by responsible metronome practice. I like to do

(1) the above routine out of tempo, waiting as long as necessary before notes

(2) the above routine at any tempo with metronome, doesn't matter how ridiculously slow

(3) increasing the tempo gradually from there, not deviating from the "note-perfect" requirement

really interesting stuff. I'm working on a tool that helps you play by ear by making you play back increasingly difficult melodies. Adding spaced repetition would probably make it a lot more efficient!

I had no idea Ogden Nash had written a poem about piano practice.

Ulrich Neisser apparently. He's also quoted by this guy who apparently had the same idea as me: https://www.modacity.co/blog/using-repetition-achieve-effect...

I believe that specific form is called a limerick.

Perhaps because of the AABBA rhyme scheme and the shorter third and fourth line. But in a limerick the first line traditionally introduces a person and a place, which doesn’t happen here.

There once was a pianist from Nantucket …

The SuperMemo scheduling algorithm is a lot easier for me to deal with than Anki: if I miss a couple days, it doesn't overwhelm me with a punishing review schedule. With that, I really wish Woz would open source the SM18 algorithm, dealing with the SM UI is really painful. I don't love the Anki UI either, but it's streets ahead of SM in that dimension.

Also, tangentially related: I really love Woz's wiki, which is filled with interesting insights on learning: https://supermemo.guru

Do you have some evidence that this particular algorithm is so effective?

I have some doubts that there is a 'perfect' forgetting curve that maps over all items equally that this would matter to such a degree.

I.e. if it comes back after 3 days then 9 in Anki, or 2 days and 10 in SM18.

What matters is showing up and spending the time with your SR tool, as well as adding high quality cards.

Just put in whatever time you have. Spaced repetition algorithms are robust against abuse. I have some spaced repetition items I've neglected for years, when I go back, the algorithms will do the appropriate thing and my time will still be used efficiently.

Years ago, I used Anki to memorize 3000 difficult words and thus get a perfect verbal score on the GRE. Great stuff!

I used Mnemosyne to memorize baby Rudin, Munkres’ Topology and large swaths of Dummit and Foote. Perfect scores on analysis and topology qualifying exams… good enough on algebra haha.

I had always been reluctant to use memorization but I found it helped immensely. Especially as a returning graduate student.

I've often heard it's not very effective for topics like math. I would love to read a write-up on how you did this - perhaps your technique is better than most people's attempts?

Also, do you continue to use it for math-like subjects?

I went a little crazy. I literally typed the whole book into cards, including ones like "prove a closed, bounded set of real numbers is compact." Then I would get a stack of scrap paper and spend maybe an hour a day writing down the answers.

Perhaps this was a reaction to my previous approach to learning, which completely shunned memorization.

Ok, so the mnemosyne doesn't contain the answers. But just a prompt/question. And then you write the solution/answers on paper every time mnemosyne shows a particular card.

From the way you described it, you change all statements/theorems/etc. in the book into questions? To put inside you mnemosyne deck. So your deck in the end has a lot of prove this statement/theorems/etc. and the problems that come with the book.

I was expecting some magical shortcut. But it's the usual tip of doing shitton of problems. Your way is especially comprehensive of course since you basically memorize/know/practice how to prove/solve everything in a book.

Impressive. This takes a lot of time and discipline.

Someone else already posted this, not sure if you will pinged from them.

Please read this article about learning math with spaced repetition. http://augmentingcognition.com/ltm.html

Could you give us an example of the type of cards you used? I'm always interested in how to use anki for math and physics. The usual advice to do well in those courses is to do a lot of problems.

Yeah, well Dummit and Foote is huge. It would take forever to get everything down.

Great job on your exams though!

The exams were more than a few years ago now. Thanks!

Show us the Anki deck! (Please?)

This was about 10 years ago, so I no longer have it.

Looking through the shared decks this one looks promising: https://ankiweb.net/shared/info/1377287053

Spaced repetition actually took off in recent years. But there's another component of SuperMemo that didn't, incremental reading: https://supermemo.guru/wiki/Incremental_reading

It seems a lot more effective than spaced repetition, but also a much higher level of commitment and a higher learning curve. It's sort of like using a power tool. I haven't tried it, but this is a good reminder that I should.


This is a tool meant to help with incremental reading, with support for generating Anki flashcards.

The problem with this tool is you can't review your extracts at the same time you are reviewing your flashcards.

For example, if you have a list of 50 PDFs and you want to incrementally read them, Polarized doesn't have a way of scheduling when you're going to read all 50 PDFs. Supermemo, you don't have worry about that. IMO, that's the key feature of Incremental Reading.

I can chuck dozens of articles at a time into Supermemo, and do not have to worry about forgetting to read them. They'll show up in my reviews over time.

I'm wary that it might not be better than the tool developed by the guy who invented incremental reading.

I'd love to have something for it on mobile, because that's where I spend all my reading time, and that was one of the main advantages of Anki.

Hello! I made a few videos that explain how incremental Reading works: 1 Topics: https://youtu.be/1ZNsn8IL-TM 2 Incremental Reading core analogy: https://youtu.be/Wme2RLm1jWY 3 How Incremental Reading makes micro learning easy: https://youtu.be/piqq1kwYL5s Spoiler alert: basically incremental reading is like using save states when playing a game with an emulator. As one might save their game state before and after a challenging section in a game, when you find a difficult to understand/explain idea, incremental reading makes it easy to pause and synthesize that idea before moving on. At the end also it’s good to summarize the ideas to yourself and simplify that explanation, and then use SuperMemo to memorize that explanation. Incremental reading is hard to explain why it is so useful, I’ve been using SuperMemo every day for 15 years and I still feel like I am learning new things

Dropping related Michael Nielsen’s excellent piece about space repitition here:

Augmenting Long Term Memory http://augmentingcognition.com/ltm.html

One past thread:

To Remember Everything You Learn, Surrender to This Algorithm (2008) - https://news.ycombinator.com/item?id=17706776 - Aug 2018 (208 comments)

Just recent enough for me to remember the shape of the article and the gist of the research, but too long ago for me to remember most of the details. :)

The thread should be posted using the SuperMemo algorithm.

It is a great algorithm. It is, however, a boring method and really takes a lot of discipline to make it a habit. For me, the best way to learn anything is have it tied to a fun challenge or a personal project. I've never been successful in the type of just-in-case learning.

The most effective trick for me is simply knowing why I need to remember something.

If I don't know its utility value, I won't even be able to recall it an hour later. On the other hand, if I can see the utility, I might be able to remember it decades later after a single exposure.

If I'm using a map with GPS on my phone to find my way around a city, I won't be able to recall any of the buildings or landmarks later and I won't be able to find my way around even after many journeys. On the other hand, if I use street signs and ask around for directions, I will remember everything. In the former case, my brain knows that it can rely on the map and GPS so it doesn't waste resources trying to remember streets, buildings and landmarks.

My wife has tried to explain to me the difference between hand-towels, wash-clothes, dish-towels, face-cloths, and so on, and so on, and... It's not that I don't want to understand; I'd like to help her fold them, put them out, etc. It's just that I'm constitutionally unable to retain that information. I think I can't, at a profound level, generate any interest in learning this, no matter how firmly I have consciously decided I want to.

I can kind of tell the difference between 'towels' and 'rags'.

I have similar 'issues'. My wife tends to fuss over tiny details that seem completely irrelevant to me. It's not that I don't want to make an effort to comply with her more stringent expectations, it's that I don't even notice those details to begin with. It's distracting to think about them. My mind is always trying to minimize its mental load by focusing only on things that it deems necessary. I guess it's a more 'entrepreneurial' mindset?

Most mathematicians seem to be able to remember anything regardless of utility value. I guess that's why I chose a more engineering/business career.

Dropping related Nicky Case's’s excellent piece about space repetition here: How to Remember Anything Forever-ish https://ncase.me/remember/

Can spaced repetition be used effectively for things beyond languages?

This would have been a godsend had I known about it during college.

Absolutely, it's a general concept.

To give some examples, it's used heavily in US med schools, and I achieved great academic results using it to study CS in university.

I've used it for some math concepts. The memorized concepts give you something to mentally play with, and give anchor points on which you can build more knowledge.

Sort of? It depends on the "thing", but spaced repetition is pretty standard for physical activities like martial arts. In essence, you're attending a class that focuses on drilling a particular technique on a regular interval. There isn't much research from sports pedagogy, though I think its more because its just a "known" requirement that doesn't require much new research. However, I would also say that spaced repetition goes beyond simply reviewing Anki flash cards and that there is an element of kinesthetic learning that is necessary to strengthen the learning gains.

I use it for encyclopedia entries and general glossaries, which I gather from PDFs or just online. There is IMO no better way to learn a topic than to memorize an encyclopedia about it.

As I understand, it's used in elementary school math teaching

It's very useful for studying for the ham radio tests.

I’ve used it effectively for history classes

I have noticed another learning dynamic and am curious to read other thoughts and experiences.

Often, I will read about something. Say designing a PCB, or to play an instrument. I will then watch a video, read, and or attempt the task just to get the lay of the land.

At that moment, if I really power through and reach a goal, the learning is useful. I can retain and recall very reasonably.

However, say I attempt and do not reach a goal. Maybe just exploring, or for lack of time. If I remain interested, I will continue to read and or listen to audio, or video or talk to people about it.

On a future attempt not made too far in the future from last visit to the subject matter, many more things seem familiar! My ability to navigate new skills and or recall rules, details is better than I would expect. Sometimes I feel as if it is actually good to do that, let just a little time pass and then I will feel like I want to do it, and maybe that want is my mind having assembled new info seeks expression of it, a use.

Usually, I am overall more effective than I was on the initial attempt despite not being hands on.

I’ve been using SuperMemo for 15 years, in fact I just finished up my flashcards for today :) It definitely is useful but it isn’t the grand answer to all of life’s problems. It makes it very easy to remember stuff you want to remember but takes some training when it comes to how to structure desirable knowledge as a simple flashcard. Over the years I’ve rubber handed in how much I use SuperMemo, but the happy medium for me seems to be 1. I usually finish my flashcards first thing in the morning, 2. Get in the habit of “thinking with flashcards” by summarizing ideas into single sentences and 3. I keep an ongoing text document in my Notes app on my phone of information I find that is worth remembering. If SuperMemo had a robust mobile app that synced with the Windows program, it would be ideal. Using my Notes app seems to be the easiest alternative that I’ve found.

Would Anki be an option for you? Been using it for some time and both Desktop and Mobile Apps are great and the anki notes are synced.

I am inclined to give his software a go. Very interesting ideas that resonate with my own learning experiences.

This sales model is also interesting and notable:[1]

Older versions of SuperMemo are successively released as freeware.

By purchasing any of these freeware items you get registered in this store and become eligible to instantly upgrade to SuperMemo 17 (see Freeware Upgrade). In the future, you will also be able to upgrade to any SuperMemo at a big discount.

Your cheapest pathway to new SuperMemos is: order SuperMemo 15 Freeware below (or any other collector item) if you like SuperMemo 15, use Freeware Upgrade to get SuperMemo 17 in the future, keep upgrading to new versions of your choice (once registered in this store, you are perpetually eligible)

Well done!

[1] https://super-memo.com/supermemo18.html

Is this the same technique used in Quantum Country?

Yes. And here is one of the author's discussion of how he uses Anki (at least at one point in time): http://augmentingcognition.com/ltm.html

Does there exist an app/website/program that implements this technique for learning languages, combined with actual lesson plans? I understand you can use the SuperMemo software or something like Anki to learn languages, but you need to decide what you want to learn and put it into the program. What I'd like is something like Duolingo with an existing set of lessons and a progression, but that makes use of this reminder timing. (Duolingo does do reminders to practice old material, but it's clearly not advanced. It regularly has me practicing stuff that is far below my level and is effectively known forever.)

The closest thing you're probably going to find for languages is the Pimsleur method, which is also based on a specific formula of spaced repetition. I don't know how effective it would be for trying to achieve true conversational fluency, but I've found it very effective for quickly learning a basic set of words and phrases for a trip to another country.


I'm making an app like this for Japanese, however it doesn't have preset lessons yet. (I'm planning to add more structured lessons in the future.) For now what I have are a bunch of vocabulary lists for various shows/novels/etc. which you can pick and learn the vocabulary from, and then you can use the app to find the next show/novel/etc. where you'll have to learn the least amount of new vocabulary to be able to consume it.

Yeah, I agree completely.

Pre-made Anki decks vary in quality. Some are great. But none that I've found have the "themes" that personally keep me motivated to language learn long term.

Anki allows you to download preset content. SuperMemo appears to have paid content for languages, but I haven't seen how effective it is.

Haven't finished the article, but the human-interest part so far:

Man refines memory techniques; dissapears beneath event-horizon of information density.

Does someone have a link to some example code that implements this idea? I am working on my own memory-assistance software, I would like to use the algorithm.


Not exactly example code, but the algorithm is well-described near the end. It should be sufficiently detailed to implement that particular variation.


Is one of the SuperMemo algorithms, particularly SM-2, described in pseudocode form. Again, should be enough to implement it yourself.

Other algorithms can be seen as variations on these themes. Like with Anki's implementation (and I believe also SuperMemo's) there is jitter added to the schedule so that too many cards don't show up together over and over. This is to avoid the issue of remembering something because of what it's with and not on its own. With language learning, say you enter a dozen cards on colors and review them all in one day, strictly speaking they could all end up recurring at the same time in the future. By adding jitter they get spread out so you can avoid accidental "topic" days and end up with a proper mix of cards for study.

I put together a page that lets you try out different spaced repetition algorithms (with javascript code included), and you can try out your own algorithms as well: https://www.ussherpress.com/freshcards/srs/

There's a link on the page to some info on how to use it as well, since it may not be obvious!

I've used this npm package to implement spaced repetition review of bookmarked words in a dictionary app. It's nicely put together and very readable.


roam/sr has code at https://github.com/aidam38/roamsr

This is a plugin for Roam Research that allows you to do spaced repetition inline with your Roam graph. It keeps its state alongside your notes, and if you want to change your notes while doing a review session that's seamless.

Just look at an image search for "spaced repetition" and you can create it from the graphs you see.

Essentially you're creating a set of reminders/events with linear or exponential spacing between. A common pattern appears to be 1, 3, and 6 days after first learning, which is also 1, 2, and 3 days apart from each other.

This should probably be marked 2008.

Added. Thanks!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact